Ba2han commited on
Commit
eedc841
·
verified ·
1 Parent(s): 270c1e5

Upload model trained with Unsloth

Browse files

Upload model trained with Unsloth 2x faster

Files changed (4) hide show
  1. README.md +1 -0
  2. config.json +107 -0
  3. generation_config.json +10 -0
  4. model.safetensors +3 -0
README.md CHANGED
@@ -6,6 +6,7 @@ tags:
6
  - unsloth
7
  - lfm2_vl
8
  - trl
 
9
  license: apache-2.0
10
  language:
11
  - en
 
6
  - unsloth
7
  - lfm2_vl
8
  - trl
9
+ - sft
10
  license: apache-2.0
11
  language:
12
  - en
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Lfm2VlForConditionalGeneration"
4
+ ],
5
+ "bos_token_id": 1,
6
+ "do_image_splitting": true,
7
+ "downsample_factor": 2,
8
+ "dtype": "bfloat16",
9
+ "encoder_patch_size": 16,
10
+ "eos_token_id": 7,
11
+ "image_token_id": 396,
12
+ "max_image_tokens": 256,
13
+ "max_pixels_tolerance": 2.0,
14
+ "max_tiles": 10,
15
+ "min_image_tokens": 64,
16
+ "min_tiles": 2,
17
+ "model_name": "tr-sft/checkpoint-2398",
18
+ "model_type": "lfm2_vl",
19
+ "pad_token_id": 0,
20
+ "projector_bias": true,
21
+ "projector_hidden_act": "gelu",
22
+ "projector_hidden_size": 2048,
23
+ "projector_use_layernorm": false,
24
+ "text_config": {
25
+ "_name_or_path": "LiquidAI/LFM2-1.2B",
26
+ "architectures": [
27
+ "Lfm2ForCausalLM"
28
+ ],
29
+ "block_auto_adjust_ff_dim": true,
30
+ "block_dim": 2048,
31
+ "block_ffn_dim_multiplier": 1.0,
32
+ "block_mlp_init_scale": 1.0,
33
+ "block_multiple_of": 256,
34
+ "block_norm_eps": 1e-05,
35
+ "block_out_init_scale": 1.0,
36
+ "block_use_swiglu": true,
37
+ "block_use_xavier_init": true,
38
+ "bos_token_id": 1,
39
+ "conv_L_cache": 3,
40
+ "conv_bias": false,
41
+ "conv_dim": 2048,
42
+ "conv_dim_out": 2048,
43
+ "conv_use_xavier_init": true,
44
+ "dtype": "bfloat16",
45
+ "eos_token_id": 7,
46
+ "full_attn_idxs": null,
47
+ "hidden_size": 2048,
48
+ "initializer_range": 0.02,
49
+ "intermediate_size": 12288,
50
+ "layer_types": [
51
+ "conv",
52
+ "conv",
53
+ "full_attention",
54
+ "conv",
55
+ "conv",
56
+ "full_attention",
57
+ "conv",
58
+ "conv",
59
+ "full_attention",
60
+ "conv",
61
+ "full_attention",
62
+ "conv",
63
+ "full_attention",
64
+ "conv",
65
+ "full_attention",
66
+ "conv"
67
+ ],
68
+ "max_position_embeddings": 128000,
69
+ "model_type": "lfm2",
70
+ "norm_eps": 1e-05,
71
+ "num_attention_heads": 32,
72
+ "num_heads": 32,
73
+ "num_hidden_layers": 16,
74
+ "num_key_value_heads": 8,
75
+ "pad_token_id": 0,
76
+ "rope_parameters": {
77
+ "rope_theta": 1000000.0,
78
+ "rope_type": "default"
79
+ },
80
+ "tie_word_embeddings": true,
81
+ "use_cache": true,
82
+ "use_pos_enc": true,
83
+ "vocab_size": 65536
84
+ },
85
+ "tie_word_embeddings": true,
86
+ "tile_size": 512,
87
+ "transformers_version": "5.5.0",
88
+ "unsloth_version": "2026.4.4",
89
+ "use_cache": false,
90
+ "use_image_special_tokens": true,
91
+ "use_thumbnail": true,
92
+ "vision_config": {
93
+ "attention_dropout": 0.0,
94
+ "dtype": "bfloat16",
95
+ "hidden_act": "gelu_pytorch_tanh",
96
+ "hidden_size": 1152,
97
+ "intermediate_size": 4304,
98
+ "layer_norm_eps": 1e-06,
99
+ "model_type": "siglip2_vision_model",
100
+ "num_attention_heads": 16,
101
+ "num_channels": 3,
102
+ "num_hidden_layers": 27,
103
+ "num_patches": 256,
104
+ "patch_size": 16,
105
+ "vision_use_head": false
106
+ }
107
+ }
generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": [
5
+ 7,
6
+ 7
7
+ ],
8
+ "pad_token_id": 0,
9
+ "transformers_version": "5.5.0"
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92b57e95828b38cb02bdd8e10fee4271621fdb0097617e00c03d53779b274084
3
+ size 3193334216