karoladelk commited on
Commit
21211e7
·
verified ·
1 Parent(s): dc677eb

karoladelk/bachh

Browse files
README.md ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: google/vit-base-patch16-224-in21k
4
+ tags:
5
+ - image-classification
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: bachh
9
+ results: []
10
+ ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ # bachh
16
+
17
+ This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the Pranavkpba2000/skin_cancer_dataset dataset.
18
+
19
+ ## Model description
20
+
21
+ More information needed
22
+
23
+ ## Intended uses & limitations
24
+
25
+ More information needed
26
+
27
+ ## Training and evaluation data
28
+
29
+ More information needed
30
+
31
+ ## Training procedure
32
+
33
+ ### Training hyperparameters
34
+
35
+ The following hyperparameters were used during training:
36
+ - learning_rate: 0.0002
37
+ - train_batch_size: 32
38
+ - eval_batch_size: 16
39
+ - seed: 42
40
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
41
+ - lr_scheduler_type: linear
42
+ - num_epochs: 4
43
+
44
+ ### Framework versions
45
+
46
+ - Transformers 4.39.3
47
+ - Pytorch 2.1.2
48
+ - Datasets 2.18.0
49
+ - Tokenizers 0.15.2
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_accuracy": 0.8441942294159043,
4
+ "eval_loss": 0.4731413722038269,
5
+ "eval_runtime": 127.8463,
6
+ "eval_samples_per_second": 55.575,
7
+ "eval_steps_per_second": 3.481,
8
+ "total_flos": 8.839521632856048e+18,
9
+ "train_loss": 0.3317156129854944,
10
+ "train_runtime": 7997.833,
11
+ "train_samples_per_second": 14.262,
12
+ "train_steps_per_second": 0.446
13
+ }
config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "google/vit-base-patch16-224-in21k",
3
+ "architectures": [
4
+ "ViTForImageClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "encoder_stride": 16,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "AK",
13
+ "1": "BCC",
14
+ "2": "BKL",
15
+ "3": "DF",
16
+ "4": "MEL",
17
+ "5": "NV",
18
+ "6": "SCC",
19
+ "7": "VASC"
20
+ },
21
+ "image_size": 224,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 3072,
24
+ "label2id": {
25
+ "AK": "0",
26
+ "BCC": "1",
27
+ "BKL": "2",
28
+ "DF": "3",
29
+ "MEL": "4",
30
+ "NV": "5",
31
+ "SCC": "6",
32
+ "VASC": "7"
33
+ },
34
+ "layer_norm_eps": 1e-12,
35
+ "model_type": "vit",
36
+ "num_attention_heads": 12,
37
+ "num_channels": 3,
38
+ "num_hidden_layers": 12,
39
+ "patch_size": 16,
40
+ "problem_type": "single_label_classification",
41
+ "qkv_bias": true,
42
+ "torch_dtype": "float32",
43
+ "transformers_version": "4.39.3"
44
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "eval_accuracy": 0.8441942294159043,
4
+ "eval_loss": 0.4731413722038269,
5
+ "eval_runtime": 127.8463,
6
+ "eval_samples_per_second": 55.575,
7
+ "eval_steps_per_second": 3.481
8
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76164f8387f163fb92f9330b42d31bf17486d5385803d11bd9c4e165f62e95ca
3
+ size 343242432
preprocessor_config.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_valid_processor_keys": [
3
+ "images",
4
+ "do_resize",
5
+ "size",
6
+ "resample",
7
+ "do_rescale",
8
+ "rescale_factor",
9
+ "do_normalize",
10
+ "image_mean",
11
+ "image_std",
12
+ "return_tensors",
13
+ "data_format",
14
+ "input_data_format"
15
+ ],
16
+ "do_normalize": true,
17
+ "do_rescale": true,
18
+ "do_resize": true,
19
+ "image_mean": [
20
+ 0.5,
21
+ 0.5,
22
+ 0.5
23
+ ],
24
+ "image_processor_type": "ViTFeatureExtractor",
25
+ "image_std": [
26
+ 0.5,
27
+ 0.5,
28
+ 0.5
29
+ ],
30
+ "resample": 2,
31
+ "rescale_factor": 0.00392156862745098,
32
+ "size": {
33
+ "height": 224,
34
+ "width": 224
35
+ }
36
+ }
runs/Apr21_17-30-20_67314e3e4708/events.out.tfevents.1713720627.67314e3e4708.91.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b607ae1d45d15592821fb7b2bb56cb04f87cb132454eb1f0e930a72d2e73468
3
+ size 91507
runs/Apr21_17-30-20_67314e3e4708/events.out.tfevents.1713728775.67314e3e4708.91.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61973e21d0e9e47d1cfb9cf3bf0da85919801dafe9ee66d9f16845f84ef39424
3
+ size 411
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 4.0,
3
+ "total_flos": 8.839521632856048e+18,
4
+ "train_loss": 0.3317156129854944,
5
+ "train_runtime": 7997.833,
6
+ "train_samples_per_second": 14.262,
7
+ "train_steps_per_second": 0.446
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,2837 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.4731413722038269,
3
+ "best_model_checkpoint": "./vit-base-beans-demo-v5/checkpoint-2200",
4
+ "epoch": 4.0,
5
+ "eval_steps": 100,
6
+ "global_step": 3568,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "grad_norm": 1.2656996250152588,
14
+ "learning_rate": 0.00019943946188340808,
15
+ "loss": 1.7641,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.02,
20
+ "grad_norm": 1.702761173248291,
21
+ "learning_rate": 0.00019887892376681615,
22
+ "loss": 1.5342,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.03,
27
+ "grad_norm": 1.4954216480255127,
28
+ "learning_rate": 0.00019831838565022422,
29
+ "loss": 1.2414,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.04,
34
+ "grad_norm": 1.647926926612854,
35
+ "learning_rate": 0.0001977578475336323,
36
+ "loss": 1.1185,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.06,
41
+ "grad_norm": 2.192228078842163,
42
+ "learning_rate": 0.00019719730941704039,
43
+ "loss": 1.1626,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "grad_norm": 2.8315672874450684,
49
+ "learning_rate": 0.00019663677130044843,
50
+ "loss": 1.1362,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.08,
55
+ "grad_norm": 2.4982831478118896,
56
+ "learning_rate": 0.0001960762331838565,
57
+ "loss": 1.0978,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.09,
62
+ "grad_norm": 3.8782095909118652,
63
+ "learning_rate": 0.0001955156950672646,
64
+ "loss": 1.0034,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.1,
69
+ "grad_norm": 2.299464464187622,
70
+ "learning_rate": 0.00019495515695067267,
71
+ "loss": 0.9887,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.11,
76
+ "grad_norm": 2.7226407527923584,
77
+ "learning_rate": 0.0001943946188340807,
78
+ "loss": 0.8874,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.11,
83
+ "eval_accuracy": 0.6571428571428571,
84
+ "eval_loss": 0.9671773314476013,
85
+ "eval_runtime": 129.3363,
86
+ "eval_samples_per_second": 54.934,
87
+ "eval_steps_per_second": 3.441,
88
+ "step": 100
89
+ },
90
+ {
91
+ "epoch": 0.12,
92
+ "grad_norm": 1.4987040758132935,
93
+ "learning_rate": 0.0001938340807174888,
94
+ "loss": 0.8796,
95
+ "step": 110
96
+ },
97
+ {
98
+ "epoch": 0.13,
99
+ "grad_norm": 1.5715798139572144,
100
+ "learning_rate": 0.00019327354260089688,
101
+ "loss": 0.9511,
102
+ "step": 120
103
+ },
104
+ {
105
+ "epoch": 0.15,
106
+ "grad_norm": 1.1380341053009033,
107
+ "learning_rate": 0.00019271300448430495,
108
+ "loss": 0.8983,
109
+ "step": 130
110
+ },
111
+ {
112
+ "epoch": 0.16,
113
+ "grad_norm": 2.5433120727539062,
114
+ "learning_rate": 0.00019215246636771302,
115
+ "loss": 0.8984,
116
+ "step": 140
117
+ },
118
+ {
119
+ "epoch": 0.17,
120
+ "grad_norm": 2.033275604248047,
121
+ "learning_rate": 0.00019159192825112109,
122
+ "loss": 0.824,
123
+ "step": 150
124
+ },
125
+ {
126
+ "epoch": 0.18,
127
+ "grad_norm": 2.4264872074127197,
128
+ "learning_rate": 0.00019103139013452916,
129
+ "loss": 0.8358,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.19,
134
+ "grad_norm": 1.856844186782837,
135
+ "learning_rate": 0.00019047085201793723,
136
+ "loss": 0.8706,
137
+ "step": 170
138
+ },
139
+ {
140
+ "epoch": 0.2,
141
+ "grad_norm": 2.4621365070343018,
142
+ "learning_rate": 0.0001899103139013453,
143
+ "loss": 0.8521,
144
+ "step": 180
145
+ },
146
+ {
147
+ "epoch": 0.21,
148
+ "grad_norm": 1.4839106798171997,
149
+ "learning_rate": 0.00018934977578475337,
150
+ "loss": 0.8314,
151
+ "step": 190
152
+ },
153
+ {
154
+ "epoch": 0.22,
155
+ "grad_norm": 2.008270502090454,
156
+ "learning_rate": 0.00018878923766816143,
157
+ "loss": 0.7806,
158
+ "step": 200
159
+ },
160
+ {
161
+ "epoch": 0.22,
162
+ "eval_accuracy": 0.7096410978184378,
163
+ "eval_loss": 0.8030331134796143,
164
+ "eval_runtime": 127.3215,
165
+ "eval_samples_per_second": 55.804,
166
+ "eval_steps_per_second": 3.495,
167
+ "step": 200
168
+ },
169
+ {
170
+ "epoch": 0.24,
171
+ "grad_norm": 2.759716272354126,
172
+ "learning_rate": 0.00018822869955156953,
173
+ "loss": 0.7414,
174
+ "step": 210
175
+ },
176
+ {
177
+ "epoch": 0.25,
178
+ "grad_norm": 2.2896201610565186,
179
+ "learning_rate": 0.00018766816143497757,
180
+ "loss": 0.7788,
181
+ "step": 220
182
+ },
183
+ {
184
+ "epoch": 0.26,
185
+ "grad_norm": 1.522684097290039,
186
+ "learning_rate": 0.00018710762331838564,
187
+ "loss": 0.7765,
188
+ "step": 230
189
+ },
190
+ {
191
+ "epoch": 0.27,
192
+ "grad_norm": 2.6483938694000244,
193
+ "learning_rate": 0.00018654708520179374,
194
+ "loss": 0.8428,
195
+ "step": 240
196
+ },
197
+ {
198
+ "epoch": 0.28,
199
+ "grad_norm": 1.5204224586486816,
200
+ "learning_rate": 0.0001859865470852018,
201
+ "loss": 0.7756,
202
+ "step": 250
203
+ },
204
+ {
205
+ "epoch": 0.29,
206
+ "grad_norm": 1.352989912033081,
207
+ "learning_rate": 0.00018542600896860985,
208
+ "loss": 0.8085,
209
+ "step": 260
210
+ },
211
+ {
212
+ "epoch": 0.3,
213
+ "grad_norm": 3.1435911655426025,
214
+ "learning_rate": 0.00018486547085201795,
215
+ "loss": 0.8336,
216
+ "step": 270
217
+ },
218
+ {
219
+ "epoch": 0.31,
220
+ "grad_norm": 1.863233208656311,
221
+ "learning_rate": 0.00018430493273542602,
222
+ "loss": 0.7351,
223
+ "step": 280
224
+ },
225
+ {
226
+ "epoch": 0.33,
227
+ "grad_norm": 1.6672592163085938,
228
+ "learning_rate": 0.0001837443946188341,
229
+ "loss": 0.7169,
230
+ "step": 290
231
+ },
232
+ {
233
+ "epoch": 0.34,
234
+ "grad_norm": 1.852493405342102,
235
+ "learning_rate": 0.00018318385650224216,
236
+ "loss": 0.7404,
237
+ "step": 300
238
+ },
239
+ {
240
+ "epoch": 0.34,
241
+ "eval_accuracy": 0.696551724137931,
242
+ "eval_loss": 0.8076898455619812,
243
+ "eval_runtime": 127.361,
244
+ "eval_samples_per_second": 55.786,
245
+ "eval_steps_per_second": 3.494,
246
+ "step": 300
247
+ },
248
+ {
249
+ "epoch": 0.35,
250
+ "grad_norm": 2.066432476043701,
251
+ "learning_rate": 0.00018262331838565023,
252
+ "loss": 0.7422,
253
+ "step": 310
254
+ },
255
+ {
256
+ "epoch": 0.36,
257
+ "grad_norm": 1.5433382987976074,
258
+ "learning_rate": 0.0001820627802690583,
259
+ "loss": 0.7509,
260
+ "step": 320
261
+ },
262
+ {
263
+ "epoch": 0.37,
264
+ "grad_norm": 2.6319210529327393,
265
+ "learning_rate": 0.00018150224215246637,
266
+ "loss": 0.7168,
267
+ "step": 330
268
+ },
269
+ {
270
+ "epoch": 0.38,
271
+ "grad_norm": 1.8435016870498657,
272
+ "learning_rate": 0.00018094170403587444,
273
+ "loss": 0.7421,
274
+ "step": 340
275
+ },
276
+ {
277
+ "epoch": 0.39,
278
+ "grad_norm": 1.946035623550415,
279
+ "learning_rate": 0.0001803811659192825,
280
+ "loss": 0.7062,
281
+ "step": 350
282
+ },
283
+ {
284
+ "epoch": 0.4,
285
+ "grad_norm": 1.926882028579712,
286
+ "learning_rate": 0.0001798206278026906,
287
+ "loss": 0.7228,
288
+ "step": 360
289
+ },
290
+ {
291
+ "epoch": 0.41,
292
+ "grad_norm": 3.0743374824523926,
293
+ "learning_rate": 0.00017926008968609868,
294
+ "loss": 0.7722,
295
+ "step": 370
296
+ },
297
+ {
298
+ "epoch": 0.43,
299
+ "grad_norm": 1.2202589511871338,
300
+ "learning_rate": 0.00017869955156950672,
301
+ "loss": 0.7334,
302
+ "step": 380
303
+ },
304
+ {
305
+ "epoch": 0.44,
306
+ "grad_norm": 1.696330189704895,
307
+ "learning_rate": 0.00017813901345291482,
308
+ "loss": 0.7306,
309
+ "step": 390
310
+ },
311
+ {
312
+ "epoch": 0.45,
313
+ "grad_norm": 2.4689078330993652,
314
+ "learning_rate": 0.0001775784753363229,
315
+ "loss": 0.7224,
316
+ "step": 400
317
+ },
318
+ {
319
+ "epoch": 0.45,
320
+ "eval_accuracy": 0.741027445460943,
321
+ "eval_loss": 0.6990236043930054,
322
+ "eval_runtime": 127.2157,
323
+ "eval_samples_per_second": 55.85,
324
+ "eval_steps_per_second": 3.498,
325
+ "step": 400
326
+ },
327
+ {
328
+ "epoch": 0.46,
329
+ "grad_norm": 1.5020831823349,
330
+ "learning_rate": 0.00017701793721973096,
331
+ "loss": 0.7291,
332
+ "step": 410
333
+ },
334
+ {
335
+ "epoch": 0.47,
336
+ "grad_norm": 2.5034234523773193,
337
+ "learning_rate": 0.00017645739910313903,
338
+ "loss": 0.6754,
339
+ "step": 420
340
+ },
341
+ {
342
+ "epoch": 0.48,
343
+ "grad_norm": 1.4598332643508911,
344
+ "learning_rate": 0.0001758968609865471,
345
+ "loss": 0.6624,
346
+ "step": 430
347
+ },
348
+ {
349
+ "epoch": 0.49,
350
+ "grad_norm": 4.073904991149902,
351
+ "learning_rate": 0.00017533632286995517,
352
+ "loss": 0.6682,
353
+ "step": 440
354
+ },
355
+ {
356
+ "epoch": 0.5,
357
+ "grad_norm": 1.7306851148605347,
358
+ "learning_rate": 0.00017477578475336324,
359
+ "loss": 0.6515,
360
+ "step": 450
361
+ },
362
+ {
363
+ "epoch": 0.52,
364
+ "grad_norm": 2.59877347946167,
365
+ "learning_rate": 0.0001742152466367713,
366
+ "loss": 0.5476,
367
+ "step": 460
368
+ },
369
+ {
370
+ "epoch": 0.53,
371
+ "grad_norm": 3.6079037189483643,
372
+ "learning_rate": 0.00017365470852017938,
373
+ "loss": 0.7031,
374
+ "step": 470
375
+ },
376
+ {
377
+ "epoch": 0.54,
378
+ "grad_norm": 3.1861743927001953,
379
+ "learning_rate": 0.00017309417040358745,
380
+ "loss": 0.7141,
381
+ "step": 480
382
+ },
383
+ {
384
+ "epoch": 0.55,
385
+ "grad_norm": 1.9290772676467896,
386
+ "learning_rate": 0.00017253363228699552,
387
+ "loss": 0.668,
388
+ "step": 490
389
+ },
390
+ {
391
+ "epoch": 0.56,
392
+ "grad_norm": 2.201539993286133,
393
+ "learning_rate": 0.0001719730941704036,
394
+ "loss": 0.6969,
395
+ "step": 500
396
+ },
397
+ {
398
+ "epoch": 0.56,
399
+ "eval_accuracy": 0.7362420830401126,
400
+ "eval_loss": 0.7265785932540894,
401
+ "eval_runtime": 127.1383,
402
+ "eval_samples_per_second": 55.884,
403
+ "eval_steps_per_second": 3.5,
404
+ "step": 500
405
+ },
406
+ {
407
+ "epoch": 0.57,
408
+ "grad_norm": 1.651583194732666,
409
+ "learning_rate": 0.00017141255605381166,
410
+ "loss": 0.6099,
411
+ "step": 510
412
+ },
413
+ {
414
+ "epoch": 0.58,
415
+ "grad_norm": 1.7216840982437134,
416
+ "learning_rate": 0.00017085201793721975,
417
+ "loss": 0.6174,
418
+ "step": 520
419
+ },
420
+ {
421
+ "epoch": 0.59,
422
+ "grad_norm": 1.8105279207229614,
423
+ "learning_rate": 0.0001702914798206278,
424
+ "loss": 0.6444,
425
+ "step": 530
426
+ },
427
+ {
428
+ "epoch": 0.61,
429
+ "grad_norm": 1.837810754776001,
430
+ "learning_rate": 0.00016973094170403587,
431
+ "loss": 0.6607,
432
+ "step": 540
433
+ },
434
+ {
435
+ "epoch": 0.62,
436
+ "grad_norm": 1.6770997047424316,
437
+ "learning_rate": 0.00016917040358744396,
438
+ "loss": 0.6268,
439
+ "step": 550
440
+ },
441
+ {
442
+ "epoch": 0.63,
443
+ "grad_norm": 1.2837833166122437,
444
+ "learning_rate": 0.00016860986547085203,
445
+ "loss": 0.7084,
446
+ "step": 560
447
+ },
448
+ {
449
+ "epoch": 0.64,
450
+ "grad_norm": 1.0523972511291504,
451
+ "learning_rate": 0.00016804932735426008,
452
+ "loss": 0.6271,
453
+ "step": 570
454
+ },
455
+ {
456
+ "epoch": 0.65,
457
+ "grad_norm": 1.356562852859497,
458
+ "learning_rate": 0.00016748878923766817,
459
+ "loss": 0.6242,
460
+ "step": 580
461
+ },
462
+ {
463
+ "epoch": 0.66,
464
+ "grad_norm": 1.5814656019210815,
465
+ "learning_rate": 0.00016692825112107624,
466
+ "loss": 0.564,
467
+ "step": 590
468
+ },
469
+ {
470
+ "epoch": 0.67,
471
+ "grad_norm": 1.757988691329956,
472
+ "learning_rate": 0.0001663677130044843,
473
+ "loss": 0.5929,
474
+ "step": 600
475
+ },
476
+ {
477
+ "epoch": 0.67,
478
+ "eval_accuracy": 0.7589021815622801,
479
+ "eval_loss": 0.6735280156135559,
480
+ "eval_runtime": 126.499,
481
+ "eval_samples_per_second": 56.166,
482
+ "eval_steps_per_second": 3.518,
483
+ "step": 600
484
+ },
485
+ {
486
+ "epoch": 0.68,
487
+ "grad_norm": 1.8649027347564697,
488
+ "learning_rate": 0.00016580717488789238,
489
+ "loss": 0.6578,
490
+ "step": 610
491
+ },
492
+ {
493
+ "epoch": 0.7,
494
+ "grad_norm": 2.1447596549987793,
495
+ "learning_rate": 0.00016524663677130045,
496
+ "loss": 0.6224,
497
+ "step": 620
498
+ },
499
+ {
500
+ "epoch": 0.71,
501
+ "grad_norm": 1.7960020303726196,
502
+ "learning_rate": 0.00016468609865470852,
503
+ "loss": 0.6506,
504
+ "step": 630
505
+ },
506
+ {
507
+ "epoch": 0.72,
508
+ "grad_norm": 2.265737533569336,
509
+ "learning_rate": 0.00016412556053811662,
510
+ "loss": 0.6302,
511
+ "step": 640
512
+ },
513
+ {
514
+ "epoch": 0.73,
515
+ "grad_norm": 1.6668283939361572,
516
+ "learning_rate": 0.00016356502242152466,
517
+ "loss": 0.6518,
518
+ "step": 650
519
+ },
520
+ {
521
+ "epoch": 0.74,
522
+ "grad_norm": 2.8400986194610596,
523
+ "learning_rate": 0.00016300448430493273,
524
+ "loss": 0.619,
525
+ "step": 660
526
+ },
527
+ {
528
+ "epoch": 0.75,
529
+ "grad_norm": 4.842011451721191,
530
+ "learning_rate": 0.00016244394618834083,
531
+ "loss": 0.5984,
532
+ "step": 670
533
+ },
534
+ {
535
+ "epoch": 0.76,
536
+ "grad_norm": 2.0878114700317383,
537
+ "learning_rate": 0.0001618834080717489,
538
+ "loss": 0.5451,
539
+ "step": 680
540
+ },
541
+ {
542
+ "epoch": 0.77,
543
+ "grad_norm": 2.6710331439971924,
544
+ "learning_rate": 0.00016132286995515694,
545
+ "loss": 0.6813,
546
+ "step": 690
547
+ },
548
+ {
549
+ "epoch": 0.78,
550
+ "grad_norm": 2.501116991043091,
551
+ "learning_rate": 0.00016076233183856504,
552
+ "loss": 0.5556,
553
+ "step": 700
554
+ },
555
+ {
556
+ "epoch": 0.78,
557
+ "eval_accuracy": 0.7463757916959888,
558
+ "eval_loss": 0.6704686284065247,
559
+ "eval_runtime": 126.4929,
560
+ "eval_samples_per_second": 56.169,
561
+ "eval_steps_per_second": 3.518,
562
+ "step": 700
563
+ },
564
+ {
565
+ "epoch": 0.8,
566
+ "grad_norm": 2.3544445037841797,
567
+ "learning_rate": 0.0001602017937219731,
568
+ "loss": 0.6037,
569
+ "step": 710
570
+ },
571
+ {
572
+ "epoch": 0.81,
573
+ "grad_norm": 1.7408026456832886,
574
+ "learning_rate": 0.00015964125560538118,
575
+ "loss": 0.7128,
576
+ "step": 720
577
+ },
578
+ {
579
+ "epoch": 0.82,
580
+ "grad_norm": 3.1222403049468994,
581
+ "learning_rate": 0.00015908071748878925,
582
+ "loss": 0.5976,
583
+ "step": 730
584
+ },
585
+ {
586
+ "epoch": 0.83,
587
+ "grad_norm": 1.2981232404708862,
588
+ "learning_rate": 0.00015852017937219732,
589
+ "loss": 0.5566,
590
+ "step": 740
591
+ },
592
+ {
593
+ "epoch": 0.84,
594
+ "grad_norm": 0.9470215439796448,
595
+ "learning_rate": 0.0001579596412556054,
596
+ "loss": 0.5446,
597
+ "step": 750
598
+ },
599
+ {
600
+ "epoch": 0.85,
601
+ "grad_norm": 1.5283540487289429,
602
+ "learning_rate": 0.00015739910313901346,
603
+ "loss": 0.5127,
604
+ "step": 760
605
+ },
606
+ {
607
+ "epoch": 0.86,
608
+ "grad_norm": 1.9940904378890991,
609
+ "learning_rate": 0.00015683856502242153,
610
+ "loss": 0.5999,
611
+ "step": 770
612
+ },
613
+ {
614
+ "epoch": 0.87,
615
+ "grad_norm": 2.0824811458587646,
616
+ "learning_rate": 0.0001562780269058296,
617
+ "loss": 0.5658,
618
+ "step": 780
619
+ },
620
+ {
621
+ "epoch": 0.89,
622
+ "grad_norm": 1.9540338516235352,
623
+ "learning_rate": 0.0001557174887892377,
624
+ "loss": 0.5329,
625
+ "step": 790
626
+ },
627
+ {
628
+ "epoch": 0.9,
629
+ "grad_norm": 1.6645208597183228,
630
+ "learning_rate": 0.00015515695067264574,
631
+ "loss": 0.5831,
632
+ "step": 800
633
+ },
634
+ {
635
+ "epoch": 0.9,
636
+ "eval_accuracy": 0.7681914144968333,
637
+ "eval_loss": 0.6300484538078308,
638
+ "eval_runtime": 127.6321,
639
+ "eval_samples_per_second": 55.668,
640
+ "eval_steps_per_second": 3.487,
641
+ "step": 800
642
+ },
643
+ {
644
+ "epoch": 0.91,
645
+ "grad_norm": 1.9477753639221191,
646
+ "learning_rate": 0.0001545964125560538,
647
+ "loss": 0.5994,
648
+ "step": 810
649
+ },
650
+ {
651
+ "epoch": 0.92,
652
+ "grad_norm": 1.3609261512756348,
653
+ "learning_rate": 0.00015403587443946188,
654
+ "loss": 0.5809,
655
+ "step": 820
656
+ },
657
+ {
658
+ "epoch": 0.93,
659
+ "grad_norm": 1.7007122039794922,
660
+ "learning_rate": 0.00015347533632286998,
661
+ "loss": 0.6124,
662
+ "step": 830
663
+ },
664
+ {
665
+ "epoch": 0.94,
666
+ "grad_norm": 1.753017783164978,
667
+ "learning_rate": 0.00015291479820627804,
668
+ "loss": 0.5809,
669
+ "step": 840
670
+ },
671
+ {
672
+ "epoch": 0.95,
673
+ "grad_norm": 1.672780156135559,
674
+ "learning_rate": 0.0001523542600896861,
675
+ "loss": 0.5536,
676
+ "step": 850
677
+ },
678
+ {
679
+ "epoch": 0.96,
680
+ "grad_norm": 1.7526847124099731,
681
+ "learning_rate": 0.00015179372197309418,
682
+ "loss": 0.6337,
683
+ "step": 860
684
+ },
685
+ {
686
+ "epoch": 0.98,
687
+ "grad_norm": 1.4697840213775635,
688
+ "learning_rate": 0.00015123318385650225,
689
+ "loss": 0.5725,
690
+ "step": 870
691
+ },
692
+ {
693
+ "epoch": 0.99,
694
+ "grad_norm": 2.619101047515869,
695
+ "learning_rate": 0.00015067264573991032,
696
+ "loss": 0.7272,
697
+ "step": 880
698
+ },
699
+ {
700
+ "epoch": 1.0,
701
+ "grad_norm": 2.2301433086395264,
702
+ "learning_rate": 0.0001501121076233184,
703
+ "loss": 0.5315,
704
+ "step": 890
705
+ },
706
+ {
707
+ "epoch": 1.01,
708
+ "grad_norm": 1.2401838302612305,
709
+ "learning_rate": 0.00014955156950672646,
710
+ "loss": 0.3992,
711
+ "step": 900
712
+ },
713
+ {
714
+ "epoch": 1.01,
715
+ "eval_accuracy": 0.7884588318085856,
716
+ "eval_loss": 0.5879009366035461,
717
+ "eval_runtime": 127.0026,
718
+ "eval_samples_per_second": 55.944,
719
+ "eval_steps_per_second": 3.504,
720
+ "step": 900
721
+ },
722
+ {
723
+ "epoch": 1.02,
724
+ "grad_norm": 0.7034109830856323,
725
+ "learning_rate": 0.00014899103139013453,
726
+ "loss": 0.3054,
727
+ "step": 910
728
+ },
729
+ {
730
+ "epoch": 1.03,
731
+ "grad_norm": 1.852588176727295,
732
+ "learning_rate": 0.0001484304932735426,
733
+ "loss": 0.4751,
734
+ "step": 920
735
+ },
736
+ {
737
+ "epoch": 1.04,
738
+ "grad_norm": 2.8306503295898438,
739
+ "learning_rate": 0.00014786995515695067,
740
+ "loss": 0.4143,
741
+ "step": 930
742
+ },
743
+ {
744
+ "epoch": 1.05,
745
+ "grad_norm": 2.520498037338257,
746
+ "learning_rate": 0.00014730941704035874,
747
+ "loss": 0.4863,
748
+ "step": 940
749
+ },
750
+ {
751
+ "epoch": 1.07,
752
+ "grad_norm": 3.4684131145477295,
753
+ "learning_rate": 0.00014674887892376684,
754
+ "loss": 0.4911,
755
+ "step": 950
756
+ },
757
+ {
758
+ "epoch": 1.08,
759
+ "grad_norm": 1.4476227760314941,
760
+ "learning_rate": 0.00014618834080717488,
761
+ "loss": 0.4375,
762
+ "step": 960
763
+ },
764
+ {
765
+ "epoch": 1.09,
766
+ "grad_norm": 2.003276824951172,
767
+ "learning_rate": 0.00014562780269058295,
768
+ "loss": 0.4116,
769
+ "step": 970
770
+ },
771
+ {
772
+ "epoch": 1.1,
773
+ "grad_norm": 2.1394705772399902,
774
+ "learning_rate": 0.00014506726457399105,
775
+ "loss": 0.3873,
776
+ "step": 980
777
+ },
778
+ {
779
+ "epoch": 1.11,
780
+ "grad_norm": 2.4954771995544434,
781
+ "learning_rate": 0.00014450672645739912,
782
+ "loss": 0.3475,
783
+ "step": 990
784
+ },
785
+ {
786
+ "epoch": 1.12,
787
+ "grad_norm": 2.449598550796509,
788
+ "learning_rate": 0.00014394618834080716,
789
+ "loss": 0.4661,
790
+ "step": 1000
791
+ },
792
+ {
793
+ "epoch": 1.12,
794
+ "eval_accuracy": 0.7887403237156931,
795
+ "eval_loss": 0.5782468914985657,
796
+ "eval_runtime": 127.3385,
797
+ "eval_samples_per_second": 55.796,
798
+ "eval_steps_per_second": 3.495,
799
+ "step": 1000
800
+ },
801
+ {
802
+ "epoch": 1.13,
803
+ "grad_norm": 2.011767864227295,
804
+ "learning_rate": 0.00014338565022421526,
805
+ "loss": 0.3975,
806
+ "step": 1010
807
+ },
808
+ {
809
+ "epoch": 1.14,
810
+ "grad_norm": 1.4032831192016602,
811
+ "learning_rate": 0.00014282511210762333,
812
+ "loss": 0.4123,
813
+ "step": 1020
814
+ },
815
+ {
816
+ "epoch": 1.15,
817
+ "grad_norm": 1.5600253343582153,
818
+ "learning_rate": 0.0001422645739910314,
819
+ "loss": 0.351,
820
+ "step": 1030
821
+ },
822
+ {
823
+ "epoch": 1.17,
824
+ "grad_norm": 3.0084493160247803,
825
+ "learning_rate": 0.00014170403587443947,
826
+ "loss": 0.4234,
827
+ "step": 1040
828
+ },
829
+ {
830
+ "epoch": 1.18,
831
+ "grad_norm": 1.441857933998108,
832
+ "learning_rate": 0.00014114349775784754,
833
+ "loss": 0.4868,
834
+ "step": 1050
835
+ },
836
+ {
837
+ "epoch": 1.19,
838
+ "grad_norm": 1.2188992500305176,
839
+ "learning_rate": 0.0001405829596412556,
840
+ "loss": 0.423,
841
+ "step": 1060
842
+ },
843
+ {
844
+ "epoch": 1.2,
845
+ "grad_norm": 1.6048247814178467,
846
+ "learning_rate": 0.0001400224215246637,
847
+ "loss": 0.4204,
848
+ "step": 1070
849
+ },
850
+ {
851
+ "epoch": 1.21,
852
+ "grad_norm": 2.915587902069092,
853
+ "learning_rate": 0.00013946188340807175,
854
+ "loss": 0.3875,
855
+ "step": 1080
856
+ },
857
+ {
858
+ "epoch": 1.22,
859
+ "grad_norm": 1.629499912261963,
860
+ "learning_rate": 0.00013890134529147982,
861
+ "loss": 0.4025,
862
+ "step": 1090
863
+ },
864
+ {
865
+ "epoch": 1.23,
866
+ "grad_norm": 2.1824419498443604,
867
+ "learning_rate": 0.00013834080717488792,
868
+ "loss": 0.358,
869
+ "step": 1100
870
+ },
871
+ {
872
+ "epoch": 1.23,
873
+ "eval_accuracy": 0.7942294159042927,
874
+ "eval_loss": 0.5690400004386902,
875
+ "eval_runtime": 127.5854,
876
+ "eval_samples_per_second": 55.688,
877
+ "eval_steps_per_second": 3.488,
878
+ "step": 1100
879
+ },
880
+ {
881
+ "epoch": 1.24,
882
+ "grad_norm": 1.4124354124069214,
883
+ "learning_rate": 0.000137780269058296,
884
+ "loss": 0.3944,
885
+ "step": 1110
886
+ },
887
+ {
888
+ "epoch": 1.26,
889
+ "grad_norm": 1.3432316780090332,
890
+ "learning_rate": 0.00013721973094170403,
891
+ "loss": 0.4344,
892
+ "step": 1120
893
+ },
894
+ {
895
+ "epoch": 1.27,
896
+ "grad_norm": 2.4002068042755127,
897
+ "learning_rate": 0.0001366591928251121,
898
+ "loss": 0.3457,
899
+ "step": 1130
900
+ },
901
+ {
902
+ "epoch": 1.28,
903
+ "grad_norm": 4.019514560699463,
904
+ "learning_rate": 0.0001360986547085202,
905
+ "loss": 0.4713,
906
+ "step": 1140
907
+ },
908
+ {
909
+ "epoch": 1.29,
910
+ "grad_norm": 1.3158634901046753,
911
+ "learning_rate": 0.00013553811659192827,
912
+ "loss": 0.3754,
913
+ "step": 1150
914
+ },
915
+ {
916
+ "epoch": 1.3,
917
+ "grad_norm": 3.053358793258667,
918
+ "learning_rate": 0.0001349775784753363,
919
+ "loss": 0.4441,
920
+ "step": 1160
921
+ },
922
+ {
923
+ "epoch": 1.31,
924
+ "grad_norm": 1.6883745193481445,
925
+ "learning_rate": 0.0001344170403587444,
926
+ "loss": 0.4562,
927
+ "step": 1170
928
+ },
929
+ {
930
+ "epoch": 1.32,
931
+ "grad_norm": 2.3927807807922363,
932
+ "learning_rate": 0.00013385650224215248,
933
+ "loss": 0.4068,
934
+ "step": 1180
935
+ },
936
+ {
937
+ "epoch": 1.33,
938
+ "grad_norm": 1.4922747611999512,
939
+ "learning_rate": 0.00013329596412556055,
940
+ "loss": 0.4174,
941
+ "step": 1190
942
+ },
943
+ {
944
+ "epoch": 1.35,
945
+ "grad_norm": 2.156853199005127,
946
+ "learning_rate": 0.00013273542600896862,
947
+ "loss": 0.3812,
948
+ "step": 1200
949
+ },
950
+ {
951
+ "epoch": 1.35,
952
+ "eval_accuracy": 0.8146375791695989,
953
+ "eval_loss": 0.5108710527420044,
954
+ "eval_runtime": 127.9489,
955
+ "eval_samples_per_second": 55.53,
956
+ "eval_steps_per_second": 3.478,
957
+ "step": 1200
958
+ },
959
+ {
960
+ "epoch": 1.36,
961
+ "grad_norm": 1.6083571910858154,
962
+ "learning_rate": 0.00013217488789237669,
963
+ "loss": 0.3351,
964
+ "step": 1210
965
+ },
966
+ {
967
+ "epoch": 1.37,
968
+ "grad_norm": 2.889650583267212,
969
+ "learning_rate": 0.00013161434977578476,
970
+ "loss": 0.4398,
971
+ "step": 1220
972
+ },
973
+ {
974
+ "epoch": 1.38,
975
+ "grad_norm": 1.9212812185287476,
976
+ "learning_rate": 0.00013105381165919283,
977
+ "loss": 0.3878,
978
+ "step": 1230
979
+ },
980
+ {
981
+ "epoch": 1.39,
982
+ "grad_norm": 1.9533714056015015,
983
+ "learning_rate": 0.0001304932735426009,
984
+ "loss": 0.3447,
985
+ "step": 1240
986
+ },
987
+ {
988
+ "epoch": 1.4,
989
+ "grad_norm": 3.091277837753296,
990
+ "learning_rate": 0.00012993273542600897,
991
+ "loss": 0.3848,
992
+ "step": 1250
993
+ },
994
+ {
995
+ "epoch": 1.41,
996
+ "grad_norm": 1.8874760866165161,
997
+ "learning_rate": 0.00012937219730941706,
998
+ "loss": 0.5295,
999
+ "step": 1260
1000
+ },
1001
+ {
1002
+ "epoch": 1.42,
1003
+ "grad_norm": 2.417236804962158,
1004
+ "learning_rate": 0.0001288116591928251,
1005
+ "loss": 0.4444,
1006
+ "step": 1270
1007
+ },
1008
+ {
1009
+ "epoch": 1.43,
1010
+ "grad_norm": 2.4967589378356934,
1011
+ "learning_rate": 0.00012825112107623318,
1012
+ "loss": 0.4944,
1013
+ "step": 1280
1014
+ },
1015
+ {
1016
+ "epoch": 1.45,
1017
+ "grad_norm": 3.3610410690307617,
1018
+ "learning_rate": 0.00012769058295964127,
1019
+ "loss": 0.3398,
1020
+ "step": 1290
1021
+ },
1022
+ {
1023
+ "epoch": 1.46,
1024
+ "grad_norm": 1.5874662399291992,
1025
+ "learning_rate": 0.00012713004484304934,
1026
+ "loss": 0.3535,
1027
+ "step": 1300
1028
+ },
1029
+ {
1030
+ "epoch": 1.46,
1031
+ "eval_accuracy": 0.8147783251231527,
1032
+ "eval_loss": 0.5213413834571838,
1033
+ "eval_runtime": 127.3924,
1034
+ "eval_samples_per_second": 55.773,
1035
+ "eval_steps_per_second": 3.493,
1036
+ "step": 1300
1037
+ },
1038
+ {
1039
+ "epoch": 1.47,
1040
+ "grad_norm": 1.2904878854751587,
1041
+ "learning_rate": 0.00012656950672645739,
1042
+ "loss": 0.3868,
1043
+ "step": 1310
1044
+ },
1045
+ {
1046
+ "epoch": 1.48,
1047
+ "grad_norm": 2.38608980178833,
1048
+ "learning_rate": 0.00012600896860986548,
1049
+ "loss": 0.4082,
1050
+ "step": 1320
1051
+ },
1052
+ {
1053
+ "epoch": 1.49,
1054
+ "grad_norm": 1.6603220701217651,
1055
+ "learning_rate": 0.00012544843049327355,
1056
+ "loss": 0.3222,
1057
+ "step": 1330
1058
+ },
1059
+ {
1060
+ "epoch": 1.5,
1061
+ "grad_norm": 1.8661950826644897,
1062
+ "learning_rate": 0.00012488789237668162,
1063
+ "loss": 0.4029,
1064
+ "step": 1340
1065
+ },
1066
+ {
1067
+ "epoch": 1.51,
1068
+ "grad_norm": 2.12640643119812,
1069
+ "learning_rate": 0.0001243273542600897,
1070
+ "loss": 0.3691,
1071
+ "step": 1350
1072
+ },
1073
+ {
1074
+ "epoch": 1.52,
1075
+ "grad_norm": 2.671631097793579,
1076
+ "learning_rate": 0.00012376681614349776,
1077
+ "loss": 0.3328,
1078
+ "step": 1360
1079
+ },
1080
+ {
1081
+ "epoch": 1.54,
1082
+ "grad_norm": 2.103508949279785,
1083
+ "learning_rate": 0.00012320627802690583,
1084
+ "loss": 0.3475,
1085
+ "step": 1370
1086
+ },
1087
+ {
1088
+ "epoch": 1.55,
1089
+ "grad_norm": 1.7701743841171265,
1090
+ "learning_rate": 0.00012264573991031393,
1091
+ "loss": 0.4423,
1092
+ "step": 1380
1093
+ },
1094
+ {
1095
+ "epoch": 1.56,
1096
+ "grad_norm": 2.38301682472229,
1097
+ "learning_rate": 0.00012208520179372197,
1098
+ "loss": 0.426,
1099
+ "step": 1390
1100
+ },
1101
+ {
1102
+ "epoch": 1.57,
1103
+ "grad_norm": 2.7273807525634766,
1104
+ "learning_rate": 0.00012152466367713004,
1105
+ "loss": 0.3901,
1106
+ "step": 1400
1107
+ },
1108
+ {
1109
+ "epoch": 1.57,
1110
+ "eval_accuracy": 0.8125263898662913,
1111
+ "eval_loss": 0.5261800289154053,
1112
+ "eval_runtime": 126.367,
1113
+ "eval_samples_per_second": 56.225,
1114
+ "eval_steps_per_second": 3.521,
1115
+ "step": 1400
1116
+ },
1117
+ {
1118
+ "epoch": 1.58,
1119
+ "grad_norm": 2.7706730365753174,
1120
+ "learning_rate": 0.00012096412556053814,
1121
+ "loss": 0.4756,
1122
+ "step": 1410
1123
+ },
1124
+ {
1125
+ "epoch": 1.59,
1126
+ "grad_norm": 2.9104268550872803,
1127
+ "learning_rate": 0.0001204035874439462,
1128
+ "loss": 0.3625,
1129
+ "step": 1420
1130
+ },
1131
+ {
1132
+ "epoch": 1.6,
1133
+ "grad_norm": 1.5548855066299438,
1134
+ "learning_rate": 0.00011984304932735426,
1135
+ "loss": 0.3352,
1136
+ "step": 1430
1137
+ },
1138
+ {
1139
+ "epoch": 1.61,
1140
+ "grad_norm": 2.2463669776916504,
1141
+ "learning_rate": 0.00011928251121076232,
1142
+ "loss": 0.3944,
1143
+ "step": 1440
1144
+ },
1145
+ {
1146
+ "epoch": 1.63,
1147
+ "grad_norm": 2.0413687229156494,
1148
+ "learning_rate": 0.00011872197309417042,
1149
+ "loss": 0.3637,
1150
+ "step": 1450
1151
+ },
1152
+ {
1153
+ "epoch": 1.64,
1154
+ "grad_norm": 2.267987012863159,
1155
+ "learning_rate": 0.00011816143497757847,
1156
+ "loss": 0.3517,
1157
+ "step": 1460
1158
+ },
1159
+ {
1160
+ "epoch": 1.65,
1161
+ "grad_norm": 1.90973699092865,
1162
+ "learning_rate": 0.00011760089686098654,
1163
+ "loss": 0.2954,
1164
+ "step": 1470
1165
+ },
1166
+ {
1167
+ "epoch": 1.66,
1168
+ "grad_norm": 1.5805819034576416,
1169
+ "learning_rate": 0.00011704035874439463,
1170
+ "loss": 0.3207,
1171
+ "step": 1480
1172
+ },
1173
+ {
1174
+ "epoch": 1.67,
1175
+ "grad_norm": 2.012744903564453,
1176
+ "learning_rate": 0.0001164798206278027,
1177
+ "loss": 0.2816,
1178
+ "step": 1490
1179
+ },
1180
+ {
1181
+ "epoch": 1.68,
1182
+ "grad_norm": 1.0160799026489258,
1183
+ "learning_rate": 0.00011591928251121075,
1184
+ "loss": 0.3276,
1185
+ "step": 1500
1186
+ },
1187
+ {
1188
+ "epoch": 1.68,
1189
+ "eval_accuracy": 0.8081632653061225,
1190
+ "eval_loss": 0.5793688893318176,
1191
+ "eval_runtime": 128.0301,
1192
+ "eval_samples_per_second": 55.495,
1193
+ "eval_steps_per_second": 3.476,
1194
+ "step": 1500
1195
+ },
1196
+ {
1197
+ "epoch": 1.69,
1198
+ "grad_norm": 2.1283419132232666,
1199
+ "learning_rate": 0.00011535874439461885,
1200
+ "loss": 0.3697,
1201
+ "step": 1510
1202
+ },
1203
+ {
1204
+ "epoch": 1.7,
1205
+ "grad_norm": 2.384875774383545,
1206
+ "learning_rate": 0.00011479820627802691,
1207
+ "loss": 0.4283,
1208
+ "step": 1520
1209
+ },
1210
+ {
1211
+ "epoch": 1.72,
1212
+ "grad_norm": 1.6091313362121582,
1213
+ "learning_rate": 0.00011423766816143498,
1214
+ "loss": 0.2917,
1215
+ "step": 1530
1216
+ },
1217
+ {
1218
+ "epoch": 1.73,
1219
+ "grad_norm": 2.50388503074646,
1220
+ "learning_rate": 0.00011367713004484306,
1221
+ "loss": 0.3705,
1222
+ "step": 1540
1223
+ },
1224
+ {
1225
+ "epoch": 1.74,
1226
+ "grad_norm": 2.4337923526763916,
1227
+ "learning_rate": 0.00011311659192825113,
1228
+ "loss": 0.3724,
1229
+ "step": 1550
1230
+ },
1231
+ {
1232
+ "epoch": 1.75,
1233
+ "grad_norm": 1.677294373512268,
1234
+ "learning_rate": 0.00011255605381165919,
1235
+ "loss": 0.3492,
1236
+ "step": 1560
1237
+ },
1238
+ {
1239
+ "epoch": 1.76,
1240
+ "grad_norm": 1.634697675704956,
1241
+ "learning_rate": 0.00011199551569506727,
1242
+ "loss": 0.3633,
1243
+ "step": 1570
1244
+ },
1245
+ {
1246
+ "epoch": 1.77,
1247
+ "grad_norm": 2.2558953762054443,
1248
+ "learning_rate": 0.00011143497757847534,
1249
+ "loss": 0.3875,
1250
+ "step": 1580
1251
+ },
1252
+ {
1253
+ "epoch": 1.78,
1254
+ "grad_norm": 1.5888583660125732,
1255
+ "learning_rate": 0.00011087443946188341,
1256
+ "loss": 0.3323,
1257
+ "step": 1590
1258
+ },
1259
+ {
1260
+ "epoch": 1.79,
1261
+ "grad_norm": 2.265427827835083,
1262
+ "learning_rate": 0.0001103139013452915,
1263
+ "loss": 0.3679,
1264
+ "step": 1600
1265
+ },
1266
+ {
1267
+ "epoch": 1.79,
1268
+ "eval_accuracy": 0.8115411681914145,
1269
+ "eval_loss": 0.5365468859672546,
1270
+ "eval_runtime": 127.8881,
1271
+ "eval_samples_per_second": 55.556,
1272
+ "eval_steps_per_second": 3.48,
1273
+ "step": 1600
1274
+ },
1275
+ {
1276
+ "epoch": 1.8,
1277
+ "grad_norm": 2.279982566833496,
1278
+ "learning_rate": 0.00010975336322869956,
1279
+ "loss": 0.2996,
1280
+ "step": 1610
1281
+ },
1282
+ {
1283
+ "epoch": 1.82,
1284
+ "grad_norm": 1.905090093612671,
1285
+ "learning_rate": 0.00010919282511210762,
1286
+ "loss": 0.3268,
1287
+ "step": 1620
1288
+ },
1289
+ {
1290
+ "epoch": 1.83,
1291
+ "grad_norm": 4.301815032958984,
1292
+ "learning_rate": 0.0001086322869955157,
1293
+ "loss": 0.3312,
1294
+ "step": 1630
1295
+ },
1296
+ {
1297
+ "epoch": 1.84,
1298
+ "grad_norm": 1.8537272214889526,
1299
+ "learning_rate": 0.00010807174887892377,
1300
+ "loss": 0.3654,
1301
+ "step": 1640
1302
+ },
1303
+ {
1304
+ "epoch": 1.85,
1305
+ "grad_norm": 1.3462457656860352,
1306
+ "learning_rate": 0.00010751121076233184,
1307
+ "loss": 0.346,
1308
+ "step": 1650
1309
+ },
1310
+ {
1311
+ "epoch": 1.86,
1312
+ "grad_norm": 2.4433109760284424,
1313
+ "learning_rate": 0.00010695067264573993,
1314
+ "loss": 0.3563,
1315
+ "step": 1660
1316
+ },
1317
+ {
1318
+ "epoch": 1.87,
1319
+ "grad_norm": 1.8175091743469238,
1320
+ "learning_rate": 0.00010639013452914798,
1321
+ "loss": 0.3065,
1322
+ "step": 1670
1323
+ },
1324
+ {
1325
+ "epoch": 1.88,
1326
+ "grad_norm": 2.724806547164917,
1327
+ "learning_rate": 0.00010582959641255605,
1328
+ "loss": 0.3949,
1329
+ "step": 1680
1330
+ },
1331
+ {
1332
+ "epoch": 1.89,
1333
+ "grad_norm": 1.4651908874511719,
1334
+ "learning_rate": 0.00010526905829596414,
1335
+ "loss": 0.3451,
1336
+ "step": 1690
1337
+ },
1338
+ {
1339
+ "epoch": 1.91,
1340
+ "grad_norm": 1.2862894535064697,
1341
+ "learning_rate": 0.0001047085201793722,
1342
+ "loss": 0.3077,
1343
+ "step": 1700
1344
+ },
1345
+ {
1346
+ "epoch": 1.91,
1347
+ "eval_accuracy": 0.825615763546798,
1348
+ "eval_loss": 0.503186047077179,
1349
+ "eval_runtime": 127.5407,
1350
+ "eval_samples_per_second": 55.708,
1351
+ "eval_steps_per_second": 3.489,
1352
+ "step": 1700
1353
+ },
1354
+ {
1355
+ "epoch": 1.92,
1356
+ "grad_norm": 1.5572246313095093,
1357
+ "learning_rate": 0.00010414798206278026,
1358
+ "loss": 0.3638,
1359
+ "step": 1710
1360
+ },
1361
+ {
1362
+ "epoch": 1.93,
1363
+ "grad_norm": 3.091179847717285,
1364
+ "learning_rate": 0.00010358744394618836,
1365
+ "loss": 0.4205,
1366
+ "step": 1720
1367
+ },
1368
+ {
1369
+ "epoch": 1.94,
1370
+ "grad_norm": 2.097266435623169,
1371
+ "learning_rate": 0.00010302690582959642,
1372
+ "loss": 0.2865,
1373
+ "step": 1730
1374
+ },
1375
+ {
1376
+ "epoch": 1.95,
1377
+ "grad_norm": 3.9726853370666504,
1378
+ "learning_rate": 0.00010246636771300449,
1379
+ "loss": 0.3283,
1380
+ "step": 1740
1381
+ },
1382
+ {
1383
+ "epoch": 1.96,
1384
+ "grad_norm": 2.2275006771087646,
1385
+ "learning_rate": 0.00010190582959641257,
1386
+ "loss": 0.2865,
1387
+ "step": 1750
1388
+ },
1389
+ {
1390
+ "epoch": 1.97,
1391
+ "grad_norm": 2.571467161178589,
1392
+ "learning_rate": 0.00010134529147982064,
1393
+ "loss": 0.381,
1394
+ "step": 1760
1395
+ },
1396
+ {
1397
+ "epoch": 1.98,
1398
+ "grad_norm": 1.8339025974273682,
1399
+ "learning_rate": 0.0001007847533632287,
1400
+ "loss": 0.3399,
1401
+ "step": 1770
1402
+ },
1403
+ {
1404
+ "epoch": 2.0,
1405
+ "grad_norm": 1.96084463596344,
1406
+ "learning_rate": 0.00010022421524663677,
1407
+ "loss": 0.4104,
1408
+ "step": 1780
1409
+ },
1410
+ {
1411
+ "epoch": 2.01,
1412
+ "grad_norm": 1.5350396633148193,
1413
+ "learning_rate": 9.966367713004485e-05,
1414
+ "loss": 0.2616,
1415
+ "step": 1790
1416
+ },
1417
+ {
1418
+ "epoch": 2.02,
1419
+ "grad_norm": 1.11006760597229,
1420
+ "learning_rate": 9.910313901345292e-05,
1421
+ "loss": 0.1593,
1422
+ "step": 1800
1423
+ },
1424
+ {
1425
+ "epoch": 2.02,
1426
+ "eval_accuracy": 0.8237860661505981,
1427
+ "eval_loss": 0.4946657121181488,
1428
+ "eval_runtime": 127.5397,
1429
+ "eval_samples_per_second": 55.708,
1430
+ "eval_steps_per_second": 3.489,
1431
+ "step": 1800
1432
+ },
1433
+ {
1434
+ "epoch": 2.03,
1435
+ "grad_norm": 1.9963308572769165,
1436
+ "learning_rate": 9.854260089686099e-05,
1437
+ "loss": 0.1422,
1438
+ "step": 1810
1439
+ },
1440
+ {
1441
+ "epoch": 2.04,
1442
+ "grad_norm": 1.350595235824585,
1443
+ "learning_rate": 9.798206278026907e-05,
1444
+ "loss": 0.2012,
1445
+ "step": 1820
1446
+ },
1447
+ {
1448
+ "epoch": 2.05,
1449
+ "grad_norm": 1.9206656217575073,
1450
+ "learning_rate": 9.742152466367713e-05,
1451
+ "loss": 0.2354,
1452
+ "step": 1830
1453
+ },
1454
+ {
1455
+ "epoch": 2.06,
1456
+ "grad_norm": 0.9061004519462585,
1457
+ "learning_rate": 9.686098654708521e-05,
1458
+ "loss": 0.1971,
1459
+ "step": 1840
1460
+ },
1461
+ {
1462
+ "epoch": 2.07,
1463
+ "grad_norm": 1.090854287147522,
1464
+ "learning_rate": 9.630044843049327e-05,
1465
+ "loss": 0.2035,
1466
+ "step": 1850
1467
+ },
1468
+ {
1469
+ "epoch": 2.09,
1470
+ "grad_norm": 1.798594355583191,
1471
+ "learning_rate": 9.573991031390135e-05,
1472
+ "loss": 0.2362,
1473
+ "step": 1860
1474
+ },
1475
+ {
1476
+ "epoch": 2.1,
1477
+ "grad_norm": 0.2856987416744232,
1478
+ "learning_rate": 9.517937219730942e-05,
1479
+ "loss": 0.1448,
1480
+ "step": 1870
1481
+ },
1482
+ {
1483
+ "epoch": 2.11,
1484
+ "grad_norm": 1.9062001705169678,
1485
+ "learning_rate": 9.461883408071749e-05,
1486
+ "loss": 0.1019,
1487
+ "step": 1880
1488
+ },
1489
+ {
1490
+ "epoch": 2.12,
1491
+ "grad_norm": 1.8595843315124512,
1492
+ "learning_rate": 9.405829596412556e-05,
1493
+ "loss": 0.1844,
1494
+ "step": 1890
1495
+ },
1496
+ {
1497
+ "epoch": 2.13,
1498
+ "grad_norm": 1.6841151714324951,
1499
+ "learning_rate": 9.349775784753365e-05,
1500
+ "loss": 0.2495,
1501
+ "step": 1900
1502
+ },
1503
+ {
1504
+ "epoch": 2.13,
1505
+ "eval_accuracy": 0.8212526389866291,
1506
+ "eval_loss": 0.5188373327255249,
1507
+ "eval_runtime": 127.5963,
1508
+ "eval_samples_per_second": 55.683,
1509
+ "eval_steps_per_second": 3.488,
1510
+ "step": 1900
1511
+ },
1512
+ {
1513
+ "epoch": 2.14,
1514
+ "grad_norm": 2.3805599212646484,
1515
+ "learning_rate": 9.29372197309417e-05,
1516
+ "loss": 0.1725,
1517
+ "step": 1910
1518
+ },
1519
+ {
1520
+ "epoch": 2.15,
1521
+ "grad_norm": 1.5507524013519287,
1522
+ "learning_rate": 9.237668161434979e-05,
1523
+ "loss": 0.1951,
1524
+ "step": 1920
1525
+ },
1526
+ {
1527
+ "epoch": 2.16,
1528
+ "grad_norm": 4.487265586853027,
1529
+ "learning_rate": 9.181614349775786e-05,
1530
+ "loss": 0.1928,
1531
+ "step": 1930
1532
+ },
1533
+ {
1534
+ "epoch": 2.17,
1535
+ "grad_norm": 5.267577171325684,
1536
+ "learning_rate": 9.125560538116593e-05,
1537
+ "loss": 0.2279,
1538
+ "step": 1940
1539
+ },
1540
+ {
1541
+ "epoch": 2.19,
1542
+ "grad_norm": 1.9050337076187134,
1543
+ "learning_rate": 9.0695067264574e-05,
1544
+ "loss": 0.1408,
1545
+ "step": 1950
1546
+ },
1547
+ {
1548
+ "epoch": 2.2,
1549
+ "grad_norm": 2.7110774517059326,
1550
+ "learning_rate": 9.013452914798208e-05,
1551
+ "loss": 0.1851,
1552
+ "step": 1960
1553
+ },
1554
+ {
1555
+ "epoch": 2.21,
1556
+ "grad_norm": 2.4091663360595703,
1557
+ "learning_rate": 8.957399103139014e-05,
1558
+ "loss": 0.1961,
1559
+ "step": 1970
1560
+ },
1561
+ {
1562
+ "epoch": 2.22,
1563
+ "grad_norm": 1.0745985507965088,
1564
+ "learning_rate": 8.901345291479822e-05,
1565
+ "loss": 0.1807,
1566
+ "step": 1980
1567
+ },
1568
+ {
1569
+ "epoch": 2.23,
1570
+ "grad_norm": 1.907657504081726,
1571
+ "learning_rate": 8.845291479820629e-05,
1572
+ "loss": 0.1656,
1573
+ "step": 1990
1574
+ },
1575
+ {
1576
+ "epoch": 2.24,
1577
+ "grad_norm": 1.4196101427078247,
1578
+ "learning_rate": 8.789237668161436e-05,
1579
+ "loss": 0.1604,
1580
+ "step": 2000
1581
+ },
1582
+ {
1583
+ "epoch": 2.24,
1584
+ "eval_accuracy": 0.8457424349049965,
1585
+ "eval_loss": 0.47485658526420593,
1586
+ "eval_runtime": 127.5001,
1587
+ "eval_samples_per_second": 55.725,
1588
+ "eval_steps_per_second": 3.49,
1589
+ "step": 2000
1590
+ },
1591
+ {
1592
+ "epoch": 2.25,
1593
+ "grad_norm": 0.8819741606712341,
1594
+ "learning_rate": 8.733183856502243e-05,
1595
+ "loss": 0.1541,
1596
+ "step": 2010
1597
+ },
1598
+ {
1599
+ "epoch": 2.26,
1600
+ "grad_norm": 2.7008216381073,
1601
+ "learning_rate": 8.67713004484305e-05,
1602
+ "loss": 0.1754,
1603
+ "step": 2020
1604
+ },
1605
+ {
1606
+ "epoch": 2.28,
1607
+ "grad_norm": 1.3049780130386353,
1608
+ "learning_rate": 8.621076233183857e-05,
1609
+ "loss": 0.1474,
1610
+ "step": 2030
1611
+ },
1612
+ {
1613
+ "epoch": 2.29,
1614
+ "grad_norm": 1.7845088243484497,
1615
+ "learning_rate": 8.565022421524664e-05,
1616
+ "loss": 0.149,
1617
+ "step": 2040
1618
+ },
1619
+ {
1620
+ "epoch": 2.3,
1621
+ "grad_norm": 2.162095546722412,
1622
+ "learning_rate": 8.508968609865471e-05,
1623
+ "loss": 0.1711,
1624
+ "step": 2050
1625
+ },
1626
+ {
1627
+ "epoch": 2.31,
1628
+ "grad_norm": 2.4429993629455566,
1629
+ "learning_rate": 8.452914798206278e-05,
1630
+ "loss": 0.1278,
1631
+ "step": 2060
1632
+ },
1633
+ {
1634
+ "epoch": 2.32,
1635
+ "grad_norm": 4.209596157073975,
1636
+ "learning_rate": 8.396860986547086e-05,
1637
+ "loss": 0.2285,
1638
+ "step": 2070
1639
+ },
1640
+ {
1641
+ "epoch": 2.33,
1642
+ "grad_norm": 0.8332444429397583,
1643
+ "learning_rate": 8.340807174887892e-05,
1644
+ "loss": 0.1706,
1645
+ "step": 2080
1646
+ },
1647
+ {
1648
+ "epoch": 2.34,
1649
+ "grad_norm": 1.2180029153823853,
1650
+ "learning_rate": 8.2847533632287e-05,
1651
+ "loss": 0.1863,
1652
+ "step": 2090
1653
+ },
1654
+ {
1655
+ "epoch": 2.35,
1656
+ "grad_norm": 1.7199923992156982,
1657
+ "learning_rate": 8.228699551569507e-05,
1658
+ "loss": 0.1347,
1659
+ "step": 2100
1660
+ },
1661
+ {
1662
+ "epoch": 2.35,
1663
+ "eval_accuracy": 0.8318085855031668,
1664
+ "eval_loss": 0.4878062307834625,
1665
+ "eval_runtime": 127.098,
1666
+ "eval_samples_per_second": 55.902,
1667
+ "eval_steps_per_second": 3.501,
1668
+ "step": 2100
1669
+ },
1670
+ {
1671
+ "epoch": 2.37,
1672
+ "grad_norm": 2.4405524730682373,
1673
+ "learning_rate": 8.172645739910314e-05,
1674
+ "loss": 0.1669,
1675
+ "step": 2110
1676
+ },
1677
+ {
1678
+ "epoch": 2.38,
1679
+ "grad_norm": 3.2856411933898926,
1680
+ "learning_rate": 8.116591928251121e-05,
1681
+ "loss": 0.1526,
1682
+ "step": 2120
1683
+ },
1684
+ {
1685
+ "epoch": 2.39,
1686
+ "grad_norm": 2.642458915710449,
1687
+ "learning_rate": 8.06053811659193e-05,
1688
+ "loss": 0.1866,
1689
+ "step": 2130
1690
+ },
1691
+ {
1692
+ "epoch": 2.4,
1693
+ "grad_norm": 1.8131886720657349,
1694
+ "learning_rate": 8.004484304932735e-05,
1695
+ "loss": 0.1413,
1696
+ "step": 2140
1697
+ },
1698
+ {
1699
+ "epoch": 2.41,
1700
+ "grad_norm": 2.279311418533325,
1701
+ "learning_rate": 7.948430493273543e-05,
1702
+ "loss": 0.1764,
1703
+ "step": 2150
1704
+ },
1705
+ {
1706
+ "epoch": 2.42,
1707
+ "grad_norm": 3.409904956817627,
1708
+ "learning_rate": 7.892376681614349e-05,
1709
+ "loss": 0.1697,
1710
+ "step": 2160
1711
+ },
1712
+ {
1713
+ "epoch": 2.43,
1714
+ "grad_norm": 1.6139248609542847,
1715
+ "learning_rate": 7.836322869955157e-05,
1716
+ "loss": 0.2257,
1717
+ "step": 2170
1718
+ },
1719
+ {
1720
+ "epoch": 2.44,
1721
+ "grad_norm": 1.9628515243530273,
1722
+ "learning_rate": 7.780269058295964e-05,
1723
+ "loss": 0.1869,
1724
+ "step": 2180
1725
+ },
1726
+ {
1727
+ "epoch": 2.46,
1728
+ "grad_norm": 2.0070972442626953,
1729
+ "learning_rate": 7.724215246636771e-05,
1730
+ "loss": 0.2005,
1731
+ "step": 2190
1732
+ },
1733
+ {
1734
+ "epoch": 2.47,
1735
+ "grad_norm": 1.0854668617248535,
1736
+ "learning_rate": 7.668161434977578e-05,
1737
+ "loss": 0.1723,
1738
+ "step": 2200
1739
+ },
1740
+ {
1741
+ "epoch": 2.47,
1742
+ "eval_accuracy": 0.8441942294159043,
1743
+ "eval_loss": 0.4731413722038269,
1744
+ "eval_runtime": 127.6751,
1745
+ "eval_samples_per_second": 55.649,
1746
+ "eval_steps_per_second": 3.485,
1747
+ "step": 2200
1748
+ },
1749
+ {
1750
+ "epoch": 2.48,
1751
+ "grad_norm": 3.4323694705963135,
1752
+ "learning_rate": 7.612107623318387e-05,
1753
+ "loss": 0.1829,
1754
+ "step": 2210
1755
+ },
1756
+ {
1757
+ "epoch": 2.49,
1758
+ "grad_norm": 1.0630773305892944,
1759
+ "learning_rate": 7.556053811659192e-05,
1760
+ "loss": 0.1704,
1761
+ "step": 2220
1762
+ },
1763
+ {
1764
+ "epoch": 2.5,
1765
+ "grad_norm": 0.900248646736145,
1766
+ "learning_rate": 7.500000000000001e-05,
1767
+ "loss": 0.1428,
1768
+ "step": 2230
1769
+ },
1770
+ {
1771
+ "epoch": 2.51,
1772
+ "grad_norm": 0.8738330602645874,
1773
+ "learning_rate": 7.443946188340808e-05,
1774
+ "loss": 0.1403,
1775
+ "step": 2240
1776
+ },
1777
+ {
1778
+ "epoch": 2.52,
1779
+ "grad_norm": 2.873507499694824,
1780
+ "learning_rate": 7.387892376681615e-05,
1781
+ "loss": 0.1267,
1782
+ "step": 2250
1783
+ },
1784
+ {
1785
+ "epoch": 2.53,
1786
+ "grad_norm": 3.962599515914917,
1787
+ "learning_rate": 7.331838565022422e-05,
1788
+ "loss": 0.1783,
1789
+ "step": 2260
1790
+ },
1791
+ {
1792
+ "epoch": 2.54,
1793
+ "grad_norm": 1.1607624292373657,
1794
+ "learning_rate": 7.27578475336323e-05,
1795
+ "loss": 0.1067,
1796
+ "step": 2270
1797
+ },
1798
+ {
1799
+ "epoch": 2.56,
1800
+ "grad_norm": 3.1270833015441895,
1801
+ "learning_rate": 7.219730941704036e-05,
1802
+ "loss": 0.1542,
1803
+ "step": 2280
1804
+ },
1805
+ {
1806
+ "epoch": 2.57,
1807
+ "grad_norm": 4.381764888763428,
1808
+ "learning_rate": 7.163677130044844e-05,
1809
+ "loss": 0.1032,
1810
+ "step": 2290
1811
+ },
1812
+ {
1813
+ "epoch": 2.58,
1814
+ "grad_norm": 4.008007526397705,
1815
+ "learning_rate": 7.107623318385651e-05,
1816
+ "loss": 0.1235,
1817
+ "step": 2300
1818
+ },
1819
+ {
1820
+ "epoch": 2.58,
1821
+ "eval_accuracy": 0.8450387051372273,
1822
+ "eval_loss": 0.493280827999115,
1823
+ "eval_runtime": 127.613,
1824
+ "eval_samples_per_second": 55.676,
1825
+ "eval_steps_per_second": 3.487,
1826
+ "step": 2300
1827
+ },
1828
+ {
1829
+ "epoch": 2.59,
1830
+ "grad_norm": 2.7960009574890137,
1831
+ "learning_rate": 7.051569506726458e-05,
1832
+ "loss": 0.1695,
1833
+ "step": 2310
1834
+ },
1835
+ {
1836
+ "epoch": 2.6,
1837
+ "grad_norm": 3.1904006004333496,
1838
+ "learning_rate": 6.995515695067265e-05,
1839
+ "loss": 0.1388,
1840
+ "step": 2320
1841
+ },
1842
+ {
1843
+ "epoch": 2.61,
1844
+ "grad_norm": 3.1949515342712402,
1845
+ "learning_rate": 6.939461883408072e-05,
1846
+ "loss": 0.2264,
1847
+ "step": 2330
1848
+ },
1849
+ {
1850
+ "epoch": 2.62,
1851
+ "grad_norm": 2.386139154434204,
1852
+ "learning_rate": 6.883408071748879e-05,
1853
+ "loss": 0.1498,
1854
+ "step": 2340
1855
+ },
1856
+ {
1857
+ "epoch": 2.63,
1858
+ "grad_norm": 2.6440839767456055,
1859
+ "learning_rate": 6.827354260089687e-05,
1860
+ "loss": 0.1445,
1861
+ "step": 2350
1862
+ },
1863
+ {
1864
+ "epoch": 2.65,
1865
+ "grad_norm": 0.2900611162185669,
1866
+ "learning_rate": 6.771300448430493e-05,
1867
+ "loss": 0.1245,
1868
+ "step": 2360
1869
+ },
1870
+ {
1871
+ "epoch": 2.66,
1872
+ "grad_norm": 3.771578073501587,
1873
+ "learning_rate": 6.715246636771301e-05,
1874
+ "loss": 0.1775,
1875
+ "step": 2370
1876
+ },
1877
+ {
1878
+ "epoch": 2.67,
1879
+ "grad_norm": 0.2707236707210541,
1880
+ "learning_rate": 6.659192825112108e-05,
1881
+ "loss": 0.1317,
1882
+ "step": 2380
1883
+ },
1884
+ {
1885
+ "epoch": 2.68,
1886
+ "grad_norm": 2.4165215492248535,
1887
+ "learning_rate": 6.603139013452915e-05,
1888
+ "loss": 0.1645,
1889
+ "step": 2390
1890
+ },
1891
+ {
1892
+ "epoch": 2.69,
1893
+ "grad_norm": 2.579758882522583,
1894
+ "learning_rate": 6.547085201793722e-05,
1895
+ "loss": 0.1752,
1896
+ "step": 2400
1897
+ },
1898
+ {
1899
+ "epoch": 2.69,
1900
+ "eval_accuracy": 0.8501055594651654,
1901
+ "eval_loss": 0.47405895590782166,
1902
+ "eval_runtime": 127.5154,
1903
+ "eval_samples_per_second": 55.719,
1904
+ "eval_steps_per_second": 3.49,
1905
+ "step": 2400
1906
+ },
1907
+ {
1908
+ "epoch": 2.7,
1909
+ "grad_norm": 2.206603527069092,
1910
+ "learning_rate": 6.491031390134529e-05,
1911
+ "loss": 0.2482,
1912
+ "step": 2410
1913
+ },
1914
+ {
1915
+ "epoch": 2.71,
1916
+ "grad_norm": 1.586669921875,
1917
+ "learning_rate": 6.434977578475336e-05,
1918
+ "loss": 0.1746,
1919
+ "step": 2420
1920
+ },
1921
+ {
1922
+ "epoch": 2.72,
1923
+ "grad_norm": 2.046320676803589,
1924
+ "learning_rate": 6.378923766816143e-05,
1925
+ "loss": 0.1386,
1926
+ "step": 2430
1927
+ },
1928
+ {
1929
+ "epoch": 2.74,
1930
+ "grad_norm": 0.8042988181114197,
1931
+ "learning_rate": 6.322869955156952e-05,
1932
+ "loss": 0.1383,
1933
+ "step": 2440
1934
+ },
1935
+ {
1936
+ "epoch": 2.75,
1937
+ "grad_norm": 3.815175771713257,
1938
+ "learning_rate": 6.266816143497759e-05,
1939
+ "loss": 0.1394,
1940
+ "step": 2450
1941
+ },
1942
+ {
1943
+ "epoch": 2.76,
1944
+ "grad_norm": 2.830374002456665,
1945
+ "learning_rate": 6.210762331838566e-05,
1946
+ "loss": 0.1511,
1947
+ "step": 2460
1948
+ },
1949
+ {
1950
+ "epoch": 2.77,
1951
+ "grad_norm": 2.1348299980163574,
1952
+ "learning_rate": 6.154708520179373e-05,
1953
+ "loss": 0.1295,
1954
+ "step": 2470
1955
+ },
1956
+ {
1957
+ "epoch": 2.78,
1958
+ "grad_norm": 4.810758590698242,
1959
+ "learning_rate": 6.0986547085201795e-05,
1960
+ "loss": 0.1757,
1961
+ "step": 2480
1962
+ },
1963
+ {
1964
+ "epoch": 2.79,
1965
+ "grad_norm": 1.4163758754730225,
1966
+ "learning_rate": 6.042600896860987e-05,
1967
+ "loss": 0.0962,
1968
+ "step": 2490
1969
+ },
1970
+ {
1971
+ "epoch": 2.8,
1972
+ "grad_norm": 2.047985792160034,
1973
+ "learning_rate": 5.9865470852017935e-05,
1974
+ "loss": 0.1421,
1975
+ "step": 2500
1976
+ },
1977
+ {
1978
+ "epoch": 2.8,
1979
+ "eval_accuracy": 0.8474313863476425,
1980
+ "eval_loss": 0.4880400598049164,
1981
+ "eval_runtime": 127.7239,
1982
+ "eval_samples_per_second": 55.628,
1983
+ "eval_steps_per_second": 3.484,
1984
+ "step": 2500
1985
+ },
1986
+ {
1987
+ "epoch": 2.81,
1988
+ "grad_norm": 1.3623876571655273,
1989
+ "learning_rate": 5.930493273542601e-05,
1990
+ "loss": 0.1455,
1991
+ "step": 2510
1992
+ },
1993
+ {
1994
+ "epoch": 2.83,
1995
+ "grad_norm": 1.8722714185714722,
1996
+ "learning_rate": 5.874439461883409e-05,
1997
+ "loss": 0.1755,
1998
+ "step": 2520
1999
+ },
2000
+ {
2001
+ "epoch": 2.84,
2002
+ "grad_norm": 2.139150619506836,
2003
+ "learning_rate": 5.818385650224215e-05,
2004
+ "loss": 0.1471,
2005
+ "step": 2530
2006
+ },
2007
+ {
2008
+ "epoch": 2.85,
2009
+ "grad_norm": 1.041858434677124,
2010
+ "learning_rate": 5.762331838565023e-05,
2011
+ "loss": 0.1372,
2012
+ "step": 2540
2013
+ },
2014
+ {
2015
+ "epoch": 2.86,
2016
+ "grad_norm": 0.6558467149734497,
2017
+ "learning_rate": 5.7062780269058305e-05,
2018
+ "loss": 0.1184,
2019
+ "step": 2550
2020
+ },
2021
+ {
2022
+ "epoch": 2.87,
2023
+ "grad_norm": 3.568887233734131,
2024
+ "learning_rate": 5.650224215246637e-05,
2025
+ "loss": 0.1752,
2026
+ "step": 2560
2027
+ },
2028
+ {
2029
+ "epoch": 2.88,
2030
+ "grad_norm": 0.7773478627204895,
2031
+ "learning_rate": 5.5941704035874445e-05,
2032
+ "loss": 0.1491,
2033
+ "step": 2570
2034
+ },
2035
+ {
2036
+ "epoch": 2.89,
2037
+ "grad_norm": 2.827122688293457,
2038
+ "learning_rate": 5.5381165919282515e-05,
2039
+ "loss": 0.1893,
2040
+ "step": 2580
2041
+ },
2042
+ {
2043
+ "epoch": 2.9,
2044
+ "grad_norm": 3.533275842666626,
2045
+ "learning_rate": 5.4820627802690585e-05,
2046
+ "loss": 0.1117,
2047
+ "step": 2590
2048
+ },
2049
+ {
2050
+ "epoch": 2.91,
2051
+ "grad_norm": 0.7163364887237549,
2052
+ "learning_rate": 5.426008968609866e-05,
2053
+ "loss": 0.1549,
2054
+ "step": 2600
2055
+ },
2056
+ {
2057
+ "epoch": 2.91,
2058
+ "eval_accuracy": 0.8389866291344124,
2059
+ "eval_loss": 0.4745788276195526,
2060
+ "eval_runtime": 127.6611,
2061
+ "eval_samples_per_second": 55.655,
2062
+ "eval_steps_per_second": 3.486,
2063
+ "step": 2600
2064
+ },
2065
+ {
2066
+ "epoch": 2.93,
2067
+ "grad_norm": 2.184201717376709,
2068
+ "learning_rate": 5.369955156950673e-05,
2069
+ "loss": 0.1698,
2070
+ "step": 2610
2071
+ },
2072
+ {
2073
+ "epoch": 2.94,
2074
+ "grad_norm": 2.9737119674682617,
2075
+ "learning_rate": 5.31390134529148e-05,
2076
+ "loss": 0.1189,
2077
+ "step": 2620
2078
+ },
2079
+ {
2080
+ "epoch": 2.95,
2081
+ "grad_norm": 1.0814894437789917,
2082
+ "learning_rate": 5.257847533632287e-05,
2083
+ "loss": 0.132,
2084
+ "step": 2630
2085
+ },
2086
+ {
2087
+ "epoch": 2.96,
2088
+ "grad_norm": 0.9624450206756592,
2089
+ "learning_rate": 5.201793721973094e-05,
2090
+ "loss": 0.1032,
2091
+ "step": 2640
2092
+ },
2093
+ {
2094
+ "epoch": 2.97,
2095
+ "grad_norm": 2.6256656646728516,
2096
+ "learning_rate": 5.145739910313902e-05,
2097
+ "loss": 0.1592,
2098
+ "step": 2650
2099
+ },
2100
+ {
2101
+ "epoch": 2.98,
2102
+ "grad_norm": 3.0557103157043457,
2103
+ "learning_rate": 5.089686098654709e-05,
2104
+ "loss": 0.1327,
2105
+ "step": 2660
2106
+ },
2107
+ {
2108
+ "epoch": 2.99,
2109
+ "grad_norm": 2.0262203216552734,
2110
+ "learning_rate": 5.033632286995516e-05,
2111
+ "loss": 0.1487,
2112
+ "step": 2670
2113
+ },
2114
+ {
2115
+ "epoch": 3.0,
2116
+ "grad_norm": 0.14867419004440308,
2117
+ "learning_rate": 4.977578475336323e-05,
2118
+ "loss": 0.1005,
2119
+ "step": 2680
2120
+ },
2121
+ {
2122
+ "epoch": 3.02,
2123
+ "grad_norm": 0.12747132778167725,
2124
+ "learning_rate": 4.92152466367713e-05,
2125
+ "loss": 0.0543,
2126
+ "step": 2690
2127
+ },
2128
+ {
2129
+ "epoch": 3.03,
2130
+ "grad_norm": 0.1839972287416458,
2131
+ "learning_rate": 4.8654708520179374e-05,
2132
+ "loss": 0.0617,
2133
+ "step": 2700
2134
+ },
2135
+ {
2136
+ "epoch": 3.03,
2137
+ "eval_accuracy": 0.8496833216045039,
2138
+ "eval_loss": 0.4935864508152008,
2139
+ "eval_runtime": 127.5847,
2140
+ "eval_samples_per_second": 55.688,
2141
+ "eval_steps_per_second": 3.488,
2142
+ "step": 2700
2143
+ },
2144
+ {
2145
+ "epoch": 3.04,
2146
+ "grad_norm": 1.8856990337371826,
2147
+ "learning_rate": 4.8094170403587444e-05,
2148
+ "loss": 0.0687,
2149
+ "step": 2710
2150
+ },
2151
+ {
2152
+ "epoch": 3.05,
2153
+ "grad_norm": 0.39913854002952576,
2154
+ "learning_rate": 4.7533632286995514e-05,
2155
+ "loss": 0.0553,
2156
+ "step": 2720
2157
+ },
2158
+ {
2159
+ "epoch": 3.06,
2160
+ "grad_norm": 0.8423967957496643,
2161
+ "learning_rate": 4.697309417040359e-05,
2162
+ "loss": 0.0797,
2163
+ "step": 2730
2164
+ },
2165
+ {
2166
+ "epoch": 3.07,
2167
+ "grad_norm": 0.2557125687599182,
2168
+ "learning_rate": 4.641255605381166e-05,
2169
+ "loss": 0.08,
2170
+ "step": 2740
2171
+ },
2172
+ {
2173
+ "epoch": 3.08,
2174
+ "grad_norm": 2.2399497032165527,
2175
+ "learning_rate": 4.585201793721973e-05,
2176
+ "loss": 0.0483,
2177
+ "step": 2750
2178
+ },
2179
+ {
2180
+ "epoch": 3.09,
2181
+ "grad_norm": 0.04336220771074295,
2182
+ "learning_rate": 4.52914798206278e-05,
2183
+ "loss": 0.0339,
2184
+ "step": 2760
2185
+ },
2186
+ {
2187
+ "epoch": 3.11,
2188
+ "grad_norm": 0.29698485136032104,
2189
+ "learning_rate": 4.473094170403588e-05,
2190
+ "loss": 0.0579,
2191
+ "step": 2770
2192
+ },
2193
+ {
2194
+ "epoch": 3.12,
2195
+ "grad_norm": 2.0980567932128906,
2196
+ "learning_rate": 4.417040358744395e-05,
2197
+ "loss": 0.0996,
2198
+ "step": 2780
2199
+ },
2200
+ {
2201
+ "epoch": 3.13,
2202
+ "grad_norm": 0.4327409267425537,
2203
+ "learning_rate": 4.360986547085202e-05,
2204
+ "loss": 0.0435,
2205
+ "step": 2790
2206
+ },
2207
+ {
2208
+ "epoch": 3.14,
2209
+ "grad_norm": 0.9807620048522949,
2210
+ "learning_rate": 4.3049327354260094e-05,
2211
+ "loss": 0.0835,
2212
+ "step": 2800
2213
+ },
2214
+ {
2215
+ "epoch": 3.14,
2216
+ "eval_accuracy": 0.8554539057002111,
2217
+ "eval_loss": 0.4977756142616272,
2218
+ "eval_runtime": 127.7265,
2219
+ "eval_samples_per_second": 55.627,
2220
+ "eval_steps_per_second": 3.484,
2221
+ "step": 2800
2222
+ },
2223
+ {
2224
+ "epoch": 3.15,
2225
+ "grad_norm": 1.283894658088684,
2226
+ "learning_rate": 4.2488789237668164e-05,
2227
+ "loss": 0.0397,
2228
+ "step": 2810
2229
+ },
2230
+ {
2231
+ "epoch": 3.16,
2232
+ "grad_norm": 0.820012092590332,
2233
+ "learning_rate": 4.1928251121076234e-05,
2234
+ "loss": 0.0601,
2235
+ "step": 2820
2236
+ },
2237
+ {
2238
+ "epoch": 3.17,
2239
+ "grad_norm": 1.781630039215088,
2240
+ "learning_rate": 4.1367713004484303e-05,
2241
+ "loss": 0.0452,
2242
+ "step": 2830
2243
+ },
2244
+ {
2245
+ "epoch": 3.18,
2246
+ "grad_norm": 0.20719203352928162,
2247
+ "learning_rate": 4.080717488789238e-05,
2248
+ "loss": 0.0401,
2249
+ "step": 2840
2250
+ },
2251
+ {
2252
+ "epoch": 3.2,
2253
+ "grad_norm": 2.106254816055298,
2254
+ "learning_rate": 4.024663677130045e-05,
2255
+ "loss": 0.0358,
2256
+ "step": 2850
2257
+ },
2258
+ {
2259
+ "epoch": 3.21,
2260
+ "grad_norm": 1.9900000095367432,
2261
+ "learning_rate": 3.968609865470852e-05,
2262
+ "loss": 0.0496,
2263
+ "step": 2860
2264
+ },
2265
+ {
2266
+ "epoch": 3.22,
2267
+ "grad_norm": 0.2951858937740326,
2268
+ "learning_rate": 3.91255605381166e-05,
2269
+ "loss": 0.0406,
2270
+ "step": 2870
2271
+ },
2272
+ {
2273
+ "epoch": 3.23,
2274
+ "grad_norm": 0.3538978695869446,
2275
+ "learning_rate": 3.8565022421524667e-05,
2276
+ "loss": 0.0481,
2277
+ "step": 2880
2278
+ },
2279
+ {
2280
+ "epoch": 3.24,
2281
+ "grad_norm": 0.9406136870384216,
2282
+ "learning_rate": 3.8004484304932737e-05,
2283
+ "loss": 0.0239,
2284
+ "step": 2890
2285
+ },
2286
+ {
2287
+ "epoch": 3.25,
2288
+ "grad_norm": 1.0194897651672363,
2289
+ "learning_rate": 3.744394618834081e-05,
2290
+ "loss": 0.0477,
2291
+ "step": 2900
2292
+ },
2293
+ {
2294
+ "epoch": 3.25,
2295
+ "eval_accuracy": 0.8586910626319494,
2296
+ "eval_loss": 0.5344606637954712,
2297
+ "eval_runtime": 127.6926,
2298
+ "eval_samples_per_second": 55.641,
2299
+ "eval_steps_per_second": 3.485,
2300
+ "step": 2900
2301
+ },
2302
+ {
2303
+ "epoch": 3.26,
2304
+ "grad_norm": 0.11960559338331223,
2305
+ "learning_rate": 3.688340807174888e-05,
2306
+ "loss": 0.0511,
2307
+ "step": 2910
2308
+ },
2309
+ {
2310
+ "epoch": 3.27,
2311
+ "grad_norm": 0.57135009765625,
2312
+ "learning_rate": 3.632286995515695e-05,
2313
+ "loss": 0.0119,
2314
+ "step": 2920
2315
+ },
2316
+ {
2317
+ "epoch": 3.28,
2318
+ "grad_norm": 0.40805578231811523,
2319
+ "learning_rate": 3.576233183856502e-05,
2320
+ "loss": 0.0735,
2321
+ "step": 2930
2322
+ },
2323
+ {
2324
+ "epoch": 3.3,
2325
+ "grad_norm": 3.6458218097686768,
2326
+ "learning_rate": 3.52017937219731e-05,
2327
+ "loss": 0.0713,
2328
+ "step": 2940
2329
+ },
2330
+ {
2331
+ "epoch": 3.31,
2332
+ "grad_norm": 0.26397281885147095,
2333
+ "learning_rate": 3.464125560538117e-05,
2334
+ "loss": 0.0248,
2335
+ "step": 2950
2336
+ },
2337
+ {
2338
+ "epoch": 3.32,
2339
+ "grad_norm": 0.8868299126625061,
2340
+ "learning_rate": 3.408071748878924e-05,
2341
+ "loss": 0.0734,
2342
+ "step": 2960
2343
+ },
2344
+ {
2345
+ "epoch": 3.33,
2346
+ "grad_norm": 0.07102204859256744,
2347
+ "learning_rate": 3.3520179372197316e-05,
2348
+ "loss": 0.0176,
2349
+ "step": 2970
2350
+ },
2351
+ {
2352
+ "epoch": 3.34,
2353
+ "grad_norm": 2.244887590408325,
2354
+ "learning_rate": 3.2959641255605386e-05,
2355
+ "loss": 0.0498,
2356
+ "step": 2980
2357
+ },
2358
+ {
2359
+ "epoch": 3.35,
2360
+ "grad_norm": 0.5616236925125122,
2361
+ "learning_rate": 3.2399103139013456e-05,
2362
+ "loss": 0.0373,
2363
+ "step": 2990
2364
+ },
2365
+ {
2366
+ "epoch": 3.36,
2367
+ "grad_norm": 0.19669972360134125,
2368
+ "learning_rate": 3.1838565022421526e-05,
2369
+ "loss": 0.0287,
2370
+ "step": 3000
2371
+ },
2372
+ {
2373
+ "epoch": 3.36,
2374
+ "eval_accuracy": 0.8596762843068262,
2375
+ "eval_loss": 0.5332924723625183,
2376
+ "eval_runtime": 127.3143,
2377
+ "eval_samples_per_second": 55.807,
2378
+ "eval_steps_per_second": 3.495,
2379
+ "step": 3000
2380
+ },
2381
+ {
2382
+ "epoch": 3.37,
2383
+ "grad_norm": 1.8116190433502197,
2384
+ "learning_rate": 3.12780269058296e-05,
2385
+ "loss": 0.0203,
2386
+ "step": 3010
2387
+ },
2388
+ {
2389
+ "epoch": 3.39,
2390
+ "grad_norm": 4.554844379425049,
2391
+ "learning_rate": 3.071748878923767e-05,
2392
+ "loss": 0.0199,
2393
+ "step": 3020
2394
+ },
2395
+ {
2396
+ "epoch": 3.4,
2397
+ "grad_norm": 1.331480622291565,
2398
+ "learning_rate": 3.015695067264574e-05,
2399
+ "loss": 0.0258,
2400
+ "step": 3030
2401
+ },
2402
+ {
2403
+ "epoch": 3.41,
2404
+ "grad_norm": 0.2970045208930969,
2405
+ "learning_rate": 2.9596412556053816e-05,
2406
+ "loss": 0.0311,
2407
+ "step": 3040
2408
+ },
2409
+ {
2410
+ "epoch": 3.42,
2411
+ "grad_norm": 1.1972055435180664,
2412
+ "learning_rate": 2.9035874439461886e-05,
2413
+ "loss": 0.0559,
2414
+ "step": 3050
2415
+ },
2416
+ {
2417
+ "epoch": 3.43,
2418
+ "grad_norm": 3.563384532928467,
2419
+ "learning_rate": 2.8475336322869956e-05,
2420
+ "loss": 0.0457,
2421
+ "step": 3060
2422
+ },
2423
+ {
2424
+ "epoch": 3.44,
2425
+ "grad_norm": 1.6148489713668823,
2426
+ "learning_rate": 2.7914798206278025e-05,
2427
+ "loss": 0.0583,
2428
+ "step": 3070
2429
+ },
2430
+ {
2431
+ "epoch": 3.45,
2432
+ "grad_norm": 0.8646751642227173,
2433
+ "learning_rate": 2.7354260089686102e-05,
2434
+ "loss": 0.03,
2435
+ "step": 3080
2436
+ },
2437
+ {
2438
+ "epoch": 3.46,
2439
+ "grad_norm": 0.4915563762187958,
2440
+ "learning_rate": 2.6793721973094172e-05,
2441
+ "loss": 0.032,
2442
+ "step": 3090
2443
+ },
2444
+ {
2445
+ "epoch": 3.48,
2446
+ "grad_norm": 1.7134366035461426,
2447
+ "learning_rate": 2.6233183856502242e-05,
2448
+ "loss": 0.0242,
2449
+ "step": 3100
2450
+ },
2451
+ {
2452
+ "epoch": 3.48,
2453
+ "eval_accuracy": 0.8602392681210416,
2454
+ "eval_loss": 0.5433253645896912,
2455
+ "eval_runtime": 128.4379,
2456
+ "eval_samples_per_second": 55.319,
2457
+ "eval_steps_per_second": 3.465,
2458
+ "step": 3100
2459
+ },
2460
+ {
2461
+ "epoch": 3.49,
2462
+ "grad_norm": 0.1170569360256195,
2463
+ "learning_rate": 2.567264573991032e-05,
2464
+ "loss": 0.0349,
2465
+ "step": 3110
2466
+ },
2467
+ {
2468
+ "epoch": 3.5,
2469
+ "grad_norm": 0.28987565636634827,
2470
+ "learning_rate": 2.511210762331839e-05,
2471
+ "loss": 0.0327,
2472
+ "step": 3120
2473
+ },
2474
+ {
2475
+ "epoch": 3.51,
2476
+ "grad_norm": 0.07998673617839813,
2477
+ "learning_rate": 2.455156950672646e-05,
2478
+ "loss": 0.0483,
2479
+ "step": 3130
2480
+ },
2481
+ {
2482
+ "epoch": 3.52,
2483
+ "grad_norm": 0.4792230725288391,
2484
+ "learning_rate": 2.3991031390134532e-05,
2485
+ "loss": 0.0271,
2486
+ "step": 3140
2487
+ },
2488
+ {
2489
+ "epoch": 3.53,
2490
+ "grad_norm": 3.571005344390869,
2491
+ "learning_rate": 2.3430493273542602e-05,
2492
+ "loss": 0.0474,
2493
+ "step": 3150
2494
+ },
2495
+ {
2496
+ "epoch": 3.54,
2497
+ "grad_norm": 0.4508035182952881,
2498
+ "learning_rate": 2.286995515695067e-05,
2499
+ "loss": 0.0487,
2500
+ "step": 3160
2501
+ },
2502
+ {
2503
+ "epoch": 3.55,
2504
+ "grad_norm": 0.219608336687088,
2505
+ "learning_rate": 2.2309417040358745e-05,
2506
+ "loss": 0.0206,
2507
+ "step": 3170
2508
+ },
2509
+ {
2510
+ "epoch": 3.57,
2511
+ "grad_norm": 0.5323840975761414,
2512
+ "learning_rate": 2.1748878923766815e-05,
2513
+ "loss": 0.0461,
2514
+ "step": 3180
2515
+ },
2516
+ {
2517
+ "epoch": 3.58,
2518
+ "grad_norm": 5.022609233856201,
2519
+ "learning_rate": 2.1188340807174888e-05,
2520
+ "loss": 0.0387,
2521
+ "step": 3190
2522
+ },
2523
+ {
2524
+ "epoch": 3.59,
2525
+ "grad_norm": 0.12102050334215164,
2526
+ "learning_rate": 2.062780269058296e-05,
2527
+ "loss": 0.0196,
2528
+ "step": 3200
2529
+ },
2530
+ {
2531
+ "epoch": 3.59,
2532
+ "eval_accuracy": 0.8584095707248417,
2533
+ "eval_loss": 0.5772469639778137,
2534
+ "eval_runtime": 126.9569,
2535
+ "eval_samples_per_second": 55.964,
2536
+ "eval_steps_per_second": 3.505,
2537
+ "step": 3200
2538
+ },
2539
+ {
2540
+ "epoch": 3.6,
2541
+ "grad_norm": 0.7281592488288879,
2542
+ "learning_rate": 2.006726457399103e-05,
2543
+ "loss": 0.029,
2544
+ "step": 3210
2545
+ },
2546
+ {
2547
+ "epoch": 3.61,
2548
+ "grad_norm": 3.789141893386841,
2549
+ "learning_rate": 1.9506726457399105e-05,
2550
+ "loss": 0.0747,
2551
+ "step": 3220
2552
+ },
2553
+ {
2554
+ "epoch": 3.62,
2555
+ "grad_norm": 0.8820950388908386,
2556
+ "learning_rate": 1.8946188340807175e-05,
2557
+ "loss": 0.0254,
2558
+ "step": 3230
2559
+ },
2560
+ {
2561
+ "epoch": 3.63,
2562
+ "grad_norm": 5.991636276245117,
2563
+ "learning_rate": 1.8385650224215248e-05,
2564
+ "loss": 0.0354,
2565
+ "step": 3240
2566
+ },
2567
+ {
2568
+ "epoch": 3.64,
2569
+ "grad_norm": 0.16825184226036072,
2570
+ "learning_rate": 1.7825112107623318e-05,
2571
+ "loss": 0.0229,
2572
+ "step": 3250
2573
+ },
2574
+ {
2575
+ "epoch": 3.65,
2576
+ "grad_norm": 0.19618666172027588,
2577
+ "learning_rate": 1.726457399103139e-05,
2578
+ "loss": 0.0737,
2579
+ "step": 3260
2580
+ },
2581
+ {
2582
+ "epoch": 3.67,
2583
+ "grad_norm": 0.08360274136066437,
2584
+ "learning_rate": 1.6704035874439464e-05,
2585
+ "loss": 0.0341,
2586
+ "step": 3270
2587
+ },
2588
+ {
2589
+ "epoch": 3.68,
2590
+ "grad_norm": 0.07565028220415115,
2591
+ "learning_rate": 1.6143497757847534e-05,
2592
+ "loss": 0.0287,
2593
+ "step": 3280
2594
+ },
2595
+ {
2596
+ "epoch": 3.69,
2597
+ "grad_norm": 2.6630303859710693,
2598
+ "learning_rate": 1.5582959641255608e-05,
2599
+ "loss": 0.0322,
2600
+ "step": 3290
2601
+ },
2602
+ {
2603
+ "epoch": 3.7,
2604
+ "grad_norm": 3.4573700428009033,
2605
+ "learning_rate": 1.5022421524663678e-05,
2606
+ "loss": 0.0297,
2607
+ "step": 3300
2608
+ },
2609
+ {
2610
+ "epoch": 3.7,
2611
+ "eval_accuracy": 0.8595355383532723,
2612
+ "eval_loss": 0.5564337372779846,
2613
+ "eval_runtime": 127.7494,
2614
+ "eval_samples_per_second": 55.617,
2615
+ "eval_steps_per_second": 3.483,
2616
+ "step": 3300
2617
+ },
2618
+ {
2619
+ "epoch": 3.71,
2620
+ "grad_norm": 2.092428207397461,
2621
+ "learning_rate": 1.4461883408071749e-05,
2622
+ "loss": 0.0486,
2623
+ "step": 3310
2624
+ },
2625
+ {
2626
+ "epoch": 3.72,
2627
+ "grad_norm": 4.266282558441162,
2628
+ "learning_rate": 1.3901345291479822e-05,
2629
+ "loss": 0.0191,
2630
+ "step": 3320
2631
+ },
2632
+ {
2633
+ "epoch": 3.73,
2634
+ "grad_norm": 0.5256732106208801,
2635
+ "learning_rate": 1.3340807174887892e-05,
2636
+ "loss": 0.0101,
2637
+ "step": 3330
2638
+ },
2639
+ {
2640
+ "epoch": 3.74,
2641
+ "grad_norm": 2.7987325191497803,
2642
+ "learning_rate": 1.2780269058295966e-05,
2643
+ "loss": 0.0547,
2644
+ "step": 3340
2645
+ },
2646
+ {
2647
+ "epoch": 3.76,
2648
+ "grad_norm": 0.17688162624835968,
2649
+ "learning_rate": 1.2219730941704037e-05,
2650
+ "loss": 0.0221,
2651
+ "step": 3350
2652
+ },
2653
+ {
2654
+ "epoch": 3.77,
2655
+ "grad_norm": 1.9824228286743164,
2656
+ "learning_rate": 1.1659192825112109e-05,
2657
+ "loss": 0.0528,
2658
+ "step": 3360
2659
+ },
2660
+ {
2661
+ "epoch": 3.78,
2662
+ "grad_norm": 0.06763149797916412,
2663
+ "learning_rate": 1.109865470852018e-05,
2664
+ "loss": 0.0193,
2665
+ "step": 3370
2666
+ },
2667
+ {
2668
+ "epoch": 3.79,
2669
+ "grad_norm": 3.0181431770324707,
2670
+ "learning_rate": 1.0538116591928252e-05,
2671
+ "loss": 0.0777,
2672
+ "step": 3380
2673
+ },
2674
+ {
2675
+ "epoch": 3.8,
2676
+ "grad_norm": 1.1364251375198364,
2677
+ "learning_rate": 9.977578475336324e-06,
2678
+ "loss": 0.0376,
2679
+ "step": 3390
2680
+ },
2681
+ {
2682
+ "epoch": 3.81,
2683
+ "grad_norm": 0.19941140711307526,
2684
+ "learning_rate": 9.417040358744395e-06,
2685
+ "loss": 0.0457,
2686
+ "step": 3400
2687
+ },
2688
+ {
2689
+ "epoch": 3.81,
2690
+ "eval_accuracy": 0.8512315270935961,
2691
+ "eval_loss": 0.5806910991668701,
2692
+ "eval_runtime": 128.3848,
2693
+ "eval_samples_per_second": 55.341,
2694
+ "eval_steps_per_second": 3.466,
2695
+ "step": 3400
2696
+ },
2697
+ {
2698
+ "epoch": 3.82,
2699
+ "grad_norm": 7.786200523376465,
2700
+ "learning_rate": 8.856502242152467e-06,
2701
+ "loss": 0.0541,
2702
+ "step": 3410
2703
+ },
2704
+ {
2705
+ "epoch": 3.83,
2706
+ "grad_norm": 0.42947888374328613,
2707
+ "learning_rate": 8.295964125560539e-06,
2708
+ "loss": 0.0169,
2709
+ "step": 3420
2710
+ },
2711
+ {
2712
+ "epoch": 3.85,
2713
+ "grad_norm": 0.17804774641990662,
2714
+ "learning_rate": 7.73542600896861e-06,
2715
+ "loss": 0.0324,
2716
+ "step": 3430
2717
+ },
2718
+ {
2719
+ "epoch": 3.86,
2720
+ "grad_norm": 0.34209346771240234,
2721
+ "learning_rate": 7.174887892376682e-06,
2722
+ "loss": 0.0158,
2723
+ "step": 3440
2724
+ },
2725
+ {
2726
+ "epoch": 3.87,
2727
+ "grad_norm": 0.16426484286785126,
2728
+ "learning_rate": 6.614349775784753e-06,
2729
+ "loss": 0.0135,
2730
+ "step": 3450
2731
+ },
2732
+ {
2733
+ "epoch": 3.88,
2734
+ "grad_norm": 0.19225721061229706,
2735
+ "learning_rate": 6.053811659192826e-06,
2736
+ "loss": 0.0291,
2737
+ "step": 3460
2738
+ },
2739
+ {
2740
+ "epoch": 3.89,
2741
+ "grad_norm": 1.250550627708435,
2742
+ "learning_rate": 5.493273542600897e-06,
2743
+ "loss": 0.0214,
2744
+ "step": 3470
2745
+ },
2746
+ {
2747
+ "epoch": 3.9,
2748
+ "grad_norm": 0.08164811879396439,
2749
+ "learning_rate": 4.932735426008968e-06,
2750
+ "loss": 0.0259,
2751
+ "step": 3480
2752
+ },
2753
+ {
2754
+ "epoch": 3.91,
2755
+ "grad_norm": 0.14926199615001678,
2756
+ "learning_rate": 4.372197309417041e-06,
2757
+ "loss": 0.0299,
2758
+ "step": 3490
2759
+ },
2760
+ {
2761
+ "epoch": 3.92,
2762
+ "grad_norm": 0.0725017860531807,
2763
+ "learning_rate": 3.8116591928251122e-06,
2764
+ "loss": 0.016,
2765
+ "step": 3500
2766
+ },
2767
+ {
2768
+ "epoch": 3.92,
2769
+ "eval_accuracy": 0.8617874736101337,
2770
+ "eval_loss": 0.5601363778114319,
2771
+ "eval_runtime": 127.8408,
2772
+ "eval_samples_per_second": 55.577,
2773
+ "eval_steps_per_second": 3.481,
2774
+ "step": 3500
2775
+ },
2776
+ {
2777
+ "epoch": 3.93,
2778
+ "grad_norm": 1.6513164043426514,
2779
+ "learning_rate": 3.251121076233184e-06,
2780
+ "loss": 0.0165,
2781
+ "step": 3510
2782
+ },
2783
+ {
2784
+ "epoch": 3.95,
2785
+ "grad_norm": 0.088756263256073,
2786
+ "learning_rate": 2.690582959641256e-06,
2787
+ "loss": 0.0332,
2788
+ "step": 3520
2789
+ },
2790
+ {
2791
+ "epoch": 3.96,
2792
+ "grad_norm": 0.26848122477531433,
2793
+ "learning_rate": 2.1300448430493275e-06,
2794
+ "loss": 0.0232,
2795
+ "step": 3530
2796
+ },
2797
+ {
2798
+ "epoch": 3.97,
2799
+ "grad_norm": 0.14724156260490417,
2800
+ "learning_rate": 1.5695067264573993e-06,
2801
+ "loss": 0.02,
2802
+ "step": 3540
2803
+ },
2804
+ {
2805
+ "epoch": 3.98,
2806
+ "grad_norm": 1.9452786445617676,
2807
+ "learning_rate": 1.0089686098654709e-06,
2808
+ "loss": 0.0433,
2809
+ "step": 3550
2810
+ },
2811
+ {
2812
+ "epoch": 3.99,
2813
+ "grad_norm": 0.16432423889636993,
2814
+ "learning_rate": 4.484304932735426e-07,
2815
+ "loss": 0.0311,
2816
+ "step": 3560
2817
+ },
2818
+ {
2819
+ "epoch": 4.0,
2820
+ "step": 3568,
2821
+ "total_flos": 8.839521632856048e+18,
2822
+ "train_loss": 0.3317156129854944,
2823
+ "train_runtime": 7997.833,
2824
+ "train_samples_per_second": 14.262,
2825
+ "train_steps_per_second": 0.446
2826
+ }
2827
+ ],
2828
+ "logging_steps": 10,
2829
+ "max_steps": 3568,
2830
+ "num_input_tokens_seen": 0,
2831
+ "num_train_epochs": 4,
2832
+ "save_steps": 100,
2833
+ "total_flos": 8.839521632856048e+18,
2834
+ "train_batch_size": 32,
2835
+ "trial_name": null,
2836
+ "trial_params": null
2837
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f142d60e7d0fe050a40d06a26894d5a308a706d22230783daf74adccbffcd2e8
3
+ size 4984