Fix GeneralVLA asset repository layout
Browse files- README.md +10 -134
- added_tokens.json +0 -6
- config.json +0 -171
- generation_config.json +0 -7
- merges.txt +0 -0
- preprocessor_config.json +0 -19
- pytorch_model-00001-of-00002.bin +0 -3
- pytorch_model-00002-of-00002.bin +0 -3
- pytorch_model.bin +0 -3
- pytorch_model.bin.index.json +0 -930
- special_tokens_map.json +0 -1
- tokenizer.json +0 -0
- tokenizer.model +0 -3
- tokenizer_config.json +0 -34
- vocab.json +0 -0
- zzzmmz/SegAgent-Model/.mdl +0 -0
- zzzmmz/SegAgent-Model/.msc +0 -0
- zzzmmz/SegAgent-Model/.mv +0 -1
- zzzmmz/SegAgent-Model/README.md +0 -50
- zzzmmz/SegAgent-Model/config.json +0 -49
- zzzmmz/SegAgent-Model/configuration.json +0 -1
- zzzmmz/SegAgent-Model/configuration_qwen.py +0 -65
- zzzmmz/SegAgent-Model/generation_config.json +0 -11
- zzzmmz/SegAgent-Model/model-00001-of-00004.safetensors +0 -3
- zzzmmz/SegAgent-Model/model-00002-of-00004.safetensors +0 -3
- zzzmmz/SegAgent-Model/model-00003-of-00004.safetensors +0 -3
- zzzmmz/SegAgent-Model/model-00004-of-00004.safetensors +0 -3
- zzzmmz/SegAgent-Model/model.safetensors.index.json +0 -860
- zzzmmz/SegAgent-Model/modeling_qwen.py +0 -1172
- zzzmmz/SegAgent-Model/qwen.tiktoken +0 -0
- zzzmmz/SegAgent-Model/qwen_generation_utils.py +0 -420
- zzzmmz/SegAgent-Model/special_tokens_map.json +0 -3
- zzzmmz/SegAgent-Model/tokenization_qwen.py +0 -598
- zzzmmz/SegAgent-Model/tokenizer_config.json +0 -14
- zzzmmz/SegAgent-Model/trainer_state.json +0 -0
- zzzmmz/SegAgent-Model/visual.py +0 -472
- zzzmmz/SegAgent-Model/zero_to_fp32.py +0 -587
README.md
CHANGED
|
@@ -1,145 +1,21 @@
|
|
| 1 |
-
---
|
| 2 |
-
tags:
|
| 3 |
-
- vision
|
| 4 |
-
widget:
|
| 5 |
-
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/cat-dog-music.png
|
| 6 |
-
candidate_labels: playing music, playing sports
|
| 7 |
-
example_title: Cat & Dog
|
| 8 |
-
---
|
| 9 |
|
| 10 |
-
# Model
|
| 11 |
|
| 12 |
-
|
| 13 |
|
| 14 |
-
|
| 15 |
|
| 16 |
-
|
| 17 |
|
| 18 |
-
##
|
| 19 |
|
| 20 |
-
|
| 21 |
|
| 22 |
-
|
| 23 |
|
| 24 |
-
|
| 25 |
|
| 26 |
-
|
| 27 |
|
|
|
|
| 28 |
|
| 29 |
-
### Documents
|
| 30 |
-
|
| 31 |
-
- [Blog Post](https://openai.com/blog/clip/)
|
| 32 |
-
- [CLIP Paper](https://arxiv.org/abs/2103.00020)
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
### Use with Transformers
|
| 36 |
-
|
| 37 |
-
```python
|
| 38 |
-
from PIL import Image
|
| 39 |
-
import requests
|
| 40 |
-
|
| 41 |
-
from transformers import CLIPProcessor, CLIPModel
|
| 42 |
-
|
| 43 |
-
model = CLIPModel.from_pretrained("openai/clip-vit-large-patch14")
|
| 44 |
-
processor = CLIPProcessor.from_pretrained("openai/clip-vit-large-patch14")
|
| 45 |
-
|
| 46 |
-
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
|
| 47 |
-
image = Image.open(requests.get(url, stream=True).raw)
|
| 48 |
-
|
| 49 |
-
inputs = processor(text=["a photo of a cat", "a photo of a dog"], images=image, return_tensors="pt", padding=True)
|
| 50 |
-
|
| 51 |
-
outputs = model(**inputs)
|
| 52 |
-
logits_per_image = outputs.logits_per_image # this is the image-text similarity score
|
| 53 |
-
probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities
|
| 54 |
-
```
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
## Model Use
|
| 58 |
-
|
| 59 |
-
### Intended Use
|
| 60 |
-
|
| 61 |
-
The model is intended as a research output for research communities. We hope that this model will enable researchers to better understand and explore zero-shot, arbitrary image classification. We also hope it can be used for interdisciplinary studies of the potential impact of such models - the CLIP paper includes a discussion of potential downstream impacts to provide an example for this sort of analysis.
|
| 62 |
-
|
| 63 |
-
#### Primary intended uses
|
| 64 |
-
|
| 65 |
-
The primary intended users of these models are AI researchers.
|
| 66 |
-
|
| 67 |
-
We primarily imagine the model will be used by researchers to better understand robustness, generalization, and other capabilities, biases, and constraints of computer vision models.
|
| 68 |
-
|
| 69 |
-
### Out-of-Scope Use Cases
|
| 70 |
-
|
| 71 |
-
**Any** deployed use case of the model - whether commercial or not - is currently out of scope. Non-deployed use cases such as image search in a constrained environment, are also not recommended unless there is thorough in-domain testing of the model with a specific, fixed class taxonomy. This is because our safety assessment demonstrated a high need for task specific testing especially given the variability of CLIP’s performance with different class taxonomies. This makes untested and unconstrained deployment of the model in any use case currently potentially harmful.
|
| 72 |
-
|
| 73 |
-
Certain use cases which would fall under the domain of surveillance and facial recognition are always out-of-scope regardless of performance of the model. This is because the use of artificial intelligence for tasks such as these can be premature currently given the lack of testing norms and checks to ensure its fair use.
|
| 74 |
-
|
| 75 |
-
Since the model has not been purposefully trained in or evaluated on any languages other than English, its use should be limited to English language use cases.
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
## Data
|
| 80 |
-
|
| 81 |
-
The model was trained on publicly available image-caption data. This was done through a combination of crawling a handful of websites and using commonly-used pre-existing image datasets such as [YFCC100M](http://projects.dfki.uni-kl.de/yfcc100m/). A large portion of the data comes from our crawling of the internet. This means that the data is more representative of people and societies most connected to the internet which tend to skew towards more developed nations, and younger, male users.
|
| 82 |
-
|
| 83 |
-
### Data Mission Statement
|
| 84 |
-
|
| 85 |
-
Our goal with building this dataset was to test out robustness and generalizability in computer vision tasks. As a result, the focus was on gathering large quantities of data from different publicly-available internet data sources. The data was gathered in a mostly non-interventionist manner. However, we only crawled websites that had policies against excessively violent and adult images and allowed us to filter out such content. We do not intend for this dataset to be used as the basis for any commercial or deployed model and will not be releasing the dataset.
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
## Performance and Limitations
|
| 90 |
-
|
| 91 |
-
### Performance
|
| 92 |
-
|
| 93 |
-
We have evaluated the performance of CLIP on a wide range of benchmarks across a variety of computer vision datasets such as OCR to texture recognition to fine-grained classification. The paper describes model performance on the following datasets:
|
| 94 |
-
|
| 95 |
-
- Food101
|
| 96 |
-
- CIFAR10
|
| 97 |
-
- CIFAR100
|
| 98 |
-
- Birdsnap
|
| 99 |
-
- SUN397
|
| 100 |
-
- Stanford Cars
|
| 101 |
-
- FGVC Aircraft
|
| 102 |
-
- VOC2007
|
| 103 |
-
- DTD
|
| 104 |
-
- Oxford-IIIT Pet dataset
|
| 105 |
-
- Caltech101
|
| 106 |
-
- Flowers102
|
| 107 |
-
- MNIST
|
| 108 |
-
- SVHN
|
| 109 |
-
- IIIT5K
|
| 110 |
-
- Hateful Memes
|
| 111 |
-
- SST-2
|
| 112 |
-
- UCF101
|
| 113 |
-
- Kinetics700
|
| 114 |
-
- Country211
|
| 115 |
-
- CLEVR Counting
|
| 116 |
-
- KITTI Distance
|
| 117 |
-
- STL-10
|
| 118 |
-
- RareAct
|
| 119 |
-
- Flickr30
|
| 120 |
-
- MSCOCO
|
| 121 |
-
- ImageNet
|
| 122 |
-
- ImageNet-A
|
| 123 |
-
- ImageNet-R
|
| 124 |
-
- ImageNet Sketch
|
| 125 |
-
- ObjectNet (ImageNet Overlap)
|
| 126 |
-
- Youtube-BB
|
| 127 |
-
- ImageNet-Vid
|
| 128 |
-
|
| 129 |
-
## Limitations
|
| 130 |
-
|
| 131 |
-
CLIP and our analysis of it have a number of limitations. CLIP currently struggles with respect to certain tasks such as fine grained classification and counting objects. CLIP also poses issues with regards to fairness and bias which we discuss in the paper and briefly in the next section. Additionally, our approach to testing CLIP also has an important limitation- in many cases we have used linear probes to evaluate the performance of CLIP and there is evidence suggesting that linear probes can underestimate model performance.
|
| 132 |
-
|
| 133 |
-
### Bias and Fairness
|
| 134 |
-
|
| 135 |
-
We find that the performance of CLIP - and the specific biases it exhibits - can depend significantly on class design and the choices one makes for categories to include and exclude. We tested the risk of certain kinds of denigration with CLIP by classifying images of people from [Fairface](https://arxiv.org/abs/1908.04913) into crime-related and non-human animal categories. We found significant disparities with respect to race and gender. Additionally, we found that these disparities could shift based on how the classes were constructed. (Details captured in the Broader Impacts Section in the paper).
|
| 136 |
-
|
| 137 |
-
We also tested the performance of CLIP on gender, race and age classification using the Fairface dataset (We default to using race categories as they are constructed in the Fairface dataset.) in order to assess quality of performance across different demographics. We found accuracy >96% across all races for gender classification with ‘Middle Eastern’ having the highest accuracy (98.4%) and ‘White’ having the lowest (96.5%). Additionally, CLIP averaged ~93% for racial classification and ~63% for age classification. Our use of evaluations to test for gender, race and age classification as well as denigration harms is simply to evaluate performance of the model across people and surface potential risks and not to demonstrate an endorsement/enthusiasm for such tasks.
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
## Feedback
|
| 142 |
-
|
| 143 |
-
### Where to send questions or comments about the model
|
| 144 |
-
|
| 145 |
-
Please use [this Google Form](https://forms.gle/Uv7afRH5dvY34ZEs9)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
+
# GeneralVLA Model Assets
|
| 3 |
|
| 4 |
+
|
| 5 |
|
| 6 |
+
This repository stores pretrained model assets and checkpoints for GeneralVLA.
|
| 7 |
|
| 8 |
+
|
| 9 |
|
| 10 |
+
## Layout
|
| 11 |
|
| 12 |
+
- `LISA-7B-v1-explanatory/`
|
| 13 |
|
| 14 |
+
- `clip-vit-large-patch14/`
|
| 15 |
|
| 16 |
+
- `segagent/zzzmmz/SegAgent-Model/`
|
| 17 |
|
| 18 |
+
- `sam_vit_h_4b8939.pth`
|
| 19 |
|
| 20 |
+
- `checkpoints/v1/checkpoint-rs.tar`
|
| 21 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
added_tokens.json
DELETED
|
@@ -1,6 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"<im_end>": 32002,
|
| 3 |
-
"<im_patch>": 32000,
|
| 4 |
-
"<im_start>": 32001,
|
| 5 |
-
"[SEG]": 32003
|
| 6 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
config.json
DELETED
|
@@ -1,171 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "clip-vit-large-patch14/",
|
| 3 |
-
"architectures": [
|
| 4 |
-
"CLIPModel"
|
| 5 |
-
],
|
| 6 |
-
"initializer_factor": 1.0,
|
| 7 |
-
"logit_scale_init_value": 2.6592,
|
| 8 |
-
"model_type": "clip",
|
| 9 |
-
"projection_dim": 768,
|
| 10 |
-
"text_config": {
|
| 11 |
-
"_name_or_path": "",
|
| 12 |
-
"add_cross_attention": false,
|
| 13 |
-
"architectures": null,
|
| 14 |
-
"attention_dropout": 0.0,
|
| 15 |
-
"bad_words_ids": null,
|
| 16 |
-
"bos_token_id": 0,
|
| 17 |
-
"chunk_size_feed_forward": 0,
|
| 18 |
-
"cross_attention_hidden_size": null,
|
| 19 |
-
"decoder_start_token_id": null,
|
| 20 |
-
"diversity_penalty": 0.0,
|
| 21 |
-
"do_sample": false,
|
| 22 |
-
"dropout": 0.0,
|
| 23 |
-
"early_stopping": false,
|
| 24 |
-
"encoder_no_repeat_ngram_size": 0,
|
| 25 |
-
"eos_token_id": 2,
|
| 26 |
-
"finetuning_task": null,
|
| 27 |
-
"forced_bos_token_id": null,
|
| 28 |
-
"forced_eos_token_id": null,
|
| 29 |
-
"hidden_act": "quick_gelu",
|
| 30 |
-
"hidden_size": 768,
|
| 31 |
-
"id2label": {
|
| 32 |
-
"0": "LABEL_0",
|
| 33 |
-
"1": "LABEL_1"
|
| 34 |
-
},
|
| 35 |
-
"initializer_factor": 1.0,
|
| 36 |
-
"initializer_range": 0.02,
|
| 37 |
-
"intermediate_size": 3072,
|
| 38 |
-
"is_decoder": false,
|
| 39 |
-
"is_encoder_decoder": false,
|
| 40 |
-
"label2id": {
|
| 41 |
-
"LABEL_0": 0,
|
| 42 |
-
"LABEL_1": 1
|
| 43 |
-
},
|
| 44 |
-
"layer_norm_eps": 1e-05,
|
| 45 |
-
"length_penalty": 1.0,
|
| 46 |
-
"max_length": 20,
|
| 47 |
-
"max_position_embeddings": 77,
|
| 48 |
-
"min_length": 0,
|
| 49 |
-
"model_type": "clip_text_model",
|
| 50 |
-
"no_repeat_ngram_size": 0,
|
| 51 |
-
"num_attention_heads": 12,
|
| 52 |
-
"num_beam_groups": 1,
|
| 53 |
-
"num_beams": 1,
|
| 54 |
-
"num_hidden_layers": 12,
|
| 55 |
-
"num_return_sequences": 1,
|
| 56 |
-
"output_attentions": false,
|
| 57 |
-
"output_hidden_states": false,
|
| 58 |
-
"output_scores": false,
|
| 59 |
-
"pad_token_id": 1,
|
| 60 |
-
"prefix": null,
|
| 61 |
-
"problem_type": null,
|
| 62 |
-
"projection_dim" : 768,
|
| 63 |
-
"pruned_heads": {},
|
| 64 |
-
"remove_invalid_values": false,
|
| 65 |
-
"repetition_penalty": 1.0,
|
| 66 |
-
"return_dict": true,
|
| 67 |
-
"return_dict_in_generate": false,
|
| 68 |
-
"sep_token_id": null,
|
| 69 |
-
"task_specific_params": null,
|
| 70 |
-
"temperature": 1.0,
|
| 71 |
-
"tie_encoder_decoder": false,
|
| 72 |
-
"tie_word_embeddings": true,
|
| 73 |
-
"tokenizer_class": null,
|
| 74 |
-
"top_k": 50,
|
| 75 |
-
"top_p": 1.0,
|
| 76 |
-
"torch_dtype": null,
|
| 77 |
-
"torchscript": false,
|
| 78 |
-
"transformers_version": "4.16.0.dev0",
|
| 79 |
-
"use_bfloat16": false,
|
| 80 |
-
"vocab_size": 49408
|
| 81 |
-
},
|
| 82 |
-
"text_config_dict": {
|
| 83 |
-
"hidden_size": 768,
|
| 84 |
-
"intermediate_size": 3072,
|
| 85 |
-
"num_attention_heads": 12,
|
| 86 |
-
"num_hidden_layers": 12,
|
| 87 |
-
"projection_dim": 768
|
| 88 |
-
},
|
| 89 |
-
"torch_dtype": "float32",
|
| 90 |
-
"transformers_version": null,
|
| 91 |
-
"vision_config": {
|
| 92 |
-
"_name_or_path": "",
|
| 93 |
-
"add_cross_attention": false,
|
| 94 |
-
"architectures": null,
|
| 95 |
-
"attention_dropout": 0.0,
|
| 96 |
-
"bad_words_ids": null,
|
| 97 |
-
"bos_token_id": null,
|
| 98 |
-
"chunk_size_feed_forward": 0,
|
| 99 |
-
"cross_attention_hidden_size": null,
|
| 100 |
-
"decoder_start_token_id": null,
|
| 101 |
-
"diversity_penalty": 0.0,
|
| 102 |
-
"do_sample": false,
|
| 103 |
-
"dropout": 0.0,
|
| 104 |
-
"early_stopping": false,
|
| 105 |
-
"encoder_no_repeat_ngram_size": 0,
|
| 106 |
-
"eos_token_id": null,
|
| 107 |
-
"finetuning_task": null,
|
| 108 |
-
"forced_bos_token_id": null,
|
| 109 |
-
"forced_eos_token_id": null,
|
| 110 |
-
"hidden_act": "quick_gelu",
|
| 111 |
-
"hidden_size": 1024,
|
| 112 |
-
"id2label": {
|
| 113 |
-
"0": "LABEL_0",
|
| 114 |
-
"1": "LABEL_1"
|
| 115 |
-
},
|
| 116 |
-
"image_size": 224,
|
| 117 |
-
"initializer_factor": 1.0,
|
| 118 |
-
"initializer_range": 0.02,
|
| 119 |
-
"intermediate_size": 4096,
|
| 120 |
-
"is_decoder": false,
|
| 121 |
-
"is_encoder_decoder": false,
|
| 122 |
-
"label2id": {
|
| 123 |
-
"LABEL_0": 0,
|
| 124 |
-
"LABEL_1": 1
|
| 125 |
-
},
|
| 126 |
-
"layer_norm_eps": 1e-05,
|
| 127 |
-
"length_penalty": 1.0,
|
| 128 |
-
"max_length": 20,
|
| 129 |
-
"min_length": 0,
|
| 130 |
-
"model_type": "clip_vision_model",
|
| 131 |
-
"no_repeat_ngram_size": 0,
|
| 132 |
-
"num_attention_heads": 16,
|
| 133 |
-
"num_beam_groups": 1,
|
| 134 |
-
"num_beams": 1,
|
| 135 |
-
"num_hidden_layers": 24,
|
| 136 |
-
"num_return_sequences": 1,
|
| 137 |
-
"output_attentions": false,
|
| 138 |
-
"output_hidden_states": false,
|
| 139 |
-
"output_scores": false,
|
| 140 |
-
"pad_token_id": null,
|
| 141 |
-
"patch_size": 14,
|
| 142 |
-
"prefix": null,
|
| 143 |
-
"problem_type": null,
|
| 144 |
-
"projection_dim" : 768,
|
| 145 |
-
"pruned_heads": {},
|
| 146 |
-
"remove_invalid_values": false,
|
| 147 |
-
"repetition_penalty": 1.0,
|
| 148 |
-
"return_dict": true,
|
| 149 |
-
"return_dict_in_generate": false,
|
| 150 |
-
"sep_token_id": null,
|
| 151 |
-
"task_specific_params": null,
|
| 152 |
-
"temperature": 1.0,
|
| 153 |
-
"tie_encoder_decoder": false,
|
| 154 |
-
"tie_word_embeddings": true,
|
| 155 |
-
"tokenizer_class": null,
|
| 156 |
-
"top_k": 50,
|
| 157 |
-
"top_p": 1.0,
|
| 158 |
-
"torch_dtype": null,
|
| 159 |
-
"torchscript": false,
|
| 160 |
-
"transformers_version": "4.16.0.dev0",
|
| 161 |
-
"use_bfloat16": false
|
| 162 |
-
},
|
| 163 |
-
"vision_config_dict": {
|
| 164 |
-
"hidden_size": 1024,
|
| 165 |
-
"intermediate_size": 4096,
|
| 166 |
-
"num_attention_heads": 16,
|
| 167 |
-
"num_hidden_layers": 24,
|
| 168 |
-
"patch_size": 14,
|
| 169 |
-
"projection_dim": 768
|
| 170 |
-
}
|
| 171 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
generation_config.json
DELETED
|
@@ -1,7 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_from_model_config": true,
|
| 3 |
-
"bos_token_id": 0,
|
| 4 |
-
"eos_token_id": 1,
|
| 5 |
-
"pad_token_id": 0,
|
| 6 |
-
"transformers_version": "4.31.0"
|
| 7 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
merges.txt
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
preprocessor_config.json
DELETED
|
@@ -1,19 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"crop_size": 224,
|
| 3 |
-
"do_center_crop": true,
|
| 4 |
-
"do_normalize": true,
|
| 5 |
-
"do_resize": true,
|
| 6 |
-
"feature_extractor_type": "CLIPFeatureExtractor",
|
| 7 |
-
"image_mean": [
|
| 8 |
-
0.48145466,
|
| 9 |
-
0.4578275,
|
| 10 |
-
0.40821073
|
| 11 |
-
],
|
| 12 |
-
"image_std": [
|
| 13 |
-
0.26862954,
|
| 14 |
-
0.26130258,
|
| 15 |
-
0.27577711
|
| 16 |
-
],
|
| 17 |
-
"resample": 3,
|
| 18 |
-
"size": 224
|
| 19 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00001-of-00002.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:38ac5624a8f7b982deb37ca66c6713047b08c2c7cf828241a7e2571d7271144a
|
| 3 |
-
size 9976667326
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00002-of-00002.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:8db6d703169c626e2f83b133ac5788adfb35a05024d200e1cb687d7780c1a75d
|
| 3 |
-
size 6144646041
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:f1a17cdbe0f36fec524f5cafb1c261ea3bbbc13e346e0f74fc9eb0460dedd0d3
|
| 3 |
-
size 1710671599
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model.bin.index.json
DELETED
|
@@ -1,930 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"metadata": {
|
| 3 |
-
"total_size": 16120985792
|
| 4 |
-
},
|
| 5 |
-
"weight_map": {
|
| 6 |
-
"lm_head.weight": "pytorch_model-00002-of-00002.bin",
|
| 7 |
-
"model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
|
| 8 |
-
"model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 9 |
-
"model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 10 |
-
"model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 11 |
-
"model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 12 |
-
"model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 13 |
-
"model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 14 |
-
"model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 15 |
-
"model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 16 |
-
"model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 17 |
-
"model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 18 |
-
"model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 19 |
-
"model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 20 |
-
"model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 21 |
-
"model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 22 |
-
"model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 23 |
-
"model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 24 |
-
"model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 25 |
-
"model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 26 |
-
"model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 27 |
-
"model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 28 |
-
"model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 29 |
-
"model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 30 |
-
"model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 31 |
-
"model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 32 |
-
"model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 33 |
-
"model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 34 |
-
"model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 35 |
-
"model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 36 |
-
"model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 37 |
-
"model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 38 |
-
"model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 39 |
-
"model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 40 |
-
"model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 41 |
-
"model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 42 |
-
"model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 43 |
-
"model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 44 |
-
"model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 45 |
-
"model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 46 |
-
"model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 47 |
-
"model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 48 |
-
"model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 49 |
-
"model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 50 |
-
"model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 51 |
-
"model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 52 |
-
"model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 53 |
-
"model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 54 |
-
"model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 55 |
-
"model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 56 |
-
"model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 57 |
-
"model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 58 |
-
"model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 59 |
-
"model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 60 |
-
"model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 61 |
-
"model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 62 |
-
"model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 63 |
-
"model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 64 |
-
"model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 65 |
-
"model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 66 |
-
"model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 67 |
-
"model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 68 |
-
"model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 69 |
-
"model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 70 |
-
"model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 71 |
-
"model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 72 |
-
"model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 73 |
-
"model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 74 |
-
"model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 75 |
-
"model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 76 |
-
"model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 77 |
-
"model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 78 |
-
"model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 79 |
-
"model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 80 |
-
"model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 81 |
-
"model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 82 |
-
"model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 83 |
-
"model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 84 |
-
"model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 85 |
-
"model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 86 |
-
"model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 87 |
-
"model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 88 |
-
"model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 89 |
-
"model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 90 |
-
"model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 91 |
-
"model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 92 |
-
"model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 93 |
-
"model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 94 |
-
"model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 95 |
-
"model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 96 |
-
"model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 97 |
-
"model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 98 |
-
"model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 99 |
-
"model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 100 |
-
"model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 101 |
-
"model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 102 |
-
"model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 103 |
-
"model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 104 |
-
"model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 105 |
-
"model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 106 |
-
"model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 107 |
-
"model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 108 |
-
"model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 109 |
-
"model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 110 |
-
"model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 111 |
-
"model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 112 |
-
"model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 113 |
-
"model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 114 |
-
"model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 115 |
-
"model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 116 |
-
"model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 117 |
-
"model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 118 |
-
"model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 119 |
-
"model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 120 |
-
"model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 121 |
-
"model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 122 |
-
"model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 123 |
-
"model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 124 |
-
"model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 125 |
-
"model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 126 |
-
"model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 127 |
-
"model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 128 |
-
"model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 129 |
-
"model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 130 |
-
"model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 131 |
-
"model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 132 |
-
"model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 133 |
-
"model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 134 |
-
"model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 135 |
-
"model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 136 |
-
"model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 137 |
-
"model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 138 |
-
"model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 139 |
-
"model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 140 |
-
"model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 141 |
-
"model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 142 |
-
"model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 143 |
-
"model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 144 |
-
"model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 145 |
-
"model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 146 |
-
"model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 147 |
-
"model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 148 |
-
"model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 149 |
-
"model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 150 |
-
"model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 151 |
-
"model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 152 |
-
"model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 153 |
-
"model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 154 |
-
"model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 155 |
-
"model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 156 |
-
"model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 157 |
-
"model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 158 |
-
"model.layers.22.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 159 |
-
"model.layers.22.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 160 |
-
"model.layers.22.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 161 |
-
"model.layers.22.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 162 |
-
"model.layers.22.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 163 |
-
"model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 164 |
-
"model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 165 |
-
"model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 166 |
-
"model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 167 |
-
"model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 168 |
-
"model.layers.23.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 169 |
-
"model.layers.23.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 170 |
-
"model.layers.23.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 171 |
-
"model.layers.23.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 172 |
-
"model.layers.23.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 173 |
-
"model.layers.23.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 174 |
-
"model.layers.23.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 175 |
-
"model.layers.23.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 176 |
-
"model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 177 |
-
"model.layers.23.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 178 |
-
"model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 179 |
-
"model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 180 |
-
"model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 181 |
-
"model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 182 |
-
"model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 183 |
-
"model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 184 |
-
"model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 185 |
-
"model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 186 |
-
"model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 187 |
-
"model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 188 |
-
"model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 189 |
-
"model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 190 |
-
"model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 191 |
-
"model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 192 |
-
"model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 193 |
-
"model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 194 |
-
"model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 195 |
-
"model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 196 |
-
"model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 197 |
-
"model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 198 |
-
"model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 199 |
-
"model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 200 |
-
"model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 201 |
-
"model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 202 |
-
"model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 203 |
-
"model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 204 |
-
"model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 205 |
-
"model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 206 |
-
"model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 207 |
-
"model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 208 |
-
"model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 209 |
-
"model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 210 |
-
"model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 211 |
-
"model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 212 |
-
"model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 213 |
-
"model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 214 |
-
"model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 215 |
-
"model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 216 |
-
"model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 217 |
-
"model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 218 |
-
"model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 219 |
-
"model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 220 |
-
"model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 221 |
-
"model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 222 |
-
"model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 223 |
-
"model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 224 |
-
"model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 225 |
-
"model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 226 |
-
"model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 227 |
-
"model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 228 |
-
"model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 229 |
-
"model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 230 |
-
"model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 231 |
-
"model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 232 |
-
"model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 233 |
-
"model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 234 |
-
"model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 235 |
-
"model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 236 |
-
"model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 237 |
-
"model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 238 |
-
"model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 239 |
-
"model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 240 |
-
"model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 241 |
-
"model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 242 |
-
"model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 243 |
-
"model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 244 |
-
"model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 245 |
-
"model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 246 |
-
"model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 247 |
-
"model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 248 |
-
"model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 249 |
-
"model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 250 |
-
"model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 251 |
-
"model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 252 |
-
"model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 253 |
-
"model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 254 |
-
"model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 255 |
-
"model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 256 |
-
"model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 257 |
-
"model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 258 |
-
"model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 259 |
-
"model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 260 |
-
"model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 261 |
-
"model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 262 |
-
"model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
|
| 263 |
-
"model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 264 |
-
"model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 265 |
-
"model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 266 |
-
"model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
|
| 267 |
-
"model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 268 |
-
"model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 269 |
-
"model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 270 |
-
"model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 271 |
-
"model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 272 |
-
"model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 273 |
-
"model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 274 |
-
"model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 275 |
-
"model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 276 |
-
"model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 277 |
-
"model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 278 |
-
"model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 279 |
-
"model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 280 |
-
"model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 281 |
-
"model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 282 |
-
"model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 283 |
-
"model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 284 |
-
"model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 285 |
-
"model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 286 |
-
"model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 287 |
-
"model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 288 |
-
"model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 289 |
-
"model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 290 |
-
"model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 291 |
-
"model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 292 |
-
"model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 293 |
-
"model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 294 |
-
"model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 295 |
-
"model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 296 |
-
"model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 297 |
-
"model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 298 |
-
"model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 299 |
-
"model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 300 |
-
"model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 301 |
-
"model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 302 |
-
"model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 303 |
-
"model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 304 |
-
"model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 305 |
-
"model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 306 |
-
"model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 307 |
-
"model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 308 |
-
"model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 309 |
-
"model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 310 |
-
"model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 311 |
-
"model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 312 |
-
"model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 313 |
-
"model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 314 |
-
"model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 315 |
-
"model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 316 |
-
"model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 317 |
-
"model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 318 |
-
"model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 319 |
-
"model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 320 |
-
"model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 321 |
-
"model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 322 |
-
"model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
|
| 323 |
-
"model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 324 |
-
"model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 325 |
-
"model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 326 |
-
"model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
|
| 327 |
-
"model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
|
| 328 |
-
"model.mm_projector.bias": "pytorch_model-00002-of-00002.bin",
|
| 329 |
-
"model.mm_projector.weight": "pytorch_model-00002-of-00002.bin",
|
| 330 |
-
"model.norm.weight": "pytorch_model-00002-of-00002.bin",
|
| 331 |
-
"model.text_hidden_fcs.0.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 332 |
-
"model.text_hidden_fcs.0.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 333 |
-
"model.text_hidden_fcs.0.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 334 |
-
"model.text_hidden_fcs.0.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 335 |
-
"model.visual_model.image_encoder.blocks.0.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 336 |
-
"model.visual_model.image_encoder.blocks.0.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 337 |
-
"model.visual_model.image_encoder.blocks.0.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 338 |
-
"model.visual_model.image_encoder.blocks.0.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 339 |
-
"model.visual_model.image_encoder.blocks.0.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 340 |
-
"model.visual_model.image_encoder.blocks.0.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 341 |
-
"model.visual_model.image_encoder.blocks.0.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 342 |
-
"model.visual_model.image_encoder.blocks.0.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 343 |
-
"model.visual_model.image_encoder.blocks.0.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 344 |
-
"model.visual_model.image_encoder.blocks.0.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 345 |
-
"model.visual_model.image_encoder.blocks.0.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 346 |
-
"model.visual_model.image_encoder.blocks.0.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 347 |
-
"model.visual_model.image_encoder.blocks.0.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 348 |
-
"model.visual_model.image_encoder.blocks.0.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 349 |
-
"model.visual_model.image_encoder.blocks.1.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 350 |
-
"model.visual_model.image_encoder.blocks.1.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 351 |
-
"model.visual_model.image_encoder.blocks.1.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 352 |
-
"model.visual_model.image_encoder.blocks.1.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 353 |
-
"model.visual_model.image_encoder.blocks.1.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 354 |
-
"model.visual_model.image_encoder.blocks.1.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 355 |
-
"model.visual_model.image_encoder.blocks.1.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 356 |
-
"model.visual_model.image_encoder.blocks.1.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 357 |
-
"model.visual_model.image_encoder.blocks.1.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 358 |
-
"model.visual_model.image_encoder.blocks.1.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 359 |
-
"model.visual_model.image_encoder.blocks.1.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 360 |
-
"model.visual_model.image_encoder.blocks.1.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 361 |
-
"model.visual_model.image_encoder.blocks.1.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 362 |
-
"model.visual_model.image_encoder.blocks.1.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 363 |
-
"model.visual_model.image_encoder.blocks.10.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 364 |
-
"model.visual_model.image_encoder.blocks.10.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 365 |
-
"model.visual_model.image_encoder.blocks.10.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 366 |
-
"model.visual_model.image_encoder.blocks.10.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 367 |
-
"model.visual_model.image_encoder.blocks.10.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 368 |
-
"model.visual_model.image_encoder.blocks.10.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 369 |
-
"model.visual_model.image_encoder.blocks.10.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 370 |
-
"model.visual_model.image_encoder.blocks.10.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 371 |
-
"model.visual_model.image_encoder.blocks.10.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 372 |
-
"model.visual_model.image_encoder.blocks.10.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 373 |
-
"model.visual_model.image_encoder.blocks.10.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 374 |
-
"model.visual_model.image_encoder.blocks.10.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 375 |
-
"model.visual_model.image_encoder.blocks.10.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 376 |
-
"model.visual_model.image_encoder.blocks.10.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 377 |
-
"model.visual_model.image_encoder.blocks.11.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 378 |
-
"model.visual_model.image_encoder.blocks.11.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 379 |
-
"model.visual_model.image_encoder.blocks.11.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 380 |
-
"model.visual_model.image_encoder.blocks.11.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 381 |
-
"model.visual_model.image_encoder.blocks.11.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 382 |
-
"model.visual_model.image_encoder.blocks.11.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 383 |
-
"model.visual_model.image_encoder.blocks.11.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 384 |
-
"model.visual_model.image_encoder.blocks.11.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 385 |
-
"model.visual_model.image_encoder.blocks.11.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 386 |
-
"model.visual_model.image_encoder.blocks.11.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 387 |
-
"model.visual_model.image_encoder.blocks.11.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 388 |
-
"model.visual_model.image_encoder.blocks.11.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 389 |
-
"model.visual_model.image_encoder.blocks.11.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 390 |
-
"model.visual_model.image_encoder.blocks.11.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 391 |
-
"model.visual_model.image_encoder.blocks.12.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 392 |
-
"model.visual_model.image_encoder.blocks.12.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 393 |
-
"model.visual_model.image_encoder.blocks.12.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 394 |
-
"model.visual_model.image_encoder.blocks.12.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 395 |
-
"model.visual_model.image_encoder.blocks.12.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 396 |
-
"model.visual_model.image_encoder.blocks.12.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 397 |
-
"model.visual_model.image_encoder.blocks.12.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 398 |
-
"model.visual_model.image_encoder.blocks.12.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 399 |
-
"model.visual_model.image_encoder.blocks.12.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 400 |
-
"model.visual_model.image_encoder.blocks.12.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 401 |
-
"model.visual_model.image_encoder.blocks.12.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 402 |
-
"model.visual_model.image_encoder.blocks.12.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 403 |
-
"model.visual_model.image_encoder.blocks.12.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 404 |
-
"model.visual_model.image_encoder.blocks.12.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 405 |
-
"model.visual_model.image_encoder.blocks.13.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 406 |
-
"model.visual_model.image_encoder.blocks.13.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 407 |
-
"model.visual_model.image_encoder.blocks.13.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 408 |
-
"model.visual_model.image_encoder.blocks.13.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 409 |
-
"model.visual_model.image_encoder.blocks.13.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 410 |
-
"model.visual_model.image_encoder.blocks.13.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 411 |
-
"model.visual_model.image_encoder.blocks.13.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 412 |
-
"model.visual_model.image_encoder.blocks.13.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 413 |
-
"model.visual_model.image_encoder.blocks.13.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 414 |
-
"model.visual_model.image_encoder.blocks.13.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 415 |
-
"model.visual_model.image_encoder.blocks.13.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 416 |
-
"model.visual_model.image_encoder.blocks.13.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 417 |
-
"model.visual_model.image_encoder.blocks.13.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 418 |
-
"model.visual_model.image_encoder.blocks.13.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 419 |
-
"model.visual_model.image_encoder.blocks.14.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 420 |
-
"model.visual_model.image_encoder.blocks.14.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 421 |
-
"model.visual_model.image_encoder.blocks.14.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 422 |
-
"model.visual_model.image_encoder.blocks.14.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 423 |
-
"model.visual_model.image_encoder.blocks.14.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 424 |
-
"model.visual_model.image_encoder.blocks.14.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 425 |
-
"model.visual_model.image_encoder.blocks.14.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 426 |
-
"model.visual_model.image_encoder.blocks.14.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 427 |
-
"model.visual_model.image_encoder.blocks.14.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 428 |
-
"model.visual_model.image_encoder.blocks.14.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 429 |
-
"model.visual_model.image_encoder.blocks.14.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 430 |
-
"model.visual_model.image_encoder.blocks.14.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 431 |
-
"model.visual_model.image_encoder.blocks.14.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 432 |
-
"model.visual_model.image_encoder.blocks.14.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 433 |
-
"model.visual_model.image_encoder.blocks.15.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 434 |
-
"model.visual_model.image_encoder.blocks.15.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 435 |
-
"model.visual_model.image_encoder.blocks.15.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 436 |
-
"model.visual_model.image_encoder.blocks.15.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 437 |
-
"model.visual_model.image_encoder.blocks.15.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 438 |
-
"model.visual_model.image_encoder.blocks.15.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 439 |
-
"model.visual_model.image_encoder.blocks.15.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 440 |
-
"model.visual_model.image_encoder.blocks.15.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 441 |
-
"model.visual_model.image_encoder.blocks.15.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 442 |
-
"model.visual_model.image_encoder.blocks.15.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 443 |
-
"model.visual_model.image_encoder.blocks.15.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 444 |
-
"model.visual_model.image_encoder.blocks.15.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 445 |
-
"model.visual_model.image_encoder.blocks.15.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 446 |
-
"model.visual_model.image_encoder.blocks.15.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 447 |
-
"model.visual_model.image_encoder.blocks.16.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 448 |
-
"model.visual_model.image_encoder.blocks.16.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 449 |
-
"model.visual_model.image_encoder.blocks.16.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 450 |
-
"model.visual_model.image_encoder.blocks.16.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 451 |
-
"model.visual_model.image_encoder.blocks.16.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 452 |
-
"model.visual_model.image_encoder.blocks.16.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 453 |
-
"model.visual_model.image_encoder.blocks.16.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 454 |
-
"model.visual_model.image_encoder.blocks.16.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 455 |
-
"model.visual_model.image_encoder.blocks.16.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 456 |
-
"model.visual_model.image_encoder.blocks.16.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 457 |
-
"model.visual_model.image_encoder.blocks.16.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 458 |
-
"model.visual_model.image_encoder.blocks.16.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 459 |
-
"model.visual_model.image_encoder.blocks.16.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 460 |
-
"model.visual_model.image_encoder.blocks.16.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 461 |
-
"model.visual_model.image_encoder.blocks.17.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 462 |
-
"model.visual_model.image_encoder.blocks.17.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 463 |
-
"model.visual_model.image_encoder.blocks.17.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 464 |
-
"model.visual_model.image_encoder.blocks.17.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 465 |
-
"model.visual_model.image_encoder.blocks.17.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 466 |
-
"model.visual_model.image_encoder.blocks.17.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 467 |
-
"model.visual_model.image_encoder.blocks.17.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 468 |
-
"model.visual_model.image_encoder.blocks.17.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 469 |
-
"model.visual_model.image_encoder.blocks.17.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 470 |
-
"model.visual_model.image_encoder.blocks.17.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 471 |
-
"model.visual_model.image_encoder.blocks.17.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 472 |
-
"model.visual_model.image_encoder.blocks.17.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 473 |
-
"model.visual_model.image_encoder.blocks.17.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 474 |
-
"model.visual_model.image_encoder.blocks.17.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 475 |
-
"model.visual_model.image_encoder.blocks.18.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 476 |
-
"model.visual_model.image_encoder.blocks.18.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 477 |
-
"model.visual_model.image_encoder.blocks.18.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 478 |
-
"model.visual_model.image_encoder.blocks.18.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 479 |
-
"model.visual_model.image_encoder.blocks.18.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 480 |
-
"model.visual_model.image_encoder.blocks.18.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 481 |
-
"model.visual_model.image_encoder.blocks.18.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 482 |
-
"model.visual_model.image_encoder.blocks.18.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 483 |
-
"model.visual_model.image_encoder.blocks.18.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 484 |
-
"model.visual_model.image_encoder.blocks.18.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 485 |
-
"model.visual_model.image_encoder.blocks.18.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 486 |
-
"model.visual_model.image_encoder.blocks.18.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 487 |
-
"model.visual_model.image_encoder.blocks.18.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 488 |
-
"model.visual_model.image_encoder.blocks.18.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 489 |
-
"model.visual_model.image_encoder.blocks.19.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 490 |
-
"model.visual_model.image_encoder.blocks.19.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 491 |
-
"model.visual_model.image_encoder.blocks.19.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 492 |
-
"model.visual_model.image_encoder.blocks.19.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 493 |
-
"model.visual_model.image_encoder.blocks.19.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 494 |
-
"model.visual_model.image_encoder.blocks.19.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 495 |
-
"model.visual_model.image_encoder.blocks.19.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 496 |
-
"model.visual_model.image_encoder.blocks.19.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 497 |
-
"model.visual_model.image_encoder.blocks.19.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 498 |
-
"model.visual_model.image_encoder.blocks.19.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 499 |
-
"model.visual_model.image_encoder.blocks.19.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 500 |
-
"model.visual_model.image_encoder.blocks.19.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 501 |
-
"model.visual_model.image_encoder.blocks.19.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 502 |
-
"model.visual_model.image_encoder.blocks.19.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 503 |
-
"model.visual_model.image_encoder.blocks.2.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 504 |
-
"model.visual_model.image_encoder.blocks.2.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 505 |
-
"model.visual_model.image_encoder.blocks.2.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 506 |
-
"model.visual_model.image_encoder.blocks.2.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 507 |
-
"model.visual_model.image_encoder.blocks.2.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 508 |
-
"model.visual_model.image_encoder.blocks.2.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 509 |
-
"model.visual_model.image_encoder.blocks.2.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 510 |
-
"model.visual_model.image_encoder.blocks.2.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 511 |
-
"model.visual_model.image_encoder.blocks.2.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 512 |
-
"model.visual_model.image_encoder.blocks.2.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 513 |
-
"model.visual_model.image_encoder.blocks.2.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 514 |
-
"model.visual_model.image_encoder.blocks.2.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 515 |
-
"model.visual_model.image_encoder.blocks.2.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 516 |
-
"model.visual_model.image_encoder.blocks.2.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 517 |
-
"model.visual_model.image_encoder.blocks.20.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 518 |
-
"model.visual_model.image_encoder.blocks.20.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 519 |
-
"model.visual_model.image_encoder.blocks.20.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 520 |
-
"model.visual_model.image_encoder.blocks.20.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 521 |
-
"model.visual_model.image_encoder.blocks.20.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 522 |
-
"model.visual_model.image_encoder.blocks.20.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 523 |
-
"model.visual_model.image_encoder.blocks.20.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 524 |
-
"model.visual_model.image_encoder.blocks.20.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 525 |
-
"model.visual_model.image_encoder.blocks.20.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 526 |
-
"model.visual_model.image_encoder.blocks.20.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 527 |
-
"model.visual_model.image_encoder.blocks.20.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 528 |
-
"model.visual_model.image_encoder.blocks.20.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 529 |
-
"model.visual_model.image_encoder.blocks.20.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 530 |
-
"model.visual_model.image_encoder.blocks.20.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 531 |
-
"model.visual_model.image_encoder.blocks.21.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 532 |
-
"model.visual_model.image_encoder.blocks.21.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 533 |
-
"model.visual_model.image_encoder.blocks.21.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 534 |
-
"model.visual_model.image_encoder.blocks.21.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 535 |
-
"model.visual_model.image_encoder.blocks.21.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 536 |
-
"model.visual_model.image_encoder.blocks.21.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 537 |
-
"model.visual_model.image_encoder.blocks.21.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 538 |
-
"model.visual_model.image_encoder.blocks.21.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 539 |
-
"model.visual_model.image_encoder.blocks.21.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 540 |
-
"model.visual_model.image_encoder.blocks.21.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 541 |
-
"model.visual_model.image_encoder.blocks.21.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 542 |
-
"model.visual_model.image_encoder.blocks.21.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 543 |
-
"model.visual_model.image_encoder.blocks.21.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 544 |
-
"model.visual_model.image_encoder.blocks.21.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 545 |
-
"model.visual_model.image_encoder.blocks.22.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 546 |
-
"model.visual_model.image_encoder.blocks.22.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 547 |
-
"model.visual_model.image_encoder.blocks.22.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 548 |
-
"model.visual_model.image_encoder.blocks.22.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 549 |
-
"model.visual_model.image_encoder.blocks.22.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 550 |
-
"model.visual_model.image_encoder.blocks.22.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 551 |
-
"model.visual_model.image_encoder.blocks.22.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 552 |
-
"model.visual_model.image_encoder.blocks.22.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 553 |
-
"model.visual_model.image_encoder.blocks.22.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 554 |
-
"model.visual_model.image_encoder.blocks.22.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 555 |
-
"model.visual_model.image_encoder.blocks.22.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 556 |
-
"model.visual_model.image_encoder.blocks.22.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 557 |
-
"model.visual_model.image_encoder.blocks.22.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 558 |
-
"model.visual_model.image_encoder.blocks.22.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 559 |
-
"model.visual_model.image_encoder.blocks.23.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 560 |
-
"model.visual_model.image_encoder.blocks.23.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 561 |
-
"model.visual_model.image_encoder.blocks.23.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 562 |
-
"model.visual_model.image_encoder.blocks.23.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 563 |
-
"model.visual_model.image_encoder.blocks.23.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 564 |
-
"model.visual_model.image_encoder.blocks.23.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 565 |
-
"model.visual_model.image_encoder.blocks.23.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 566 |
-
"model.visual_model.image_encoder.blocks.23.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 567 |
-
"model.visual_model.image_encoder.blocks.23.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 568 |
-
"model.visual_model.image_encoder.blocks.23.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 569 |
-
"model.visual_model.image_encoder.blocks.23.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 570 |
-
"model.visual_model.image_encoder.blocks.23.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 571 |
-
"model.visual_model.image_encoder.blocks.23.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 572 |
-
"model.visual_model.image_encoder.blocks.23.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 573 |
-
"model.visual_model.image_encoder.blocks.24.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 574 |
-
"model.visual_model.image_encoder.blocks.24.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 575 |
-
"model.visual_model.image_encoder.blocks.24.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 576 |
-
"model.visual_model.image_encoder.blocks.24.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 577 |
-
"model.visual_model.image_encoder.blocks.24.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 578 |
-
"model.visual_model.image_encoder.blocks.24.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 579 |
-
"model.visual_model.image_encoder.blocks.24.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 580 |
-
"model.visual_model.image_encoder.blocks.24.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 581 |
-
"model.visual_model.image_encoder.blocks.24.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 582 |
-
"model.visual_model.image_encoder.blocks.24.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 583 |
-
"model.visual_model.image_encoder.blocks.24.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 584 |
-
"model.visual_model.image_encoder.blocks.24.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 585 |
-
"model.visual_model.image_encoder.blocks.24.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 586 |
-
"model.visual_model.image_encoder.blocks.24.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 587 |
-
"model.visual_model.image_encoder.blocks.25.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 588 |
-
"model.visual_model.image_encoder.blocks.25.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 589 |
-
"model.visual_model.image_encoder.blocks.25.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 590 |
-
"model.visual_model.image_encoder.blocks.25.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 591 |
-
"model.visual_model.image_encoder.blocks.25.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 592 |
-
"model.visual_model.image_encoder.blocks.25.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 593 |
-
"model.visual_model.image_encoder.blocks.25.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 594 |
-
"model.visual_model.image_encoder.blocks.25.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 595 |
-
"model.visual_model.image_encoder.blocks.25.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 596 |
-
"model.visual_model.image_encoder.blocks.25.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 597 |
-
"model.visual_model.image_encoder.blocks.25.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 598 |
-
"model.visual_model.image_encoder.blocks.25.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 599 |
-
"model.visual_model.image_encoder.blocks.25.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 600 |
-
"model.visual_model.image_encoder.blocks.25.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 601 |
-
"model.visual_model.image_encoder.blocks.26.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 602 |
-
"model.visual_model.image_encoder.blocks.26.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 603 |
-
"model.visual_model.image_encoder.blocks.26.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 604 |
-
"model.visual_model.image_encoder.blocks.26.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 605 |
-
"model.visual_model.image_encoder.blocks.26.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 606 |
-
"model.visual_model.image_encoder.blocks.26.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 607 |
-
"model.visual_model.image_encoder.blocks.26.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 608 |
-
"model.visual_model.image_encoder.blocks.26.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 609 |
-
"model.visual_model.image_encoder.blocks.26.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 610 |
-
"model.visual_model.image_encoder.blocks.26.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 611 |
-
"model.visual_model.image_encoder.blocks.26.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 612 |
-
"model.visual_model.image_encoder.blocks.26.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 613 |
-
"model.visual_model.image_encoder.blocks.26.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 614 |
-
"model.visual_model.image_encoder.blocks.26.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 615 |
-
"model.visual_model.image_encoder.blocks.27.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 616 |
-
"model.visual_model.image_encoder.blocks.27.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 617 |
-
"model.visual_model.image_encoder.blocks.27.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 618 |
-
"model.visual_model.image_encoder.blocks.27.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 619 |
-
"model.visual_model.image_encoder.blocks.27.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 620 |
-
"model.visual_model.image_encoder.blocks.27.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 621 |
-
"model.visual_model.image_encoder.blocks.27.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 622 |
-
"model.visual_model.image_encoder.blocks.27.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 623 |
-
"model.visual_model.image_encoder.blocks.27.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 624 |
-
"model.visual_model.image_encoder.blocks.27.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 625 |
-
"model.visual_model.image_encoder.blocks.27.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 626 |
-
"model.visual_model.image_encoder.blocks.27.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 627 |
-
"model.visual_model.image_encoder.blocks.27.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 628 |
-
"model.visual_model.image_encoder.blocks.27.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 629 |
-
"model.visual_model.image_encoder.blocks.28.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 630 |
-
"model.visual_model.image_encoder.blocks.28.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 631 |
-
"model.visual_model.image_encoder.blocks.28.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 632 |
-
"model.visual_model.image_encoder.blocks.28.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 633 |
-
"model.visual_model.image_encoder.blocks.28.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 634 |
-
"model.visual_model.image_encoder.blocks.28.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 635 |
-
"model.visual_model.image_encoder.blocks.28.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 636 |
-
"model.visual_model.image_encoder.blocks.28.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 637 |
-
"model.visual_model.image_encoder.blocks.28.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 638 |
-
"model.visual_model.image_encoder.blocks.28.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 639 |
-
"model.visual_model.image_encoder.blocks.28.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 640 |
-
"model.visual_model.image_encoder.blocks.28.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 641 |
-
"model.visual_model.image_encoder.blocks.28.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 642 |
-
"model.visual_model.image_encoder.blocks.28.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 643 |
-
"model.visual_model.image_encoder.blocks.29.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 644 |
-
"model.visual_model.image_encoder.blocks.29.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 645 |
-
"model.visual_model.image_encoder.blocks.29.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 646 |
-
"model.visual_model.image_encoder.blocks.29.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 647 |
-
"model.visual_model.image_encoder.blocks.29.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 648 |
-
"model.visual_model.image_encoder.blocks.29.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 649 |
-
"model.visual_model.image_encoder.blocks.29.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 650 |
-
"model.visual_model.image_encoder.blocks.29.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 651 |
-
"model.visual_model.image_encoder.blocks.29.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 652 |
-
"model.visual_model.image_encoder.blocks.29.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 653 |
-
"model.visual_model.image_encoder.blocks.29.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 654 |
-
"model.visual_model.image_encoder.blocks.29.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 655 |
-
"model.visual_model.image_encoder.blocks.29.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 656 |
-
"model.visual_model.image_encoder.blocks.29.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 657 |
-
"model.visual_model.image_encoder.blocks.3.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 658 |
-
"model.visual_model.image_encoder.blocks.3.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 659 |
-
"model.visual_model.image_encoder.blocks.3.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 660 |
-
"model.visual_model.image_encoder.blocks.3.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 661 |
-
"model.visual_model.image_encoder.blocks.3.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 662 |
-
"model.visual_model.image_encoder.blocks.3.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 663 |
-
"model.visual_model.image_encoder.blocks.3.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 664 |
-
"model.visual_model.image_encoder.blocks.3.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 665 |
-
"model.visual_model.image_encoder.blocks.3.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 666 |
-
"model.visual_model.image_encoder.blocks.3.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 667 |
-
"model.visual_model.image_encoder.blocks.3.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 668 |
-
"model.visual_model.image_encoder.blocks.3.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 669 |
-
"model.visual_model.image_encoder.blocks.3.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 670 |
-
"model.visual_model.image_encoder.blocks.3.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 671 |
-
"model.visual_model.image_encoder.blocks.30.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 672 |
-
"model.visual_model.image_encoder.blocks.30.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 673 |
-
"model.visual_model.image_encoder.blocks.30.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 674 |
-
"model.visual_model.image_encoder.blocks.30.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 675 |
-
"model.visual_model.image_encoder.blocks.30.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 676 |
-
"model.visual_model.image_encoder.blocks.30.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 677 |
-
"model.visual_model.image_encoder.blocks.30.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 678 |
-
"model.visual_model.image_encoder.blocks.30.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 679 |
-
"model.visual_model.image_encoder.blocks.30.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 680 |
-
"model.visual_model.image_encoder.blocks.30.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 681 |
-
"model.visual_model.image_encoder.blocks.30.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 682 |
-
"model.visual_model.image_encoder.blocks.30.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 683 |
-
"model.visual_model.image_encoder.blocks.30.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 684 |
-
"model.visual_model.image_encoder.blocks.30.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 685 |
-
"model.visual_model.image_encoder.blocks.31.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 686 |
-
"model.visual_model.image_encoder.blocks.31.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 687 |
-
"model.visual_model.image_encoder.blocks.31.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 688 |
-
"model.visual_model.image_encoder.blocks.31.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 689 |
-
"model.visual_model.image_encoder.blocks.31.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 690 |
-
"model.visual_model.image_encoder.blocks.31.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 691 |
-
"model.visual_model.image_encoder.blocks.31.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 692 |
-
"model.visual_model.image_encoder.blocks.31.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 693 |
-
"model.visual_model.image_encoder.blocks.31.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 694 |
-
"model.visual_model.image_encoder.blocks.31.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 695 |
-
"model.visual_model.image_encoder.blocks.31.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 696 |
-
"model.visual_model.image_encoder.blocks.31.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 697 |
-
"model.visual_model.image_encoder.blocks.31.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 698 |
-
"model.visual_model.image_encoder.blocks.31.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 699 |
-
"model.visual_model.image_encoder.blocks.4.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 700 |
-
"model.visual_model.image_encoder.blocks.4.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 701 |
-
"model.visual_model.image_encoder.blocks.4.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 702 |
-
"model.visual_model.image_encoder.blocks.4.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 703 |
-
"model.visual_model.image_encoder.blocks.4.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 704 |
-
"model.visual_model.image_encoder.blocks.4.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 705 |
-
"model.visual_model.image_encoder.blocks.4.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 706 |
-
"model.visual_model.image_encoder.blocks.4.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 707 |
-
"model.visual_model.image_encoder.blocks.4.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 708 |
-
"model.visual_model.image_encoder.blocks.4.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 709 |
-
"model.visual_model.image_encoder.blocks.4.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 710 |
-
"model.visual_model.image_encoder.blocks.4.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 711 |
-
"model.visual_model.image_encoder.blocks.4.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 712 |
-
"model.visual_model.image_encoder.blocks.4.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 713 |
-
"model.visual_model.image_encoder.blocks.5.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 714 |
-
"model.visual_model.image_encoder.blocks.5.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 715 |
-
"model.visual_model.image_encoder.blocks.5.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 716 |
-
"model.visual_model.image_encoder.blocks.5.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 717 |
-
"model.visual_model.image_encoder.blocks.5.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 718 |
-
"model.visual_model.image_encoder.blocks.5.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 719 |
-
"model.visual_model.image_encoder.blocks.5.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 720 |
-
"model.visual_model.image_encoder.blocks.5.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 721 |
-
"model.visual_model.image_encoder.blocks.5.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 722 |
-
"model.visual_model.image_encoder.blocks.5.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 723 |
-
"model.visual_model.image_encoder.blocks.5.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 724 |
-
"model.visual_model.image_encoder.blocks.5.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 725 |
-
"model.visual_model.image_encoder.blocks.5.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 726 |
-
"model.visual_model.image_encoder.blocks.5.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 727 |
-
"model.visual_model.image_encoder.blocks.6.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 728 |
-
"model.visual_model.image_encoder.blocks.6.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 729 |
-
"model.visual_model.image_encoder.blocks.6.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 730 |
-
"model.visual_model.image_encoder.blocks.6.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 731 |
-
"model.visual_model.image_encoder.blocks.6.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 732 |
-
"model.visual_model.image_encoder.blocks.6.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 733 |
-
"model.visual_model.image_encoder.blocks.6.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 734 |
-
"model.visual_model.image_encoder.blocks.6.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 735 |
-
"model.visual_model.image_encoder.blocks.6.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 736 |
-
"model.visual_model.image_encoder.blocks.6.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 737 |
-
"model.visual_model.image_encoder.blocks.6.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 738 |
-
"model.visual_model.image_encoder.blocks.6.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 739 |
-
"model.visual_model.image_encoder.blocks.6.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 740 |
-
"model.visual_model.image_encoder.blocks.6.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 741 |
-
"model.visual_model.image_encoder.blocks.7.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 742 |
-
"model.visual_model.image_encoder.blocks.7.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 743 |
-
"model.visual_model.image_encoder.blocks.7.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 744 |
-
"model.visual_model.image_encoder.blocks.7.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 745 |
-
"model.visual_model.image_encoder.blocks.7.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 746 |
-
"model.visual_model.image_encoder.blocks.7.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 747 |
-
"model.visual_model.image_encoder.blocks.7.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 748 |
-
"model.visual_model.image_encoder.blocks.7.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 749 |
-
"model.visual_model.image_encoder.blocks.7.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 750 |
-
"model.visual_model.image_encoder.blocks.7.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 751 |
-
"model.visual_model.image_encoder.blocks.7.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 752 |
-
"model.visual_model.image_encoder.blocks.7.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 753 |
-
"model.visual_model.image_encoder.blocks.7.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 754 |
-
"model.visual_model.image_encoder.blocks.7.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 755 |
-
"model.visual_model.image_encoder.blocks.8.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 756 |
-
"model.visual_model.image_encoder.blocks.8.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 757 |
-
"model.visual_model.image_encoder.blocks.8.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 758 |
-
"model.visual_model.image_encoder.blocks.8.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 759 |
-
"model.visual_model.image_encoder.blocks.8.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 760 |
-
"model.visual_model.image_encoder.blocks.8.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 761 |
-
"model.visual_model.image_encoder.blocks.8.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 762 |
-
"model.visual_model.image_encoder.blocks.8.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 763 |
-
"model.visual_model.image_encoder.blocks.8.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 764 |
-
"model.visual_model.image_encoder.blocks.8.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 765 |
-
"model.visual_model.image_encoder.blocks.8.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 766 |
-
"model.visual_model.image_encoder.blocks.8.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 767 |
-
"model.visual_model.image_encoder.blocks.8.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 768 |
-
"model.visual_model.image_encoder.blocks.8.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 769 |
-
"model.visual_model.image_encoder.blocks.9.attn.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 770 |
-
"model.visual_model.image_encoder.blocks.9.attn.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 771 |
-
"model.visual_model.image_encoder.blocks.9.attn.qkv.bias": "pytorch_model-00002-of-00002.bin",
|
| 772 |
-
"model.visual_model.image_encoder.blocks.9.attn.qkv.weight": "pytorch_model-00002-of-00002.bin",
|
| 773 |
-
"model.visual_model.image_encoder.blocks.9.attn.rel_pos_h": "pytorch_model-00002-of-00002.bin",
|
| 774 |
-
"model.visual_model.image_encoder.blocks.9.attn.rel_pos_w": "pytorch_model-00002-of-00002.bin",
|
| 775 |
-
"model.visual_model.image_encoder.blocks.9.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 776 |
-
"model.visual_model.image_encoder.blocks.9.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 777 |
-
"model.visual_model.image_encoder.blocks.9.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 778 |
-
"model.visual_model.image_encoder.blocks.9.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 779 |
-
"model.visual_model.image_encoder.blocks.9.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 780 |
-
"model.visual_model.image_encoder.blocks.9.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 781 |
-
"model.visual_model.image_encoder.blocks.9.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 782 |
-
"model.visual_model.image_encoder.blocks.9.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 783 |
-
"model.visual_model.image_encoder.neck.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 784 |
-
"model.visual_model.image_encoder.neck.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 785 |
-
"model.visual_model.image_encoder.neck.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 786 |
-
"model.visual_model.image_encoder.neck.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 787 |
-
"model.visual_model.image_encoder.neck.3.bias": "pytorch_model-00002-of-00002.bin",
|
| 788 |
-
"model.visual_model.image_encoder.neck.3.weight": "pytorch_model-00002-of-00002.bin",
|
| 789 |
-
"model.visual_model.image_encoder.patch_embed.proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 790 |
-
"model.visual_model.image_encoder.patch_embed.proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 791 |
-
"model.visual_model.image_encoder.pos_embed": "pytorch_model-00002-of-00002.bin",
|
| 792 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 793 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 794 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 795 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 796 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 797 |
-
"model.visual_model.mask_decoder.iou_prediction_head.layers.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 798 |
-
"model.visual_model.mask_decoder.iou_token.weight": "pytorch_model-00002-of-00002.bin",
|
| 799 |
-
"model.visual_model.mask_decoder.mask_tokens.weight": "pytorch_model-00002-of-00002.bin",
|
| 800 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 801 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 802 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 803 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 804 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 805 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.0.layers.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 806 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 807 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 808 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 809 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 810 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 811 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.1.layers.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 812 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 813 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 814 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 815 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 816 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 817 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.2.layers.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 818 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 819 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 820 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 821 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 822 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.2.bias": "pytorch_model-00002-of-00002.bin",
|
| 823 |
-
"model.visual_model.mask_decoder.output_hypernetworks_mlps.3.layers.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 824 |
-
"model.visual_model.mask_decoder.output_upscaling.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 825 |
-
"model.visual_model.mask_decoder.output_upscaling.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 826 |
-
"model.visual_model.mask_decoder.output_upscaling.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 827 |
-
"model.visual_model.mask_decoder.output_upscaling.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 828 |
-
"model.visual_model.mask_decoder.output_upscaling.3.bias": "pytorch_model-00002-of-00002.bin",
|
| 829 |
-
"model.visual_model.mask_decoder.output_upscaling.3.weight": "pytorch_model-00002-of-00002.bin",
|
| 830 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 831 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 832 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 833 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 834 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 835 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 836 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 837 |
-
"model.visual_model.mask_decoder.transformer.final_attn_token_to_image.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 838 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 839 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 840 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 841 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 842 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 843 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 844 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 845 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_image_to_token.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 846 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 847 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 848 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 849 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 850 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 851 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 852 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 853 |
-
"model.visual_model.mask_decoder.transformer.layers.0.cross_attn_token_to_image.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 854 |
-
"model.visual_model.mask_decoder.transformer.layers.0.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 855 |
-
"model.visual_model.mask_decoder.transformer.layers.0.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 856 |
-
"model.visual_model.mask_decoder.transformer.layers.0.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 857 |
-
"model.visual_model.mask_decoder.transformer.layers.0.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 858 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 859 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 860 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 861 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 862 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm3.bias": "pytorch_model-00002-of-00002.bin",
|
| 863 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm3.weight": "pytorch_model-00002-of-00002.bin",
|
| 864 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm4.bias": "pytorch_model-00002-of-00002.bin",
|
| 865 |
-
"model.visual_model.mask_decoder.transformer.layers.0.norm4.weight": "pytorch_model-00002-of-00002.bin",
|
| 866 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 867 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 868 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 869 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 870 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 871 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 872 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 873 |
-
"model.visual_model.mask_decoder.transformer.layers.0.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 874 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 875 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 876 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 877 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 878 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 879 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 880 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 881 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_image_to_token.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 882 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 883 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 884 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 885 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 886 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 887 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 888 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 889 |
-
"model.visual_model.mask_decoder.transformer.layers.1.cross_attn_token_to_image.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 890 |
-
"model.visual_model.mask_decoder.transformer.layers.1.mlp.lin1.bias": "pytorch_model-00002-of-00002.bin",
|
| 891 |
-
"model.visual_model.mask_decoder.transformer.layers.1.mlp.lin1.weight": "pytorch_model-00002-of-00002.bin",
|
| 892 |
-
"model.visual_model.mask_decoder.transformer.layers.1.mlp.lin2.bias": "pytorch_model-00002-of-00002.bin",
|
| 893 |
-
"model.visual_model.mask_decoder.transformer.layers.1.mlp.lin2.weight": "pytorch_model-00002-of-00002.bin",
|
| 894 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm1.bias": "pytorch_model-00002-of-00002.bin",
|
| 895 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm1.weight": "pytorch_model-00002-of-00002.bin",
|
| 896 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm2.bias": "pytorch_model-00002-of-00002.bin",
|
| 897 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm2.weight": "pytorch_model-00002-of-00002.bin",
|
| 898 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm3.bias": "pytorch_model-00002-of-00002.bin",
|
| 899 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm3.weight": "pytorch_model-00002-of-00002.bin",
|
| 900 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm4.bias": "pytorch_model-00002-of-00002.bin",
|
| 901 |
-
"model.visual_model.mask_decoder.transformer.layers.1.norm4.weight": "pytorch_model-00002-of-00002.bin",
|
| 902 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.k_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 903 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 904 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.out_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 905 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.out_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 906 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.q_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 907 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 908 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.v_proj.bias": "pytorch_model-00002-of-00002.bin",
|
| 909 |
-
"model.visual_model.mask_decoder.transformer.layers.1.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
|
| 910 |
-
"model.visual_model.mask_decoder.transformer.norm_final_attn.bias": "pytorch_model-00002-of-00002.bin",
|
| 911 |
-
"model.visual_model.mask_decoder.transformer.norm_final_attn.weight": "pytorch_model-00002-of-00002.bin",
|
| 912 |
-
"model.visual_model.prompt_encoder.mask_downscaling.0.bias": "pytorch_model-00002-of-00002.bin",
|
| 913 |
-
"model.visual_model.prompt_encoder.mask_downscaling.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 914 |
-
"model.visual_model.prompt_encoder.mask_downscaling.1.bias": "pytorch_model-00002-of-00002.bin",
|
| 915 |
-
"model.visual_model.prompt_encoder.mask_downscaling.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 916 |
-
"model.visual_model.prompt_encoder.mask_downscaling.3.bias": "pytorch_model-00002-of-00002.bin",
|
| 917 |
-
"model.visual_model.prompt_encoder.mask_downscaling.3.weight": "pytorch_model-00002-of-00002.bin",
|
| 918 |
-
"model.visual_model.prompt_encoder.mask_downscaling.4.bias": "pytorch_model-00002-of-00002.bin",
|
| 919 |
-
"model.visual_model.prompt_encoder.mask_downscaling.4.weight": "pytorch_model-00002-of-00002.bin",
|
| 920 |
-
"model.visual_model.prompt_encoder.mask_downscaling.6.bias": "pytorch_model-00002-of-00002.bin",
|
| 921 |
-
"model.visual_model.prompt_encoder.mask_downscaling.6.weight": "pytorch_model-00002-of-00002.bin",
|
| 922 |
-
"model.visual_model.prompt_encoder.no_mask_embed.weight": "pytorch_model-00002-of-00002.bin",
|
| 923 |
-
"model.visual_model.prompt_encoder.not_a_point_embed.weight": "pytorch_model-00002-of-00002.bin",
|
| 924 |
-
"model.visual_model.prompt_encoder.pe_layer.positional_encoding_gaussian_matrix": "pytorch_model-00002-of-00002.bin",
|
| 925 |
-
"model.visual_model.prompt_encoder.point_embeddings.0.weight": "pytorch_model-00002-of-00002.bin",
|
| 926 |
-
"model.visual_model.prompt_encoder.point_embeddings.1.weight": "pytorch_model-00002-of-00002.bin",
|
| 927 |
-
"model.visual_model.prompt_encoder.point_embeddings.2.weight": "pytorch_model-00002-of-00002.bin",
|
| 928 |
-
"model.visual_model.prompt_encoder.point_embeddings.3.weight": "pytorch_model-00002-of-00002.bin"
|
| 929 |
-
}
|
| 930 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
special_tokens_map.json
DELETED
|
@@ -1 +0,0 @@
|
|
| 1 |
-
{"bos_token": {"content": "<|startoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "pad_token": "<|endoftext|>"}
|
|
|
|
|
|
tokenizer.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer.model
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
-
size 499723
|
|
|
|
|
|
|
|
|
|
|
|
tokenizer_config.json
DELETED
|
@@ -1,34 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"unk_token": {
|
| 3 |
-
"content": "<|endoftext|>",
|
| 4 |
-
"single_word": false,
|
| 5 |
-
"lstrip": false,
|
| 6 |
-
"rstrip": false,
|
| 7 |
-
"normalized": true,
|
| 8 |
-
"__type": "AddedToken"
|
| 9 |
-
},
|
| 10 |
-
"bos_token": {
|
| 11 |
-
"content": "<|startoftext|>",
|
| 12 |
-
"single_word": false,
|
| 13 |
-
"lstrip": false,
|
| 14 |
-
"rstrip": false,
|
| 15 |
-
"normalized": true,
|
| 16 |
-
"__type": "AddedToken"
|
| 17 |
-
},
|
| 18 |
-
"eos_token": {
|
| 19 |
-
"content": "<|endoftext|>",
|
| 20 |
-
"single_word": false,
|
| 21 |
-
"lstrip": false,
|
| 22 |
-
"rstrip": false,
|
| 23 |
-
"normalized": true,
|
| 24 |
-
"__type": "AddedToken"
|
| 25 |
-
},
|
| 26 |
-
"pad_token": "<|endoftext|>",
|
| 27 |
-
"add_prefix_space": false,
|
| 28 |
-
"errors": "replace",
|
| 29 |
-
"do_lower_case": true,
|
| 30 |
-
"name_or_path": "openai/clip-vit-base-patch32",
|
| 31 |
-
"model_max_length": 77,
|
| 32 |
-
"special_tokens_map_file": "./special_tokens_map.json",
|
| 33 |
-
"tokenizer_class": "CLIPTokenizer"
|
| 34 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
vocab.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
zzzmmz/SegAgent-Model/.mdl
DELETED
|
Binary file (44 Bytes)
|
|
|
zzzmmz/SegAgent-Model/.msc
DELETED
|
Binary file (1.45 kB)
|
|
|
zzzmmz/SegAgent-Model/.mv
DELETED
|
@@ -1 +0,0 @@
|
|
| 1 |
-
Revision:master,CreatedAt:1754641259
|
|
|
|
|
|
zzzmmz/SegAgent-Model/README.md
DELETED
|
@@ -1,50 +0,0 @@
|
|
| 1 |
-
---
|
| 2 |
-
frameworks:
|
| 3 |
-
- Pytorch
|
| 4 |
-
license: Apache License 2.0
|
| 5 |
-
tasks:
|
| 6 |
-
- image-captioning
|
| 7 |
-
|
| 8 |
-
#model-type:
|
| 9 |
-
##such as gpt、phi、llama、chatglm、baichuan, etc.
|
| 10 |
-
#- gpt
|
| 11 |
-
|
| 12 |
-
#domain:
|
| 13 |
-
##such as nlp、cv、audio、multi-modal, etc.
|
| 14 |
-
#- nlp
|
| 15 |
-
|
| 16 |
-
#language:
|
| 17 |
-
##language code list https://help.aliyun.com/document_detail/215387.html?spm=a2c4g.11186623.0.0.9f8d7467kni6Aa
|
| 18 |
-
#- cn
|
| 19 |
-
|
| 20 |
-
#metrics:
|
| 21 |
-
##such as CIDEr、Blue、ROUGE, etc.
|
| 22 |
-
#- CIDEr
|
| 23 |
-
|
| 24 |
-
#tags:
|
| 25 |
-
##various custom tags, including pretrained, fine-tuned, instruction-tuned, RL-tuned, and others
|
| 26 |
-
#- pretrained
|
| 27 |
-
|
| 28 |
-
#tools:
|
| 29 |
-
##such as vllm、fastchat、llamacpp、AdaSeq, etc.
|
| 30 |
-
#- vllm
|
| 31 |
-
---
|
| 32 |
-
### You are viewing the default Readme template as no detailed model-card was provided by the model’s contributors. You can access the model files in the "Files and versions" tab.
|
| 33 |
-
#### Model files may be downloaded with ModelScope SDK or through git clone directly.
|
| 34 |
-
|
| 35 |
-
Download with ModelScope’s Python SDK
|
| 36 |
-
```bash
|
| 37 |
-
#Install ModelScope
|
| 38 |
-
pip install modelscope
|
| 39 |
-
```
|
| 40 |
-
```python
|
| 41 |
-
#Download with ModelScope’s Python SDK
|
| 42 |
-
from modelscope import snapshot_download
|
| 43 |
-
model_dir = snapshot_download('zzzmmz/SegAgent-Model')
|
| 44 |
-
```
|
| 45 |
-
Download with Git clone
|
| 46 |
-
```
|
| 47 |
-
git clone https://www.modelscope.cn/zzzmmz/SegAgent-Model.git
|
| 48 |
-
```
|
| 49 |
-
|
| 50 |
-
<p style="color: lightgrey;">If you are a contributor to this model, we invite you to promptly update the model card content according to <a href="https://modelscope.cn/docs/ModelScope%E6%A8%A1%E5%9E%8B%E6%8E%A5%E5%85%A5%E6%B5%81%E7%A8%8B%E6%A6%82%E8%A7%88" style="color: lightgrey; text-decoration: underline;">the model contribution documentation</a>.</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/config.json
DELETED
|
@@ -1,49 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "/mnt/input/zhumuzhi.zmz/weight/models--Qwen--Qwen-VL-Chat",
|
| 3 |
-
"architectures": [
|
| 4 |
-
"QWenLMHeadModel"
|
| 5 |
-
],
|
| 6 |
-
"attn_dropout_prob": 0.0,
|
| 7 |
-
"auto_map": {
|
| 8 |
-
"AutoConfig": "configuration_qwen.QWenConfig",
|
| 9 |
-
"AutoModelForCausalLM": "modeling_qwen.QWenLMHeadModel"
|
| 10 |
-
},
|
| 11 |
-
"bf16": true,
|
| 12 |
-
"emb_dropout_prob": 0.0,
|
| 13 |
-
"fp16": false,
|
| 14 |
-
"fp32": false,
|
| 15 |
-
"hidden_size": 4096,
|
| 16 |
-
"initializer_range": 0.02,
|
| 17 |
-
"intermediate_size": 22016,
|
| 18 |
-
"kv_channels": 128,
|
| 19 |
-
"layer_norm_epsilon": 1e-06,
|
| 20 |
-
"max_position_embeddings": 8192,
|
| 21 |
-
"model_type": "qwen",
|
| 22 |
-
"no_bias": true,
|
| 23 |
-
"num_attention_heads": 32,
|
| 24 |
-
"num_hidden_layers": 32,
|
| 25 |
-
"onnx_safe": null,
|
| 26 |
-
"rotary_emb_base": 10000,
|
| 27 |
-
"rotary_pct": 1.0,
|
| 28 |
-
"scale_attn_weights": true,
|
| 29 |
-
"seq_length": 2048,
|
| 30 |
-
"tie_word_embeddings": false,
|
| 31 |
-
"tokenizer_type": "QWenTokenizer",
|
| 32 |
-
"torch_dtype": "bfloat16",
|
| 33 |
-
"transformers_version": "4.37.2",
|
| 34 |
-
"use_cache": false,
|
| 35 |
-
"use_dynamic_ntk": true,
|
| 36 |
-
"use_flash_attn": false,
|
| 37 |
-
"use_logn_attn": true,
|
| 38 |
-
"visual": {
|
| 39 |
-
"heads": 16,
|
| 40 |
-
"image_size": 448,
|
| 41 |
-
"image_start_id": 151857,
|
| 42 |
-
"layers": 48,
|
| 43 |
-
"mlp_ratio": 4.9231,
|
| 44 |
-
"output_dim": 4096,
|
| 45 |
-
"patch_size": 14,
|
| 46 |
-
"width": 1664
|
| 47 |
-
},
|
| 48 |
-
"vocab_size": 151936
|
| 49 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/configuration.json
DELETED
|
@@ -1 +0,0 @@
|
|
| 1 |
-
{"framework":"Pytorch","task":"image-captioning"}
|
|
|
|
|
|
zzzmmz/SegAgent-Model/configuration_qwen.py
DELETED
|
@@ -1,65 +0,0 @@
|
|
| 1 |
-
# Copyright (c) Alibaba Cloud.
|
| 2 |
-
#
|
| 3 |
-
# This source code is licensed under the license found in the
|
| 4 |
-
# LICENSE file in the root directory of this source tree.
|
| 5 |
-
|
| 6 |
-
from transformers import PretrainedConfig
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
class QWenConfig(PretrainedConfig):
|
| 10 |
-
model_type = "qwen"
|
| 11 |
-
keys_to_ignore_at_inference = ["past_key_values"]
|
| 12 |
-
|
| 13 |
-
def __init__(
|
| 14 |
-
self,
|
| 15 |
-
vocab_size=151936,
|
| 16 |
-
hidden_size=4096,
|
| 17 |
-
num_hidden_layers=32,
|
| 18 |
-
num_attention_heads=32,
|
| 19 |
-
emb_dropout_prob=0.0,
|
| 20 |
-
attn_dropout_prob=0.0,
|
| 21 |
-
layer_norm_epsilon=1e-6,
|
| 22 |
-
initializer_range=0.02,
|
| 23 |
-
max_position_embeddings=8192,
|
| 24 |
-
scale_attn_weights=True,
|
| 25 |
-
use_cache=True,
|
| 26 |
-
bf16=False,
|
| 27 |
-
fp16=False,
|
| 28 |
-
fp32=False,
|
| 29 |
-
kv_channels=128,
|
| 30 |
-
rotary_pct=1.0,
|
| 31 |
-
rotary_emb_base=10000,
|
| 32 |
-
use_dynamic_ntk=True,
|
| 33 |
-
use_logn_attn=True,
|
| 34 |
-
use_flash_attn="auto",
|
| 35 |
-
intermediate_size=22016,
|
| 36 |
-
no_bias=True,
|
| 37 |
-
tie_word_embeddings=False,
|
| 38 |
-
**kwargs,
|
| 39 |
-
):
|
| 40 |
-
self.vocab_size = vocab_size
|
| 41 |
-
self.hidden_size = hidden_size
|
| 42 |
-
self.intermediate_size = intermediate_size
|
| 43 |
-
self.num_hidden_layers = num_hidden_layers
|
| 44 |
-
self.num_attention_heads = num_attention_heads
|
| 45 |
-
self.emb_dropout_prob = emb_dropout_prob
|
| 46 |
-
self.attn_dropout_prob = attn_dropout_prob
|
| 47 |
-
self.layer_norm_epsilon = layer_norm_epsilon
|
| 48 |
-
self.initializer_range = initializer_range
|
| 49 |
-
self.scale_attn_weights = scale_attn_weights
|
| 50 |
-
self.use_cache = use_cache
|
| 51 |
-
self.max_position_embeddings = max_position_embeddings
|
| 52 |
-
self.bf16 = bf16
|
| 53 |
-
self.fp16 = fp16
|
| 54 |
-
self.fp32 = fp32
|
| 55 |
-
self.kv_channels = kv_channels
|
| 56 |
-
self.rotary_pct = rotary_pct
|
| 57 |
-
self.rotary_emb_base = rotary_emb_base
|
| 58 |
-
self.use_dynamic_ntk = use_dynamic_ntk
|
| 59 |
-
self.use_logn_attn = use_logn_attn
|
| 60 |
-
self.use_flash_attn = use_flash_attn
|
| 61 |
-
self.no_bias = no_bias
|
| 62 |
-
super().__init__(
|
| 63 |
-
tie_word_embeddings=tie_word_embeddings,
|
| 64 |
-
**kwargs
|
| 65 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/generation_config.json
DELETED
|
@@ -1,11 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"chat_format": "chatml",
|
| 3 |
-
"do_sample": true,
|
| 4 |
-
"eos_token_id": 151643,
|
| 5 |
-
"max_new_tokens": 512,
|
| 6 |
-
"max_window_size": 6144,
|
| 7 |
-
"pad_token_id": 151643,
|
| 8 |
-
"top_k": 0,
|
| 9 |
-
"top_p": 0.3,
|
| 10 |
-
"transformers_version": "4.37.2"
|
| 11 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/model-00001-of-00004.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:470ade4cefc50a771662a41077ea08f0d166c701eacb6a399652ccc383a5f386
|
| 3 |
-
size 4988485656
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/model-00002-of-00004.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d2ee9a55ff2501ed625d914fa21e16ff6751058d74f96e4d408cb7fc9718f348
|
| 3 |
-
size 4981246520
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/model-00003-of-00004.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d665efb246e1fe8a80f2a26e6f4af2256c2a78f94b532486a751d5ee99b70599
|
| 3 |
-
size 4977360088
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/model-00004-of-00004.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:8b975b9b3b9dcbf70e093f81684cf36b3c2e3c1aacf19546e6cb838addaa3446
|
| 3 |
-
size 4366885504
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/model.safetensors.index.json
DELETED
|
@@ -1,860 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"metadata": {
|
| 3 |
-
"total_size": 19313870336
|
| 4 |
-
},
|
| 5 |
-
"weight_map": {
|
| 6 |
-
"lm_head.weight": "model-00004-of-00004.safetensors",
|
| 7 |
-
"transformer.h.0.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 8 |
-
"transformer.h.0.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 9 |
-
"transformer.h.0.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 10 |
-
"transformer.h.0.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 11 |
-
"transformer.h.0.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 12 |
-
"transformer.h.0.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 13 |
-
"transformer.h.0.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 14 |
-
"transformer.h.0.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 15 |
-
"transformer.h.1.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 16 |
-
"transformer.h.1.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 17 |
-
"transformer.h.1.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 18 |
-
"transformer.h.1.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 19 |
-
"transformer.h.1.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 20 |
-
"transformer.h.1.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 21 |
-
"transformer.h.1.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 22 |
-
"transformer.h.1.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 23 |
-
"transformer.h.10.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 24 |
-
"transformer.h.10.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 25 |
-
"transformer.h.10.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 26 |
-
"transformer.h.10.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 27 |
-
"transformer.h.10.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 28 |
-
"transformer.h.10.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 29 |
-
"transformer.h.10.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 30 |
-
"transformer.h.10.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 31 |
-
"transformer.h.11.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 32 |
-
"transformer.h.11.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 33 |
-
"transformer.h.11.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 34 |
-
"transformer.h.11.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 35 |
-
"transformer.h.11.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 36 |
-
"transformer.h.11.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 37 |
-
"transformer.h.11.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 38 |
-
"transformer.h.11.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 39 |
-
"transformer.h.12.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 40 |
-
"transformer.h.12.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 41 |
-
"transformer.h.12.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 42 |
-
"transformer.h.12.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 43 |
-
"transformer.h.12.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 44 |
-
"transformer.h.12.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 45 |
-
"transformer.h.12.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 46 |
-
"transformer.h.12.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 47 |
-
"transformer.h.13.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 48 |
-
"transformer.h.13.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 49 |
-
"transformer.h.13.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 50 |
-
"transformer.h.13.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 51 |
-
"transformer.h.13.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 52 |
-
"transformer.h.13.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 53 |
-
"transformer.h.13.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 54 |
-
"transformer.h.13.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 55 |
-
"transformer.h.14.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 56 |
-
"transformer.h.14.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 57 |
-
"transformer.h.14.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 58 |
-
"transformer.h.14.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 59 |
-
"transformer.h.14.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 60 |
-
"transformer.h.14.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 61 |
-
"transformer.h.14.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 62 |
-
"transformer.h.14.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 63 |
-
"transformer.h.15.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 64 |
-
"transformer.h.15.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 65 |
-
"transformer.h.15.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 66 |
-
"transformer.h.15.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 67 |
-
"transformer.h.15.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 68 |
-
"transformer.h.15.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 69 |
-
"transformer.h.15.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 70 |
-
"transformer.h.15.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 71 |
-
"transformer.h.16.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 72 |
-
"transformer.h.16.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 73 |
-
"transformer.h.16.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 74 |
-
"transformer.h.16.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 75 |
-
"transformer.h.16.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 76 |
-
"transformer.h.16.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 77 |
-
"transformer.h.16.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 78 |
-
"transformer.h.16.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 79 |
-
"transformer.h.17.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 80 |
-
"transformer.h.17.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 81 |
-
"transformer.h.17.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 82 |
-
"transformer.h.17.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 83 |
-
"transformer.h.17.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 84 |
-
"transformer.h.17.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 85 |
-
"transformer.h.17.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 86 |
-
"transformer.h.17.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 87 |
-
"transformer.h.18.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 88 |
-
"transformer.h.18.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 89 |
-
"transformer.h.18.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 90 |
-
"transformer.h.18.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 91 |
-
"transformer.h.18.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 92 |
-
"transformer.h.18.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 93 |
-
"transformer.h.18.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 94 |
-
"transformer.h.18.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 95 |
-
"transformer.h.19.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 96 |
-
"transformer.h.19.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 97 |
-
"transformer.h.19.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 98 |
-
"transformer.h.19.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 99 |
-
"transformer.h.19.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 100 |
-
"transformer.h.19.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 101 |
-
"transformer.h.19.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 102 |
-
"transformer.h.19.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 103 |
-
"transformer.h.2.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 104 |
-
"transformer.h.2.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 105 |
-
"transformer.h.2.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 106 |
-
"transformer.h.2.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 107 |
-
"transformer.h.2.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 108 |
-
"transformer.h.2.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 109 |
-
"transformer.h.2.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 110 |
-
"transformer.h.2.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 111 |
-
"transformer.h.20.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 112 |
-
"transformer.h.20.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 113 |
-
"transformer.h.20.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 114 |
-
"transformer.h.20.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 115 |
-
"transformer.h.20.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 116 |
-
"transformer.h.20.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 117 |
-
"transformer.h.20.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 118 |
-
"transformer.h.20.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 119 |
-
"transformer.h.21.attn.c_attn.bias": "model-00002-of-00004.safetensors",
|
| 120 |
-
"transformer.h.21.attn.c_attn.weight": "model-00002-of-00004.safetensors",
|
| 121 |
-
"transformer.h.21.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 122 |
-
"transformer.h.21.ln_1.weight": "model-00002-of-00004.safetensors",
|
| 123 |
-
"transformer.h.21.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 124 |
-
"transformer.h.21.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 125 |
-
"transformer.h.21.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 126 |
-
"transformer.h.21.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 127 |
-
"transformer.h.22.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 128 |
-
"transformer.h.22.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 129 |
-
"transformer.h.22.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 130 |
-
"transformer.h.22.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 131 |
-
"transformer.h.22.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 132 |
-
"transformer.h.22.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 133 |
-
"transformer.h.22.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 134 |
-
"transformer.h.22.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 135 |
-
"transformer.h.23.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 136 |
-
"transformer.h.23.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 137 |
-
"transformer.h.23.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 138 |
-
"transformer.h.23.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 139 |
-
"transformer.h.23.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 140 |
-
"transformer.h.23.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 141 |
-
"transformer.h.23.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 142 |
-
"transformer.h.23.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 143 |
-
"transformer.h.24.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 144 |
-
"transformer.h.24.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 145 |
-
"transformer.h.24.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 146 |
-
"transformer.h.24.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 147 |
-
"transformer.h.24.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 148 |
-
"transformer.h.24.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 149 |
-
"transformer.h.24.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 150 |
-
"transformer.h.24.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 151 |
-
"transformer.h.25.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 152 |
-
"transformer.h.25.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 153 |
-
"transformer.h.25.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 154 |
-
"transformer.h.25.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 155 |
-
"transformer.h.25.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 156 |
-
"transformer.h.25.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 157 |
-
"transformer.h.25.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 158 |
-
"transformer.h.25.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 159 |
-
"transformer.h.26.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 160 |
-
"transformer.h.26.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 161 |
-
"transformer.h.26.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 162 |
-
"transformer.h.26.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 163 |
-
"transformer.h.26.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 164 |
-
"transformer.h.26.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 165 |
-
"transformer.h.26.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 166 |
-
"transformer.h.26.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 167 |
-
"transformer.h.27.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 168 |
-
"transformer.h.27.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 169 |
-
"transformer.h.27.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 170 |
-
"transformer.h.27.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 171 |
-
"transformer.h.27.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 172 |
-
"transformer.h.27.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 173 |
-
"transformer.h.27.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 174 |
-
"transformer.h.27.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 175 |
-
"transformer.h.28.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 176 |
-
"transformer.h.28.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 177 |
-
"transformer.h.28.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 178 |
-
"transformer.h.28.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 179 |
-
"transformer.h.28.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 180 |
-
"transformer.h.28.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 181 |
-
"transformer.h.28.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 182 |
-
"transformer.h.28.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 183 |
-
"transformer.h.29.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 184 |
-
"transformer.h.29.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 185 |
-
"transformer.h.29.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 186 |
-
"transformer.h.29.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 187 |
-
"transformer.h.29.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 188 |
-
"transformer.h.29.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 189 |
-
"transformer.h.29.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 190 |
-
"transformer.h.29.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 191 |
-
"transformer.h.3.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 192 |
-
"transformer.h.3.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 193 |
-
"transformer.h.3.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 194 |
-
"transformer.h.3.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 195 |
-
"transformer.h.3.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 196 |
-
"transformer.h.3.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 197 |
-
"transformer.h.3.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 198 |
-
"transformer.h.3.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 199 |
-
"transformer.h.30.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 200 |
-
"transformer.h.30.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 201 |
-
"transformer.h.30.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 202 |
-
"transformer.h.30.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 203 |
-
"transformer.h.30.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 204 |
-
"transformer.h.30.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 205 |
-
"transformer.h.30.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 206 |
-
"transformer.h.30.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 207 |
-
"transformer.h.31.attn.c_attn.bias": "model-00003-of-00004.safetensors",
|
| 208 |
-
"transformer.h.31.attn.c_attn.weight": "model-00003-of-00004.safetensors",
|
| 209 |
-
"transformer.h.31.attn.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 210 |
-
"transformer.h.31.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 211 |
-
"transformer.h.31.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 212 |
-
"transformer.h.31.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 213 |
-
"transformer.h.31.mlp.w1.weight": "model-00003-of-00004.safetensors",
|
| 214 |
-
"transformer.h.31.mlp.w2.weight": "model-00003-of-00004.safetensors",
|
| 215 |
-
"transformer.h.4.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 216 |
-
"transformer.h.4.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 217 |
-
"transformer.h.4.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 218 |
-
"transformer.h.4.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 219 |
-
"transformer.h.4.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 220 |
-
"transformer.h.4.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 221 |
-
"transformer.h.4.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 222 |
-
"transformer.h.4.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 223 |
-
"transformer.h.5.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 224 |
-
"transformer.h.5.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 225 |
-
"transformer.h.5.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 226 |
-
"transformer.h.5.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 227 |
-
"transformer.h.5.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 228 |
-
"transformer.h.5.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 229 |
-
"transformer.h.5.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 230 |
-
"transformer.h.5.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 231 |
-
"transformer.h.6.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 232 |
-
"transformer.h.6.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 233 |
-
"transformer.h.6.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 234 |
-
"transformer.h.6.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 235 |
-
"transformer.h.6.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 236 |
-
"transformer.h.6.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 237 |
-
"transformer.h.6.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 238 |
-
"transformer.h.6.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 239 |
-
"transformer.h.7.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 240 |
-
"transformer.h.7.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 241 |
-
"transformer.h.7.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 242 |
-
"transformer.h.7.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 243 |
-
"transformer.h.7.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 244 |
-
"transformer.h.7.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 245 |
-
"transformer.h.7.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 246 |
-
"transformer.h.7.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 247 |
-
"transformer.h.8.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 248 |
-
"transformer.h.8.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 249 |
-
"transformer.h.8.attn.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 250 |
-
"transformer.h.8.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 251 |
-
"transformer.h.8.ln_2.weight": "model-00001-of-00004.safetensors",
|
| 252 |
-
"transformer.h.8.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
|
| 253 |
-
"transformer.h.8.mlp.w1.weight": "model-00001-of-00004.safetensors",
|
| 254 |
-
"transformer.h.8.mlp.w2.weight": "model-00001-of-00004.safetensors",
|
| 255 |
-
"transformer.h.9.attn.c_attn.bias": "model-00001-of-00004.safetensors",
|
| 256 |
-
"transformer.h.9.attn.c_attn.weight": "model-00001-of-00004.safetensors",
|
| 257 |
-
"transformer.h.9.attn.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 258 |
-
"transformer.h.9.ln_1.weight": "model-00001-of-00004.safetensors",
|
| 259 |
-
"transformer.h.9.ln_2.weight": "model-00002-of-00004.safetensors",
|
| 260 |
-
"transformer.h.9.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
|
| 261 |
-
"transformer.h.9.mlp.w1.weight": "model-00002-of-00004.safetensors",
|
| 262 |
-
"transformer.h.9.mlp.w2.weight": "model-00002-of-00004.safetensors",
|
| 263 |
-
"transformer.ln_f.weight": "model-00003-of-00004.safetensors",
|
| 264 |
-
"transformer.visual.attn_pool.attn.in_proj_bias": "model-00004-of-00004.safetensors",
|
| 265 |
-
"transformer.visual.attn_pool.attn.in_proj_weight": "model-00004-of-00004.safetensors",
|
| 266 |
-
"transformer.visual.attn_pool.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 267 |
-
"transformer.visual.attn_pool.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 268 |
-
"transformer.visual.attn_pool.kv_proj.weight": "model-00004-of-00004.safetensors",
|
| 269 |
-
"transformer.visual.attn_pool.ln_kv.bias": "model-00004-of-00004.safetensors",
|
| 270 |
-
"transformer.visual.attn_pool.ln_kv.weight": "model-00004-of-00004.safetensors",
|
| 271 |
-
"transformer.visual.attn_pool.ln_q.bias": "model-00004-of-00004.safetensors",
|
| 272 |
-
"transformer.visual.attn_pool.ln_q.weight": "model-00004-of-00004.safetensors",
|
| 273 |
-
"transformer.visual.attn_pool.pos_embed": "model-00004-of-00004.safetensors",
|
| 274 |
-
"transformer.visual.attn_pool.query": "model-00004-of-00004.safetensors",
|
| 275 |
-
"transformer.visual.conv1.weight": "model-00003-of-00004.safetensors",
|
| 276 |
-
"transformer.visual.ln_post.bias": "model-00004-of-00004.safetensors",
|
| 277 |
-
"transformer.visual.ln_post.weight": "model-00004-of-00004.safetensors",
|
| 278 |
-
"transformer.visual.ln_pre.bias": "model-00003-of-00004.safetensors",
|
| 279 |
-
"transformer.visual.ln_pre.weight": "model-00003-of-00004.safetensors",
|
| 280 |
-
"transformer.visual.positional_embedding": "model-00003-of-00004.safetensors",
|
| 281 |
-
"transformer.visual.proj": "model-00003-of-00004.safetensors",
|
| 282 |
-
"transformer.visual.transformer.resblocks.0.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 283 |
-
"transformer.visual.transformer.resblocks.0.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 284 |
-
"transformer.visual.transformer.resblocks.0.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 285 |
-
"transformer.visual.transformer.resblocks.0.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 286 |
-
"transformer.visual.transformer.resblocks.0.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 287 |
-
"transformer.visual.transformer.resblocks.0.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 288 |
-
"transformer.visual.transformer.resblocks.0.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 289 |
-
"transformer.visual.transformer.resblocks.0.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 290 |
-
"transformer.visual.transformer.resblocks.0.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 291 |
-
"transformer.visual.transformer.resblocks.0.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 292 |
-
"transformer.visual.transformer.resblocks.0.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 293 |
-
"transformer.visual.transformer.resblocks.0.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 294 |
-
"transformer.visual.transformer.resblocks.1.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 295 |
-
"transformer.visual.transformer.resblocks.1.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 296 |
-
"transformer.visual.transformer.resblocks.1.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 297 |
-
"transformer.visual.transformer.resblocks.1.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 298 |
-
"transformer.visual.transformer.resblocks.1.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 299 |
-
"transformer.visual.transformer.resblocks.1.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 300 |
-
"transformer.visual.transformer.resblocks.1.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 301 |
-
"transformer.visual.transformer.resblocks.1.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 302 |
-
"transformer.visual.transformer.resblocks.1.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 303 |
-
"transformer.visual.transformer.resblocks.1.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 304 |
-
"transformer.visual.transformer.resblocks.1.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 305 |
-
"transformer.visual.transformer.resblocks.1.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 306 |
-
"transformer.visual.transformer.resblocks.10.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 307 |
-
"transformer.visual.transformer.resblocks.10.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 308 |
-
"transformer.visual.transformer.resblocks.10.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 309 |
-
"transformer.visual.transformer.resblocks.10.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 310 |
-
"transformer.visual.transformer.resblocks.10.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 311 |
-
"transformer.visual.transformer.resblocks.10.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 312 |
-
"transformer.visual.transformer.resblocks.10.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 313 |
-
"transformer.visual.transformer.resblocks.10.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 314 |
-
"transformer.visual.transformer.resblocks.10.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 315 |
-
"transformer.visual.transformer.resblocks.10.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 316 |
-
"transformer.visual.transformer.resblocks.10.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 317 |
-
"transformer.visual.transformer.resblocks.10.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 318 |
-
"transformer.visual.transformer.resblocks.11.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 319 |
-
"transformer.visual.transformer.resblocks.11.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 320 |
-
"transformer.visual.transformer.resblocks.11.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 321 |
-
"transformer.visual.transformer.resblocks.11.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 322 |
-
"transformer.visual.transformer.resblocks.11.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 323 |
-
"transformer.visual.transformer.resblocks.11.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 324 |
-
"transformer.visual.transformer.resblocks.11.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 325 |
-
"transformer.visual.transformer.resblocks.11.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 326 |
-
"transformer.visual.transformer.resblocks.11.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 327 |
-
"transformer.visual.transformer.resblocks.11.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 328 |
-
"transformer.visual.transformer.resblocks.11.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 329 |
-
"transformer.visual.transformer.resblocks.11.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 330 |
-
"transformer.visual.transformer.resblocks.12.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 331 |
-
"transformer.visual.transformer.resblocks.12.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 332 |
-
"transformer.visual.transformer.resblocks.12.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 333 |
-
"transformer.visual.transformer.resblocks.12.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 334 |
-
"transformer.visual.transformer.resblocks.12.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 335 |
-
"transformer.visual.transformer.resblocks.12.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 336 |
-
"transformer.visual.transformer.resblocks.12.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 337 |
-
"transformer.visual.transformer.resblocks.12.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 338 |
-
"transformer.visual.transformer.resblocks.12.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 339 |
-
"transformer.visual.transformer.resblocks.12.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 340 |
-
"transformer.visual.transformer.resblocks.12.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 341 |
-
"transformer.visual.transformer.resblocks.12.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 342 |
-
"transformer.visual.transformer.resblocks.13.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 343 |
-
"transformer.visual.transformer.resblocks.13.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 344 |
-
"transformer.visual.transformer.resblocks.13.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 345 |
-
"transformer.visual.transformer.resblocks.13.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 346 |
-
"transformer.visual.transformer.resblocks.13.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 347 |
-
"transformer.visual.transformer.resblocks.13.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 348 |
-
"transformer.visual.transformer.resblocks.13.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 349 |
-
"transformer.visual.transformer.resblocks.13.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 350 |
-
"transformer.visual.transformer.resblocks.13.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 351 |
-
"transformer.visual.transformer.resblocks.13.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 352 |
-
"transformer.visual.transformer.resblocks.13.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 353 |
-
"transformer.visual.transformer.resblocks.13.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 354 |
-
"transformer.visual.transformer.resblocks.14.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 355 |
-
"transformer.visual.transformer.resblocks.14.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 356 |
-
"transformer.visual.transformer.resblocks.14.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 357 |
-
"transformer.visual.transformer.resblocks.14.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 358 |
-
"transformer.visual.transformer.resblocks.14.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 359 |
-
"transformer.visual.transformer.resblocks.14.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 360 |
-
"transformer.visual.transformer.resblocks.14.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 361 |
-
"transformer.visual.transformer.resblocks.14.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 362 |
-
"transformer.visual.transformer.resblocks.14.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 363 |
-
"transformer.visual.transformer.resblocks.14.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 364 |
-
"transformer.visual.transformer.resblocks.14.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 365 |
-
"transformer.visual.transformer.resblocks.14.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 366 |
-
"transformer.visual.transformer.resblocks.15.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 367 |
-
"transformer.visual.transformer.resblocks.15.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 368 |
-
"transformer.visual.transformer.resblocks.15.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 369 |
-
"transformer.visual.transformer.resblocks.15.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 370 |
-
"transformer.visual.transformer.resblocks.15.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 371 |
-
"transformer.visual.transformer.resblocks.15.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 372 |
-
"transformer.visual.transformer.resblocks.15.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 373 |
-
"transformer.visual.transformer.resblocks.15.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 374 |
-
"transformer.visual.transformer.resblocks.15.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 375 |
-
"transformer.visual.transformer.resblocks.15.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 376 |
-
"transformer.visual.transformer.resblocks.15.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 377 |
-
"transformer.visual.transformer.resblocks.15.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 378 |
-
"transformer.visual.transformer.resblocks.16.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 379 |
-
"transformer.visual.transformer.resblocks.16.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 380 |
-
"transformer.visual.transformer.resblocks.16.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 381 |
-
"transformer.visual.transformer.resblocks.16.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 382 |
-
"transformer.visual.transformer.resblocks.16.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 383 |
-
"transformer.visual.transformer.resblocks.16.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 384 |
-
"transformer.visual.transformer.resblocks.16.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 385 |
-
"transformer.visual.transformer.resblocks.16.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 386 |
-
"transformer.visual.transformer.resblocks.16.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 387 |
-
"transformer.visual.transformer.resblocks.16.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 388 |
-
"transformer.visual.transformer.resblocks.16.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 389 |
-
"transformer.visual.transformer.resblocks.16.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 390 |
-
"transformer.visual.transformer.resblocks.17.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 391 |
-
"transformer.visual.transformer.resblocks.17.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 392 |
-
"transformer.visual.transformer.resblocks.17.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 393 |
-
"transformer.visual.transformer.resblocks.17.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 394 |
-
"transformer.visual.transformer.resblocks.17.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 395 |
-
"transformer.visual.transformer.resblocks.17.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 396 |
-
"transformer.visual.transformer.resblocks.17.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 397 |
-
"transformer.visual.transformer.resblocks.17.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 398 |
-
"transformer.visual.transformer.resblocks.17.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 399 |
-
"transformer.visual.transformer.resblocks.17.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 400 |
-
"transformer.visual.transformer.resblocks.17.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 401 |
-
"transformer.visual.transformer.resblocks.17.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 402 |
-
"transformer.visual.transformer.resblocks.18.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 403 |
-
"transformer.visual.transformer.resblocks.18.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 404 |
-
"transformer.visual.transformer.resblocks.18.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 405 |
-
"transformer.visual.transformer.resblocks.18.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 406 |
-
"transformer.visual.transformer.resblocks.18.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 407 |
-
"transformer.visual.transformer.resblocks.18.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 408 |
-
"transformer.visual.transformer.resblocks.18.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 409 |
-
"transformer.visual.transformer.resblocks.18.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 410 |
-
"transformer.visual.transformer.resblocks.18.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 411 |
-
"transformer.visual.transformer.resblocks.18.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 412 |
-
"transformer.visual.transformer.resblocks.18.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 413 |
-
"transformer.visual.transformer.resblocks.18.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 414 |
-
"transformer.visual.transformer.resblocks.19.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 415 |
-
"transformer.visual.transformer.resblocks.19.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 416 |
-
"transformer.visual.transformer.resblocks.19.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 417 |
-
"transformer.visual.transformer.resblocks.19.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 418 |
-
"transformer.visual.transformer.resblocks.19.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 419 |
-
"transformer.visual.transformer.resblocks.19.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 420 |
-
"transformer.visual.transformer.resblocks.19.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 421 |
-
"transformer.visual.transformer.resblocks.19.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 422 |
-
"transformer.visual.transformer.resblocks.19.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 423 |
-
"transformer.visual.transformer.resblocks.19.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 424 |
-
"transformer.visual.transformer.resblocks.19.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 425 |
-
"transformer.visual.transformer.resblocks.19.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 426 |
-
"transformer.visual.transformer.resblocks.2.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 427 |
-
"transformer.visual.transformer.resblocks.2.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 428 |
-
"transformer.visual.transformer.resblocks.2.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 429 |
-
"transformer.visual.transformer.resblocks.2.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 430 |
-
"transformer.visual.transformer.resblocks.2.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 431 |
-
"transformer.visual.transformer.resblocks.2.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 432 |
-
"transformer.visual.transformer.resblocks.2.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 433 |
-
"transformer.visual.transformer.resblocks.2.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 434 |
-
"transformer.visual.transformer.resblocks.2.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 435 |
-
"transformer.visual.transformer.resblocks.2.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 436 |
-
"transformer.visual.transformer.resblocks.2.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 437 |
-
"transformer.visual.transformer.resblocks.2.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 438 |
-
"transformer.visual.transformer.resblocks.20.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 439 |
-
"transformer.visual.transformer.resblocks.20.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 440 |
-
"transformer.visual.transformer.resblocks.20.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 441 |
-
"transformer.visual.transformer.resblocks.20.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 442 |
-
"transformer.visual.transformer.resblocks.20.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 443 |
-
"transformer.visual.transformer.resblocks.20.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 444 |
-
"transformer.visual.transformer.resblocks.20.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 445 |
-
"transformer.visual.transformer.resblocks.20.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 446 |
-
"transformer.visual.transformer.resblocks.20.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 447 |
-
"transformer.visual.transformer.resblocks.20.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 448 |
-
"transformer.visual.transformer.resblocks.20.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 449 |
-
"transformer.visual.transformer.resblocks.20.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 450 |
-
"transformer.visual.transformer.resblocks.21.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 451 |
-
"transformer.visual.transformer.resblocks.21.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 452 |
-
"transformer.visual.transformer.resblocks.21.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 453 |
-
"transformer.visual.transformer.resblocks.21.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 454 |
-
"transformer.visual.transformer.resblocks.21.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 455 |
-
"transformer.visual.transformer.resblocks.21.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 456 |
-
"transformer.visual.transformer.resblocks.21.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 457 |
-
"transformer.visual.transformer.resblocks.21.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 458 |
-
"transformer.visual.transformer.resblocks.21.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 459 |
-
"transformer.visual.transformer.resblocks.21.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 460 |
-
"transformer.visual.transformer.resblocks.21.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 461 |
-
"transformer.visual.transformer.resblocks.21.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 462 |
-
"transformer.visual.transformer.resblocks.22.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 463 |
-
"transformer.visual.transformer.resblocks.22.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 464 |
-
"transformer.visual.transformer.resblocks.22.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 465 |
-
"transformer.visual.transformer.resblocks.22.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 466 |
-
"transformer.visual.transformer.resblocks.22.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 467 |
-
"transformer.visual.transformer.resblocks.22.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 468 |
-
"transformer.visual.transformer.resblocks.22.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 469 |
-
"transformer.visual.transformer.resblocks.22.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 470 |
-
"transformer.visual.transformer.resblocks.22.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 471 |
-
"transformer.visual.transformer.resblocks.22.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 472 |
-
"transformer.visual.transformer.resblocks.22.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 473 |
-
"transformer.visual.transformer.resblocks.22.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 474 |
-
"transformer.visual.transformer.resblocks.23.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 475 |
-
"transformer.visual.transformer.resblocks.23.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 476 |
-
"transformer.visual.transformer.resblocks.23.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 477 |
-
"transformer.visual.transformer.resblocks.23.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 478 |
-
"transformer.visual.transformer.resblocks.23.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 479 |
-
"transformer.visual.transformer.resblocks.23.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 480 |
-
"transformer.visual.transformer.resblocks.23.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 481 |
-
"transformer.visual.transformer.resblocks.23.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 482 |
-
"transformer.visual.transformer.resblocks.23.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 483 |
-
"transformer.visual.transformer.resblocks.23.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 484 |
-
"transformer.visual.transformer.resblocks.23.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 485 |
-
"transformer.visual.transformer.resblocks.23.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 486 |
-
"transformer.visual.transformer.resblocks.24.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 487 |
-
"transformer.visual.transformer.resblocks.24.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 488 |
-
"transformer.visual.transformer.resblocks.24.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 489 |
-
"transformer.visual.transformer.resblocks.24.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 490 |
-
"transformer.visual.transformer.resblocks.24.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 491 |
-
"transformer.visual.transformer.resblocks.24.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 492 |
-
"transformer.visual.transformer.resblocks.24.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 493 |
-
"transformer.visual.transformer.resblocks.24.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 494 |
-
"transformer.visual.transformer.resblocks.24.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 495 |
-
"transformer.visual.transformer.resblocks.24.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 496 |
-
"transformer.visual.transformer.resblocks.24.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 497 |
-
"transformer.visual.transformer.resblocks.24.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 498 |
-
"transformer.visual.transformer.resblocks.25.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 499 |
-
"transformer.visual.transformer.resblocks.25.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 500 |
-
"transformer.visual.transformer.resblocks.25.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 501 |
-
"transformer.visual.transformer.resblocks.25.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 502 |
-
"transformer.visual.transformer.resblocks.25.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 503 |
-
"transformer.visual.transformer.resblocks.25.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 504 |
-
"transformer.visual.transformer.resblocks.25.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 505 |
-
"transformer.visual.transformer.resblocks.25.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 506 |
-
"transformer.visual.transformer.resblocks.25.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 507 |
-
"transformer.visual.transformer.resblocks.25.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 508 |
-
"transformer.visual.transformer.resblocks.25.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 509 |
-
"transformer.visual.transformer.resblocks.25.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 510 |
-
"transformer.visual.transformer.resblocks.26.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 511 |
-
"transformer.visual.transformer.resblocks.26.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 512 |
-
"transformer.visual.transformer.resblocks.26.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 513 |
-
"transformer.visual.transformer.resblocks.26.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 514 |
-
"transformer.visual.transformer.resblocks.26.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 515 |
-
"transformer.visual.transformer.resblocks.26.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 516 |
-
"transformer.visual.transformer.resblocks.26.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 517 |
-
"transformer.visual.transformer.resblocks.26.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 518 |
-
"transformer.visual.transformer.resblocks.26.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 519 |
-
"transformer.visual.transformer.resblocks.26.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 520 |
-
"transformer.visual.transformer.resblocks.26.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 521 |
-
"transformer.visual.transformer.resblocks.26.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 522 |
-
"transformer.visual.transformer.resblocks.27.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 523 |
-
"transformer.visual.transformer.resblocks.27.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 524 |
-
"transformer.visual.transformer.resblocks.27.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 525 |
-
"transformer.visual.transformer.resblocks.27.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 526 |
-
"transformer.visual.transformer.resblocks.27.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 527 |
-
"transformer.visual.transformer.resblocks.27.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 528 |
-
"transformer.visual.transformer.resblocks.27.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 529 |
-
"transformer.visual.transformer.resblocks.27.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 530 |
-
"transformer.visual.transformer.resblocks.27.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 531 |
-
"transformer.visual.transformer.resblocks.27.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 532 |
-
"transformer.visual.transformer.resblocks.27.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 533 |
-
"transformer.visual.transformer.resblocks.27.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 534 |
-
"transformer.visual.transformer.resblocks.28.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 535 |
-
"transformer.visual.transformer.resblocks.28.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 536 |
-
"transformer.visual.transformer.resblocks.28.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 537 |
-
"transformer.visual.transformer.resblocks.28.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 538 |
-
"transformer.visual.transformer.resblocks.28.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 539 |
-
"transformer.visual.transformer.resblocks.28.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 540 |
-
"transformer.visual.transformer.resblocks.28.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 541 |
-
"transformer.visual.transformer.resblocks.28.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 542 |
-
"transformer.visual.transformer.resblocks.28.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 543 |
-
"transformer.visual.transformer.resblocks.28.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 544 |
-
"transformer.visual.transformer.resblocks.28.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 545 |
-
"transformer.visual.transformer.resblocks.28.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 546 |
-
"transformer.visual.transformer.resblocks.29.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 547 |
-
"transformer.visual.transformer.resblocks.29.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 548 |
-
"transformer.visual.transformer.resblocks.29.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 549 |
-
"transformer.visual.transformer.resblocks.29.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 550 |
-
"transformer.visual.transformer.resblocks.29.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 551 |
-
"transformer.visual.transformer.resblocks.29.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 552 |
-
"transformer.visual.transformer.resblocks.29.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 553 |
-
"transformer.visual.transformer.resblocks.29.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 554 |
-
"transformer.visual.transformer.resblocks.29.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 555 |
-
"transformer.visual.transformer.resblocks.29.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 556 |
-
"transformer.visual.transformer.resblocks.29.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 557 |
-
"transformer.visual.transformer.resblocks.29.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 558 |
-
"transformer.visual.transformer.resblocks.3.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 559 |
-
"transformer.visual.transformer.resblocks.3.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 560 |
-
"transformer.visual.transformer.resblocks.3.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 561 |
-
"transformer.visual.transformer.resblocks.3.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 562 |
-
"transformer.visual.transformer.resblocks.3.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 563 |
-
"transformer.visual.transformer.resblocks.3.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 564 |
-
"transformer.visual.transformer.resblocks.3.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 565 |
-
"transformer.visual.transformer.resblocks.3.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 566 |
-
"transformer.visual.transformer.resblocks.3.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 567 |
-
"transformer.visual.transformer.resblocks.3.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 568 |
-
"transformer.visual.transformer.resblocks.3.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 569 |
-
"transformer.visual.transformer.resblocks.3.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 570 |
-
"transformer.visual.transformer.resblocks.30.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 571 |
-
"transformer.visual.transformer.resblocks.30.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 572 |
-
"transformer.visual.transformer.resblocks.30.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 573 |
-
"transformer.visual.transformer.resblocks.30.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 574 |
-
"transformer.visual.transformer.resblocks.30.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 575 |
-
"transformer.visual.transformer.resblocks.30.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 576 |
-
"transformer.visual.transformer.resblocks.30.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 577 |
-
"transformer.visual.transformer.resblocks.30.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 578 |
-
"transformer.visual.transformer.resblocks.30.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 579 |
-
"transformer.visual.transformer.resblocks.30.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 580 |
-
"transformer.visual.transformer.resblocks.30.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 581 |
-
"transformer.visual.transformer.resblocks.30.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 582 |
-
"transformer.visual.transformer.resblocks.31.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 583 |
-
"transformer.visual.transformer.resblocks.31.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 584 |
-
"transformer.visual.transformer.resblocks.31.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 585 |
-
"transformer.visual.transformer.resblocks.31.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 586 |
-
"transformer.visual.transformer.resblocks.31.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 587 |
-
"transformer.visual.transformer.resblocks.31.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 588 |
-
"transformer.visual.transformer.resblocks.31.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 589 |
-
"transformer.visual.transformer.resblocks.31.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 590 |
-
"transformer.visual.transformer.resblocks.31.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 591 |
-
"transformer.visual.transformer.resblocks.31.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 592 |
-
"transformer.visual.transformer.resblocks.31.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 593 |
-
"transformer.visual.transformer.resblocks.31.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 594 |
-
"transformer.visual.transformer.resblocks.32.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 595 |
-
"transformer.visual.transformer.resblocks.32.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 596 |
-
"transformer.visual.transformer.resblocks.32.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 597 |
-
"transformer.visual.transformer.resblocks.32.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 598 |
-
"transformer.visual.transformer.resblocks.32.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 599 |
-
"transformer.visual.transformer.resblocks.32.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 600 |
-
"transformer.visual.transformer.resblocks.32.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 601 |
-
"transformer.visual.transformer.resblocks.32.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 602 |
-
"transformer.visual.transformer.resblocks.32.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 603 |
-
"transformer.visual.transformer.resblocks.32.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 604 |
-
"transformer.visual.transformer.resblocks.32.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 605 |
-
"transformer.visual.transformer.resblocks.32.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 606 |
-
"transformer.visual.transformer.resblocks.33.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 607 |
-
"transformer.visual.transformer.resblocks.33.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 608 |
-
"transformer.visual.transformer.resblocks.33.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 609 |
-
"transformer.visual.transformer.resblocks.33.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 610 |
-
"transformer.visual.transformer.resblocks.33.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 611 |
-
"transformer.visual.transformer.resblocks.33.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 612 |
-
"transformer.visual.transformer.resblocks.33.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 613 |
-
"transformer.visual.transformer.resblocks.33.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 614 |
-
"transformer.visual.transformer.resblocks.33.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 615 |
-
"transformer.visual.transformer.resblocks.33.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 616 |
-
"transformer.visual.transformer.resblocks.33.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 617 |
-
"transformer.visual.transformer.resblocks.33.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 618 |
-
"transformer.visual.transformer.resblocks.34.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 619 |
-
"transformer.visual.transformer.resblocks.34.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 620 |
-
"transformer.visual.transformer.resblocks.34.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 621 |
-
"transformer.visual.transformer.resblocks.34.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 622 |
-
"transformer.visual.transformer.resblocks.34.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 623 |
-
"transformer.visual.transformer.resblocks.34.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 624 |
-
"transformer.visual.transformer.resblocks.34.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 625 |
-
"transformer.visual.transformer.resblocks.34.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 626 |
-
"transformer.visual.transformer.resblocks.34.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 627 |
-
"transformer.visual.transformer.resblocks.34.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 628 |
-
"transformer.visual.transformer.resblocks.34.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 629 |
-
"transformer.visual.transformer.resblocks.34.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 630 |
-
"transformer.visual.transformer.resblocks.35.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 631 |
-
"transformer.visual.transformer.resblocks.35.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 632 |
-
"transformer.visual.transformer.resblocks.35.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 633 |
-
"transformer.visual.transformer.resblocks.35.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 634 |
-
"transformer.visual.transformer.resblocks.35.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 635 |
-
"transformer.visual.transformer.resblocks.35.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 636 |
-
"transformer.visual.transformer.resblocks.35.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 637 |
-
"transformer.visual.transformer.resblocks.35.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 638 |
-
"transformer.visual.transformer.resblocks.35.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 639 |
-
"transformer.visual.transformer.resblocks.35.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 640 |
-
"transformer.visual.transformer.resblocks.35.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 641 |
-
"transformer.visual.transformer.resblocks.35.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 642 |
-
"transformer.visual.transformer.resblocks.36.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 643 |
-
"transformer.visual.transformer.resblocks.36.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 644 |
-
"transformer.visual.transformer.resblocks.36.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 645 |
-
"transformer.visual.transformer.resblocks.36.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 646 |
-
"transformer.visual.transformer.resblocks.36.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 647 |
-
"transformer.visual.transformer.resblocks.36.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 648 |
-
"transformer.visual.transformer.resblocks.36.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 649 |
-
"transformer.visual.transformer.resblocks.36.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 650 |
-
"transformer.visual.transformer.resblocks.36.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 651 |
-
"transformer.visual.transformer.resblocks.36.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 652 |
-
"transformer.visual.transformer.resblocks.36.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 653 |
-
"transformer.visual.transformer.resblocks.36.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 654 |
-
"transformer.visual.transformer.resblocks.37.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 655 |
-
"transformer.visual.transformer.resblocks.37.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 656 |
-
"transformer.visual.transformer.resblocks.37.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 657 |
-
"transformer.visual.transformer.resblocks.37.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 658 |
-
"transformer.visual.transformer.resblocks.37.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 659 |
-
"transformer.visual.transformer.resblocks.37.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 660 |
-
"transformer.visual.transformer.resblocks.37.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 661 |
-
"transformer.visual.transformer.resblocks.37.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 662 |
-
"transformer.visual.transformer.resblocks.37.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 663 |
-
"transformer.visual.transformer.resblocks.37.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 664 |
-
"transformer.visual.transformer.resblocks.37.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 665 |
-
"transformer.visual.transformer.resblocks.37.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 666 |
-
"transformer.visual.transformer.resblocks.38.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 667 |
-
"transformer.visual.transformer.resblocks.38.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 668 |
-
"transformer.visual.transformer.resblocks.38.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 669 |
-
"transformer.visual.transformer.resblocks.38.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 670 |
-
"transformer.visual.transformer.resblocks.38.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 671 |
-
"transformer.visual.transformer.resblocks.38.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 672 |
-
"transformer.visual.transformer.resblocks.38.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 673 |
-
"transformer.visual.transformer.resblocks.38.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 674 |
-
"transformer.visual.transformer.resblocks.38.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 675 |
-
"transformer.visual.transformer.resblocks.38.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 676 |
-
"transformer.visual.transformer.resblocks.38.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 677 |
-
"transformer.visual.transformer.resblocks.38.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 678 |
-
"transformer.visual.transformer.resblocks.39.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 679 |
-
"transformer.visual.transformer.resblocks.39.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 680 |
-
"transformer.visual.transformer.resblocks.39.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 681 |
-
"transformer.visual.transformer.resblocks.39.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 682 |
-
"transformer.visual.transformer.resblocks.39.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 683 |
-
"transformer.visual.transformer.resblocks.39.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 684 |
-
"transformer.visual.transformer.resblocks.39.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 685 |
-
"transformer.visual.transformer.resblocks.39.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 686 |
-
"transformer.visual.transformer.resblocks.39.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 687 |
-
"transformer.visual.transformer.resblocks.39.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 688 |
-
"transformer.visual.transformer.resblocks.39.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 689 |
-
"transformer.visual.transformer.resblocks.39.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 690 |
-
"transformer.visual.transformer.resblocks.4.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 691 |
-
"transformer.visual.transformer.resblocks.4.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 692 |
-
"transformer.visual.transformer.resblocks.4.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 693 |
-
"transformer.visual.transformer.resblocks.4.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 694 |
-
"transformer.visual.transformer.resblocks.4.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 695 |
-
"transformer.visual.transformer.resblocks.4.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 696 |
-
"transformer.visual.transformer.resblocks.4.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 697 |
-
"transformer.visual.transformer.resblocks.4.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 698 |
-
"transformer.visual.transformer.resblocks.4.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 699 |
-
"transformer.visual.transformer.resblocks.4.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 700 |
-
"transformer.visual.transformer.resblocks.4.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 701 |
-
"transformer.visual.transformer.resblocks.4.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 702 |
-
"transformer.visual.transformer.resblocks.40.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 703 |
-
"transformer.visual.transformer.resblocks.40.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 704 |
-
"transformer.visual.transformer.resblocks.40.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 705 |
-
"transformer.visual.transformer.resblocks.40.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 706 |
-
"transformer.visual.transformer.resblocks.40.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 707 |
-
"transformer.visual.transformer.resblocks.40.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 708 |
-
"transformer.visual.transformer.resblocks.40.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 709 |
-
"transformer.visual.transformer.resblocks.40.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 710 |
-
"transformer.visual.transformer.resblocks.40.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 711 |
-
"transformer.visual.transformer.resblocks.40.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 712 |
-
"transformer.visual.transformer.resblocks.40.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 713 |
-
"transformer.visual.transformer.resblocks.40.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 714 |
-
"transformer.visual.transformer.resblocks.41.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 715 |
-
"transformer.visual.transformer.resblocks.41.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 716 |
-
"transformer.visual.transformer.resblocks.41.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 717 |
-
"transformer.visual.transformer.resblocks.41.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 718 |
-
"transformer.visual.transformer.resblocks.41.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 719 |
-
"transformer.visual.transformer.resblocks.41.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 720 |
-
"transformer.visual.transformer.resblocks.41.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 721 |
-
"transformer.visual.transformer.resblocks.41.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 722 |
-
"transformer.visual.transformer.resblocks.41.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 723 |
-
"transformer.visual.transformer.resblocks.41.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 724 |
-
"transformer.visual.transformer.resblocks.41.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 725 |
-
"transformer.visual.transformer.resblocks.41.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 726 |
-
"transformer.visual.transformer.resblocks.42.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 727 |
-
"transformer.visual.transformer.resblocks.42.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 728 |
-
"transformer.visual.transformer.resblocks.42.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 729 |
-
"transformer.visual.transformer.resblocks.42.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 730 |
-
"transformer.visual.transformer.resblocks.42.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 731 |
-
"transformer.visual.transformer.resblocks.42.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 732 |
-
"transformer.visual.transformer.resblocks.42.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 733 |
-
"transformer.visual.transformer.resblocks.42.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 734 |
-
"transformer.visual.transformer.resblocks.42.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 735 |
-
"transformer.visual.transformer.resblocks.42.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 736 |
-
"transformer.visual.transformer.resblocks.42.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 737 |
-
"transformer.visual.transformer.resblocks.42.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 738 |
-
"transformer.visual.transformer.resblocks.43.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 739 |
-
"transformer.visual.transformer.resblocks.43.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 740 |
-
"transformer.visual.transformer.resblocks.43.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 741 |
-
"transformer.visual.transformer.resblocks.43.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 742 |
-
"transformer.visual.transformer.resblocks.43.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 743 |
-
"transformer.visual.transformer.resblocks.43.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 744 |
-
"transformer.visual.transformer.resblocks.43.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 745 |
-
"transformer.visual.transformer.resblocks.43.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 746 |
-
"transformer.visual.transformer.resblocks.43.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 747 |
-
"transformer.visual.transformer.resblocks.43.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 748 |
-
"transformer.visual.transformer.resblocks.43.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 749 |
-
"transformer.visual.transformer.resblocks.43.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 750 |
-
"transformer.visual.transformer.resblocks.44.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 751 |
-
"transformer.visual.transformer.resblocks.44.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 752 |
-
"transformer.visual.transformer.resblocks.44.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 753 |
-
"transformer.visual.transformer.resblocks.44.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 754 |
-
"transformer.visual.transformer.resblocks.44.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 755 |
-
"transformer.visual.transformer.resblocks.44.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 756 |
-
"transformer.visual.transformer.resblocks.44.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 757 |
-
"transformer.visual.transformer.resblocks.44.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 758 |
-
"transformer.visual.transformer.resblocks.44.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 759 |
-
"transformer.visual.transformer.resblocks.44.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 760 |
-
"transformer.visual.transformer.resblocks.44.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 761 |
-
"transformer.visual.transformer.resblocks.44.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 762 |
-
"transformer.visual.transformer.resblocks.45.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 763 |
-
"transformer.visual.transformer.resblocks.45.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 764 |
-
"transformer.visual.transformer.resblocks.45.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 765 |
-
"transformer.visual.transformer.resblocks.45.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 766 |
-
"transformer.visual.transformer.resblocks.45.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 767 |
-
"transformer.visual.transformer.resblocks.45.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 768 |
-
"transformer.visual.transformer.resblocks.45.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 769 |
-
"transformer.visual.transformer.resblocks.45.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 770 |
-
"transformer.visual.transformer.resblocks.45.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 771 |
-
"transformer.visual.transformer.resblocks.45.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 772 |
-
"transformer.visual.transformer.resblocks.45.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 773 |
-
"transformer.visual.transformer.resblocks.45.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 774 |
-
"transformer.visual.transformer.resblocks.46.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 775 |
-
"transformer.visual.transformer.resblocks.46.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 776 |
-
"transformer.visual.transformer.resblocks.46.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 777 |
-
"transformer.visual.transformer.resblocks.46.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 778 |
-
"transformer.visual.transformer.resblocks.46.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 779 |
-
"transformer.visual.transformer.resblocks.46.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 780 |
-
"transformer.visual.transformer.resblocks.46.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 781 |
-
"transformer.visual.transformer.resblocks.46.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 782 |
-
"transformer.visual.transformer.resblocks.46.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 783 |
-
"transformer.visual.transformer.resblocks.46.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 784 |
-
"transformer.visual.transformer.resblocks.46.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 785 |
-
"transformer.visual.transformer.resblocks.46.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 786 |
-
"transformer.visual.transformer.resblocks.47.attn.in_proj.bias": "model-00004-of-00004.safetensors",
|
| 787 |
-
"transformer.visual.transformer.resblocks.47.attn.in_proj.weight": "model-00004-of-00004.safetensors",
|
| 788 |
-
"transformer.visual.transformer.resblocks.47.attn.out_proj.bias": "model-00004-of-00004.safetensors",
|
| 789 |
-
"transformer.visual.transformer.resblocks.47.attn.out_proj.weight": "model-00004-of-00004.safetensors",
|
| 790 |
-
"transformer.visual.transformer.resblocks.47.ln_1.bias": "model-00004-of-00004.safetensors",
|
| 791 |
-
"transformer.visual.transformer.resblocks.47.ln_1.weight": "model-00004-of-00004.safetensors",
|
| 792 |
-
"transformer.visual.transformer.resblocks.47.ln_2.bias": "model-00004-of-00004.safetensors",
|
| 793 |
-
"transformer.visual.transformer.resblocks.47.ln_2.weight": "model-00004-of-00004.safetensors",
|
| 794 |
-
"transformer.visual.transformer.resblocks.47.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 795 |
-
"transformer.visual.transformer.resblocks.47.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 796 |
-
"transformer.visual.transformer.resblocks.47.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 797 |
-
"transformer.visual.transformer.resblocks.47.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 798 |
-
"transformer.visual.transformer.resblocks.5.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 799 |
-
"transformer.visual.transformer.resblocks.5.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 800 |
-
"transformer.visual.transformer.resblocks.5.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 801 |
-
"transformer.visual.transformer.resblocks.5.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 802 |
-
"transformer.visual.transformer.resblocks.5.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 803 |
-
"transformer.visual.transformer.resblocks.5.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 804 |
-
"transformer.visual.transformer.resblocks.5.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 805 |
-
"transformer.visual.transformer.resblocks.5.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 806 |
-
"transformer.visual.transformer.resblocks.5.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 807 |
-
"transformer.visual.transformer.resblocks.5.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 808 |
-
"transformer.visual.transformer.resblocks.5.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 809 |
-
"transformer.visual.transformer.resblocks.5.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 810 |
-
"transformer.visual.transformer.resblocks.6.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 811 |
-
"transformer.visual.transformer.resblocks.6.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 812 |
-
"transformer.visual.transformer.resblocks.6.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 813 |
-
"transformer.visual.transformer.resblocks.6.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 814 |
-
"transformer.visual.transformer.resblocks.6.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 815 |
-
"transformer.visual.transformer.resblocks.6.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 816 |
-
"transformer.visual.transformer.resblocks.6.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 817 |
-
"transformer.visual.transformer.resblocks.6.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 818 |
-
"transformer.visual.transformer.resblocks.6.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 819 |
-
"transformer.visual.transformer.resblocks.6.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 820 |
-
"transformer.visual.transformer.resblocks.6.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 821 |
-
"transformer.visual.transformer.resblocks.6.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 822 |
-
"transformer.visual.transformer.resblocks.7.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 823 |
-
"transformer.visual.transformer.resblocks.7.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 824 |
-
"transformer.visual.transformer.resblocks.7.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 825 |
-
"transformer.visual.transformer.resblocks.7.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 826 |
-
"transformer.visual.transformer.resblocks.7.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 827 |
-
"transformer.visual.transformer.resblocks.7.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 828 |
-
"transformer.visual.transformer.resblocks.7.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 829 |
-
"transformer.visual.transformer.resblocks.7.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 830 |
-
"transformer.visual.transformer.resblocks.7.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 831 |
-
"transformer.visual.transformer.resblocks.7.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 832 |
-
"transformer.visual.transformer.resblocks.7.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 833 |
-
"transformer.visual.transformer.resblocks.7.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 834 |
-
"transformer.visual.transformer.resblocks.8.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 835 |
-
"transformer.visual.transformer.resblocks.8.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 836 |
-
"transformer.visual.transformer.resblocks.8.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 837 |
-
"transformer.visual.transformer.resblocks.8.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 838 |
-
"transformer.visual.transformer.resblocks.8.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 839 |
-
"transformer.visual.transformer.resblocks.8.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 840 |
-
"transformer.visual.transformer.resblocks.8.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 841 |
-
"transformer.visual.transformer.resblocks.8.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 842 |
-
"transformer.visual.transformer.resblocks.8.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
|
| 843 |
-
"transformer.visual.transformer.resblocks.8.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
|
| 844 |
-
"transformer.visual.transformer.resblocks.8.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
|
| 845 |
-
"transformer.visual.transformer.resblocks.8.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
|
| 846 |
-
"transformer.visual.transformer.resblocks.9.attn.in_proj.bias": "model-00003-of-00004.safetensors",
|
| 847 |
-
"transformer.visual.transformer.resblocks.9.attn.in_proj.weight": "model-00003-of-00004.safetensors",
|
| 848 |
-
"transformer.visual.transformer.resblocks.9.attn.out_proj.bias": "model-00003-of-00004.safetensors",
|
| 849 |
-
"transformer.visual.transformer.resblocks.9.attn.out_proj.weight": "model-00003-of-00004.safetensors",
|
| 850 |
-
"transformer.visual.transformer.resblocks.9.ln_1.bias": "model-00003-of-00004.safetensors",
|
| 851 |
-
"transformer.visual.transformer.resblocks.9.ln_1.weight": "model-00003-of-00004.safetensors",
|
| 852 |
-
"transformer.visual.transformer.resblocks.9.ln_2.bias": "model-00003-of-00004.safetensors",
|
| 853 |
-
"transformer.visual.transformer.resblocks.9.ln_2.weight": "model-00003-of-00004.safetensors",
|
| 854 |
-
"transformer.visual.transformer.resblocks.9.mlp.c_fc.bias": "model-00004-of-00004.safetensors",
|
| 855 |
-
"transformer.visual.transformer.resblocks.9.mlp.c_fc.weight": "model-00004-of-00004.safetensors",
|
| 856 |
-
"transformer.visual.transformer.resblocks.9.mlp.c_proj.bias": "model-00004-of-00004.safetensors",
|
| 857 |
-
"transformer.visual.transformer.resblocks.9.mlp.c_proj.weight": "model-00004-of-00004.safetensors",
|
| 858 |
-
"transformer.wte.weight": "model-00001-of-00004.safetensors"
|
| 859 |
-
}
|
| 860 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/modeling_qwen.py
DELETED
|
@@ -1,1172 +0,0 @@
|
|
| 1 |
-
# Copyright (c) Alibaba Cloud.
|
| 2 |
-
#
|
| 3 |
-
# This source code is licensed under the license found in the
|
| 4 |
-
# LICENSE file in the root directory of this source tree.
|
| 5 |
-
|
| 6 |
-
import importlib
|
| 7 |
-
import math
|
| 8 |
-
from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List, Any, Generator
|
| 9 |
-
|
| 10 |
-
import torch
|
| 11 |
-
import torch.nn.functional as F
|
| 12 |
-
import torch.utils.checkpoint
|
| 13 |
-
from torch.cuda.amp import autocast
|
| 14 |
-
|
| 15 |
-
from torch.nn import CrossEntropyLoss
|
| 16 |
-
from transformers import PreTrainedTokenizer, GenerationConfig, StoppingCriteriaList
|
| 17 |
-
from transformers.generation.logits_process import LogitsProcessorList
|
| 18 |
-
|
| 19 |
-
if TYPE_CHECKING:
|
| 20 |
-
from transformers.generation.streamers import BaseStreamer
|
| 21 |
-
from transformers.generation.utils import GenerateOutput
|
| 22 |
-
from transformers.modeling_outputs import (
|
| 23 |
-
BaseModelOutputWithPast,
|
| 24 |
-
CausalLMOutputWithPast,
|
| 25 |
-
)
|
| 26 |
-
from transformers.modeling_utils import PreTrainedModel
|
| 27 |
-
from transformers.utils import logging
|
| 28 |
-
|
| 29 |
-
try:
|
| 30 |
-
from einops import rearrange
|
| 31 |
-
except ImportError:
|
| 32 |
-
rearrange = None
|
| 33 |
-
from torch import nn
|
| 34 |
-
|
| 35 |
-
SUPPORT_CUDA = torch.cuda.is_available()
|
| 36 |
-
SUPPORT_BF16 = SUPPORT_CUDA and torch.cuda.is_bf16_supported()
|
| 37 |
-
SUPPORT_FP16 = SUPPORT_CUDA and torch.cuda.get_device_capability(0)[0] >= 7
|
| 38 |
-
|
| 39 |
-
from .configuration_qwen import QWenConfig
|
| 40 |
-
from .qwen_generation_utils import (
|
| 41 |
-
HistoryType,
|
| 42 |
-
make_context,
|
| 43 |
-
decode_tokens,
|
| 44 |
-
get_stop_words_ids,
|
| 45 |
-
StopWordsLogitsProcessor,
|
| 46 |
-
)
|
| 47 |
-
from .visual import VisionTransformer
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
logger = logging.get_logger(__name__)
|
| 51 |
-
|
| 52 |
-
_CHECKPOINT_FOR_DOC = "qwen"
|
| 53 |
-
_CONFIG_FOR_DOC = "QWenConfig"
|
| 54 |
-
|
| 55 |
-
QWen_PRETRAINED_MODEL_ARCHIVE_LIST = ["qwen-7b"]
|
| 56 |
-
|
| 57 |
-
_ERROR_BAD_CHAT_FORMAT = """\
|
| 58 |
-
We detect you are probably using the pretrained model (rather than chat model) for chatting, since the chat_format in generation_config is not "chatml".
|
| 59 |
-
If you are directly using the model downloaded from Huggingface, please make sure you are using our "Qwen/Qwen-7B-Chat" Huggingface model (rather than "Qwen/Qwen-7B") when you call model.chat().
|
| 60 |
-
我们检测到您可能在使用预训练模型（而非chat模型）进行多轮chat，因为您当前在generation_config指定的chat_format，并未设置为我们在对话中所支持的"chatml"格式。
|
| 61 |
-
如果您在直接使用我们从Huggingface提供的模型，请确保您在调用model.chat()时，使用的是"Qwen/Qwen-7B-Chat"模型（而非"Qwen/Qwen-7B"预训练模型）。
|
| 62 |
-
"""
|
| 63 |
-
|
| 64 |
-
_SENTINEL = object()
|
| 65 |
-
_ERROR_STREAM_IN_CHAT = """\
|
| 66 |
-
Pass argument `stream` to model.chat() is buggy, deprecated, and marked for removal. Please use model.chat_stream(...) instead of model.chat(..., stream=True).
|
| 67 |
-
向model.chat()传入参数stream的用法可能存在Bug，该用法已被废弃，将在未来被移除。请使用model.chat_stream(...)代替model.chat(..., stream=True)。
|
| 68 |
-
"""
|
| 69 |
-
|
| 70 |
-
apply_rotary_emb_func = None
|
| 71 |
-
rms_norm = None
|
| 72 |
-
def int_list_to_str(lst: List[int]) -> str:
|
| 73 |
-
"""将整数列表转换回字符串"""
|
| 74 |
-
return ''.join(chr(i) for i in lst)
|
| 75 |
-
|
| 76 |
-
# Copied from transformers.models.bart.modeling_bart._make_causal_mask
|
| 77 |
-
def _make_causal_mask(
|
| 78 |
-
input_ids_shape: torch.Size, dtype: torch.dtype, device: torch.device, past_key_values_length: int = 0
|
| 79 |
-
):
|
| 80 |
-
"""
|
| 81 |
-
Make causal mask used for bi-directional self-attention.
|
| 82 |
-
"""
|
| 83 |
-
bsz, tgt_len = input_ids_shape
|
| 84 |
-
mask = torch.full((tgt_len, tgt_len), torch.finfo(dtype).min, device=device)
|
| 85 |
-
mask_cond = torch.arange(mask.size(-1), device=device)
|
| 86 |
-
mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
|
| 87 |
-
mask = mask.to(dtype)
|
| 88 |
-
|
| 89 |
-
if past_key_values_length > 0:
|
| 90 |
-
mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
|
| 91 |
-
return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
# Copied from transformers.models.bart.modeling_bart._expand_mask
|
| 95 |
-
def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
|
| 96 |
-
"""
|
| 97 |
-
Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
|
| 98 |
-
"""
|
| 99 |
-
bsz, src_len = mask.size()
|
| 100 |
-
tgt_len = tgt_len if tgt_len is not None else src_len
|
| 101 |
-
|
| 102 |
-
expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
|
| 103 |
-
|
| 104 |
-
inverted_mask = 1.0 - expanded_mask
|
| 105 |
-
|
| 106 |
-
return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
class QWenAttention(nn.Module):
|
| 110 |
-
def __init__(self, config):
|
| 111 |
-
super().__init__()
|
| 112 |
-
|
| 113 |
-
self.register_buffer("masked_bias", torch.tensor(-1e4), persistent=False)
|
| 114 |
-
self.seq_length = config.seq_length
|
| 115 |
-
|
| 116 |
-
self.hidden_size = config.hidden_size
|
| 117 |
-
self.split_size = config.hidden_size
|
| 118 |
-
self.num_heads = config.num_attention_heads
|
| 119 |
-
self.head_dim = self.hidden_size // self.num_heads
|
| 120 |
-
|
| 121 |
-
self.scale_attn_weights = True
|
| 122 |
-
|
| 123 |
-
self.projection_size = config.kv_channels * config.num_attention_heads
|
| 124 |
-
|
| 125 |
-
assert self.projection_size % config.num_attention_heads == 0
|
| 126 |
-
self.hidden_size_per_attention_head = (
|
| 127 |
-
self.projection_size // config.num_attention_heads
|
| 128 |
-
)
|
| 129 |
-
|
| 130 |
-
self.c_attn = nn.Linear(config.hidden_size, 3 * self.projection_size)
|
| 131 |
-
|
| 132 |
-
self.c_proj = nn.Linear(
|
| 133 |
-
config.hidden_size, self.projection_size, bias=not config.no_bias
|
| 134 |
-
)
|
| 135 |
-
|
| 136 |
-
self.is_fp32 = not (config.bf16 or config.fp16)
|
| 137 |
-
self.bf16 = config.bf16
|
| 138 |
-
|
| 139 |
-
self.use_dynamic_ntk = config.use_dynamic_ntk
|
| 140 |
-
self.use_logn_attn = config.use_logn_attn
|
| 141 |
-
|
| 142 |
-
logn_list = [
|
| 143 |
-
math.log(i, self.seq_length) if i > self.seq_length else 1
|
| 144 |
-
for i in range(1, 32768)
|
| 145 |
-
]
|
| 146 |
-
self.logn_tensor = torch.tensor(logn_list)[None, :, None, None]
|
| 147 |
-
|
| 148 |
-
self.attn_dropout = nn.Dropout(config.attn_dropout_prob)
|
| 149 |
-
|
| 150 |
-
def _attn(self, query, key, value, registered_causal_mask, attention_mask=None, head_mask=None):
|
| 151 |
-
attn_weights = torch.matmul(query, key.transpose(-1, -2))
|
| 152 |
-
|
| 153 |
-
if self.scale_attn_weights:
|
| 154 |
-
attn_weights = attn_weights / torch.full(
|
| 155 |
-
[],
|
| 156 |
-
value.size(-1) ** 0.5,
|
| 157 |
-
dtype=attn_weights.dtype,
|
| 158 |
-
device=attn_weights.device,
|
| 159 |
-
)
|
| 160 |
-
|
| 161 |
-
query_length, key_length = query.size(-2), key.size(-2)
|
| 162 |
-
# causal_mask = self.bias[
|
| 163 |
-
# :, :, key_length - query_length : key_length, :key_length
|
| 164 |
-
# ]
|
| 165 |
-
# mask_value = torch.finfo(attn_weights.dtype).min
|
| 166 |
-
# mask_value = torch.full([], mask_value, dtype=attn_weights.dtype).to(
|
| 167 |
-
# attn_weights.device
|
| 168 |
-
# )
|
| 169 |
-
# attn_weights = torch.where(
|
| 170 |
-
# causal_mask, attn_weights.to(attn_weights.dtype), mask_value
|
| 171 |
-
# )
|
| 172 |
-
attn_weights = attn_weights + attention_mask
|
| 173 |
-
|
| 174 |
-
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
|
| 175 |
-
|
| 176 |
-
attn_weights = attn_weights.type(value.dtype)
|
| 177 |
-
attn_weights = self.attn_dropout(attn_weights)
|
| 178 |
-
|
| 179 |
-
if head_mask is not None:
|
| 180 |
-
attn_weights = attn_weights * head_mask
|
| 181 |
-
|
| 182 |
-
attn_output = torch.matmul(attn_weights, value)
|
| 183 |
-
attn_output = attn_output.transpose(1, 2)
|
| 184 |
-
|
| 185 |
-
return attn_output, attn_weights
|
| 186 |
-
|
| 187 |
-
def _upcast_and_reordered_attn(
|
| 188 |
-
self, query, key, value, registered_causal_mask, attention_mask=None, head_mask=None
|
| 189 |
-
):
|
| 190 |
-
bsz, num_heads, q_seq_len, dk = query.size()
|
| 191 |
-
_, _, k_seq_len, _ = key.size()
|
| 192 |
-
|
| 193 |
-
attn_weights = torch.empty(
|
| 194 |
-
bsz * num_heads,
|
| 195 |
-
q_seq_len,
|
| 196 |
-
k_seq_len,
|
| 197 |
-
dtype=torch.float32,
|
| 198 |
-
device=query.device,
|
| 199 |
-
)
|
| 200 |
-
|
| 201 |
-
scale_factor = 1.0
|
| 202 |
-
if self.scale_attn_weights:
|
| 203 |
-
scale_factor /= float(value.size(-1)) ** 0.5
|
| 204 |
-
|
| 205 |
-
with autocast(enabled=False):
|
| 206 |
-
q, k = query.reshape(-1, q_seq_len, dk), key.transpose(-1, -2).reshape(
|
| 207 |
-
-1, dk, k_seq_len
|
| 208 |
-
)
|
| 209 |
-
attn_weights = torch.baddbmm(
|
| 210 |
-
attn_weights, q.float(), k.float(), beta=0, alpha=scale_factor
|
| 211 |
-
)
|
| 212 |
-
attn_weights = attn_weights.reshape(bsz, num_heads, q_seq_len, k_seq_len)
|
| 213 |
-
|
| 214 |
-
query_length, key_length = query.size(-2), key.size(-2)
|
| 215 |
-
causal_mask = registered_causal_mask[
|
| 216 |
-
:, :, key_length - query_length : key_length, :key_length
|
| 217 |
-
]
|
| 218 |
-
mask_value = torch.finfo(attn_weights.dtype).min
|
| 219 |
-
mask_value = torch.tensor(mask_value, dtype=attn_weights.dtype).to(
|
| 220 |
-
attn_weights.device
|
| 221 |
-
)
|
| 222 |
-
attn_weights = torch.where(causal_mask, attn_weights, mask_value)
|
| 223 |
-
|
| 224 |
-
if attention_mask is not None:
|
| 225 |
-
attn_weights = attn_weights + attention_mask
|
| 226 |
-
|
| 227 |
-
attn_weights = nn.functional.softmax(attn_weights, dim=-1)
|
| 228 |
-
|
| 229 |
-
if attn_weights.dtype != torch.float32:
|
| 230 |
-
raise RuntimeError(
|
| 231 |
-
"Error with upcasting, attn_weights does not have dtype torch.float32"
|
| 232 |
-
)
|
| 233 |
-
attn_weights = attn_weights.type(value.dtype)
|
| 234 |
-
attn_weights = self.attn_dropout(attn_weights)
|
| 235 |
-
|
| 236 |
-
if head_mask is not None:
|
| 237 |
-
attn_weights = attn_weights * head_mask
|
| 238 |
-
|
| 239 |
-
attn_output = torch.matmul(attn_weights, value)
|
| 240 |
-
|
| 241 |
-
return attn_output, attn_weights
|
| 242 |
-
|
| 243 |
-
def _split_heads(self, tensor, num_heads, attn_head_size):
|
| 244 |
-
new_shape = tensor.size()[:-1] + (num_heads, attn_head_size)
|
| 245 |
-
tensor = tensor.view(new_shape)
|
| 246 |
-
return tensor
|
| 247 |
-
|
| 248 |
-
def _merge_heads(self, tensor, num_heads, attn_head_size):
|
| 249 |
-
tensor = tensor.contiguous()
|
| 250 |
-
new_shape = tensor.size()[:-2] + (num_heads * attn_head_size,)
|
| 251 |
-
return tensor.view(new_shape)
|
| 252 |
-
|
| 253 |
-
def forward(
|
| 254 |
-
self,
|
| 255 |
-
hidden_states: Optional[Tuple[torch.FloatTensor]],
|
| 256 |
-
rotary_pos_emb: Optional[List[torch.Tensor]] = None,
|
| 257 |
-
registered_causal_mask: Optional[torch.Tensor] = None,
|
| 258 |
-
layer_past: Optional[Tuple[torch.Tensor]] = None,
|
| 259 |
-
attention_mask: Optional[torch.FloatTensor] = None,
|
| 260 |
-
head_mask: Optional[torch.FloatTensor] = None,
|
| 261 |
-
encoder_hidden_states: Optional[torch.Tensor] = None,
|
| 262 |
-
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
| 263 |
-
output_attentions: Optional[bool] = False,
|
| 264 |
-
use_cache: Optional[bool] = False,
|
| 265 |
-
):
|
| 266 |
-
|
| 267 |
-
mixed_x_layer = self.c_attn(hidden_states)
|
| 268 |
-
|
| 269 |
-
query, key, value = mixed_x_layer.split(self.split_size, dim=2)
|
| 270 |
-
|
| 271 |
-
query = self._split_heads(query, self.num_heads, self.head_dim)
|
| 272 |
-
key = self._split_heads(key, self.num_heads, self.head_dim)
|
| 273 |
-
value = self._split_heads(value, self.num_heads, self.head_dim)
|
| 274 |
-
|
| 275 |
-
if rotary_pos_emb is not None:
|
| 276 |
-
cur_len = query.shape[1]
|
| 277 |
-
rotary_pos_emb = [i[:, -cur_len:, :, :] for i in rotary_pos_emb]
|
| 278 |
-
rotary_pos_emb = (rotary_pos_emb,) * 2
|
| 279 |
-
q_pos_emb, k_pos_emb = rotary_pos_emb
|
| 280 |
-
# Slice the pos emb for current inference
|
| 281 |
-
query = apply_rotary_pos_emb(query, q_pos_emb)
|
| 282 |
-
key = apply_rotary_pos_emb(key, k_pos_emb)
|
| 283 |
-
|
| 284 |
-
if layer_past is not None:
|
| 285 |
-
past_key, past_value = layer_past[0], layer_past[1]
|
| 286 |
-
key = torch.cat((past_key, key), dim=1)
|
| 287 |
-
value = torch.cat((past_value, value), dim=1)
|
| 288 |
-
|
| 289 |
-
if use_cache:
|
| 290 |
-
present = (key, value)
|
| 291 |
-
else:
|
| 292 |
-
present = None
|
| 293 |
-
|
| 294 |
-
if self.use_logn_attn and not self.training:
|
| 295 |
-
if self.logn_tensor.device != query.device or self.logn_tensor.dtype != query.dtype:
|
| 296 |
-
self.logn_tensor = self.logn_tensor.to(query.device).type_as(query)
|
| 297 |
-
seq_start = key.size(1) - query.size(1)
|
| 298 |
-
seq_end = key.size(1)
|
| 299 |
-
logn_tensor = self.logn_tensor[:, seq_start:seq_end, :, :]
|
| 300 |
-
query = query * logn_tensor.expand_as(query)
|
| 301 |
-
|
| 302 |
-
query = query.permute(0, 2, 1, 3)
|
| 303 |
-
key = key.permute(0, 2, 1, 3)
|
| 304 |
-
value = value.permute(0, 2, 1, 3)
|
| 305 |
-
attn_output, attn_weight = self._attn(
|
| 306 |
-
query, key, value, registered_causal_mask, attention_mask, head_mask
|
| 307 |
-
)
|
| 308 |
-
context_layer = self._merge_heads(
|
| 309 |
-
attn_output, self.num_heads, self.head_dim
|
| 310 |
-
)
|
| 311 |
-
|
| 312 |
-
attn_output = self.c_proj(context_layer)
|
| 313 |
-
|
| 314 |
-
outputs = (attn_output, present)
|
| 315 |
-
if output_attentions:
|
| 316 |
-
outputs += (attn_weight,)
|
| 317 |
-
|
| 318 |
-
return outputs
|
| 319 |
-
|
| 320 |
-
|
| 321 |
-
class QWenMLP(nn.Module):
|
| 322 |
-
def __init__(self, config):
|
| 323 |
-
super().__init__()
|
| 324 |
-
self.w1 = nn.Linear(
|
| 325 |
-
config.hidden_size, config.intermediate_size // 2, bias=not config.no_bias
|
| 326 |
-
)
|
| 327 |
-
self.w2 = nn.Linear(
|
| 328 |
-
config.hidden_size, config.intermediate_size // 2, bias=not config.no_bias
|
| 329 |
-
)
|
| 330 |
-
ff_dim_in = config.intermediate_size // 2
|
| 331 |
-
self.c_proj = nn.Linear(ff_dim_in, config.hidden_size, bias=not config.no_bias)
|
| 332 |
-
|
| 333 |
-
def forward(self, hidden_states):
|
| 334 |
-
a1 = self.w1(hidden_states)
|
| 335 |
-
a2 = self.w2(hidden_states)
|
| 336 |
-
intermediate_parallel = a1 * F.silu(a2)
|
| 337 |
-
output = self.c_proj(intermediate_parallel)
|
| 338 |
-
return output
|
| 339 |
-
|
| 340 |
-
class QWenBlock(nn.Module):
|
| 341 |
-
def __init__(self, config):
|
| 342 |
-
super().__init__()
|
| 343 |
-
hidden_size = config.hidden_size
|
| 344 |
-
self.bf16 = config.bf16
|
| 345 |
-
|
| 346 |
-
self.ln_1 = RMSNorm(
|
| 347 |
-
hidden_size,
|
| 348 |
-
eps=config.layer_norm_epsilon,
|
| 349 |
-
)
|
| 350 |
-
self.attn = QWenAttention(config)
|
| 351 |
-
self.ln_2 = RMSNorm(
|
| 352 |
-
hidden_size,
|
| 353 |
-
eps=config.layer_norm_epsilon,
|
| 354 |
-
)
|
| 355 |
-
|
| 356 |
-
self.mlp = QWenMLP(config)
|
| 357 |
-
|
| 358 |
-
def forward(
|
| 359 |
-
self,
|
| 360 |
-
hidden_states: Optional[Tuple[torch.FloatTensor]],
|
| 361 |
-
rotary_pos_emb: Optional[List[torch.Tensor]] = None,
|
| 362 |
-
registered_causal_mask: Optional[torch.Tensor] = None,
|
| 363 |
-
layer_past: Optional[Tuple[torch.Tensor]] = None,
|
| 364 |
-
attention_mask: Optional[torch.FloatTensor] = None,
|
| 365 |
-
head_mask: Optional[torch.FloatTensor] = None,
|
| 366 |
-
encoder_hidden_states: Optional[torch.Tensor] = None,
|
| 367 |
-
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
| 368 |
-
use_cache: Optional[bool] = False,
|
| 369 |
-
output_attentions: Optional[bool] = False,
|
| 370 |
-
):
|
| 371 |
-
layernorm_output = self.ln_1(hidden_states)
|
| 372 |
-
|
| 373 |
-
attn_outputs = self.attn(
|
| 374 |
-
layernorm_output,
|
| 375 |
-
rotary_pos_emb,
|
| 376 |
-
registered_causal_mask=registered_causal_mask,
|
| 377 |
-
layer_past=layer_past,
|
| 378 |
-
attention_mask=attention_mask,
|
| 379 |
-
head_mask=head_mask,
|
| 380 |
-
use_cache=use_cache,
|
| 381 |
-
output_attentions=output_attentions,
|
| 382 |
-
)
|
| 383 |
-
attn_output = attn_outputs[0]
|
| 384 |
-
|
| 385 |
-
outputs = attn_outputs[1:]
|
| 386 |
-
|
| 387 |
-
residual = hidden_states
|
| 388 |
-
layernorm_input = attn_output + residual
|
| 389 |
-
|
| 390 |
-
layernorm_output = self.ln_2(layernorm_input)
|
| 391 |
-
|
| 392 |
-
residual = layernorm_input
|
| 393 |
-
mlp_output = self.mlp(layernorm_output)
|
| 394 |
-
hidden_states = residual + mlp_output
|
| 395 |
-
|
| 396 |
-
if use_cache:
|
| 397 |
-
outputs = (hidden_states,) + outputs
|
| 398 |
-
else:
|
| 399 |
-
outputs = (hidden_states,) + outputs[1:]
|
| 400 |
-
|
| 401 |
-
return outputs
|
| 402 |
-
|
| 403 |
-
|
| 404 |
-
class QWenPreTrainedModel(PreTrainedModel):
|
| 405 |
-
config_class = QWenConfig
|
| 406 |
-
base_model_prefix = "transformer"
|
| 407 |
-
is_parallelizable = False
|
| 408 |
-
supports_gradient_checkpointing = True
|
| 409 |
-
_no_split_modules = ["QWenBlock"]
|
| 410 |
-
|
| 411 |
-
def __init__(self, *inputs, **kwargs):
|
| 412 |
-
super().__init__(*inputs, **kwargs)
|
| 413 |
-
|
| 414 |
-
def _init_weights(self, module):
|
| 415 |
-
"""Initialize the weights."""
|
| 416 |
-
if isinstance(module, nn.Linear):
|
| 417 |
-
module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
|
| 418 |
-
if module.bias is not None:
|
| 419 |
-
module.bias.data.zero_()
|
| 420 |
-
elif isinstance(module, nn.Embedding):
|
| 421 |
-
module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
|
| 422 |
-
if module.padding_idx is not None:
|
| 423 |
-
module.weight.data[module.padding_idx].zero_()
|
| 424 |
-
elif isinstance(module, RMSNorm):
|
| 425 |
-
module.weight.data.fill_(1.0)
|
| 426 |
-
|
| 427 |
-
for name, p in module.named_parameters():
|
| 428 |
-
if name == "c_proj.weight":
|
| 429 |
-
p.data.normal_(
|
| 430 |
-
mean=0.0,
|
| 431 |
-
std=(
|
| 432 |
-
self.config.initializer_range
|
| 433 |
-
/ math.sqrt(2 * self.config.num_hidden_layers)
|
| 434 |
-
),
|
| 435 |
-
)
|
| 436 |
-
|
| 437 |
-
def _set_gradient_checkpointing(self, module, value=False):
|
| 438 |
-
if isinstance(module, QWenModel):
|
| 439 |
-
module.gradient_checkpointing = value
|
| 440 |
-
|
| 441 |
-
|
| 442 |
-
class QWenModel(QWenPreTrainedModel):
|
| 443 |
-
_keys_to_ignore_on_load_missing = ["attn.masked_bias"]
|
| 444 |
-
|
| 445 |
-
def __init__(self, config):
|
| 446 |
-
super().__init__(config)
|
| 447 |
-
self.vocab_size = config.vocab_size
|
| 448 |
-
self.num_hidden_layers = config.num_hidden_layers
|
| 449 |
-
self.embed_dim = config.hidden_size
|
| 450 |
-
|
| 451 |
-
self.gradient_checkpointing = False
|
| 452 |
-
self.use_dynamic_ntk = config.use_dynamic_ntk
|
| 453 |
-
self.seq_length = config.seq_length
|
| 454 |
-
|
| 455 |
-
self.wte = nn.Embedding(self.vocab_size, self.embed_dim)
|
| 456 |
-
|
| 457 |
-
self.drop = nn.Dropout(config.emb_dropout_prob)
|
| 458 |
-
|
| 459 |
-
if config.rotary_pct == 1.0:
|
| 460 |
-
self.rotary_ndims = None
|
| 461 |
-
else:
|
| 462 |
-
assert config.rotary_pct < 1
|
| 463 |
-
self.rotary_ndims = int(
|
| 464 |
-
config.kv_channels * config.rotary_pct
|
| 465 |
-
)
|
| 466 |
-
dim = (
|
| 467 |
-
self.rotary_ndims
|
| 468 |
-
if self.rotary_ndims is not None
|
| 469 |
-
else config.kv_channels
|
| 470 |
-
)
|
| 471 |
-
self.rotary_emb = RotaryEmbedding(dim, base=config.rotary_emb_base)
|
| 472 |
-
|
| 473 |
-
self.use_flash_attn = config.use_flash_attn
|
| 474 |
-
self.is_fp32 = not (config.bf16 or config.fp16)
|
| 475 |
-
self.registered_causal_mask = None
|
| 476 |
-
# if (
|
| 477 |
-
# self.use_flash_attn
|
| 478 |
-
# and flash_attn_unpadded_func is not None
|
| 479 |
-
# and not self.is_fp32
|
| 480 |
-
# ):
|
| 481 |
-
# self.registered_causal_mask = None
|
| 482 |
-
# else:
|
| 483 |
-
# max_positions = config.max_position_embeddings
|
| 484 |
-
# self.register_buffer(
|
| 485 |
-
# "registered_causal_mask",
|
| 486 |
-
# torch.tril(
|
| 487 |
-
# torch.ones((max_positions, max_positions), dtype=torch.bool)
|
| 488 |
-
# ).view(1, 1, max_positions, max_positions),
|
| 489 |
-
# persistent=False,
|
| 490 |
-
# )
|
| 491 |
-
|
| 492 |
-
self.h = nn.ModuleList(
|
| 493 |
-
[
|
| 494 |
-
QWenBlock(
|
| 495 |
-
config
|
| 496 |
-
)
|
| 497 |
-
for i in range(config.num_hidden_layers)
|
| 498 |
-
]
|
| 499 |
-
)
|
| 500 |
-
self.ln_f = RMSNorm(
|
| 501 |
-
self.embed_dim,
|
| 502 |
-
eps=config.layer_norm_epsilon,
|
| 503 |
-
)
|
| 504 |
-
|
| 505 |
-
self.visual = VisionTransformer(**config.visual)
|
| 506 |
-
|
| 507 |
-
self.post_init()
|
| 508 |
-
|
| 509 |
-
def get_input_embeddings(self):
|
| 510 |
-
return self.wte
|
| 511 |
-
|
| 512 |
-
def set_input_embeddings(self, new_embeddings):
|
| 513 |
-
self.wte = new_embeddings
|
| 514 |
-
|
| 515 |
-
# Copied from transformers.models.bart.modeling_bart.BartDecoder._prepare_decoder_attention_mask
|
| 516 |
-
def _prepare_decoder_attention_mask(self, attention_mask, input_shape, inputs_embeds, past_key_values_length):
|
| 517 |
-
# create causal mask
|
| 518 |
-
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
|
| 519 |
-
combined_attention_mask = None
|
| 520 |
-
if input_shape[-1] > 1:
|
| 521 |
-
combined_attention_mask = _make_causal_mask(
|
| 522 |
-
input_shape,
|
| 523 |
-
inputs_embeds.dtype,
|
| 524 |
-
device=inputs_embeds.device,
|
| 525 |
-
past_key_values_length=past_key_values_length,
|
| 526 |
-
)
|
| 527 |
-
|
| 528 |
-
if attention_mask is not None:
|
| 529 |
-
# [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len]
|
| 530 |
-
expanded_attn_mask = _expand_mask(attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1]).to(
|
| 531 |
-
inputs_embeds.device
|
| 532 |
-
)
|
| 533 |
-
combined_attention_mask = (
|
| 534 |
-
expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask + combined_attention_mask
|
| 535 |
-
)
|
| 536 |
-
|
| 537 |
-
return combined_attention_mask
|
| 538 |
-
|
| 539 |
-
|
| 540 |
-
def forward(
|
| 541 |
-
self,
|
| 542 |
-
input_ids: Optional[torch.LongTensor] = None,
|
| 543 |
-
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
|
| 544 |
-
attention_mask: Optional[torch.FloatTensor] = None,
|
| 545 |
-
token_type_ids: Optional[torch.LongTensor] = None,
|
| 546 |
-
position_ids: Optional[torch.LongTensor] = None,
|
| 547 |
-
head_mask: Optional[torch.FloatTensor] = None,
|
| 548 |
-
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 549 |
-
encoder_hidden_states: Optional[torch.Tensor] = None,
|
| 550 |
-
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
| 551 |
-
use_cache: Optional[bool] = None,
|
| 552 |
-
output_attentions: Optional[bool] = None,
|
| 553 |
-
output_hidden_states: Optional[bool] = None,
|
| 554 |
-
return_dict: Optional[bool] = None,
|
| 555 |
-
masks_ids: Optional[torch.LongTensor] = None,
|
| 556 |
-
):
|
| 557 |
-
if past_key_values is None and torch.any(input_ids == self.config.visual['image_start_id']):
|
| 558 |
-
bos_pos = torch.where(input_ids == self.config.visual['image_start_id'])
|
| 559 |
-
eos_pos = torch.where(input_ids == self.config.visual['image_start_id'] + 1)
|
| 560 |
-
assert (bos_pos[0] == eos_pos[0]).all()
|
| 561 |
-
img_pos = torch.stack((bos_pos[0], bos_pos[1], eos_pos[1]), dim=1)
|
| 562 |
-
images = []
|
| 563 |
-
masks = []
|
| 564 |
-
for i, a, b in img_pos:
|
| 565 |
-
image = input_ids[i][a + 1 : b - 1].tolist()
|
| 566 |
-
image = image[ : image.index(self.config.visual['image_start_id'] + 2)]
|
| 567 |
-
images.append(bytes(image).decode('utf-8'))
|
| 568 |
-
if masks_ids is not None:
|
| 569 |
-
mask = int_list_to_str(masks_ids[i][masks_ids[i] != -1].tolist())
|
| 570 |
-
masks.append(mask)
|
| 571 |
-
else:
|
| 572 |
-
masks.append('')
|
| 573 |
-
images = self.visual.encode(images,masks)
|
| 574 |
-
assert images.shape[0] == len(images)
|
| 575 |
-
fake_images = None
|
| 576 |
-
elif self.training:
|
| 577 |
-
fake_images=torch.zeros(1,3,224,224).to(
|
| 578 |
-
dtype=self.visual.conv1.weight.dtype, device=self.visual.conv1.weight.device)
|
| 579 |
-
images = self.visual(fake_images)
|
| 580 |
-
else:
|
| 581 |
-
fake_images = None
|
| 582 |
-
images = None
|
| 583 |
-
|
| 584 |
-
output_attentions = (
|
| 585 |
-
output_attentions
|
| 586 |
-
if output_attentions is not None
|
| 587 |
-
else self.config.output_attentions
|
| 588 |
-
)
|
| 589 |
-
output_hidden_states = (
|
| 590 |
-
output_hidden_states
|
| 591 |
-
if output_hidden_states is not None
|
| 592 |
-
else self.config.output_hidden_states
|
| 593 |
-
)
|
| 594 |
-
use_cache = use_cache if use_cache is not None else self.config.use_cache
|
| 595 |
-
return_dict = (
|
| 596 |
-
return_dict if return_dict is not None else self.config.use_return_dict
|
| 597 |
-
)
|
| 598 |
-
|
| 599 |
-
if input_ids is not None and inputs_embeds is not None:
|
| 600 |
-
raise ValueError(
|
| 601 |
-
"You cannot specify both input_ids and inputs_embeds at the same time"
|
| 602 |
-
)
|
| 603 |
-
elif input_ids is not None:
|
| 604 |
-
input_shape = input_ids.size()
|
| 605 |
-
input_ids = input_ids.view(-1, input_shape[-1])
|
| 606 |
-
batch_size = input_ids.shape[0]
|
| 607 |
-
elif inputs_embeds is not None:
|
| 608 |
-
input_shape = inputs_embeds.size()[:-1]
|
| 609 |
-
batch_size = inputs_embeds.shape[0]
|
| 610 |
-
else:
|
| 611 |
-
raise ValueError("You have to specify either input_ids or inputs_embeds")
|
| 612 |
-
|
| 613 |
-
device = input_ids.device if input_ids is not None else inputs_embeds.device
|
| 614 |
-
|
| 615 |
-
if token_type_ids is not None:
|
| 616 |
-
token_type_ids = token_type_ids.view(-1, input_shape[-1])
|
| 617 |
-
if position_ids is not None:
|
| 618 |
-
position_ids = position_ids.view(-1, input_shape[-1])
|
| 619 |
-
|
| 620 |
-
if past_key_values is None:
|
| 621 |
-
past_length = 0
|
| 622 |
-
past_key_values = tuple([None] * len(self.h))
|
| 623 |
-
else:
|
| 624 |
-
past_length = past_key_values[0][0].size(-2)
|
| 625 |
-
|
| 626 |
-
if position_ids is None:
|
| 627 |
-
position_ids = torch.arange(
|
| 628 |
-
past_length,
|
| 629 |
-
input_shape[-1] + past_length,
|
| 630 |
-
dtype=torch.long,
|
| 631 |
-
device=device,
|
| 632 |
-
)
|
| 633 |
-
position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1])
|
| 634 |
-
|
| 635 |
-
encoder_attention_mask = None
|
| 636 |
-
head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
|
| 637 |
-
|
| 638 |
-
if inputs_embeds is None:
|
| 639 |
-
inputs_embeds = self.wte(input_ids)
|
| 640 |
-
|
| 641 |
-
if batch_size <= 0:
|
| 642 |
-
raise ValueError("batch_size has to be defined and > 0")
|
| 643 |
-
attention_mask = self._prepare_decoder_attention_mask(
|
| 644 |
-
attention_mask, input_shape, inputs_embeds, past_length
|
| 645 |
-
)
|
| 646 |
-
|
| 647 |
-
hidden_states = inputs_embeds
|
| 648 |
-
|
| 649 |
-
kv_seq_len = hidden_states.size()[1]
|
| 650 |
-
if past_key_values[0] is not None:
|
| 651 |
-
# past key values[0][0] shape: bs * seq_len * head_num * dim
|
| 652 |
-
kv_seq_len += past_key_values[0][0].shape[1]
|
| 653 |
-
if (
|
| 654 |
-
self.use_dynamic_ntk
|
| 655 |
-
and kv_seq_len == hidden_states.size()[1]
|
| 656 |
-
and not self.training
|
| 657 |
-
):
|
| 658 |
-
context_value = math.log(kv_seq_len / self.seq_length, 2) + 1
|
| 659 |
-
ntk_alpha = 2 ** math.ceil(context_value) - 1
|
| 660 |
-
ntk_alpha = max(ntk_alpha, 1)
|
| 661 |
-
else:
|
| 662 |
-
ntk_alpha = self.rotary_emb._ntk_alpha_cached
|
| 663 |
-
|
| 664 |
-
rotary_pos_emb = self.rotary_emb(kv_seq_len, ntk_alpha=ntk_alpha)
|
| 665 |
-
for idx in range(len(rotary_pos_emb)):
|
| 666 |
-
rotary_pos_emb[idx] = rotary_pos_emb[idx].to(hidden_states.device)
|
| 667 |
-
|
| 668 |
-
hidden_states = self.drop(hidden_states).clone()
|
| 669 |
-
if fake_images is not None:
|
| 670 |
-
hidden_states = hidden_states + images.mean()*0
|
| 671 |
-
elif images is not None:
|
| 672 |
-
for idx, (i, a, b) in enumerate(img_pos):
|
| 673 |
-
hidden_states[i][a + 1 : b] = images[idx]
|
| 674 |
-
output_shape = input_shape + (hidden_states.size(-1),)
|
| 675 |
-
|
| 676 |
-
if self.gradient_checkpointing and self.training:
|
| 677 |
-
if use_cache:
|
| 678 |
-
logger.warning_once(
|
| 679 |
-
"`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
|
| 680 |
-
)
|
| 681 |
-
use_cache = False
|
| 682 |
-
|
| 683 |
-
presents = () if use_cache else None
|
| 684 |
-
all_self_attentions = () if output_attentions else None
|
| 685 |
-
all_hidden_states = () if output_hidden_states else None
|
| 686 |
-
for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
|
| 687 |
-
|
| 688 |
-
if output_hidden_states:
|
| 689 |
-
all_hidden_states = all_hidden_states + (hidden_states,)
|
| 690 |
-
|
| 691 |
-
if self.gradient_checkpointing and self.training:
|
| 692 |
-
|
| 693 |
-
def create_custom_forward(module):
|
| 694 |
-
def custom_forward(*inputs):
|
| 695 |
-
# None for past_key_value
|
| 696 |
-
return module(*inputs, use_cache, output_attentions)
|
| 697 |
-
|
| 698 |
-
return custom_forward
|
| 699 |
-
|
| 700 |
-
outputs = torch.utils.checkpoint.checkpoint(
|
| 701 |
-
create_custom_forward(block),
|
| 702 |
-
hidden_states,
|
| 703 |
-
rotary_pos_emb,
|
| 704 |
-
self.registered_causal_mask,
|
| 705 |
-
None,
|
| 706 |
-
attention_mask,
|
| 707 |
-
head_mask[i],
|
| 708 |
-
encoder_hidden_states,
|
| 709 |
-
encoder_attention_mask,
|
| 710 |
-
)
|
| 711 |
-
else:
|
| 712 |
-
outputs = block(
|
| 713 |
-
hidden_states,
|
| 714 |
-
layer_past=layer_past,
|
| 715 |
-
rotary_pos_emb=rotary_pos_emb,
|
| 716 |
-
registered_causal_mask=self.registered_causal_mask,
|
| 717 |
-
attention_mask=attention_mask,
|
| 718 |
-
head_mask=head_mask[i],
|
| 719 |
-
encoder_hidden_states=encoder_hidden_states,
|
| 720 |
-
encoder_attention_mask=encoder_attention_mask,
|
| 721 |
-
use_cache=use_cache,
|
| 722 |
-
output_attentions=output_attentions,
|
| 723 |
-
)
|
| 724 |
-
|
| 725 |
-
hidden_states = outputs[0]
|
| 726 |
-
if use_cache is True:
|
| 727 |
-
presents = presents + (outputs[1],)
|
| 728 |
-
|
| 729 |
-
if output_attentions:
|
| 730 |
-
all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],)
|
| 731 |
-
|
| 732 |
-
hidden_states = self.ln_f(hidden_states)
|
| 733 |
-
hidden_states = hidden_states.view(output_shape)
|
| 734 |
-
# Add last hidden state
|
| 735 |
-
if output_hidden_states:
|
| 736 |
-
all_hidden_states = all_hidden_states + (hidden_states,)
|
| 737 |
-
|
| 738 |
-
if not return_dict:
|
| 739 |
-
return tuple(
|
| 740 |
-
v for v in [hidden_states, presents, all_hidden_states] if v is not None
|
| 741 |
-
)
|
| 742 |
-
|
| 743 |
-
return BaseModelOutputWithPast(
|
| 744 |
-
last_hidden_state=hidden_states,
|
| 745 |
-
past_key_values=presents,
|
| 746 |
-
hidden_states=all_hidden_states,
|
| 747 |
-
attentions=all_self_attentions,
|
| 748 |
-
)
|
| 749 |
-
|
| 750 |
-
|
| 751 |
-
class QWenLMHeadModel(QWenPreTrainedModel):
|
| 752 |
-
_keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.rotary_emb\.inv_freq"]
|
| 753 |
-
_keys_to_ignore_on_load_unexpected = [r"h\.\d+\.attn\.masked_bias"]
|
| 754 |
-
|
| 755 |
-
def __init__(self, config):
|
| 756 |
-
super().__init__(config)
|
| 757 |
-
assert (
|
| 758 |
-
config.bf16 + config.fp16 + config.fp32 <= 1
|
| 759 |
-
), "Only one of \"bf16\", \"fp16\", \"fp32\" can be true"
|
| 760 |
-
|
| 761 |
-
autoset_precision = config.bf16 + config.fp16 + config.fp32 == 0
|
| 762 |
-
|
| 763 |
-
if autoset_precision:
|
| 764 |
-
if SUPPORT_BF16:
|
| 765 |
-
logger.warn(
|
| 766 |
-
"The model is automatically converting to bf16 for faster inference. "
|
| 767 |
-
"If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"."
|
| 768 |
-
)
|
| 769 |
-
config.bf16 = True
|
| 770 |
-
elif SUPPORT_FP16:
|
| 771 |
-
logger.warn(
|
| 772 |
-
"The model is automatically converting to fp16 for faster inference. "
|
| 773 |
-
"If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"."
|
| 774 |
-
)
|
| 775 |
-
config.fp16 = True
|
| 776 |
-
else:
|
| 777 |
-
config.fp32 = True
|
| 778 |
-
|
| 779 |
-
if config.bf16 and SUPPORT_CUDA and not SUPPORT_BF16:
|
| 780 |
-
logger.warn("Your device does NOT seem to support bf16, you can switch to fp16 or fp32 by by passing fp16/fp32=True in \"AutoModelForCausalLM.from_pretrained\".")
|
| 781 |
-
if config.fp16 and SUPPORT_CUDA and not SUPPORT_FP16:
|
| 782 |
-
logger.warn("Your device does NOT support faster inference with fp16, please switch to fp32 which is likely to be faster")
|
| 783 |
-
if config.fp32:
|
| 784 |
-
if SUPPORT_BF16:
|
| 785 |
-
logger.warn("Your device support faster inference by passing bf16=True in \"AutoModelForCausalLM.from_pretrained\".")
|
| 786 |
-
elif SUPPORT_FP16:
|
| 787 |
-
logger.warn("Your device support faster inference by passing fp16=True in \"AutoModelForCausalLM.from_pretrained\".")
|
| 788 |
-
|
| 789 |
-
self.transformer = QWenModel(config)
|
| 790 |
-
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
|
| 791 |
-
|
| 792 |
-
if config.bf16:
|
| 793 |
-
self.transformer.bfloat16()
|
| 794 |
-
self.lm_head.bfloat16()
|
| 795 |
-
if config.fp16:
|
| 796 |
-
self.transformer.half()
|
| 797 |
-
self.lm_head.half()
|
| 798 |
-
self.post_init()
|
| 799 |
-
|
| 800 |
-
def get_output_embeddings(self):
|
| 801 |
-
return self.lm_head
|
| 802 |
-
|
| 803 |
-
def set_output_embeddings(self, new_embeddings):
|
| 804 |
-
self.lm_head = new_embeddings
|
| 805 |
-
|
| 806 |
-
def prepare_inputs_for_generation(
|
| 807 |
-
self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs
|
| 808 |
-
):
|
| 809 |
-
token_type_ids = kwargs.get("token_type_ids", None)
|
| 810 |
-
if past_key_values:
|
| 811 |
-
input_ids = input_ids[:, -1].unsqueeze(-1)
|
| 812 |
-
if token_type_ids is not None:
|
| 813 |
-
token_type_ids = token_type_ids[:, -1].unsqueeze(-1)
|
| 814 |
-
|
| 815 |
-
attention_mask = kwargs.get("attention_mask", None)
|
| 816 |
-
position_ids = kwargs.get("position_ids", None)
|
| 817 |
-
|
| 818 |
-
if attention_mask is not None and position_ids is None:
|
| 819 |
-
position_ids = attention_mask.long().cumsum(-1) - 1
|
| 820 |
-
position_ids.masked_fill_(attention_mask == 0, 1)
|
| 821 |
-
if past_key_values:
|
| 822 |
-
position_ids = position_ids[:, -1].unsqueeze(-1)
|
| 823 |
-
else:
|
| 824 |
-
position_ids = None
|
| 825 |
-
|
| 826 |
-
if inputs_embeds is not None and past_key_values is None:
|
| 827 |
-
model_inputs = {"inputs_embeds": inputs_embeds}
|
| 828 |
-
else:
|
| 829 |
-
model_inputs = {"input_ids": input_ids}
|
| 830 |
-
|
| 831 |
-
model_inputs.update(
|
| 832 |
-
{
|
| 833 |
-
"past_key_values": past_key_values,
|
| 834 |
-
"use_cache": kwargs.get("use_cache"),
|
| 835 |
-
"position_ids": position_ids,
|
| 836 |
-
"attention_mask": attention_mask,
|
| 837 |
-
"token_type_ids": token_type_ids,
|
| 838 |
-
}
|
| 839 |
-
)
|
| 840 |
-
return model_inputs
|
| 841 |
-
|
| 842 |
-
def forward(
|
| 843 |
-
self,
|
| 844 |
-
input_ids: Optional[torch.LongTensor] = None,
|
| 845 |
-
past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None,
|
| 846 |
-
attention_mask: Optional[torch.FloatTensor] = None,
|
| 847 |
-
token_type_ids: Optional[torch.LongTensor] = None,
|
| 848 |
-
position_ids: Optional[torch.LongTensor] = None,
|
| 849 |
-
head_mask: Optional[torch.FloatTensor] = None,
|
| 850 |
-
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 851 |
-
encoder_hidden_states: Optional[torch.Tensor] = None,
|
| 852 |
-
encoder_attention_mask: Optional[torch.FloatTensor] = None,
|
| 853 |
-
labels: Optional[torch.LongTensor] = None,
|
| 854 |
-
use_cache: Optional[bool] = None,
|
| 855 |
-
output_attentions: Optional[bool] = None,
|
| 856 |
-
output_hidden_states: Optional[bool] = None,
|
| 857 |
-
return_dict: Optional[bool] = None,
|
| 858 |
-
masks_ids: Optional[torch.LongTensor] = None,
|
| 859 |
-
) -> Union[Tuple, CausalLMOutputWithPast]:
|
| 860 |
-
|
| 861 |
-
return_dict = (
|
| 862 |
-
return_dict if return_dict is not None else self.config.use_return_dict
|
| 863 |
-
)
|
| 864 |
-
|
| 865 |
-
transformer_outputs = self.transformer(
|
| 866 |
-
input_ids,
|
| 867 |
-
past_key_values=past_key_values,
|
| 868 |
-
attention_mask=attention_mask,
|
| 869 |
-
token_type_ids=token_type_ids,
|
| 870 |
-
position_ids=position_ids,
|
| 871 |
-
head_mask=head_mask,
|
| 872 |
-
inputs_embeds=inputs_embeds,
|
| 873 |
-
encoder_hidden_states=encoder_hidden_states,
|
| 874 |
-
encoder_attention_mask=encoder_attention_mask,
|
| 875 |
-
use_cache=use_cache,
|
| 876 |
-
output_attentions=output_attentions,
|
| 877 |
-
output_hidden_states=output_hidden_states,
|
| 878 |
-
return_dict=return_dict,
|
| 879 |
-
masks_ids=masks_ids,
|
| 880 |
-
)
|
| 881 |
-
hidden_states = transformer_outputs[0]
|
| 882 |
-
|
| 883 |
-
lm_logits = self.lm_head(hidden_states)
|
| 884 |
-
|
| 885 |
-
loss = None
|
| 886 |
-
if labels is not None:
|
| 887 |
-
labels = labels.to(lm_logits.device)
|
| 888 |
-
shift_logits = lm_logits[..., :-1, :].contiguous()
|
| 889 |
-
shift_labels = labels[..., 1:].contiguous()
|
| 890 |
-
loss_fct = CrossEntropyLoss()
|
| 891 |
-
loss = loss_fct(
|
| 892 |
-
shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1)
|
| 893 |
-
)
|
| 894 |
-
|
| 895 |
-
if not return_dict:
|
| 896 |
-
output = (lm_logits,) + transformer_outputs[1:]
|
| 897 |
-
return ((loss,) + output) if loss is not None else output
|
| 898 |
-
|
| 899 |
-
return CausalLMOutputWithPast(
|
| 900 |
-
loss=loss,
|
| 901 |
-
logits=lm_logits,
|
| 902 |
-
past_key_values=transformer_outputs.past_key_values,
|
| 903 |
-
hidden_states=transformer_outputs.hidden_states,
|
| 904 |
-
attentions=transformer_outputs.attentions,
|
| 905 |
-
)
|
| 906 |
-
|
| 907 |
-
@staticmethod
|
| 908 |
-
def _reorder_cache(
|
| 909 |
-
past_key_values: Tuple[Tuple[torch.Tensor]], beam_idx: torch.Tensor
|
| 910 |
-
) -> Tuple[Tuple[torch.Tensor]]:
|
| 911 |
-
|
| 912 |
-
return tuple(
|
| 913 |
-
tuple(
|
| 914 |
-
past_state.index_select(0, beam_idx.to(past_state.device))
|
| 915 |
-
for past_state in layer_past
|
| 916 |
-
)
|
| 917 |
-
for layer_past in past_key_values
|
| 918 |
-
)
|
| 919 |
-
|
| 920 |
-
def chat(
|
| 921 |
-
self,
|
| 922 |
-
tokenizer: PreTrainedTokenizer,
|
| 923 |
-
query: str,
|
| 924 |
-
history: Optional[HistoryType],
|
| 925 |
-
system: str = "You are a helpful assistant.",
|
| 926 |
-
append_history: bool = True,
|
| 927 |
-
stream: Optional[bool] = _SENTINEL,
|
| 928 |
-
stop_words_ids: Optional[List[List[int]]] = None,
|
| 929 |
-
generation_config: Optional[GenerationConfig] = None,
|
| 930 |
-
**kwargs,
|
| 931 |
-
) -> Tuple[str, HistoryType]:
|
| 932 |
-
generation_config = generation_config if generation_config is not None else self.generation_config
|
| 933 |
-
|
| 934 |
-
assert stream is _SENTINEL, _ERROR_STREAM_IN_CHAT
|
| 935 |
-
assert generation_config.chat_format == 'chatml', _ERROR_BAD_CHAT_FORMAT
|
| 936 |
-
if history is None:
|
| 937 |
-
history = []
|
| 938 |
-
if stop_words_ids is None:
|
| 939 |
-
stop_words_ids = []
|
| 940 |
-
|
| 941 |
-
max_window_size = kwargs.get('max_window_size', None)
|
| 942 |
-
if max_window_size is None:
|
| 943 |
-
max_window_size = generation_config.max_window_size
|
| 944 |
-
raw_text, context_tokens = make_context(
|
| 945 |
-
tokenizer,
|
| 946 |
-
query,
|
| 947 |
-
history=history,
|
| 948 |
-
system=system,
|
| 949 |
-
max_window_size=max_window_size,
|
| 950 |
-
chat_format=generation_config.chat_format,
|
| 951 |
-
)
|
| 952 |
-
|
| 953 |
-
stop_words_ids.extend(get_stop_words_ids(
|
| 954 |
-
generation_config.chat_format, tokenizer
|
| 955 |
-
))
|
| 956 |
-
input_ids = torch.tensor([context_tokens]).to(self.device)
|
| 957 |
-
outputs = self.generate(
|
| 958 |
-
input_ids,
|
| 959 |
-
stop_words_ids=stop_words_ids,
|
| 960 |
-
return_dict_in_generate=False,
|
| 961 |
-
generation_config=generation_config,
|
| 962 |
-
**kwargs,
|
| 963 |
-
)
|
| 964 |
-
|
| 965 |
-
response = decode_tokens(
|
| 966 |
-
outputs[0],
|
| 967 |
-
tokenizer,
|
| 968 |
-
raw_text_len=len(raw_text),
|
| 969 |
-
context_length=len(context_tokens),
|
| 970 |
-
chat_format=generation_config.chat_format,
|
| 971 |
-
verbose=False,
|
| 972 |
-
errors='replace'
|
| 973 |
-
)
|
| 974 |
-
|
| 975 |
-
if append_history:
|
| 976 |
-
history.append((query, response))
|
| 977 |
-
|
| 978 |
-
return response, history
|
| 979 |
-
|
| 980 |
-
def chat_stream(
|
| 981 |
-
self,
|
| 982 |
-
tokenizer: PreTrainedTokenizer,
|
| 983 |
-
query: str,
|
| 984 |
-
history: Optional[HistoryType],
|
| 985 |
-
system: str = "You are a helpful assistant.",
|
| 986 |
-
stop_words_ids: Optional[List[List[int]]] = None,
|
| 987 |
-
logits_processor: Optional[LogitsProcessorList] = None,
|
| 988 |
-
generation_config: Optional[GenerationConfig] = None,
|
| 989 |
-
**kwargs,
|
| 990 |
-
) -> Generator[str, Any, None]:
|
| 991 |
-
generation_config = generation_config if generation_config is not None else self.generation_config
|
| 992 |
-
assert generation_config.chat_format == 'chatml', _ERROR_BAD_CHAT_FORMAT
|
| 993 |
-
if history is None:
|
| 994 |
-
history = []
|
| 995 |
-
if stop_words_ids is None:
|
| 996 |
-
stop_words_ids = []
|
| 997 |
-
|
| 998 |
-
max_window_size = kwargs.get('max_window_size', None)
|
| 999 |
-
if max_window_size is None:
|
| 1000 |
-
max_window_size = generation_config.max_window_size
|
| 1001 |
-
raw_text, context_tokens = make_context(
|
| 1002 |
-
tokenizer,
|
| 1003 |
-
query,
|
| 1004 |
-
history=history,
|
| 1005 |
-
system=system,
|
| 1006 |
-
max_window_size=max_window_size,
|
| 1007 |
-
chat_format=generation_config.chat_format,
|
| 1008 |
-
)
|
| 1009 |
-
|
| 1010 |
-
stop_words_ids.extend(get_stop_words_ids(
|
| 1011 |
-
generation_config.chat_format, tokenizer
|
| 1012 |
-
))
|
| 1013 |
-
if stop_words_ids is not None:
|
| 1014 |
-
stop_words_logits_processor = StopWordsLogitsProcessor(
|
| 1015 |
-
stop_words_ids=stop_words_ids,
|
| 1016 |
-
eos_token_id=generation_config.eos_token_id,
|
| 1017 |
-
)
|
| 1018 |
-
if logits_processor is None:
|
| 1019 |
-
logits_processor = LogitsProcessorList([stop_words_logits_processor])
|
| 1020 |
-
else:
|
| 1021 |
-
logits_processor.append(stop_words_logits_processor)
|
| 1022 |
-
input_ids = torch.tensor([context_tokens]).to(self.device)
|
| 1023 |
-
|
| 1024 |
-
from transformers_stream_generator.main import NewGenerationMixin, StreamGenerationConfig
|
| 1025 |
-
self.__class__.generate_stream = NewGenerationMixin.generate
|
| 1026 |
-
self.__class__.sample_stream = NewGenerationMixin.sample_stream
|
| 1027 |
-
stream_config = StreamGenerationConfig(**generation_config.to_dict(), do_stream=True)
|
| 1028 |
-
|
| 1029 |
-
def stream_generator():
|
| 1030 |
-
outputs = []
|
| 1031 |
-
for token in self.generate_stream(
|
| 1032 |
-
input_ids,
|
| 1033 |
-
return_dict_in_generate=False,
|
| 1034 |
-
generation_config=stream_config,
|
| 1035 |
-
logits_processor=logits_processor,
|
| 1036 |
-
seed=-1,
|
| 1037 |
-
**kwargs):
|
| 1038 |
-
outputs.append(token.item())
|
| 1039 |
-
yield tokenizer.decode(outputs, skip_special_tokens=True, errors='ignore', keep_image_special=True)
|
| 1040 |
-
|
| 1041 |
-
return stream_generator()
|
| 1042 |
-
|
| 1043 |
-
def generate(
|
| 1044 |
-
self,
|
| 1045 |
-
inputs: Optional[torch.Tensor] = None,
|
| 1046 |
-
generation_config: Optional[GenerationConfig] = None,
|
| 1047 |
-
logits_processor: Optional[LogitsProcessorList] = None,
|
| 1048 |
-
stopping_criteria: Optional[StoppingCriteriaList] = None,
|
| 1049 |
-
prefix_allowed_tokens_fn: Optional[
|
| 1050 |
-
Callable[[int, torch.Tensor], List[int]]
|
| 1051 |
-
] = None,
|
| 1052 |
-
synced_gpus: Optional[bool] = None,
|
| 1053 |
-
assistant_model: Optional["PreTrainedModel"] = None,
|
| 1054 |
-
streamer: Optional["BaseStreamer"] = None,
|
| 1055 |
-
**kwargs,
|
| 1056 |
-
) -> Union[GenerateOutput, torch.LongTensor]:
|
| 1057 |
-
generation_config = generation_config if generation_config is not None else self.generation_config
|
| 1058 |
-
|
| 1059 |
-
# Process stop_words_ids.
|
| 1060 |
-
stop_words_ids = kwargs.pop("stop_words_ids", None)
|
| 1061 |
-
if stop_words_ids is None and generation_config is not None:
|
| 1062 |
-
stop_words_ids = getattr(generation_config, "stop_words_ids", None)
|
| 1063 |
-
if stop_words_ids is None:
|
| 1064 |
-
stop_words_ids = getattr(generation_config, "stop_words_ids", None)
|
| 1065 |
-
|
| 1066 |
-
if stop_words_ids is not None:
|
| 1067 |
-
stop_words_logits_processor = StopWordsLogitsProcessor(
|
| 1068 |
-
stop_words_ids=stop_words_ids,
|
| 1069 |
-
eos_token_id=generation_config.eos_token_id,
|
| 1070 |
-
)
|
| 1071 |
-
if logits_processor is None:
|
| 1072 |
-
logits_processor = LogitsProcessorList([stop_words_logits_processor])
|
| 1073 |
-
else:
|
| 1074 |
-
logits_processor.append(stop_words_logits_processor)
|
| 1075 |
-
|
| 1076 |
-
return super().generate(
|
| 1077 |
-
inputs,
|
| 1078 |
-
generation_config=generation_config,
|
| 1079 |
-
logits_processor=logits_processor,
|
| 1080 |
-
stopping_criteria=stopping_criteria,
|
| 1081 |
-
prefix_allowed_tokens_fn=prefix_allowed_tokens_fn,
|
| 1082 |
-
synced_gpus=synced_gpus,
|
| 1083 |
-
assistant_model=assistant_model,
|
| 1084 |
-
streamer=streamer,
|
| 1085 |
-
**kwargs,
|
| 1086 |
-
)
|
| 1087 |
-
|
| 1088 |
-
|
| 1089 |
-
class RotaryEmbedding(torch.nn.Module):
|
| 1090 |
-
def __init__(self, dim, base=10000):
|
| 1091 |
-
super().__init__()
|
| 1092 |
-
self.dim = dim
|
| 1093 |
-
self.base = base
|
| 1094 |
-
self.inv_freq = 1.0 / (base ** (torch.arange(0, dim, 2).float() / dim))
|
| 1095 |
-
if importlib.util.find_spec("einops") is None:
|
| 1096 |
-
raise RuntimeError("einops is required for Rotary Embedding")
|
| 1097 |
-
|
| 1098 |
-
self._rotary_pos_emb_cache = None
|
| 1099 |
-
self._seq_len_cached = 0
|
| 1100 |
-
self._ntk_alpha_cached = 1.0
|
| 1101 |
-
|
| 1102 |
-
def update_rotary_pos_emb_cache(self, max_seq_len, offset=0, ntk_alpha=1.0):
|
| 1103 |
-
seqlen = max_seq_len + offset
|
| 1104 |
-
if seqlen > self._seq_len_cached or ntk_alpha != self._ntk_alpha_cached:
|
| 1105 |
-
base = self.base * ntk_alpha ** (self.dim / (self.dim - 2))
|
| 1106 |
-
self.inv_freq = 1.0 / (
|
| 1107 |
-
base
|
| 1108 |
-
** (
|
| 1109 |
-
torch.arange(0, self.dim, 2, device=self.inv_freq.device).float()
|
| 1110 |
-
/ self.dim
|
| 1111 |
-
)
|
| 1112 |
-
)
|
| 1113 |
-
self._seq_len_cached = max(2 * seqlen, 16)
|
| 1114 |
-
self._ntk_alpha_cached = ntk_alpha
|
| 1115 |
-
seq = torch.arange(self._seq_len_cached, device=self.inv_freq.device)
|
| 1116 |
-
freqs = torch.outer(seq.type_as(self.inv_freq), self.inv_freq)
|
| 1117 |
-
|
| 1118 |
-
emb = torch.cat((freqs, freqs), dim=-1)
|
| 1119 |
-
from einops import rearrange
|
| 1120 |
-
|
| 1121 |
-
emb = rearrange(emb, "n d -> 1 n 1 d")
|
| 1122 |
-
|
| 1123 |
-
cos, sin = emb.cos(), emb.sin()
|
| 1124 |
-
self._rotary_pos_emb_cache = [cos, sin]
|
| 1125 |
-
|
| 1126 |
-
def forward(self, max_seq_len, offset=0, ntk_alpha=1.0):
|
| 1127 |
-
self.update_rotary_pos_emb_cache(max_seq_len, offset, ntk_alpha)
|
| 1128 |
-
cos, sin = self._rotary_pos_emb_cache
|
| 1129 |
-
return [cos[:, offset : offset + max_seq_len], sin[:, offset : offset + max_seq_len]]
|
| 1130 |
-
|
| 1131 |
-
|
| 1132 |
-
def _rotate_half(x):
|
| 1133 |
-
from einops import rearrange
|
| 1134 |
-
|
| 1135 |
-
x = rearrange(x, "... (j d) -> ... j d", j=2)
|
| 1136 |
-
x1, x2 = x.unbind(dim=-2)
|
| 1137 |
-
return torch.cat((-x2, x1), dim=-1)
|
| 1138 |
-
|
| 1139 |
-
|
| 1140 |
-
def apply_rotary_pos_emb(t, freqs):
|
| 1141 |
-
cos, sin = freqs
|
| 1142 |
-
if apply_rotary_emb_func is not None and t.is_cuda:
|
| 1143 |
-
t_ = t.float()
|
| 1144 |
-
cos = cos.squeeze(0).squeeze(1)[:, : cos.shape[-1] // 2]
|
| 1145 |
-
sin = sin.squeeze(0).squeeze(1)[:, : sin.shape[-1] // 2]
|
| 1146 |
-
output = apply_rotary_emb_func(t_, cos, sin).type_as(t)
|
| 1147 |
-
return output
|
| 1148 |
-
else:
|
| 1149 |
-
rot_dim = freqs[0].shape[-1]
|
| 1150 |
-
cos, sin = freqs
|
| 1151 |
-
t_, t_pass_ = t[..., :rot_dim], t[..., rot_dim:]
|
| 1152 |
-
t_ = t_.float()
|
| 1153 |
-
t_pass_ = t_pass_.float()
|
| 1154 |
-
t_ = (t_ * cos) + (_rotate_half(t_) * sin)
|
| 1155 |
-
return torch.cat((t_, t_pass_), dim=-1).type_as(t)
|
| 1156 |
-
|
| 1157 |
-
|
| 1158 |
-
class RMSNorm(torch.nn.Module):
|
| 1159 |
-
def __init__(self, dim: int, eps: float = 1e-6):
|
| 1160 |
-
super().__init__()
|
| 1161 |
-
self.eps = eps
|
| 1162 |
-
self.weight = nn.Parameter(torch.ones(dim))
|
| 1163 |
-
|
| 1164 |
-
def _norm(self, x):
|
| 1165 |
-
return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
|
| 1166 |
-
|
| 1167 |
-
def forward(self, x):
|
| 1168 |
-
if rms_norm is not None and x.is_cuda:
|
| 1169 |
-
return rms_norm(x, self.weight, self.eps)
|
| 1170 |
-
else:
|
| 1171 |
-
output = self._norm(x.float()).type_as(x)
|
| 1172 |
-
return output * self.weight
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/qwen.tiktoken
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
zzzmmz/SegAgent-Model/qwen_generation_utils.py
DELETED
|
@@ -1,420 +0,0 @@
|
|
| 1 |
-
# Copyright (c) Alibaba Cloud.
|
| 2 |
-
#
|
| 3 |
-
# This source code is licensed under the license found in the
|
| 4 |
-
# LICENSE file in the root directory of this source tree.
|
| 5 |
-
|
| 6 |
-
"""Generation support."""
|
| 7 |
-
|
| 8 |
-
from typing import Tuple, List, Union, Iterable
|
| 9 |
-
|
| 10 |
-
import numpy as np
|
| 11 |
-
import torch
|
| 12 |
-
import torch.nn.functional as F
|
| 13 |
-
from transformers import PreTrainedTokenizer
|
| 14 |
-
from transformers import logging
|
| 15 |
-
from transformers.generation import LogitsProcessor
|
| 16 |
-
|
| 17 |
-
logger = logging.get_logger(__name__)
|
| 18 |
-
|
| 19 |
-
# Types.
|
| 20 |
-
HistoryType = List[Tuple[str, str]]
|
| 21 |
-
TokensType = List[int]
|
| 22 |
-
BatchTokensType = List[List[int]]
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
def pad_batch(batch: BatchTokensType, pad_id: int, seq_length: int) -> BatchTokensType:
|
| 26 |
-
for tokens in batch:
|
| 27 |
-
context_length = len(tokens)
|
| 28 |
-
if context_length < seq_length:
|
| 29 |
-
tokens.extend([pad_id] * (seq_length - context_length))
|
| 30 |
-
return batch
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
def get_ltor_masks_and_position_ids(
|
| 34 |
-
data,
|
| 35 |
-
eod_token,
|
| 36 |
-
reset_position_ids,
|
| 37 |
-
reset_attention_mask,
|
| 38 |
-
eod_mask_loss,
|
| 39 |
-
):
|
| 40 |
-
"""Build masks and position id for left to right model."""
|
| 41 |
-
|
| 42 |
-
# Extract batch size and sequence length.
|
| 43 |
-
micro_batch_size, seq_length = data.size()
|
| 44 |
-
|
| 45 |
-
# Attention mask (lower triangular).
|
| 46 |
-
if reset_attention_mask:
|
| 47 |
-
att_mask_batch = micro_batch_size
|
| 48 |
-
else:
|
| 49 |
-
att_mask_batch = 1
|
| 50 |
-
attention_mask = torch.tril(
|
| 51 |
-
torch.ones((att_mask_batch, seq_length, seq_length), device=data.device)
|
| 52 |
-
).view(att_mask_batch, 1, seq_length, seq_length)
|
| 53 |
-
|
| 54 |
-
# Loss mask.
|
| 55 |
-
loss_mask = torch.ones(data.size(), dtype=torch.float, device=data.device)
|
| 56 |
-
if eod_mask_loss:
|
| 57 |
-
loss_mask[data == eod_token] = 0.0
|
| 58 |
-
|
| 59 |
-
# Position ids.
|
| 60 |
-
position_ids = torch.arange(seq_length, dtype=torch.long, device=data.device)
|
| 61 |
-
position_ids = position_ids.unsqueeze(0).expand_as(data)
|
| 62 |
-
# We need to clone as the ids will be modifed based on batch index.
|
| 63 |
-
if reset_position_ids:
|
| 64 |
-
position_ids = position_ids.clone()
|
| 65 |
-
|
| 66 |
-
if reset_position_ids or reset_attention_mask:
|
| 67 |
-
# Loop through the batches:
|
| 68 |
-
for b in range(micro_batch_size):
|
| 69 |
-
|
| 70 |
-
# Find indecies where EOD token is.
|
| 71 |
-
eod_index = position_ids[b, data[b] == eod_token]
|
| 72 |
-
# Detach indecies from positions if going to modify positions.
|
| 73 |
-
if reset_position_ids:
|
| 74 |
-
eod_index = eod_index.clone()
|
| 75 |
-
|
| 76 |
-
# Loop through EOD indecies:
|
| 77 |
-
prev_index = 0
|
| 78 |
-
for j in range(eod_index.size()[0]):
|
| 79 |
-
i = eod_index[j]
|
| 80 |
-
# Mask attention loss.
|
| 81 |
-
if reset_attention_mask:
|
| 82 |
-
attention_mask[b, 0, (i + 1) :, : (i + 1)] = 0
|
| 83 |
-
# Reset positions.
|
| 84 |
-
if reset_position_ids:
|
| 85 |
-
position_ids[b, (i + 1) :] -= i + 1 - prev_index
|
| 86 |
-
prev_index = i + 1
|
| 87 |
-
|
| 88 |
-
# Convert attention mask to binary:
|
| 89 |
-
attention_mask = attention_mask < 0.5
|
| 90 |
-
|
| 91 |
-
return attention_mask, loss_mask, position_ids
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
def get_batch(context_tokens: torch.LongTensor, eod_id: int):
|
| 95 |
-
"""Generate batch from context tokens."""
|
| 96 |
-
# Move to GPU.
|
| 97 |
-
tokens = context_tokens.contiguous().to(context_tokens.device)
|
| 98 |
-
# Get the attention mask and postition ids.
|
| 99 |
-
attention_mask, _, position_ids = get_ltor_masks_and_position_ids(
|
| 100 |
-
tokens,
|
| 101 |
-
eod_id,
|
| 102 |
-
reset_position_ids=False,
|
| 103 |
-
reset_attention_mask=False,
|
| 104 |
-
eod_mask_loss=False,
|
| 105 |
-
)
|
| 106 |
-
return tokens, attention_mask, position_ids
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
def get_stop_words_ids(chat_format, tokenizer):
|
| 110 |
-
if chat_format == "raw":
|
| 111 |
-
stop_words_ids = [tokenizer.encode("Human:"), [tokenizer.eod_id]]
|
| 112 |
-
elif chat_format == "chatml":
|
| 113 |
-
stop_words_ids = [[tokenizer.im_end_id], [tokenizer.im_start_id]]
|
| 114 |
-
else:
|
| 115 |
-
raise NotImplementedError(f"Unknown chat format {chat_format!r}")
|
| 116 |
-
return stop_words_ids
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
def make_context(
|
| 120 |
-
tokenizer: PreTrainedTokenizer,
|
| 121 |
-
query: str,
|
| 122 |
-
history: List[Tuple[str, str]] = None,
|
| 123 |
-
system: str = "",
|
| 124 |
-
max_window_size: int = 6144,
|
| 125 |
-
chat_format: str = "chatml",
|
| 126 |
-
):
|
| 127 |
-
if history is None:
|
| 128 |
-
history = []
|
| 129 |
-
|
| 130 |
-
if chat_format == "chatml":
|
| 131 |
-
im_start, im_end = "<|im_start|>", "<|im_end|>"
|
| 132 |
-
im_start_tokens = [tokenizer.im_start_id]
|
| 133 |
-
im_end_tokens = [tokenizer.im_end_id]
|
| 134 |
-
nl_tokens = tokenizer.encode("\n")
|
| 135 |
-
|
| 136 |
-
def _tokenize_str(role, content):
|
| 137 |
-
return f"{role}\n{content}", tokenizer.encode(
|
| 138 |
-
role, allowed_special=set(tokenizer.IMAGE_ST)
|
| 139 |
-
) + nl_tokens + tokenizer.encode(content, allowed_special=set(tokenizer.IMAGE_ST))
|
| 140 |
-
|
| 141 |
-
system_text, system_tokens_part = _tokenize_str("system", system)
|
| 142 |
-
system_tokens = im_start_tokens + system_tokens_part + im_end_tokens
|
| 143 |
-
|
| 144 |
-
raw_text = ""
|
| 145 |
-
context_tokens = []
|
| 146 |
-
|
| 147 |
-
for turn_query, turn_response in reversed(history):
|
| 148 |
-
query_text, query_tokens_part = _tokenize_str("user", turn_query)
|
| 149 |
-
query_tokens = im_start_tokens + query_tokens_part + im_end_tokens
|
| 150 |
-
if turn_response is not None:
|
| 151 |
-
response_text, response_tokens_part = _tokenize_str(
|
| 152 |
-
"assistant", turn_response
|
| 153 |
-
)
|
| 154 |
-
response_tokens = im_start_tokens + response_tokens_part + im_end_tokens
|
| 155 |
-
|
| 156 |
-
next_context_tokens = nl_tokens + query_tokens + nl_tokens + response_tokens
|
| 157 |
-
prev_chat = (
|
| 158 |
-
f"\n{im_start}{query_text}{im_end}\n{im_start}{response_text}{im_end}"
|
| 159 |
-
)
|
| 160 |
-
else:
|
| 161 |
-
next_context_tokens = nl_tokens + query_tokens + nl_tokens
|
| 162 |
-
prev_chat = f"\n{im_start}{query_text}{im_end}\n"
|
| 163 |
-
|
| 164 |
-
current_context_size = (
|
| 165 |
-
len(system_tokens) + len(next_context_tokens) + len(context_tokens)
|
| 166 |
-
)
|
| 167 |
-
if current_context_size < max_window_size:
|
| 168 |
-
context_tokens = next_context_tokens + context_tokens
|
| 169 |
-
raw_text = prev_chat + raw_text
|
| 170 |
-
else:
|
| 171 |
-
break
|
| 172 |
-
|
| 173 |
-
context_tokens = system_tokens + context_tokens
|
| 174 |
-
raw_text = f"{im_start}{system_text}{im_end}" + raw_text
|
| 175 |
-
context_tokens += (
|
| 176 |
-
nl_tokens
|
| 177 |
-
+ im_start_tokens
|
| 178 |
-
+ _tokenize_str("user", query)[1]
|
| 179 |
-
+ im_end_tokens
|
| 180 |
-
+ nl_tokens
|
| 181 |
-
+ im_start_tokens
|
| 182 |
-
+ tokenizer.encode("assistant")
|
| 183 |
-
+ nl_tokens
|
| 184 |
-
)
|
| 185 |
-
raw_text += f"\n{im_start}user\n{query}{im_end}\n{im_start}assistant\n"
|
| 186 |
-
|
| 187 |
-
elif chat_format == "raw":
|
| 188 |
-
raw_text = query
|
| 189 |
-
context_tokens = tokenizer.encode(raw_text)
|
| 190 |
-
else:
|
| 191 |
-
raise NotImplementedError(f"Unknown chat format {chat_format!r}")
|
| 192 |
-
|
| 193 |
-
return raw_text, context_tokens
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
def _decode_default(
|
| 197 |
-
tokens: List[int],
|
| 198 |
-
*,
|
| 199 |
-
stop_words: List[str],
|
| 200 |
-
eod_words: List[str],
|
| 201 |
-
tokenizer: PreTrainedTokenizer,
|
| 202 |
-
raw_text_len: int,
|
| 203 |
-
verbose: bool = False,
|
| 204 |
-
return_end_reason: bool = False,
|
| 205 |
-
errors: str='replace',
|
| 206 |
-
):
|
| 207 |
-
trim_decode_tokens = tokenizer.decode(tokens, errors=errors)[raw_text_len:]
|
| 208 |
-
if verbose:
|
| 209 |
-
print("\nRaw Generate: ", trim_decode_tokens)
|
| 210 |
-
|
| 211 |
-
end_reason = f"Gen length {len(tokens)}"
|
| 212 |
-
for stop_word in stop_words:
|
| 213 |
-
trim_decode_tokens = trim_decode_tokens.replace(stop_word, "").strip()
|
| 214 |
-
for eod_word in eod_words:
|
| 215 |
-
if eod_word in trim_decode_tokens:
|
| 216 |
-
end_reason = f"Gen {eod_word!r}"
|
| 217 |
-
trim_decode_tokens = trim_decode_tokens.split(eod_word)[0]
|
| 218 |
-
trim_decode_tokens = trim_decode_tokens.strip()
|
| 219 |
-
if verbose:
|
| 220 |
-
print("\nEnd Reason:", end_reason)
|
| 221 |
-
print("\nGenerate: ", trim_decode_tokens)
|
| 222 |
-
|
| 223 |
-
if return_end_reason:
|
| 224 |
-
return trim_decode_tokens, end_reason
|
| 225 |
-
else:
|
| 226 |
-
return trim_decode_tokens
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
def _decode_chatml(
|
| 230 |
-
tokens: List[int],
|
| 231 |
-
*,
|
| 232 |
-
stop_words: List[str],
|
| 233 |
-
eod_token_ids: List[int],
|
| 234 |
-
tokenizer: PreTrainedTokenizer,
|
| 235 |
-
raw_text_len: int,
|
| 236 |
-
context_length: int,
|
| 237 |
-
verbose: bool = False,
|
| 238 |
-
return_end_reason: bool = False,
|
| 239 |
-
errors: str='replace'
|
| 240 |
-
):
|
| 241 |
-
end_reason = f"Gen length {len(tokens)}"
|
| 242 |
-
eod_token_idx = context_length
|
| 243 |
-
for eod_token_idx in range(context_length, len(tokens)):
|
| 244 |
-
if tokens[eod_token_idx] in eod_token_ids:
|
| 245 |
-
end_reason = f"Gen {tokenizer.decode([tokens[eod_token_idx]])!r}"
|
| 246 |
-
break
|
| 247 |
-
|
| 248 |
-
trim_decode_tokens = tokenizer.decode(tokens[:eod_token_idx], errors=errors)[raw_text_len:]
|
| 249 |
-
if verbose:
|
| 250 |
-
print("\nRaw Generate w/o EOD:", tokenizer.decode(tokens, errors=errors)[raw_text_len:])
|
| 251 |
-
print("\nRaw Generate:", trim_decode_tokens)
|
| 252 |
-
print("\nEnd Reason:", end_reason)
|
| 253 |
-
for stop_word in stop_words:
|
| 254 |
-
trim_decode_tokens = trim_decode_tokens.replace(stop_word, "").strip()
|
| 255 |
-
trim_decode_tokens = trim_decode_tokens.strip()
|
| 256 |
-
if verbose:
|
| 257 |
-
print("\nGenerate:", trim_decode_tokens)
|
| 258 |
-
|
| 259 |
-
if return_end_reason:
|
| 260 |
-
return trim_decode_tokens, end_reason
|
| 261 |
-
else:
|
| 262 |
-
return trim_decode_tokens
|
| 263 |
-
|
| 264 |
-
|
| 265 |
-
def decode_tokens(
|
| 266 |
-
tokens: Union[torch.LongTensor, TokensType],
|
| 267 |
-
tokenizer: PreTrainedTokenizer,
|
| 268 |
-
raw_text_len: int,
|
| 269 |
-
context_length: int,
|
| 270 |
-
chat_format: str,
|
| 271 |
-
verbose: bool = False,
|
| 272 |
-
return_end_reason: bool = False,
|
| 273 |
-
errors: str="replace",
|
| 274 |
-
) -> str:
|
| 275 |
-
if torch.is_tensor(tokens):
|
| 276 |
-
tokens = tokens.cpu().numpy().tolist()
|
| 277 |
-
|
| 278 |
-
if chat_format == "chatml":
|
| 279 |
-
return _decode_chatml(
|
| 280 |
-
tokens,
|
| 281 |
-
stop_words=[],
|
| 282 |
-
eod_token_ids=[tokenizer.im_start_id, tokenizer.im_end_id],
|
| 283 |
-
tokenizer=tokenizer,
|
| 284 |
-
raw_text_len=raw_text_len,
|
| 285 |
-
context_length=context_length,
|
| 286 |
-
verbose=verbose,
|
| 287 |
-
return_end_reason=return_end_reason,
|
| 288 |
-
errors=errors,
|
| 289 |
-
)
|
| 290 |
-
elif chat_format == "raw":
|
| 291 |
-
return _decode_default(
|
| 292 |
-
tokens,
|
| 293 |
-
stop_words=["<|endoftext|>"],
|
| 294 |
-
eod_words=["<|endoftext|>"],
|
| 295 |
-
tokenizer=tokenizer,
|
| 296 |
-
raw_text_len=raw_text_len,
|
| 297 |
-
verbose=verbose,
|
| 298 |
-
return_end_reason=return_end_reason,
|
| 299 |
-
errors=errors,
|
| 300 |
-
)
|
| 301 |
-
else:
|
| 302 |
-
raise NotImplementedError(f"Unknown chat format {chat_format!r}")
|
| 303 |
-
|
| 304 |
-
|
| 305 |
-
class StopWordsLogitsProcessor(LogitsProcessor):
|
| 306 |
-
"""
|
| 307 |
-
:class:`transformers.LogitsProcessor` that enforces that when specified sequences appear, stop geration.
|
| 308 |
-
|
| 309 |
-
Args:
|
| 310 |
-
stop_words_ids (:obj:`List[List[int]]`):
|
| 311 |
-
List of list of token ids of stop ids. In order to get the tokens of the words
|
| 312 |
-
that should not appear in the generated text, use :obj:`tokenizer(bad_word,
|
| 313 |
-
add_prefix_space=True).input_ids`.
|
| 314 |
-
eos_token_id (:obj:`int`):
|
| 315 |
-
The id of the `end-of-sequence` token.
|
| 316 |
-
"""
|
| 317 |
-
|
| 318 |
-
def __init__(self, stop_words_ids: Iterable[Iterable[int]], eos_token_id: int):
|
| 319 |
-
|
| 320 |
-
if not isinstance(stop_words_ids, List) or len(stop_words_ids) == 0:
|
| 321 |
-
raise ValueError(
|
| 322 |
-
f"`stop_words_ids` has to be a non-emtpy list, but is {stop_words_ids}."
|
| 323 |
-
)
|
| 324 |
-
if any(not isinstance(bad_word_ids, list) for bad_word_ids in stop_words_ids):
|
| 325 |
-
raise ValueError(
|
| 326 |
-
f"`stop_words_ids` has to be a list of lists, but is {stop_words_ids}."
|
| 327 |
-
)
|
| 328 |
-
if any(
|
| 329 |
-
any(
|
| 330 |
-
(not isinstance(token_id, (int, np.integer)) or token_id < 0)
|
| 331 |
-
for token_id in stop_word_ids
|
| 332 |
-
)
|
| 333 |
-
for stop_word_ids in stop_words_ids
|
| 334 |
-
):
|
| 335 |
-
raise ValueError(
|
| 336 |
-
f"Each list in `stop_words_ids` has to be a list of positive integers, but is {stop_words_ids}."
|
| 337 |
-
)
|
| 338 |
-
|
| 339 |
-
self.stop_words_ids = list(
|
| 340 |
-
filter(
|
| 341 |
-
lambda bad_token_seq: bad_token_seq != [eos_token_id], stop_words_ids
|
| 342 |
-
)
|
| 343 |
-
)
|
| 344 |
-
self.eos_token_id = eos_token_id
|
| 345 |
-
for stop_token_seq in self.stop_words_ids:
|
| 346 |
-
assert (
|
| 347 |
-
len(stop_token_seq) > 0
|
| 348 |
-
), "Stop words token sequences {} cannot have an empty list".format(
|
| 349 |
-
stop_words_ids
|
| 350 |
-
)
|
| 351 |
-
|
| 352 |
-
def __call__(
|
| 353 |
-
self, input_ids: torch.LongTensor, scores: torch.FloatTensor
|
| 354 |
-
) -> torch.FloatTensor:
|
| 355 |
-
stopped_samples = self._calc_stopped_samples(input_ids)
|
| 356 |
-
for i, should_stop in enumerate(stopped_samples):
|
| 357 |
-
if should_stop:
|
| 358 |
-
scores[i, self.eos_token_id] = float(2**15)
|
| 359 |
-
return scores
|
| 360 |
-
|
| 361 |
-
def _tokens_match(self, prev_tokens: torch.LongTensor, tokens: List[int]) -> bool:
|
| 362 |
-
if len(tokens) == 0:
|
| 363 |
-
# if bad word tokens is just one token always ban it
|
| 364 |
-
return True
|
| 365 |
-
elif len(tokens) > len(prev_tokens):
|
| 366 |
-
# if bad word tokens are longer then prev input_ids they can't be equal
|
| 367 |
-
return False
|
| 368 |
-
elif prev_tokens[-len(tokens) :].tolist() == tokens:
|
| 369 |
-
# if tokens match
|
| 370 |
-
return True
|
| 371 |
-
else:
|
| 372 |
-
return False
|
| 373 |
-
|
| 374 |
-
def _calc_stopped_samples(self, prev_input_ids: Iterable[int]) -> Iterable[int]:
|
| 375 |
-
stopped_samples = []
|
| 376 |
-
for prev_input_ids_slice in prev_input_ids:
|
| 377 |
-
match = False
|
| 378 |
-
for stop_token_seq in self.stop_words_ids:
|
| 379 |
-
if self._tokens_match(prev_input_ids_slice, stop_token_seq):
|
| 380 |
-
# if tokens do not match continue
|
| 381 |
-
match = True
|
| 382 |
-
break
|
| 383 |
-
stopped_samples.append(match)
|
| 384 |
-
|
| 385 |
-
return stopped_samples
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
def top_k_logits(logits, top_k=0, top_p=0.0, filter_value=-float("Inf")):
|
| 389 |
-
"""This function has been mostly taken from huggingface conversational
|
| 390 |
-
ai code at
|
| 391 |
-
https://medium.com/huggingface/how-to-build-a-state-of-the-art-
|
| 392 |
-
conversational-ai-with-transfer-learning-2d818ac26313"""
|
| 393 |
-
|
| 394 |
-
if top_k > 0:
|
| 395 |
-
# Remove all tokens with a probability less than the
|
| 396 |
-
# last token of the top-k
|
| 397 |
-
indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
|
| 398 |
-
logits[indices_to_remove] = filter_value
|
| 399 |
-
|
| 400 |
-
if top_p > 0.0:
|
| 401 |
-
# Cconvert to 1D
|
| 402 |
-
sorted_logits, sorted_indices = torch.sort(logits, descending=True, dim=-1)
|
| 403 |
-
cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
|
| 404 |
-
|
| 405 |
-
# Remove tokens with cumulative probability above the threshold
|
| 406 |
-
sorted_indices_to_remove = cumulative_probs > top_p
|
| 407 |
-
# Shift the indices to the right to keep also the first token
|
| 408 |
-
# above the threshold
|
| 409 |
-
sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
|
| 410 |
-
sorted_indices_to_remove[..., 0] = 0
|
| 411 |
-
for i in range(sorted_indices.size(0)):
|
| 412 |
-
indices_to_remove = sorted_indices[i][sorted_indices_to_remove[i]]
|
| 413 |
-
logits[i][indices_to_remove] = filter_value
|
| 414 |
-
|
| 415 |
-
return logits
|
| 416 |
-
|
| 417 |
-
|
| 418 |
-
def switch(val1, val2, boolean):
|
| 419 |
-
boolean = boolean.type_as(val1)
|
| 420 |
-
return (1 - boolean) * val1 + boolean * val2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/special_tokens_map.json
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"pad_token": "<|endoftext|>"
|
| 3 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/tokenization_qwen.py
DELETED
|
@@ -1,598 +0,0 @@
|
|
| 1 |
-
# Copyright (c) Alibaba Cloud.
|
| 2 |
-
#
|
| 3 |
-
# This source code is licensed under the license found in the
|
| 4 |
-
# LICENSE file in the root directory of this source tree.
|
| 5 |
-
|
| 6 |
-
"""Tokenization classes for QWen."""
|
| 7 |
-
|
| 8 |
-
import base64
|
| 9 |
-
import logging
|
| 10 |
-
import os
|
| 11 |
-
import requests
|
| 12 |
-
import unicodedata
|
| 13 |
-
from typing import Collection, Dict, List, Set, Tuple, Union, Any, Callable, Optional
|
| 14 |
-
|
| 15 |
-
import tiktoken
|
| 16 |
-
import numpy as np
|
| 17 |
-
from PIL import Image
|
| 18 |
-
from PIL import ImageFont
|
| 19 |
-
from PIL import ImageDraw
|
| 20 |
-
from transformers import PreTrainedTokenizer, AddedToken
|
| 21 |
-
from transformers.utils import try_to_load_from_cache
|
| 22 |
-
|
| 23 |
-
import matplotlib.colors as mcolors
|
| 24 |
-
from matplotlib.font_manager import FontProperties
|
| 25 |
-
|
| 26 |
-
logger = logging.getLogger(__name__)
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
VOCAB_FILES_NAMES = {"vocab_file": "qwen.tiktoken", "ttf": "SimSun.ttf"}
|
| 30 |
-
FONT_PATH = try_to_load_from_cache("Qwen/Qwen-VL-Chat", "SimSun.ttf")
|
| 31 |
-
if FONT_PATH is None:
|
| 32 |
-
if not os.path.exists("SimSun.ttf"):
|
| 33 |
-
ttf = requests.get("https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/SimSun.ttf")
|
| 34 |
-
open("SimSun.ttf", "wb").write(ttf.content)
|
| 35 |
-
FONT_PATH = "SimSun.ttf"
|
| 36 |
-
|
| 37 |
-
PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
|
| 38 |
-
ENDOFTEXT = "<|endoftext|>"
|
| 39 |
-
IMSTART = "<|im_start|>"
|
| 40 |
-
IMEND = "<|im_end|>"
|
| 41 |
-
# as the default behavior is changed to allow special tokens in
|
| 42 |
-
# regular texts, the surface forms of special tokens need to be
|
| 43 |
-
# as different as possible to minimize the impact
|
| 44 |
-
EXTRAS = tuple((f"<|extra_{i}|>" for i in range(205)))
|
| 45 |
-
SPECIAL_TOKENS = (
|
| 46 |
-
ENDOFTEXT,
|
| 47 |
-
IMSTART,
|
| 48 |
-
IMEND,
|
| 49 |
-
) + EXTRAS
|
| 50 |
-
IMG_TOKEN_SPAN = 256
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
def _load_tiktoken_bpe(tiktoken_bpe_file: str) -> Dict[bytes, int]:
|
| 54 |
-
with open(tiktoken_bpe_file, "rb") as f:
|
| 55 |
-
contents = f.read()
|
| 56 |
-
return {
|
| 57 |
-
base64.b64decode(token): int(rank)
|
| 58 |
-
for token, rank in (line.split() for line in contents.splitlines() if line)
|
| 59 |
-
}
|
| 60 |
-
|
| 61 |
-
def _list_find(
|
| 62 |
-
input_list: List[Any],
|
| 63 |
-
candidates: Tuple[Any],
|
| 64 |
-
start: int = 0,
|
| 65 |
-
):
|
| 66 |
-
for i in range(start, len(input_list)):
|
| 67 |
-
if input_list[i] in candidates:
|
| 68 |
-
return i
|
| 69 |
-
return -1
|
| 70 |
-
|
| 71 |
-
def _replace_closed_tag(
|
| 72 |
-
input_tokens: List[Any],
|
| 73 |
-
start_tags: Union[Any, Tuple[Any]],
|
| 74 |
-
end_tags: Union[Any, Tuple[Any]],
|
| 75 |
-
inclusive_replace_func: Callable,
|
| 76 |
-
exclusive_replace_func: Callable = lambda x: x,
|
| 77 |
-
):
|
| 78 |
-
if isinstance(start_tags, (str, int)):
|
| 79 |
-
start_tags = (start_tags,)
|
| 80 |
-
if isinstance(end_tags, (str, int)):
|
| 81 |
-
end_tags = (end_tags,)
|
| 82 |
-
assert len(start_tags) == len(end_tags)
|
| 83 |
-
|
| 84 |
-
output_tokens = []
|
| 85 |
-
end = 0
|
| 86 |
-
while True:
|
| 87 |
-
start = _list_find(input_tokens, start_tags, end)
|
| 88 |
-
if start == -1:
|
| 89 |
-
break
|
| 90 |
-
output_tokens.extend(exclusive_replace_func(input_tokens[end : start]))
|
| 91 |
-
tag_idx = start_tags.index(input_tokens[start])
|
| 92 |
-
end = _list_find(input_tokens, (end_tags[tag_idx],), start)
|
| 93 |
-
if end == -1:
|
| 94 |
-
raise ValueError("Unclosed image token")
|
| 95 |
-
output_tokens.extend(inclusive_replace_func(input_tokens[start : end + 1]))
|
| 96 |
-
end += 1
|
| 97 |
-
output_tokens.extend(exclusive_replace_func(input_tokens[end : ]))
|
| 98 |
-
return output_tokens
|
| 99 |
-
|
| 100 |
-
class QWenTokenizer(PreTrainedTokenizer):
|
| 101 |
-
"""QWen tokenizer."""
|
| 102 |
-
|
| 103 |
-
vocab_files_names = VOCAB_FILES_NAMES
|
| 104 |
-
|
| 105 |
-
def __init__(
|
| 106 |
-
self,
|
| 107 |
-
vocab_file,
|
| 108 |
-
errors="replace",
|
| 109 |
-
image_start_tag='<img>',
|
| 110 |
-
image_end_tag='</img>',
|
| 111 |
-
image_pad_tag='<imgpad>',
|
| 112 |
-
ref_start_tag='<ref>',
|
| 113 |
-
ref_end_tag='</ref>',
|
| 114 |
-
box_start_tag='<box>',
|
| 115 |
-
box_end_tag='</box>',
|
| 116 |
-
quad_start_tag='<quad>',
|
| 117 |
-
quad_end_tag='</quad>',
|
| 118 |
-
**kwargs,
|
| 119 |
-
):
|
| 120 |
-
super().__init__(**kwargs)
|
| 121 |
-
self.image_start_tag = image_start_tag
|
| 122 |
-
self.image_end_tag = image_end_tag
|
| 123 |
-
self.image_pad_tag = image_pad_tag
|
| 124 |
-
self.ref_start_tag = ref_start_tag
|
| 125 |
-
self.ref_end_tag = ref_end_tag
|
| 126 |
-
self.box_start_tag = box_start_tag
|
| 127 |
-
self.box_end_tag = box_end_tag
|
| 128 |
-
self.quad_start_tag = quad_start_tag
|
| 129 |
-
self.quad_end_tag = quad_end_tag
|
| 130 |
-
self.IMAGE_ST = (
|
| 131 |
-
ref_start_tag, ref_end_tag,
|
| 132 |
-
box_start_tag, box_end_tag,
|
| 133 |
-
quad_start_tag, quad_end_tag,
|
| 134 |
-
image_start_tag, image_end_tag,
|
| 135 |
-
image_pad_tag
|
| 136 |
-
)
|
| 137 |
-
|
| 138 |
-
self.errors = errors # how to handle errors in decoding
|
| 139 |
-
|
| 140 |
-
self.mergeable_ranks = _load_tiktoken_bpe(vocab_file) # type: dict[bytes, int]
|
| 141 |
-
self.special_tokens = {
|
| 142 |
-
token: index
|
| 143 |
-
for index, token in enumerate(
|
| 144 |
-
SPECIAL_TOKENS + self.IMAGE_ST, start=len(self.mergeable_ranks)
|
| 145 |
-
)
|
| 146 |
-
}
|
| 147 |
-
self.img_start_id = self.special_tokens[self.image_start_tag]
|
| 148 |
-
self.img_end_id = self.special_tokens[self.image_end_tag]
|
| 149 |
-
self.img_pad_id = self.special_tokens[self.image_pad_tag]
|
| 150 |
-
self.ref_start_id = self.special_tokens[self.ref_start_tag]
|
| 151 |
-
self.ref_end_id = self.special_tokens[self.ref_end_tag]
|
| 152 |
-
self.box_start_id = self.special_tokens[self.box_start_tag]
|
| 153 |
-
self.box_end_id = self.special_tokens[self.box_end_tag]
|
| 154 |
-
self.quad_start_id = self.special_tokens[self.quad_start_tag]
|
| 155 |
-
self.quad_end_id = self.special_tokens[self.quad_end_tag]
|
| 156 |
-
self.image_special_tokens = set([
|
| 157 |
-
self.ref_start_id, self.ref_end_id, self.box_start_id, self.box_end_id,
|
| 158 |
-
self.quad_start_id, self.quad_end_id,
|
| 159 |
-
])
|
| 160 |
-
|
| 161 |
-
enc = tiktoken.Encoding(
|
| 162 |
-
"Qwen",
|
| 163 |
-
pat_str=PAT_STR,
|
| 164 |
-
mergeable_ranks=self.mergeable_ranks,
|
| 165 |
-
special_tokens=self.special_tokens,
|
| 166 |
-
)
|
| 167 |
-
assert (
|
| 168 |
-
len(self.mergeable_ranks) + len(self.special_tokens) == enc.n_vocab
|
| 169 |
-
), f"{len(self.mergeable_ranks) + len(self.special_tokens)} != {enc.n_vocab} in encoding"
|
| 170 |
-
|
| 171 |
-
self.decoder = {
|
| 172 |
-
v: k for k, v in self.mergeable_ranks.items()
|
| 173 |
-
} # type: dict[int, bytes|str]
|
| 174 |
-
self.decoder.update({v: k for k, v in self.special_tokens.items()})
|
| 175 |
-
|
| 176 |
-
self.tokenizer = enc # type: tiktoken.Encoding
|
| 177 |
-
|
| 178 |
-
self.eod_id = self.tokenizer.eot_token
|
| 179 |
-
self.im_start_id = self.special_tokens[IMSTART]
|
| 180 |
-
self.im_end_id = self.special_tokens[IMEND]
|
| 181 |
-
|
| 182 |
-
def __getstate__(self):
|
| 183 |
-
# for pickle lovers
|
| 184 |
-
state = self.__dict__.copy()
|
| 185 |
-
del state['tokenizer']
|
| 186 |
-
return state
|
| 187 |
-
|
| 188 |
-
def __setstate__(self, state):
|
| 189 |
-
# tokenizer is not python native; don't pass it; rebuild it
|
| 190 |
-
self.__dict__.update(state)
|
| 191 |
-
enc = tiktoken.Encoding(
|
| 192 |
-
"Qwen",
|
| 193 |
-
pat_str=PAT_STR,
|
| 194 |
-
mergeable_ranks=self.mergeable_ranks,
|
| 195 |
-
special_tokens=self.special_tokens,
|
| 196 |
-
)
|
| 197 |
-
self.tokenizer = enc
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
def __len__(self) -> int:
|
| 201 |
-
return self.tokenizer.n_vocab
|
| 202 |
-
|
| 203 |
-
def get_vocab(self) -> Dict[bytes, int]:
|
| 204 |
-
return self.mergeable_ranks
|
| 205 |
-
|
| 206 |
-
def convert_tokens_to_ids(
|
| 207 |
-
self, tokens: Union[bytes, str, List[Union[bytes, str]]]
|
| 208 |
-
) -> List[int]:
|
| 209 |
-
ids = []
|
| 210 |
-
if isinstance(tokens, (str, bytes)):
|
| 211 |
-
if tokens in self.special_tokens:
|
| 212 |
-
return self.special_tokens[tokens]
|
| 213 |
-
else:
|
| 214 |
-
return self.mergeable_ranks.get(tokens)
|
| 215 |
-
for token in tokens:
|
| 216 |
-
if token in self.special_tokens:
|
| 217 |
-
ids.append(self.special_tokens[token])
|
| 218 |
-
else:
|
| 219 |
-
ids.append(self.mergeable_ranks.get(token))
|
| 220 |
-
return ids
|
| 221 |
-
|
| 222 |
-
def _add_tokens(self, new_tokens: Union[List[str], List[AddedToken]], special_tokens: bool = False) -> int:
|
| 223 |
-
if not special_tokens and new_tokens:
|
| 224 |
-
raise ValueError('Adding regular tokens is not supported')
|
| 225 |
-
for token in new_tokens:
|
| 226 |
-
surface_form = token.content if isinstance(token, AddedToken) else token
|
| 227 |
-
if surface_form not in SPECIAL_TOKENS + self.IMAGE_ST:
|
| 228 |
-
raise ValueError('Adding unknown special tokens is not supported')
|
| 229 |
-
return 0
|
| 230 |
-
|
| 231 |
-
def save_vocabulary(self, save_directory: str, **kwargs) -> Tuple[str]:
|
| 232 |
-
"""
|
| 233 |
-
Save only the vocabulary of the tokenizer (vocabulary).
|
| 234 |
-
|
| 235 |
-
Returns:
|
| 236 |
-
`Tuple(str)`: Paths to the files saved.
|
| 237 |
-
"""
|
| 238 |
-
file_path = os.path.join(save_directory, "qwen.tiktoken")
|
| 239 |
-
with open(file_path, "w", encoding="utf8") as w:
|
| 240 |
-
for k, v in self.mergeable_ranks.items():
|
| 241 |
-
line = base64.b64encode(k).decode("utf8") + " " + str(v) + "\n"
|
| 242 |
-
w.write(line)
|
| 243 |
-
return (file_path,)
|
| 244 |
-
|
| 245 |
-
def tokenize(
|
| 246 |
-
self,
|
| 247 |
-
text: str,
|
| 248 |
-
allowed_special: Union[Set, str] = "all",
|
| 249 |
-
disallowed_special: Union[Collection, str] = (),
|
| 250 |
-
**kwargs,
|
| 251 |
-
) -> List[Union[bytes, str]]:
|
| 252 |
-
"""
|
| 253 |
-
Converts a string in a sequence of tokens.
|
| 254 |
-
|
| 255 |
-
Args:
|
| 256 |
-
text (`str`):
|
| 257 |
-
The sequence to be encoded.
|
| 258 |
-
allowed_special (`Literal["all"]` or `set`):
|
| 259 |
-
The surface forms of the tokens to be encoded as special tokens in regular texts.
|
| 260 |
-
Default to "all".
|
| 261 |
-
disallowed_special (`Literal["all"]` or `Collection`):
|
| 262 |
-
The surface forms of the tokens that should not be in regular texts and trigger errors.
|
| 263 |
-
Default to an empty tuple.
|
| 264 |
-
|
| 265 |
-
kwargs (additional keyword arguments, *optional*):
|
| 266 |
-
Will be passed to the underlying model specific encode method.
|
| 267 |
-
|
| 268 |
-
Returns:
|
| 269 |
-
`List[bytes|str]`: The list of tokens.
|
| 270 |
-
"""
|
| 271 |
-
tokens = []
|
| 272 |
-
text = unicodedata.normalize("NFC", text)
|
| 273 |
-
|
| 274 |
-
# this implementation takes a detour: text -> token id -> token surface forms
|
| 275 |
-
for t in self.tokenizer.encode(
|
| 276 |
-
text, allowed_special=allowed_special, disallowed_special=disallowed_special
|
| 277 |
-
):
|
| 278 |
-
tokens.append(self.decoder[t])
|
| 279 |
-
|
| 280 |
-
def _encode_imgurl(img_tokens):
|
| 281 |
-
assert img_tokens[0] == self.image_start_tag and img_tokens[-1] == self.image_end_tag
|
| 282 |
-
img_tokens = img_tokens[1:-1]
|
| 283 |
-
img_url = b''.join(img_tokens)
|
| 284 |
-
out_img_tokens = list(map(self.decoder.get, img_url))
|
| 285 |
-
if len(out_img_tokens) > IMG_TOKEN_SPAN:
|
| 286 |
-
raise ValueError("The content in {}..{} is too long".format(
|
| 287 |
-
self.image_start_tag, self.image_end_tag))
|
| 288 |
-
out_img_tokens.extend([self.image_pad_tag] * (IMG_TOKEN_SPAN - len(out_img_tokens)))
|
| 289 |
-
out_img_tokens = [self.image_start_tag] + out_img_tokens + [self.image_end_tag]
|
| 290 |
-
return out_img_tokens
|
| 291 |
-
|
| 292 |
-
return _replace_closed_tag(tokens, self.image_start_tag, self.image_end_tag, _encode_imgurl)
|
| 293 |
-
|
| 294 |
-
def convert_tokens_to_string(self, tokens: List[Union[bytes, str]]) -> str:
|
| 295 |
-
"""
|
| 296 |
-
Converts a sequence of tokens in a single string.
|
| 297 |
-
"""
|
| 298 |
-
text = ""
|
| 299 |
-
temp = b""
|
| 300 |
-
for t in tokens:
|
| 301 |
-
if isinstance(t, str):
|
| 302 |
-
if temp:
|
| 303 |
-
text += temp.decode("utf-8", errors=self.errors)
|
| 304 |
-
temp = b""
|
| 305 |
-
text += t
|
| 306 |
-
elif isinstance(t, bytes):
|
| 307 |
-
temp += t
|
| 308 |
-
else:
|
| 309 |
-
raise TypeError("token should only be of type types or str")
|
| 310 |
-
if temp:
|
| 311 |
-
text += temp.decode("utf-8", errors=self.errors)
|
| 312 |
-
return text
|
| 313 |
-
|
| 314 |
-
@property
|
| 315 |
-
def vocab_size(self):
|
| 316 |
-
return self.tokenizer.n_vocab
|
| 317 |
-
|
| 318 |
-
def _convert_id_to_token(self, index: int) -> Union[bytes, str]:
|
| 319 |
-
"""Converts an id to a token, special tokens included"""
|
| 320 |
-
if index in self.decoder:
|
| 321 |
-
return self.decoder[index]
|
| 322 |
-
raise ValueError("unknown ids")
|
| 323 |
-
|
| 324 |
-
def _convert_token_to_id(self, token: Union[bytes, str]) -> int:
|
| 325 |
-
"""Converts a token to an id using the vocab, special tokens included"""
|
| 326 |
-
if token in self.special_tokens:
|
| 327 |
-
return self.special_tokens[token]
|
| 328 |
-
if token in self.mergeable_ranks:
|
| 329 |
-
return self.mergeable_ranks[token]
|
| 330 |
-
raise ValueError("unknown token")
|
| 331 |
-
|
| 332 |
-
def _tokenize(self, text: str, **kwargs):
|
| 333 |
-
"""
|
| 334 |
-
Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
|
| 335 |
-
vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
|
| 336 |
-
|
| 337 |
-
Do NOT take care of added tokens.
|
| 338 |
-
"""
|
| 339 |
-
raise NotImplementedError
|
| 340 |
-
|
| 341 |
-
def _decode(
|
| 342 |
-
self,
|
| 343 |
-
token_ids: Union[int, List[int]],
|
| 344 |
-
skip_special_tokens: bool = False,
|
| 345 |
-
errors: str = None,
|
| 346 |
-
**kwargs,
|
| 347 |
-
) -> str:
|
| 348 |
-
if isinstance(token_ids, int):
|
| 349 |
-
token_ids = [token_ids]
|
| 350 |
-
|
| 351 |
-
def _decode_imgurl(img_token_ids):
|
| 352 |
-
assert img_token_ids[0] == self.img_start_id and img_token_ids[-1] == self.img_end_id
|
| 353 |
-
img_token_ids = img_token_ids[1:-1]
|
| 354 |
-
img_token_ids = img_token_ids[ : img_token_ids.index(self.img_pad_id)]
|
| 355 |
-
img_url = bytes(img_token_ids).decode('utf-8')
|
| 356 |
-
return [self.img_start_id] + self.tokenizer.encode(img_url) + [self.img_end_id]
|
| 357 |
-
|
| 358 |
-
token_ids = _replace_closed_tag(token_ids, self.img_start_id, self.img_end_id, _decode_imgurl)
|
| 359 |
-
|
| 360 |
-
if skip_special_tokens:
|
| 361 |
-
if kwargs.get('keep_image_special', False):
|
| 362 |
-
token_ids = [i for i in token_ids if i < self.eod_id
|
| 363 |
-
or i in self.image_special_tokens]
|
| 364 |
-
else:
|
| 365 |
-
token_ids = [i for i in token_ids if i < self.eod_id]
|
| 366 |
-
return self.tokenizer.decode(token_ids, errors=errors or self.errors)
|
| 367 |
-
|
| 368 |
-
def to_list_format(self, text: str):
|
| 369 |
-
text = unicodedata.normalize("NFC", text)
|
| 370 |
-
token_ids = self.tokenizer.encode(
|
| 371 |
-
text, allowed_special=set(self.IMAGE_ST + (ENDOFTEXT,)))
|
| 372 |
-
|
| 373 |
-
def _encode_vl_info(tokens):
|
| 374 |
-
if len(tokens) == 0:
|
| 375 |
-
return []
|
| 376 |
-
if tokens[0] == self.img_start_id and tokens[-1] == self.img_end_id:
|
| 377 |
-
key = 'image'
|
| 378 |
-
elif tokens[0] == self.ref_start_id and tokens[-1] == self.ref_end_id:
|
| 379 |
-
key = 'ref'
|
| 380 |
-
elif tokens[0] == self.box_start_id and tokens[-1] == self.box_end_id:
|
| 381 |
-
key = 'box'
|
| 382 |
-
elif tokens[0] == self.quad_start_id and tokens[-1] == self.quad_end_id:
|
| 383 |
-
key = 'quad'
|
| 384 |
-
else:
|
| 385 |
-
_tobytes = lambda x: x.encode('utf-8') if isinstance(x, str) else x
|
| 386 |
-
return [{'text': b''.join(map(_tobytes, map(self.decoder.get, tokens))).decode('utf-8')}]
|
| 387 |
-
_tobytes = lambda x: x.encode('utf-8') if isinstance(x, str) else x
|
| 388 |
-
val = b''.join(map(_tobytes, map(self.decoder.get, tokens[1:-1]))).decode('utf-8')
|
| 389 |
-
return [{key: val}]
|
| 390 |
-
|
| 391 |
-
return _replace_closed_tag(
|
| 392 |
-
token_ids,
|
| 393 |
-
(self.img_start_id, self.ref_start_id, self.box_start_id, self.quad_start_id),
|
| 394 |
-
(self.img_end_id, self.ref_end_id, self.box_end_id, self.quad_end_id),
|
| 395 |
-
_encode_vl_info,
|
| 396 |
-
_encode_vl_info,
|
| 397 |
-
)
|
| 398 |
-
|
| 399 |
-
def from_list_format(self, list_format: List[Dict]):
|
| 400 |
-
text = ''
|
| 401 |
-
num_images = 0
|
| 402 |
-
for ele in list_format:
|
| 403 |
-
if 'image' in ele:
|
| 404 |
-
num_images += 1
|
| 405 |
-
text += f'Picture {num_images}: '
|
| 406 |
-
text += self.image_start_tag + ele['image'] + self.image_end_tag
|
| 407 |
-
text += '\n'
|
| 408 |
-
elif 'text' in ele:
|
| 409 |
-
text += ele['text']
|
| 410 |
-
elif 'box' in ele:
|
| 411 |
-
if 'ref' in ele:
|
| 412 |
-
text += self.ref_start_tag + ele['ref'] + self.ref_end_tag
|
| 413 |
-
for box in ele['box']:
|
| 414 |
-
text += self.box_start_tag + '(%d,%d),(%d,%d)' % (box[0], box[1], box[2], box[3]) + self.box_end_tag
|
| 415 |
-
else:
|
| 416 |
-
raise ValueError("Unsupport element: " + str(ele))
|
| 417 |
-
return text
|
| 418 |
-
|
| 419 |
-
def _fetch_latest_picture(self, response, history):
|
| 420 |
-
if history is None:
|
| 421 |
-
history = []
|
| 422 |
-
_history = history + [(response, None)]
|
| 423 |
-
for q, r in _history[::-1]:
|
| 424 |
-
for ele in self.to_list_format(q)[::-1]:
|
| 425 |
-
if 'image' in ele:
|
| 426 |
-
return ele['image']
|
| 427 |
-
return None
|
| 428 |
-
|
| 429 |
-
def _fetch_all_box_with_ref(self, text):
|
| 430 |
-
list_format = self.to_list_format(text)
|
| 431 |
-
output = []
|
| 432 |
-
for i, ele in enumerate(list_format):
|
| 433 |
-
if 'box' in ele:
|
| 434 |
-
bbox = tuple(map(int, ele['box'].replace('(', '').replace(')', '').split(',')))
|
| 435 |
-
assert len(bbox) == 4
|
| 436 |
-
output.append({'box': bbox})
|
| 437 |
-
if i > 0 and 'ref' in list_format[i-1]:
|
| 438 |
-
output[-1]['ref'] = list_format[i-1]['ref'].strip()
|
| 439 |
-
return output
|
| 440 |
-
|
| 441 |
-
def draw_bbox_on_latest_picture(
|
| 442 |
-
self,
|
| 443 |
-
response,
|
| 444 |
-
history=None,
|
| 445 |
-
) -> Optional[Image.Image]:
|
| 446 |
-
image = self._fetch_latest_picture(response, history)
|
| 447 |
-
if image is None:
|
| 448 |
-
return None
|
| 449 |
-
if image.startswith("http://") or image.startswith("https://"):
|
| 450 |
-
image = Image.open(requests.get(image, stream=True).raw).convert("RGB")
|
| 451 |
-
h, w = image.height, image.width
|
| 452 |
-
else:
|
| 453 |
-
image = np.asarray(Image.open(image).convert("RGB"))
|
| 454 |
-
h, w = image.shape[0], image.shape[1]
|
| 455 |
-
visualizer = Visualizer(image)
|
| 456 |
-
|
| 457 |
-
boxes = self._fetch_all_box_with_ref(response)
|
| 458 |
-
if not boxes:
|
| 459 |
-
return None
|
| 460 |
-
color = random.choice([_ for _ in mcolors.TABLEAU_COLORS.keys()]) # init color
|
| 461 |
-
for box in boxes:
|
| 462 |
-
if 'ref' in box: # random new color for new refexps
|
| 463 |
-
color = random.choice([_ for _ in mcolors.TABLEAU_COLORS.keys()])
|
| 464 |
-
x1, y1, x2, y2 = box['box']
|
| 465 |
-
x1, y1, x2, y2 = (int(x1 / 1000 * w), int(y1 / 1000 * h), int(x2 / 1000 * w), int(y2 / 1000 * h))
|
| 466 |
-
visualizer.draw_box((x1, y1, x2, y2), alpha=1, edge_color=color)
|
| 467 |
-
if 'ref' in box:
|
| 468 |
-
visualizer.draw_text(box['ref'], (x1, y1), color=color, horizontal_alignment="left")
|
| 469 |
-
return visualizer.output
|
| 470 |
-
|
| 471 |
-
|
| 472 |
-
import colorsys
|
| 473 |
-
import logging
|
| 474 |
-
import math
|
| 475 |
-
import numpy as np
|
| 476 |
-
import matplotlib as mpl
|
| 477 |
-
import matplotlib.colors as mplc
|
| 478 |
-
import matplotlib.figure as mplfigure
|
| 479 |
-
import torch
|
| 480 |
-
from matplotlib.backends.backend_agg import FigureCanvasAgg
|
| 481 |
-
from PIL import Image
|
| 482 |
-
import random
|
| 483 |
-
|
| 484 |
-
logger = logging.getLogger(__name__)
|
| 485 |
-
|
| 486 |
-
|
| 487 |
-
class VisImage:
|
| 488 |
-
def __init__(self, img, scale=1.0):
|
| 489 |
-
self.img = img
|
| 490 |
-
self.scale = scale
|
| 491 |
-
self.width, self.height = img.shape[1], img.shape[0]
|
| 492 |
-
self._setup_figure(img)
|
| 493 |
-
|
| 494 |
-
def _setup_figure(self, img):
|
| 495 |
-
fig = mplfigure.Figure(frameon=False)
|
| 496 |
-
self.dpi = fig.get_dpi()
|
| 497 |
-
# add a small 1e-2 to avoid precision lost due to matplotlib's truncation
|
| 498 |
-
# (https://github.com/matplotlib/matplotlib/issues/15363)
|
| 499 |
-
fig.set_size_inches(
|
| 500 |
-
(self.width * self.scale + 1e-2) / self.dpi,
|
| 501 |
-
(self.height * self.scale + 1e-2) / self.dpi,
|
| 502 |
-
)
|
| 503 |
-
self.canvas = FigureCanvasAgg(fig)
|
| 504 |
-
# self.canvas = mpl.backends.backend_cairo.FigureCanvasCairo(fig)
|
| 505 |
-
ax = fig.add_axes([0.0, 0.0, 1.0, 1.0])
|
| 506 |
-
ax.axis("off")
|
| 507 |
-
self.fig = fig
|
| 508 |
-
self.ax = ax
|
| 509 |
-
self.reset_image(img)
|
| 510 |
-
|
| 511 |
-
def reset_image(self, img):
|
| 512 |
-
img = img.astype("uint8")
|
| 513 |
-
self.ax.imshow(img, extent=(0, self.width, self.height, 0), interpolation="nearest")
|
| 514 |
-
|
| 515 |
-
def save(self, filepath):
|
| 516 |
-
self.fig.savefig(filepath)
|
| 517 |
-
|
| 518 |
-
def get_image(self):
|
| 519 |
-
canvas = self.canvas
|
| 520 |
-
s, (width, height) = canvas.print_to_buffer()
|
| 521 |
-
|
| 522 |
-
buffer = np.frombuffer(s, dtype="uint8")
|
| 523 |
-
|
| 524 |
-
img_rgba = buffer.reshape(height, width, 4)
|
| 525 |
-
rgb, alpha = np.split(img_rgba, [3], axis=2)
|
| 526 |
-
return rgb.astype("uint8")
|
| 527 |
-
|
| 528 |
-
|
| 529 |
-
class Visualizer:
|
| 530 |
-
def __init__(self, img_rgb, metadata=None, scale=1.0):
|
| 531 |
-
self.img = np.asarray(img_rgb).clip(0, 255).astype(np.uint8)
|
| 532 |
-
self.font_path = FONT_PATH
|
| 533 |
-
self.output = VisImage(self.img, scale=scale)
|
| 534 |
-
self.cpu_device = torch.device("cpu")
|
| 535 |
-
|
| 536 |
-
# too small texts are useless, therefore clamp to 14
|
| 537 |
-
self._default_font_size = max(
|
| 538 |
-
np.sqrt(self.output.height * self.output.width) // 30, 15 // scale
|
| 539 |
-
)
|
| 540 |
-
|
| 541 |
-
def draw_text(
|
| 542 |
-
self,
|
| 543 |
-
text,
|
| 544 |
-
position,
|
| 545 |
-
*,
|
| 546 |
-
font_size=None,
|
| 547 |
-
color="g",
|
| 548 |
-
horizontal_alignment="center",
|
| 549 |
-
rotation=0,
|
| 550 |
-
):
|
| 551 |
-
if not font_size:
|
| 552 |
-
font_size = self._default_font_size
|
| 553 |
-
|
| 554 |
-
# since the text background is dark, we don't want the text to be dark
|
| 555 |
-
color = np.maximum(list(mplc.to_rgb(color)), 0.2)
|
| 556 |
-
color[np.argmax(color)] = max(0.8, np.max(color))
|
| 557 |
-
|
| 558 |
-
x, y = position
|
| 559 |
-
self.output.ax.text(
|
| 560 |
-
x,
|
| 561 |
-
y,
|
| 562 |
-
text,
|
| 563 |
-
size=font_size * self.output.scale,
|
| 564 |
-
fontproperties=FontProperties(fname=self.font_path),
|
| 565 |
-
bbox={"facecolor": "black", "alpha": 0.8, "pad": 0.7, "edgecolor": "none"},
|
| 566 |
-
verticalalignment="top",
|
| 567 |
-
horizontalalignment=horizontal_alignment,
|
| 568 |
-
color=color,
|
| 569 |
-
zorder=10,
|
| 570 |
-
rotation=rotation,
|
| 571 |
-
)
|
| 572 |
-
return self.output
|
| 573 |
-
|
| 574 |
-
def draw_box(self, box_coord, alpha=0.5, edge_color="g", line_style="-"):
|
| 575 |
-
|
| 576 |
-
x0, y0, x1, y1 = box_coord
|
| 577 |
-
width = x1 - x0
|
| 578 |
-
height = y1 - y0
|
| 579 |
-
|
| 580 |
-
linewidth = max(self._default_font_size / 4, 1)
|
| 581 |
-
|
| 582 |
-
self.output.ax.add_patch(
|
| 583 |
-
mpl.patches.Rectangle(
|
| 584 |
-
(x0, y0),
|
| 585 |
-
width,
|
| 586 |
-
height,
|
| 587 |
-
fill=False,
|
| 588 |
-
edgecolor=edge_color,
|
| 589 |
-
linewidth=linewidth * self.output.scale,
|
| 590 |
-
alpha=alpha,
|
| 591 |
-
linestyle=line_style,
|
| 592 |
-
)
|
| 593 |
-
)
|
| 594 |
-
return self.output
|
| 595 |
-
|
| 596 |
-
def get_output(self):
|
| 597 |
-
|
| 598 |
-
return self.output
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/tokenizer_config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"added_tokens_decoder": {},
|
| 3 |
-
"auto_map": {
|
| 4 |
-
"AutoTokenizer": [
|
| 5 |
-
"tokenization_qwen.QWenTokenizer",
|
| 6 |
-
null
|
| 7 |
-
]
|
| 8 |
-
},
|
| 9 |
-
"clean_up_tokenization_spaces": true,
|
| 10 |
-
"model_max_length": 2048,
|
| 11 |
-
"pad_token": "<|endoftext|>",
|
| 12 |
-
"padding_side": "right",
|
| 13 |
-
"tokenizer_class": "QWenTokenizer"
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/trainer_state.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
zzzmmz/SegAgent-Model/visual.py
DELETED
|
@@ -1,472 +0,0 @@
|
|
| 1 |
-
# Copyright (c) Alibaba Cloud.
|
| 2 |
-
#
|
| 3 |
-
# This source code is licensed under the license found in the
|
| 4 |
-
# LICENSE file in the root directory of this source tree.
|
| 5 |
-
|
| 6 |
-
from collections import OrderedDict
|
| 7 |
-
import math
|
| 8 |
-
import requests
|
| 9 |
-
from io import BytesIO
|
| 10 |
-
from functools import partial
|
| 11 |
-
from PIL import Image,ImageOps
|
| 12 |
-
from typing import Callable, Optional, Sequence, Tuple, List
|
| 13 |
-
import numpy as np
|
| 14 |
-
|
| 15 |
-
import torch
|
| 16 |
-
from torch import nn
|
| 17 |
-
from torch.nn import functional as F
|
| 18 |
-
from torch.nn.init import normal_
|
| 19 |
-
from torchvision import transforms
|
| 20 |
-
from torchvision.transforms import InterpolationMode
|
| 21 |
-
from pycocotools import mask as maskUtils
|
| 22 |
-
import os
|
| 23 |
-
import torchshow
|
| 24 |
-
def overlay_mask(image, mask, color=(0, 255, 0), alpha=0.5):
|
| 25 |
-
overlay = image.copy()
|
| 26 |
-
binary_mask = mask
|
| 27 |
-
color_mask = np.zeros_like(image)
|
| 28 |
-
color_mask[binary_mask > 0] = color
|
| 29 |
-
for c in range(0, 3):
|
| 30 |
-
overlay[:, :, c] = np.where(binary_mask > 0,
|
| 31 |
-
overlay[:, :, c] * (1 - alpha) + color_mask[:, :, c] * alpha,
|
| 32 |
-
overlay[:, :, c])
|
| 33 |
-
return overlay
|
| 34 |
-
|
| 35 |
-
def get_abs_pos(abs_pos, tgt_size):
|
| 36 |
-
# abs_pos: L, C
|
| 37 |
-
# tgt_size: M
|
| 38 |
-
# return: M, C
|
| 39 |
-
src_size = int(math.sqrt(abs_pos.size(0)))
|
| 40 |
-
tgt_size = int(math.sqrt(tgt_size))
|
| 41 |
-
dtype = abs_pos.dtype
|
| 42 |
-
|
| 43 |
-
if src_size != tgt_size:
|
| 44 |
-
return F.interpolate(
|
| 45 |
-
abs_pos.float().reshape(1, src_size, src_size, -1).permute(0, 3, 1, 2),
|
| 46 |
-
size=(tgt_size, tgt_size),
|
| 47 |
-
mode="bicubic",
|
| 48 |
-
align_corners=False,
|
| 49 |
-
).permute(0, 2, 3, 1).flatten(0, 2).to(dtype=dtype)
|
| 50 |
-
else:
|
| 51 |
-
return abs_pos
|
| 52 |
-
|
| 53 |
-
# https://github.com/facebookresearch/mae/blob/efb2a8062c206524e35e47d04501ed4f544c0ae8/util/pos_embed.py#L20
|
| 54 |
-
def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False):
|
| 55 |
-
"""
|
| 56 |
-
grid_size: int of the grid height and width
|
| 57 |
-
return:
|
| 58 |
-
pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/ or w/o cls_token)
|
| 59 |
-
"""
|
| 60 |
-
grid_h = np.arange(grid_size, dtype=np.float32)
|
| 61 |
-
grid_w = np.arange(grid_size, dtype=np.float32)
|
| 62 |
-
grid = np.meshgrid(grid_w, grid_h) # here w goes first
|
| 63 |
-
grid = np.stack(grid, axis=0)
|
| 64 |
-
|
| 65 |
-
grid = grid.reshape([2, 1, grid_size, grid_size])
|
| 66 |
-
pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
|
| 67 |
-
if cls_token:
|
| 68 |
-
pos_embed = np.concatenate([np.zeros([1, embed_dim]), pos_embed], axis=0)
|
| 69 |
-
return pos_embed
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
def get_2d_sincos_pos_embed_from_grid(embed_dim, grid):
|
| 73 |
-
assert embed_dim % 2 == 0
|
| 74 |
-
|
| 75 |
-
# use half of dimensions to encode grid_h
|
| 76 |
-
emb_h = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0]) # (H*W, D/2)
|
| 77 |
-
emb_w = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1]) # (H*W, D/2)
|
| 78 |
-
|
| 79 |
-
emb = np.concatenate([emb_h, emb_w], axis=1) # (H*W, D)
|
| 80 |
-
return emb
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
def get_1d_sincos_pos_embed_from_grid(embed_dim, pos):
|
| 84 |
-
"""
|
| 85 |
-
embed_dim: output dimension for each position
|
| 86 |
-
pos: a list of positions to be encoded: size (M,)
|
| 87 |
-
out: (M, D)
|
| 88 |
-
"""
|
| 89 |
-
assert embed_dim % 2 == 0
|
| 90 |
-
omega = np.arange(embed_dim // 2, dtype=np.float32)
|
| 91 |
-
omega /= embed_dim / 2.
|
| 92 |
-
omega = 1. / 10000**omega # (D/2,)
|
| 93 |
-
|
| 94 |
-
pos = pos.reshape(-1) # (M,)
|
| 95 |
-
out = np.einsum('m,d->md', pos, omega) # (M, D/2), outer product
|
| 96 |
-
|
| 97 |
-
emb_sin = np.sin(out) # (M, D/2)
|
| 98 |
-
emb_cos = np.cos(out) # (M, D/2)
|
| 99 |
-
|
| 100 |
-
emb = np.concatenate([emb_sin, emb_cos], axis=1) # (M, D)
|
| 101 |
-
return emb
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
class Resampler(nn.Module):
|
| 105 |
-
"""
|
| 106 |
-
A 2D perceiver-resampler network with one cross attention layers by
|
| 107 |
-
(grid_size**2) learnable queries and 2d sincos pos_emb
|
| 108 |
-
Outputs:
|
| 109 |
-
A tensor with the shape of (grid_size**2, embed_dim)
|
| 110 |
-
"""
|
| 111 |
-
def __init__(
|
| 112 |
-
self,
|
| 113 |
-
grid_size,
|
| 114 |
-
embed_dim,
|
| 115 |
-
num_heads,
|
| 116 |
-
kv_dim=None,
|
| 117 |
-
norm_layer=nn.LayerNorm
|
| 118 |
-
):
|
| 119 |
-
super().__init__()
|
| 120 |
-
self.num_queries = grid_size ** 2
|
| 121 |
-
self.embed_dim = embed_dim
|
| 122 |
-
self.num_heads = num_heads
|
| 123 |
-
|
| 124 |
-
self.pos_embed = nn.Parameter(
|
| 125 |
-
torch.from_numpy(get_2d_sincos_pos_embed(embed_dim, grid_size)).float()
|
| 126 |
-
).requires_grad_(False)
|
| 127 |
-
|
| 128 |
-
self.query = nn.Parameter(torch.zeros(self.num_queries, embed_dim))
|
| 129 |
-
normal_(self.query, std=.02)
|
| 130 |
-
|
| 131 |
-
if kv_dim is not None and kv_dim != embed_dim:
|
| 132 |
-
self.kv_proj = nn.Linear(kv_dim, embed_dim, bias=False)
|
| 133 |
-
else:
|
| 134 |
-
self.kv_proj = nn.Identity()
|
| 135 |
-
|
| 136 |
-
self.attn = nn.MultiheadAttention(embed_dim, num_heads)
|
| 137 |
-
self.ln_q = norm_layer(embed_dim)
|
| 138 |
-
self.ln_kv = norm_layer(embed_dim)
|
| 139 |
-
|
| 140 |
-
# self.apply(self._init_weights)
|
| 141 |
-
|
| 142 |
-
def _init_weights(self, m):
|
| 143 |
-
if isinstance(m, nn.Linear):
|
| 144 |
-
normal_(m.weight, std=.02)
|
| 145 |
-
if isinstance(m, nn.Linear) and m.bias is not None:
|
| 146 |
-
nn.init.constant_(m.bias, 0)
|
| 147 |
-
elif isinstance(m, nn.LayerNorm):
|
| 148 |
-
nn.init.constant_(m.bias, 0)
|
| 149 |
-
nn.init.constant_(m.weight, 1.0)
|
| 150 |
-
|
| 151 |
-
def forward(self, x, attn_mask=None):
|
| 152 |
-
|
| 153 |
-
pos_embed = get_abs_pos(self.pos_embed, x.size(1))
|
| 154 |
-
|
| 155 |
-
x = self.kv_proj(x)
|
| 156 |
-
x = self.ln_kv(x).permute(1, 0, 2)
|
| 157 |
-
|
| 158 |
-
N = x.shape[1]
|
| 159 |
-
q = self.ln_q(self.query)
|
| 160 |
-
out = self.attn(
|
| 161 |
-
self._repeat(q, N) + self.pos_embed.unsqueeze(1),
|
| 162 |
-
x + pos_embed.unsqueeze(1),
|
| 163 |
-
x,
|
| 164 |
-
attn_mask=attn_mask)[0]
|
| 165 |
-
return out.permute(1, 0, 2)
|
| 166 |
-
|
| 167 |
-
def _repeat(self, query, N: int):
|
| 168 |
-
return query.unsqueeze(1).repeat(1, N, 1)
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
class VisualAttention(nn.Module):
|
| 172 |
-
"""self-attention layer class.
|
| 173 |
-
|
| 174 |
-
Self-attention layer takes input with size [s, b, h]
|
| 175 |
-
and returns output of the same size.
|
| 176 |
-
"""
|
| 177 |
-
|
| 178 |
-
def __init__(self, embed_dim, num_heads,
|
| 179 |
-
bias=True, kdim=None, vdim=None):
|
| 180 |
-
super(VisualAttention, self).__init__()
|
| 181 |
-
self.embed_dim = embed_dim
|
| 182 |
-
self.kdim = kdim if kdim is not None else embed_dim
|
| 183 |
-
self.vdim = vdim if vdim is not None else embed_dim
|
| 184 |
-
self._qkv_same_embed_dim = self.kdim == embed_dim and self.vdim == embed_dim
|
| 185 |
-
|
| 186 |
-
self.num_heads = num_heads
|
| 187 |
-
|
| 188 |
-
# Per attention head and per partition values.
|
| 189 |
-
assert embed_dim % num_heads == 0
|
| 190 |
-
self.hidden_size_per_attention_head = embed_dim // num_heads
|
| 191 |
-
self.num_attention_heads_per_partition = num_heads
|
| 192 |
-
self.hidden_size_per_partition = embed_dim
|
| 193 |
-
|
| 194 |
-
# Strided linear layer.
|
| 195 |
-
assert self._qkv_same_embed_dim, 'Only Support SelfAttention Currently'
|
| 196 |
-
self.in_proj = nn.Linear(embed_dim, 3 * embed_dim)
|
| 197 |
-
self.out_proj = nn.Linear(embed_dim, embed_dim)
|
| 198 |
-
self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
|
| 199 |
-
|
| 200 |
-
def forward(self, query, key, value, attn_mask = None):
|
| 201 |
-
# query/key/value: [sq, b, h]
|
| 202 |
-
sq, b, _ = query.size()
|
| 203 |
-
|
| 204 |
-
assert torch.allclose(query, key), 'Only Support Self-Attention Currently'
|
| 205 |
-
sk = sq
|
| 206 |
-
mixed_x_layer = self.in_proj(query)
|
| 207 |
-
|
| 208 |
-
# [sq, b, (np * 3 * hn)] --> [sq, b, np, 3 * hn]
|
| 209 |
-
new_tensor_shape = mixed_x_layer.size()[:-1] + \
|
| 210 |
-
(self.num_attention_heads_per_partition,
|
| 211 |
-
3 * self.hidden_size_per_attention_head)
|
| 212 |
-
mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
|
| 213 |
-
|
| 214 |
-
# [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
|
| 215 |
-
query_layer, key_layer, value_layer = mixed_x_layer.split(
|
| 216 |
-
self.hidden_size_per_attention_head, dim=-1)
|
| 217 |
-
|
| 218 |
-
# [sq, b, np, hn] -> [sq, b * np, hn]
|
| 219 |
-
query_layer = query_layer.view(sq,
|
| 220 |
-
b * self.num_attention_heads_per_partition,
|
| 221 |
-
self.hidden_size_per_attention_head).transpose(0, 1)
|
| 222 |
-
# [sk, b, np, hn] -> [sk, b * np, hn]
|
| 223 |
-
key_layer = key_layer.view(sk,
|
| 224 |
-
b * self.num_attention_heads_per_partition,
|
| 225 |
-
self.hidden_size_per_attention_head).transpose(0, 1)
|
| 226 |
-
|
| 227 |
-
q_scaled = query_layer / self.norm_factor
|
| 228 |
-
if attn_mask is not None:
|
| 229 |
-
attention_probs = torch.baddbmm(attn_mask, q_scaled, key_layer.transpose(-2, -1))
|
| 230 |
-
else:
|
| 231 |
-
attention_probs = torch.bmm(q_scaled, key_layer.transpose(-2, -1))
|
| 232 |
-
attention_probs = attention_probs.softmax(dim=-1)
|
| 233 |
-
|
| 234 |
-
value_layer = value_layer.view(sk,
|
| 235 |
-
b * self.num_attention_heads_per_partition,
|
| 236 |
-
self.hidden_size_per_attention_head).transpose(0, 1)
|
| 237 |
-
|
| 238 |
-
# matmul: [b * np, sq, hn]
|
| 239 |
-
context_layer = torch.bmm(attention_probs, value_layer)
|
| 240 |
-
|
| 241 |
-
# change view [b, np, sq, hn]
|
| 242 |
-
context_layer = context_layer.view(b,
|
| 243 |
-
self.num_attention_heads_per_partition,
|
| 244 |
-
sq, self.hidden_size_per_attention_head)
|
| 245 |
-
|
| 246 |
-
# [b, np, sq, hn] --> [sq, b, np, hn]
|
| 247 |
-
context_layer = context_layer.permute(2, 0, 1, 3).contiguous()
|
| 248 |
-
|
| 249 |
-
# [sq, b, np, hn] --> [sq, b, hp]
|
| 250 |
-
new_context_layer_shape = context_layer.size()[:-2] + \
|
| 251 |
-
(self.hidden_size_per_partition,)
|
| 252 |
-
context_layer = context_layer.view(*new_context_layer_shape)
|
| 253 |
-
|
| 254 |
-
output = self.out_proj(context_layer)
|
| 255 |
-
|
| 256 |
-
return output
|
| 257 |
-
|
| 258 |
-
|
| 259 |
-
class VisualAttentionBlock(nn.Module):
|
| 260 |
-
def __init__(
|
| 261 |
-
self,
|
| 262 |
-
d_model: int,
|
| 263 |
-
n_head: int,
|
| 264 |
-
mlp_ratio: float = 4.0,
|
| 265 |
-
act_layer: Callable = nn.GELU,
|
| 266 |
-
norm_layer: Callable = nn.LayerNorm,
|
| 267 |
-
is_cross_attention: bool = False,
|
| 268 |
-
):
|
| 269 |
-
super().__init__()
|
| 270 |
-
|
| 271 |
-
self.ln_1 = norm_layer(d_model)
|
| 272 |
-
if is_cross_attention:
|
| 273 |
-
self.ln_1_kv = norm_layer(d_model)
|
| 274 |
-
|
| 275 |
-
self.ln_2 = norm_layer(d_model)
|
| 276 |
-
mlp_width = int(d_model * mlp_ratio)
|
| 277 |
-
self.attn = VisualAttention(d_model, n_head)
|
| 278 |
-
self.mlp = nn.Sequential(OrderedDict([
|
| 279 |
-
("c_fc", nn.Linear(d_model, mlp_width)),
|
| 280 |
-
("gelu", act_layer()),
|
| 281 |
-
("c_proj", nn.Linear(mlp_width, d_model))
|
| 282 |
-
]))
|
| 283 |
-
|
| 284 |
-
def attention(
|
| 285 |
-
self,
|
| 286 |
-
q_x: torch.Tensor,
|
| 287 |
-
k_x: Optional[torch.Tensor] = None,
|
| 288 |
-
v_x: Optional[torch.Tensor] = None,
|
| 289 |
-
attn_mask: Optional[torch.Tensor] = None,
|
| 290 |
-
):
|
| 291 |
-
k_x = k_x if k_x is not None else q_x
|
| 292 |
-
v_x = v_x if v_x is not None else q_x
|
| 293 |
-
|
| 294 |
-
attn_mask = attn_mask.to(q_x.dtype) if attn_mask is not None else None
|
| 295 |
-
return self.attn(q_x, k_x, v_x, attn_mask=attn_mask)
|
| 296 |
-
|
| 297 |
-
def forward(
|
| 298 |
-
self,
|
| 299 |
-
q_x: torch.Tensor,
|
| 300 |
-
k_x: Optional[torch.Tensor] = None,
|
| 301 |
-
v_x: Optional[torch.Tensor] = None,
|
| 302 |
-
attn_mask: Optional[torch.Tensor] = None,
|
| 303 |
-
):
|
| 304 |
-
k_x = self.ln_1_kv(k_x) if hasattr(self, "ln_1_kv") and k_x is not None else None
|
| 305 |
-
v_x = self.ln_1_kv(v_x) if hasattr(self, "ln_1_kv") and v_x is not None else None
|
| 306 |
-
|
| 307 |
-
x = q_x + self.attention(q_x=self.ln_1(q_x), k_x=k_x, v_x=v_x, attn_mask=attn_mask)
|
| 308 |
-
x = x + self.mlp(self.ln_2(x))
|
| 309 |
-
return x
|
| 310 |
-
|
| 311 |
-
|
| 312 |
-
class TransformerBlock(nn.Module):
|
| 313 |
-
def __init__(
|
| 314 |
-
self,
|
| 315 |
-
width: int,
|
| 316 |
-
layers: int,
|
| 317 |
-
heads: int,
|
| 318 |
-
mlp_ratio: float = 4.0,
|
| 319 |
-
act_layer: Callable = nn.GELU,
|
| 320 |
-
norm_layer: Callable = nn.LayerNorm,
|
| 321 |
-
):
|
| 322 |
-
super().__init__()
|
| 323 |
-
self.width = width
|
| 324 |
-
self.layers = layers
|
| 325 |
-
|
| 326 |
-
self.resblocks = nn.ModuleList([
|
| 327 |
-
VisualAttentionBlock(
|
| 328 |
-
width, heads, mlp_ratio, act_layer=act_layer, norm_layer=norm_layer)
|
| 329 |
-
for _ in range(layers)
|
| 330 |
-
])
|
| 331 |
-
|
| 332 |
-
def get_cast_dtype(self) -> torch.dtype:
|
| 333 |
-
return self.resblocks[0].mlp.c_fc.weight.dtype
|
| 334 |
-
|
| 335 |
-
def get_cast_device(self) -> torch.device:
|
| 336 |
-
return self.resblocks[0].mlp.c_fc.weight.device
|
| 337 |
-
|
| 338 |
-
def forward(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None):
|
| 339 |
-
for r in self.resblocks:
|
| 340 |
-
x = r(x, attn_mask=attn_mask)
|
| 341 |
-
return x
|
| 342 |
-
|
| 343 |
-
|
| 344 |
-
class VisionTransformer(nn.Module):
|
| 345 |
-
|
| 346 |
-
def __init__(
|
| 347 |
-
self,
|
| 348 |
-
image_size: int,
|
| 349 |
-
patch_size: int,
|
| 350 |
-
width: int,
|
| 351 |
-
layers: int,
|
| 352 |
-
heads: int,
|
| 353 |
-
mlp_ratio: float,
|
| 354 |
-
n_queries: int = 256,
|
| 355 |
-
output_dim: int = 512,
|
| 356 |
-
**kwargs
|
| 357 |
-
):
|
| 358 |
-
super().__init__()
|
| 359 |
-
image_height, image_width = self.image_size = (image_size, image_size)
|
| 360 |
-
patch_height, patch_width = self.patch_size = (patch_size, patch_size)
|
| 361 |
-
self.grid_size = (image_height // patch_height, image_width // patch_width)
|
| 362 |
-
self.output_dim = output_dim
|
| 363 |
-
|
| 364 |
-
mean = (0.48145466, 0.4578275, 0.40821073)
|
| 365 |
-
std = (0.26862954, 0.26130258, 0.27577711)
|
| 366 |
-
self.image_transform = transforms.Compose([
|
| 367 |
-
transforms.Resize(
|
| 368 |
-
(image_size, image_size),
|
| 369 |
-
interpolation=InterpolationMode.BICUBIC
|
| 370 |
-
),
|
| 371 |
-
transforms.ToTensor(),
|
| 372 |
-
transforms.Normalize(mean=mean, std=std),
|
| 373 |
-
])
|
| 374 |
-
|
| 375 |
-
self.conv1 = nn.Conv2d(in_channels=3, out_channels=width, kernel_size=patch_size, stride=patch_size, bias=False)
|
| 376 |
-
|
| 377 |
-
# class embeddings and positional embeddings
|
| 378 |
-
scale = width ** -0.5
|
| 379 |
-
self.positional_embedding = nn.Parameter(scale * torch.randn(256, width))
|
| 380 |
-
|
| 381 |
-
norm_layer = partial(nn.LayerNorm, eps=1e-6)
|
| 382 |
-
act_layer = nn.GELU
|
| 383 |
-
|
| 384 |
-
self.ln_pre = norm_layer(width)
|
| 385 |
-
self.transformer = TransformerBlock(
|
| 386 |
-
width,
|
| 387 |
-
layers,
|
| 388 |
-
heads,
|
| 389 |
-
mlp_ratio,
|
| 390 |
-
act_layer=act_layer,
|
| 391 |
-
norm_layer=norm_layer,
|
| 392 |
-
)
|
| 393 |
-
|
| 394 |
-
self.attn_pool = Resampler(
|
| 395 |
-
grid_size=int(math.sqrt(n_queries)),
|
| 396 |
-
embed_dim=output_dim,
|
| 397 |
-
num_heads=output_dim // 128,
|
| 398 |
-
kv_dim=width,
|
| 399 |
-
norm_layer=norm_layer,
|
| 400 |
-
)
|
| 401 |
-
self.ln_post = norm_layer(output_dim)
|
| 402 |
-
self.proj = nn.Parameter((output_dim** -0.5) * torch.randn(output_dim, output_dim))
|
| 403 |
-
self.data_path = os.environ.get("DATA", "./")
|
| 404 |
-
self.visualize = os.environ.get("VISUALIZE", False)
|
| 405 |
-
|
| 406 |
-
def forward(self, x: torch.Tensor):
|
| 407 |
-
x = x.to(
|
| 408 |
-
dtype=self.transformer.get_cast_dtype(),
|
| 409 |
-
device=self.transformer.get_cast_device(),
|
| 410 |
-
)
|
| 411 |
-
# to patches
|
| 412 |
-
x = self.conv1(x) # shape = [*, width, grid, grid]
|
| 413 |
-
x = x.reshape(x.shape[0], x.shape[1], -1) # shape = [*, width, grid ** 2]
|
| 414 |
-
x = x.permute(0, 2, 1) # shape = [*, grid ** 2, width]
|
| 415 |
-
|
| 416 |
-
x = x + get_abs_pos(self.positional_embedding, x.size(1))
|
| 417 |
-
|
| 418 |
-
x = self.ln_pre(x)
|
| 419 |
-
|
| 420 |
-
x = x.permute(1, 0, 2) # NLD -> LND
|
| 421 |
-
x = self.transformer(x)
|
| 422 |
-
x = x.permute(1, 0, 2) # LND -> NLD
|
| 423 |
-
|
| 424 |
-
x = self.attn_pool(x)
|
| 425 |
-
x = self.ln_post(x)
|
| 426 |
-
x = x @ self.proj
|
| 427 |
-
|
| 428 |
-
return x
|
| 429 |
-
|
| 430 |
-
def encode(self, image_paths: List[str], masks:List[str] ):
|
| 431 |
-
images = []
|
| 432 |
-
for image_path, mask in zip(image_paths, masks):
|
| 433 |
-
if image_path.startswith("http://") or image_path.startswith("https://"):
|
| 434 |
-
image = Image.open(requests.get(image_path, stream=True).raw)
|
| 435 |
-
else:
|
| 436 |
-
image = Image.open(image_path)
|
| 437 |
-
image = ImageOps.exif_transpose(image)
|
| 438 |
-
image = image.convert("RGB")
|
| 439 |
-
if len(mask)>0:
|
| 440 |
-
print(len(mask))
|
| 441 |
-
seg = {'counts': mask, 'size': [image.size[1], image.size[0]]}
|
| 442 |
-
else:
|
| 443 |
-
seg = None
|
| 444 |
-
if seg is not None:
|
| 445 |
-
mask = self.annToMask(seg, image.size[1], image.size[0])
|
| 446 |
-
image = overlay_mask(np.array(image), mask)
|
| 447 |
-
image = Image.fromarray(image)
|
| 448 |
-
visulize = self.visualize
|
| 449 |
-
if visulize:
|
| 450 |
-
self.data_path = "visualize"
|
| 451 |
-
os.makedirs(self.data_path, exist_ok=True)
|
| 452 |
-
torchshow.save(image, os.path.join(self.data_path, image_path.split("/")[-1]))
|
| 453 |
-
images.append(self.image_transform(image))
|
| 454 |
-
images = torch.stack(images, dim=0)
|
| 455 |
-
#torchshow.save(images)
|
| 456 |
-
return self(images)
|
| 457 |
-
|
| 458 |
-
def annToMask(self, mask_ann, h, w):
|
| 459 |
-
if mask_ann is None:
|
| 460 |
-
return np.zeros((h, w), dtype=np.uint8)
|
| 461 |
-
|
| 462 |
-
if isinstance(mask_ann, list):
|
| 463 |
-
rles = maskUtils.frPyObjects(mask_ann, h, w)
|
| 464 |
-
rle = maskUtils.merge(rles)
|
| 465 |
-
elif isinstance(mask_ann['counts'], list):
|
| 466 |
-
# uncompressed RLE
|
| 467 |
-
rle = maskUtils.frPyObjects(mask_ann, h, w)
|
| 468 |
-
else:
|
| 469 |
-
# rle
|
| 470 |
-
rle = mask_ann
|
| 471 |
-
mask = maskUtils.decode(rle)
|
| 472 |
-
return mask
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
zzzmmz/SegAgent-Model/zero_to_fp32.py
DELETED
|
@@ -1,587 +0,0 @@
|
|
| 1 |
-
#!/usr/bin/env python
|
| 2 |
-
|
| 3 |
-
# Copyright (c) Microsoft Corporation.
|
| 4 |
-
# SPDX-License-Identifier: Apache-2.0
|
| 5 |
-
|
| 6 |
-
# DeepSpeed Team
|
| 7 |
-
|
| 8 |
-
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
| 9 |
-
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
| 10 |
-
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
| 11 |
-
# application.
|
| 12 |
-
#
|
| 13 |
-
# example: python zero_to_fp32.py . pytorch_model.bin
|
| 14 |
-
|
| 15 |
-
import argparse
|
| 16 |
-
import torch
|
| 17 |
-
import glob
|
| 18 |
-
import math
|
| 19 |
-
import os
|
| 20 |
-
import re
|
| 21 |
-
from collections import OrderedDict
|
| 22 |
-
from dataclasses import dataclass
|
| 23 |
-
|
| 24 |
-
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 25 |
-
# DeepSpeed data structures it has to be available in the current python environment.
|
| 26 |
-
from deepspeed.utils import logger
|
| 27 |
-
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 28 |
-
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 29 |
-
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
@dataclass
|
| 33 |
-
class zero_model_state:
|
| 34 |
-
buffers: dict()
|
| 35 |
-
param_shapes: dict()
|
| 36 |
-
shared_params: list
|
| 37 |
-
ds_version: int
|
| 38 |
-
frozen_param_shapes: dict()
|
| 39 |
-
frozen_param_fragments: dict()
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
debug = 0
|
| 43 |
-
|
| 44 |
-
# load to cpu
|
| 45 |
-
device = torch.device('cpu')
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
def atoi(text):
|
| 49 |
-
return int(text) if text.isdigit() else text
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
def natural_keys(text):
|
| 53 |
-
'''
|
| 54 |
-
alist.sort(key=natural_keys) sorts in human order
|
| 55 |
-
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 56 |
-
(See Toothy's implementation in the comments)
|
| 57 |
-
'''
|
| 58 |
-
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 62 |
-
if not os.path.isdir(checkpoint_dir):
|
| 63 |
-
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 64 |
-
|
| 65 |
-
# there should be only one file
|
| 66 |
-
if zero_stage <= 2:
|
| 67 |
-
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 68 |
-
elif zero_stage == 3:
|
| 69 |
-
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 70 |
-
|
| 71 |
-
if not os.path.exists(file):
|
| 72 |
-
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 73 |
-
|
| 74 |
-
return file
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 78 |
-
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 79 |
-
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 80 |
-
|
| 81 |
-
if len(ckpt_files) == 0:
|
| 82 |
-
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 83 |
-
|
| 84 |
-
return ckpt_files
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
def get_optim_files(checkpoint_dir):
|
| 88 |
-
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
def get_model_state_files(checkpoint_dir):
|
| 92 |
-
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
def parse_model_states(files):
|
| 96 |
-
zero_model_states = []
|
| 97 |
-
for file in files:
|
| 98 |
-
state_dict = torch.load(file, map_location=device)
|
| 99 |
-
|
| 100 |
-
if BUFFER_NAMES not in state_dict:
|
| 101 |
-
raise ValueError(f"{file} is not a model state checkpoint")
|
| 102 |
-
buffer_names = state_dict[BUFFER_NAMES]
|
| 103 |
-
if debug:
|
| 104 |
-
print("Found buffers:", buffer_names)
|
| 105 |
-
|
| 106 |
-
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 107 |
-
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 108 |
-
param_shapes = state_dict[PARAM_SHAPES]
|
| 109 |
-
|
| 110 |
-
# collect parameters that are included in param_shapes
|
| 111 |
-
param_names = []
|
| 112 |
-
for s in param_shapes:
|
| 113 |
-
for name in s.keys():
|
| 114 |
-
param_names.append(name)
|
| 115 |
-
|
| 116 |
-
# update with frozen parameters
|
| 117 |
-
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 118 |
-
if frozen_param_shapes is not None:
|
| 119 |
-
if debug:
|
| 120 |
-
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 121 |
-
param_names += list(frozen_param_shapes.keys())
|
| 122 |
-
|
| 123 |
-
# handle shared params
|
| 124 |
-
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 125 |
-
|
| 126 |
-
ds_version = state_dict.get(DS_VERSION, None)
|
| 127 |
-
|
| 128 |
-
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 129 |
-
|
| 130 |
-
z_model_state = zero_model_state(buffers=buffers,
|
| 131 |
-
param_shapes=param_shapes,
|
| 132 |
-
shared_params=shared_params,
|
| 133 |
-
ds_version=ds_version,
|
| 134 |
-
frozen_param_shapes=frozen_param_shapes,
|
| 135 |
-
frozen_param_fragments=frozen_param_fragments)
|
| 136 |
-
zero_model_states.append(z_model_state)
|
| 137 |
-
|
| 138 |
-
return zero_model_states
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
def parse_optim_states(files, ds_checkpoint_dir):
|
| 142 |
-
|
| 143 |
-
total_files = len(files)
|
| 144 |
-
state_dicts = []
|
| 145 |
-
for f in files:
|
| 146 |
-
state_dict = torch.load(f, map_location=device)
|
| 147 |
-
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 148 |
-
# and also handle the case where it was already removed by another helper script
|
| 149 |
-
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 150 |
-
state_dicts.append(state_dict)
|
| 151 |
-
|
| 152 |
-
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 153 |
-
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 154 |
-
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 155 |
-
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 156 |
-
|
| 157 |
-
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 158 |
-
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 159 |
-
# use the max of the partition_count to get the dp world_size.
|
| 160 |
-
|
| 161 |
-
if type(world_size) is list:
|
| 162 |
-
world_size = max(world_size)
|
| 163 |
-
|
| 164 |
-
if world_size != total_files:
|
| 165 |
-
raise ValueError(
|
| 166 |
-
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 167 |
-
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 168 |
-
)
|
| 169 |
-
|
| 170 |
-
# the groups are named differently in each stage
|
| 171 |
-
if zero_stage <= 2:
|
| 172 |
-
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 173 |
-
elif zero_stage == 3:
|
| 174 |
-
fp32_groups_key = FP32_FLAT_GROUPS
|
| 175 |
-
else:
|
| 176 |
-
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 177 |
-
|
| 178 |
-
if zero_stage <= 2:
|
| 179 |
-
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 180 |
-
elif zero_stage == 3:
|
| 181 |
-
# if there is more than one param group, there will be multiple flattened tensors - one
|
| 182 |
-
# flattened tensor per group - for simplicity merge them into a single tensor
|
| 183 |
-
#
|
| 184 |
-
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
| 185 |
-
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
| 186 |
-
|
| 187 |
-
fp32_flat_groups = [
|
| 188 |
-
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
| 189 |
-
]
|
| 190 |
-
|
| 191 |
-
return zero_stage, world_size, fp32_flat_groups
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
|
| 195 |
-
"""
|
| 196 |
-
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 197 |
-
|
| 198 |
-
Args:
|
| 199 |
-
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 200 |
-
|
| 201 |
-
"""
|
| 202 |
-
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 203 |
-
|
| 204 |
-
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 205 |
-
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 206 |
-
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 207 |
-
|
| 208 |
-
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 209 |
-
|
| 210 |
-
zero_model_states = parse_model_states(model_files)
|
| 211 |
-
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 212 |
-
|
| 213 |
-
if zero_stage <= 2:
|
| 214 |
-
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
| 215 |
-
elif zero_stage == 3:
|
| 216 |
-
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
|
| 217 |
-
|
| 218 |
-
|
| 219 |
-
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 220 |
-
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 221 |
-
return
|
| 222 |
-
|
| 223 |
-
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 224 |
-
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 225 |
-
|
| 226 |
-
if debug:
|
| 227 |
-
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 228 |
-
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 229 |
-
|
| 230 |
-
wanted_params = len(frozen_param_shapes)
|
| 231 |
-
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 232 |
-
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 233 |
-
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 234 |
-
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 235 |
-
|
| 236 |
-
total_params = 0
|
| 237 |
-
total_numel = 0
|
| 238 |
-
for name, shape in frozen_param_shapes.items():
|
| 239 |
-
total_params += 1
|
| 240 |
-
unpartitioned_numel = shape.numel()
|
| 241 |
-
total_numel += unpartitioned_numel
|
| 242 |
-
|
| 243 |
-
state_dict[name] = frozen_param_fragments[name]
|
| 244 |
-
|
| 245 |
-
if debug:
|
| 246 |
-
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 247 |
-
|
| 248 |
-
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 249 |
-
|
| 250 |
-
|
| 251 |
-
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 252 |
-
param_shapes = zero_model_states[0].param_shapes
|
| 253 |
-
|
| 254 |
-
# Reconstruction protocol:
|
| 255 |
-
#
|
| 256 |
-
# XXX: document this
|
| 257 |
-
|
| 258 |
-
if debug:
|
| 259 |
-
for i in range(world_size):
|
| 260 |
-
for j in range(len(fp32_flat_groups[0])):
|
| 261 |
-
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 262 |
-
|
| 263 |
-
# XXX: memory usage doubles here (zero2)
|
| 264 |
-
num_param_groups = len(fp32_flat_groups[0])
|
| 265 |
-
merged_single_partition_of_fp32_groups = []
|
| 266 |
-
for i in range(num_param_groups):
|
| 267 |
-
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 268 |
-
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 269 |
-
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 270 |
-
avail_numel = sum(
|
| 271 |
-
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 272 |
-
|
| 273 |
-
if debug:
|
| 274 |
-
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 275 |
-
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 276 |
-
# not asserting if there is a mismatch due to possible padding
|
| 277 |
-
print(f"Have {avail_numel} numels to process.")
|
| 278 |
-
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 279 |
-
|
| 280 |
-
# params
|
| 281 |
-
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 282 |
-
# out-of-core computing solution
|
| 283 |
-
total_numel = 0
|
| 284 |
-
total_params = 0
|
| 285 |
-
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 286 |
-
offset = 0
|
| 287 |
-
avail_numel = full_single_fp32_vector.numel()
|
| 288 |
-
for name, shape in shapes.items():
|
| 289 |
-
|
| 290 |
-
unpartitioned_numel = shape.numel()
|
| 291 |
-
total_numel += unpartitioned_numel
|
| 292 |
-
total_params += 1
|
| 293 |
-
|
| 294 |
-
if debug:
|
| 295 |
-
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 296 |
-
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 297 |
-
offset += unpartitioned_numel
|
| 298 |
-
|
| 299 |
-
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 300 |
-
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 301 |
-
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 302 |
-
# live optimizer object, so we are checking that the numbers are within the right range
|
| 303 |
-
align_to = 2 * world_size
|
| 304 |
-
|
| 305 |
-
def zero2_align(x):
|
| 306 |
-
return align_to * math.ceil(x / align_to)
|
| 307 |
-
|
| 308 |
-
if debug:
|
| 309 |
-
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 310 |
-
|
| 311 |
-
offset = zero2_align(offset)
|
| 312 |
-
avail_numel = zero2_align(avail_numel)
|
| 313 |
-
|
| 314 |
-
if debug:
|
| 315 |
-
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 316 |
-
|
| 317 |
-
# Sanity check
|
| 318 |
-
if offset != avail_numel:
|
| 319 |
-
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 320 |
-
|
| 321 |
-
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 322 |
-
|
| 323 |
-
|
| 324 |
-
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
| 325 |
-
state_dict = OrderedDict()
|
| 326 |
-
|
| 327 |
-
# buffers
|
| 328 |
-
buffers = zero_model_states[0].buffers
|
| 329 |
-
state_dict.update(buffers)
|
| 330 |
-
if debug:
|
| 331 |
-
print(f"added {len(buffers)} buffers")
|
| 332 |
-
|
| 333 |
-
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 334 |
-
|
| 335 |
-
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 336 |
-
|
| 337 |
-
# recover shared parameters
|
| 338 |
-
for pair in zero_model_states[0].shared_params:
|
| 339 |
-
if pair[1] in state_dict:
|
| 340 |
-
state_dict[pair[0]] = state_dict[pair[1]]
|
| 341 |
-
|
| 342 |
-
return state_dict
|
| 343 |
-
|
| 344 |
-
|
| 345 |
-
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 346 |
-
remainder = unpartitioned_numel % world_size
|
| 347 |
-
padding_numel = (world_size - remainder) if remainder else 0
|
| 348 |
-
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 349 |
-
return partitioned_numel, padding_numel
|
| 350 |
-
|
| 351 |
-
|
| 352 |
-
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 353 |
-
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 354 |
-
return
|
| 355 |
-
|
| 356 |
-
if debug:
|
| 357 |
-
for i in range(world_size):
|
| 358 |
-
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 359 |
-
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 360 |
-
|
| 361 |
-
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 362 |
-
wanted_params = len(frozen_param_shapes)
|
| 363 |
-
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 364 |
-
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 365 |
-
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 366 |
-
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 367 |
-
|
| 368 |
-
total_params = 0
|
| 369 |
-
total_numel = 0
|
| 370 |
-
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 371 |
-
total_params += 1
|
| 372 |
-
unpartitioned_numel = shape.numel()
|
| 373 |
-
total_numel += unpartitioned_numel
|
| 374 |
-
|
| 375 |
-
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 376 |
-
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 377 |
-
|
| 378 |
-
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 379 |
-
|
| 380 |
-
if debug:
|
| 381 |
-
print(
|
| 382 |
-
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 383 |
-
)
|
| 384 |
-
|
| 385 |
-
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 389 |
-
param_shapes = zero_model_states[0].param_shapes
|
| 390 |
-
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 391 |
-
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 392 |
-
# param, re-consolidating each param, while dealing with padding if any
|
| 393 |
-
|
| 394 |
-
# merge list of dicts, preserving order
|
| 395 |
-
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 396 |
-
|
| 397 |
-
if debug:
|
| 398 |
-
for i in range(world_size):
|
| 399 |
-
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 400 |
-
|
| 401 |
-
wanted_params = len(param_shapes)
|
| 402 |
-
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 403 |
-
# not asserting if there is a mismatch due to possible padding
|
| 404 |
-
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 405 |
-
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 406 |
-
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 407 |
-
|
| 408 |
-
# params
|
| 409 |
-
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 410 |
-
# out-of-core computing solution
|
| 411 |
-
offset = 0
|
| 412 |
-
total_numel = 0
|
| 413 |
-
total_params = 0
|
| 414 |
-
for name, shape in param_shapes.items():
|
| 415 |
-
|
| 416 |
-
unpartitioned_numel = shape.numel()
|
| 417 |
-
total_numel += unpartitioned_numel
|
| 418 |
-
total_params += 1
|
| 419 |
-
|
| 420 |
-
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 421 |
-
|
| 422 |
-
if debug:
|
| 423 |
-
print(
|
| 424 |
-
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 425 |
-
)
|
| 426 |
-
|
| 427 |
-
# XXX: memory usage doubles here
|
| 428 |
-
state_dict[name] = torch.cat(
|
| 429 |
-
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
| 430 |
-
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 431 |
-
offset += partitioned_numel
|
| 432 |
-
|
| 433 |
-
offset *= world_size
|
| 434 |
-
|
| 435 |
-
# Sanity check
|
| 436 |
-
if offset != avail_numel:
|
| 437 |
-
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 438 |
-
|
| 439 |
-
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 440 |
-
|
| 441 |
-
|
| 442 |
-
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
|
| 443 |
-
state_dict = OrderedDict()
|
| 444 |
-
|
| 445 |
-
# buffers
|
| 446 |
-
buffers = zero_model_states[0].buffers
|
| 447 |
-
state_dict.update(buffers)
|
| 448 |
-
if debug:
|
| 449 |
-
print(f"added {len(buffers)} buffers")
|
| 450 |
-
|
| 451 |
-
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 452 |
-
|
| 453 |
-
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 454 |
-
|
| 455 |
-
# recover shared parameters
|
| 456 |
-
for pair in zero_model_states[0].shared_params:
|
| 457 |
-
if pair[1] in state_dict:
|
| 458 |
-
state_dict[pair[0]] = state_dict[pair[1]]
|
| 459 |
-
|
| 460 |
-
return state_dict
|
| 461 |
-
|
| 462 |
-
|
| 463 |
-
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
|
| 464 |
-
"""
|
| 465 |
-
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 466 |
-
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 467 |
-
via a model hub.
|
| 468 |
-
|
| 469 |
-
Args:
|
| 470 |
-
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 471 |
-
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
|
| 472 |
-
|
| 473 |
-
Returns:
|
| 474 |
-
- pytorch ``state_dict``
|
| 475 |
-
|
| 476 |
-
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
| 477 |
-
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 478 |
-
the checkpoint.
|
| 479 |
-
|
| 480 |
-
A typical usage might be ::
|
| 481 |
-
|
| 482 |
-
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 483 |
-
# do the training and checkpoint saving
|
| 484 |
-
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 485 |
-
model = model.cpu() # move to cpu
|
| 486 |
-
model.load_state_dict(state_dict)
|
| 487 |
-
# submit to model hub or save the model to share with others
|
| 488 |
-
|
| 489 |
-
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 490 |
-
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 491 |
-
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 492 |
-
|
| 493 |
-
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 494 |
-
|
| 495 |
-
"""
|
| 496 |
-
if tag is None:
|
| 497 |
-
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 498 |
-
if os.path.isfile(latest_path):
|
| 499 |
-
with open(latest_path, 'r') as fd:
|
| 500 |
-
tag = fd.read().strip()
|
| 501 |
-
else:
|
| 502 |
-
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 503 |
-
|
| 504 |
-
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 505 |
-
|
| 506 |
-
if not os.path.isdir(ds_checkpoint_dir):
|
| 507 |
-
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 508 |
-
|
| 509 |
-
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
|
| 510 |
-
|
| 511 |
-
|
| 512 |
-
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
|
| 513 |
-
"""
|
| 514 |
-
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 515 |
-
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 516 |
-
|
| 517 |
-
Args:
|
| 518 |
-
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 519 |
-
- ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
|
| 520 |
-
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 521 |
-
"""
|
| 522 |
-
|
| 523 |
-
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 524 |
-
print(f"Saving fp32 state dict to {output_file}")
|
| 525 |
-
torch.save(state_dict, output_file)
|
| 526 |
-
|
| 527 |
-
|
| 528 |
-
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 529 |
-
"""
|
| 530 |
-
1. Put the provided model to cpu
|
| 531 |
-
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 532 |
-
3. Load it into the provided model
|
| 533 |
-
|
| 534 |
-
Args:
|
| 535 |
-
- ``model``: the model object to update
|
| 536 |
-
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 537 |
-
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 538 |
-
|
| 539 |
-
Returns:
|
| 540 |
-
- ``model`: modified model
|
| 541 |
-
|
| 542 |
-
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 543 |
-
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 544 |
-
conveniently placed for you in the checkpoint folder.
|
| 545 |
-
|
| 546 |
-
A typical usage might be ::
|
| 547 |
-
|
| 548 |
-
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 549 |
-
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 550 |
-
# submit to model hub or save the model to share with others
|
| 551 |
-
|
| 552 |
-
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 553 |
-
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 554 |
-
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 555 |
-
|
| 556 |
-
"""
|
| 557 |
-
logger.info(f"Extracting fp32 weights")
|
| 558 |
-
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 559 |
-
|
| 560 |
-
logger.info(f"Overwriting model with fp32 weights")
|
| 561 |
-
model = model.cpu()
|
| 562 |
-
model.load_state_dict(state_dict, strict=False)
|
| 563 |
-
|
| 564 |
-
return model
|
| 565 |
-
|
| 566 |
-
|
| 567 |
-
if __name__ == "__main__":
|
| 568 |
-
|
| 569 |
-
parser = argparse.ArgumentParser()
|
| 570 |
-
parser.add_argument("checkpoint_dir",
|
| 571 |
-
type=str,
|
| 572 |
-
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 573 |
-
parser.add_argument(
|
| 574 |
-
"output_file",
|
| 575 |
-
type=str,
|
| 576 |
-
help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
|
| 577 |
-
parser.add_argument("-t",
|
| 578 |
-
"--tag",
|
| 579 |
-
type=str,
|
| 580 |
-
default=None,
|
| 581 |
-
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 582 |
-
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 583 |
-
args = parser.parse_args()
|
| 584 |
-
|
| 585 |
-
debug = args.debug
|
| 586 |
-
|
| 587 |
-
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|