Hugging Face's logo

Seed42Lab
/
trabank-vl

Safetensors
Model card Files
xet
Community
wonderwind271 commited on
Commit
e092dbd
·
verified ·
1 Parent(s): 75ff7f0

seed 42-242, resize+no mormalization

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +65 -0
  2. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/config.json +53 -0
  3. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/generation_config.json +7 -0
  4. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/model.safetensors +3 -0
  5. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/optimizer.pt +3 -0
  6. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/rng_state_0.pth +3 -0
  7. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/rng_state_1.pth +3 -0
  8. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/scheduler.pt +3 -0
  9. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/special_tokens_map.json +24 -0
  10. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/tokenizer_config.json +45 -0
  11. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/trainer_state.json +0 -0
  12. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/training_args.bin +3 -0
  13. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/vocab.json +0 -0
  14. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/config.json +53 -0
  15. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/generation_config.json +7 -0
  16. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/model.safetensors +3 -0
  17. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/optimizer.pt +3 -0
  18. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/rng_state_0.pth +3 -0
  19. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/rng_state_1.pth +3 -0
  20. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/scheduler.pt +3 -0
  21. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/special_tokens_map.json +24 -0
  22. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/tokenizer_config.json +45 -0
  23. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/trainer_state.json +3 -0
  24. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/training_args.bin +3 -0
  25. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/vocab.json +0 -0
  26. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/config.json +53 -0
  27. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/generation_config.json +7 -0
  28. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/model.safetensors +3 -0
  29. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/optimizer.pt +3 -0
  30. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/rng_state_0.pth +3 -0
  31. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/rng_state_1.pth +3 -0
  32. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/scheduler.pt +3 -0
  33. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/special_tokens_map.json +24 -0
  34. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/tokenizer_config.json +45 -0
  35. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/trainer_state.json +3 -0
  36. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/training_args.bin +3 -0
  37. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/vocab.json +0 -0
  38. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/config.json +53 -0
  39. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/generation_config.json +7 -0
  40. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/model.safetensors +3 -0
  41. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/optimizer.pt +3 -0
  42. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/rng_state_0.pth +3 -0
  43. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/rng_state_1.pth +3 -0
  44. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/scheduler.pt +3 -0
  45. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/special_tokens_map.json +24 -0
  46. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/tokenizer_config.json +45 -0
  47. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/trainer_state.json +3 -0
  48. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/training_args.bin +3 -0
  49. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/vocab.json +0 -0
  50. trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-130000/config.json +53 -0
.gitattributes CHANGED
@@ -133,3 +133,68 @@ trabank_vl_dino_pretrain_resize_s42/checkpoint-70000/trainer_state.json filter=l
133
  trabank_vl_dino_pretrain_resize_s42/checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
134
  trabank_vl_dino_pretrain_resize_s42/checkpoint-90000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
135
  trabank_vl_dino_pretrain_resize_s42/trainer_state.json filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
133
  trabank_vl_dino_pretrain_resize_s42/checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
134
  trabank_vl_dino_pretrain_resize_s42/checkpoint-90000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
135
  trabank_vl_dino_pretrain_resize_s42/trainer_state.json filter=lfs diff=lfs merge=lfs -text
136
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
137
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
138
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
139
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-130000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
140
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-140000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
141
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-150000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
142
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-160000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
143
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-170000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
144
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-180000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
145
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-190000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
146
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-200000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
147
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-210000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
148
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-220000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
149
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-230000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
150
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-240000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
151
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-250000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
152
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-260000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
153
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-270000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
154
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-60000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
155
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-70000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
156
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
157
+ trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-90000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
158
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-100000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
159
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-110000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
160
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-120000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
161
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-130000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
162
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-140000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
163
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-150000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
164
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-160000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
165
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-170000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
166
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-180000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
167
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-190000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
168
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-200000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
169
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-210000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
170
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-220000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
171
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-230000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
172
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-240000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
173
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-250000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
174
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-60000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
175
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-70000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
176
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
177
+ trabank_vl_dino_pretrain_resize_pure_s242/checkpoint-90000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
178
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-100000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
179
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-110000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
180
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-120000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
181
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-130000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
182
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-140000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
183
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-150000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
184
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-160000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
185
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-170000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
186
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-180000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
187
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-190000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
188
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-200000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
189
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-210000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
190
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-220000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
191
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-230000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
192
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-240000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
193
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-250000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
194
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-260000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
195
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-270000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
196
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-280000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
197
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-60000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
198
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-70000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
199
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
200
+ trabank_vl_dino_pretrain_resize_pure_s42/checkpoint-90000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "LlavaGPTForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 1,
10
+ "detect_loss": false,
11
+ "embd_pdrop": 0.1,
12
+ "eos_token_id": 2,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "image_aspect_ratio": "pad",
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 3072,
18
+ "is_decoder": true,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "mm_projector_lr": null,
21
+ "mm_use_im_patch_token": false,
22
+ "mm_use_im_start_end": false,
23
+ "mm_vision_tower": "dino",
24
+ "model_type": "gpt2",
25
+ "n_embd": 768,
26
+ "n_head": 12,
27
+ "n_inner": null,
28
+ "n_layer": 12,
29
+ "n_positions": 1024,
30
+ "num_key_value_heads": 12,
31
+ "reorder_and_upcast_attn": false,
32
+ "resid_pdrop": 0.1,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": null,
35
+ "rope_theta": 10000.0,
36
+ "scale_attn_by_inverse_layer_idx": false,
37
+ "scale_attn_weights": true,
38
+ "summary_activation": null,
39
+ "summary_first_dropout": 0.1,
40
+ "summary_proj_to_labels": true,
41
+ "summary_type": "cls_index",
42
+ "summary_use_proj": true,
43
+ "tie_word_embeddings": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "tokenizer_model_max_length": 2048,
46
+ "tokenizer_padding_side": "right",
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.48.1",
49
+ "tune_mm_mlp_adapter": false,
50
+ "use_cache": false,
51
+ "vision_tower_type": "dino",
52
+ "vocab_size": 25005
53
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.48.1",
6
+ "use_cache": false
7
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13afbf3c23ea9a3e301f29827278582a83327f4cff8a7cae91a05d0b216b1360
3
+ size 670210056
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62dc17ee3fadb7806c0e3c7ecda2c0aa97c24ac9198ebe70567c1515ef58fe7e
3
+ size 994126027
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15d66358286f683760a8b00931084a4e9740ab3fd44ff3417d806db27d37d9ed
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d64ec443c3250d190531330e0c0eb150b7032a051bb979eb15c22c563fb57ee4
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4aeaedc1e6e4ac9ac4f28a8691779e4a906da7b051e36d0bbe385033f066178d
3
+ size 1465
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<SOS>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<EOS>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<UNK>",
17
+ "unk_token": {
18
+ "content": "<UNK>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<PAD>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<UNK>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<SOS>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<EOS>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<SOS>",
37
+ "clean_up_tokenization_spaces": false,
38
+ "eos_token": "<EOS>",
39
+ "extra_special_tokens": {},
40
+ "model_max_length": 2048,
41
+ "pad_token": "<UNK>",
42
+ "padding_side": "right",
43
+ "tokenizer_class": "CustomWordTokenizer",
44
+ "unk_token": "<UNK>"
45
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcdf880640b0f1f0dd22f89b8dc6b4044f277625466d23e0f6cb69077a5681ed
3
+ size 6353
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-10000/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "LlavaGPTForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 1,
10
+ "detect_loss": false,
11
+ "embd_pdrop": 0.1,
12
+ "eos_token_id": 2,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "image_aspect_ratio": "pad",
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 3072,
18
+ "is_decoder": true,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "mm_projector_lr": null,
21
+ "mm_use_im_patch_token": false,
22
+ "mm_use_im_start_end": false,
23
+ "mm_vision_tower": "dino",
24
+ "model_type": "gpt2",
25
+ "n_embd": 768,
26
+ "n_head": 12,
27
+ "n_inner": null,
28
+ "n_layer": 12,
29
+ "n_positions": 1024,
30
+ "num_key_value_heads": 12,
31
+ "reorder_and_upcast_attn": false,
32
+ "resid_pdrop": 0.1,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": null,
35
+ "rope_theta": 10000.0,
36
+ "scale_attn_by_inverse_layer_idx": false,
37
+ "scale_attn_weights": true,
38
+ "summary_activation": null,
39
+ "summary_first_dropout": 0.1,
40
+ "summary_proj_to_labels": true,
41
+ "summary_type": "cls_index",
42
+ "summary_use_proj": true,
43
+ "tie_word_embeddings": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "tokenizer_model_max_length": 2048,
46
+ "tokenizer_padding_side": "right",
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.48.1",
49
+ "tune_mm_mlp_adapter": false,
50
+ "use_cache": false,
51
+ "vision_tower_type": "dino",
52
+ "vocab_size": 25005
53
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.48.1",
6
+ "use_cache": false
7
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b68da138e37141dbac0966e39d6c0384a5e6d0bf15c50f1c337105fdeabf7dc
3
+ size 670210056
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6c41964d902e3c21495d790d6b77aa853e962bc7fa89ddd0ac47770e7d9bc0d
3
+ size 994126027
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06e5de7729cf767b3f212a746582f8ae26a3dad63bd1c52c2d8d299d4396ba86
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c23c350a480f32f399e5bfa314cfe2c36e25c8b05873a1c90c0c433b624d392b
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae0a341cc06fce0d21bf7169c4964d449cefbf6ff943711fe1a78667c32ac1df
3
+ size 1465
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<SOS>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<EOS>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<UNK>",
17
+ "unk_token": {
18
+ "content": "<UNK>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<PAD>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<UNK>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<SOS>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<EOS>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<SOS>",
37
+ "clean_up_tokenization_spaces": false,
38
+ "eos_token": "<EOS>",
39
+ "extra_special_tokens": {},
40
+ "model_max_length": 2048,
41
+ "pad_token": "<UNK>",
42
+ "padding_side": "right",
43
+ "tokenizer_class": "CustomWordTokenizer",
44
+ "unk_token": "<UNK>"
45
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77e924976c14c22bd50ff2a77fd3e13276df11f9c8897e2f22bd74af36ef6a7f
3
+ size 17489462
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcdf880640b0f1f0dd22f89b8dc6b4044f277625466d23e0f6cb69077a5681ed
3
+ size 6353
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-100000/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "LlavaGPTForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 1,
10
+ "detect_loss": false,
11
+ "embd_pdrop": 0.1,
12
+ "eos_token_id": 2,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "image_aspect_ratio": "pad",
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 3072,
18
+ "is_decoder": true,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "mm_projector_lr": null,
21
+ "mm_use_im_patch_token": false,
22
+ "mm_use_im_start_end": false,
23
+ "mm_vision_tower": "dino",
24
+ "model_type": "gpt2",
25
+ "n_embd": 768,
26
+ "n_head": 12,
27
+ "n_inner": null,
28
+ "n_layer": 12,
29
+ "n_positions": 1024,
30
+ "num_key_value_heads": 12,
31
+ "reorder_and_upcast_attn": false,
32
+ "resid_pdrop": 0.1,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": null,
35
+ "rope_theta": 10000.0,
36
+ "scale_attn_by_inverse_layer_idx": false,
37
+ "scale_attn_weights": true,
38
+ "summary_activation": null,
39
+ "summary_first_dropout": 0.1,
40
+ "summary_proj_to_labels": true,
41
+ "summary_type": "cls_index",
42
+ "summary_use_proj": true,
43
+ "tie_word_embeddings": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "tokenizer_model_max_length": 2048,
46
+ "tokenizer_padding_side": "right",
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.48.1",
49
+ "tune_mm_mlp_adapter": false,
50
+ "use_cache": false,
51
+ "vision_tower_type": "dino",
52
+ "vocab_size": 25005
53
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.48.1",
6
+ "use_cache": false
7
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68816d2d04d4ca8c3162509edec82b4761612aee35a8bbd83cb8c6e08508e44b
3
+ size 670210056
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d6e08876fc29ce99465e622ca3b36833e47a9ad92d1a25d7796e485119f3ba1
3
+ size 994126027
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b83ff1a186d8d04437067fb334aead772bc50611c7e9f24a9b3ca0d48159b70a
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56c2244f2f9b564de102a52074cfd40669d0113612da1ea66ed1e4261b4e563e
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:db6067b09a272acaf5c85ca7f6a9e4741d56ce0b4071a298c7b399949ec03a20
3
+ size 1465
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<SOS>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<EOS>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<UNK>",
17
+ "unk_token": {
18
+ "content": "<UNK>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<PAD>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<UNK>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<SOS>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<EOS>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<SOS>",
37
+ "clean_up_tokenization_spaces": false,
38
+ "eos_token": "<EOS>",
39
+ "extra_special_tokens": {},
40
+ "model_max_length": 2048,
41
+ "pad_token": "<UNK>",
42
+ "padding_side": "right",
43
+ "tokenizer_class": "CustomWordTokenizer",
44
+ "unk_token": "<UNK>"
45
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:897f6c74d5b1a5334ffb9c38e099e1871e1a6859d4f962d4e45838fa860132e9
3
+ size 19247368
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcdf880640b0f1f0dd22f89b8dc6b4044f277625466d23e0f6cb69077a5681ed
3
+ size 6353
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-110000/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "LlavaGPTForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 1,
10
+ "detect_loss": false,
11
+ "embd_pdrop": 0.1,
12
+ "eos_token_id": 2,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "image_aspect_ratio": "pad",
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 3072,
18
+ "is_decoder": true,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "mm_projector_lr": null,
21
+ "mm_use_im_patch_token": false,
22
+ "mm_use_im_start_end": false,
23
+ "mm_vision_tower": "dino",
24
+ "model_type": "gpt2",
25
+ "n_embd": 768,
26
+ "n_head": 12,
27
+ "n_inner": null,
28
+ "n_layer": 12,
29
+ "n_positions": 1024,
30
+ "num_key_value_heads": 12,
31
+ "reorder_and_upcast_attn": false,
32
+ "resid_pdrop": 0.1,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": null,
35
+ "rope_theta": 10000.0,
36
+ "scale_attn_by_inverse_layer_idx": false,
37
+ "scale_attn_weights": true,
38
+ "summary_activation": null,
39
+ "summary_first_dropout": 0.1,
40
+ "summary_proj_to_labels": true,
41
+ "summary_type": "cls_index",
42
+ "summary_use_proj": true,
43
+ "tie_word_embeddings": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "tokenizer_model_max_length": 2048,
46
+ "tokenizer_padding_side": "right",
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.48.1",
49
+ "tune_mm_mlp_adapter": false,
50
+ "use_cache": false,
51
+ "vision_tower_type": "dino",
52
+ "vocab_size": 25005
53
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.48.1",
6
+ "use_cache": false
7
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1dabfae3ae6f59671e2195a7096e07ee2d81a587d7bbe2d031024d65d3352178
3
+ size 670210056
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:647d808fd8fd4cadb66c1a57aad1ca82d9a7d4a4800b822f4e195a03f0cdaa06
3
+ size 994126027
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ff2ff6268633b053dcf48425b70cd4b49b4f1e6626daf711a55b615a4a11f90
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9405c0c28d83b343b383b2658b698af060e277813034e732a2bc4b1cb89fab06
3
+ size 14981
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba0eb79c069ba3b4ea5167de610e72a035c58c9eda7c3683881e185425cc6ffd
3
+ size 1465
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<SOS>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<EOS>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<UNK>",
17
+ "unk_token": {
18
+ "content": "<UNK>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<PAD>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<UNK>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "<SOS>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "3": {
28
+ "content": "<EOS>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<SOS>",
37
+ "clean_up_tokenization_spaces": false,
38
+ "eos_token": "<EOS>",
39
+ "extra_special_tokens": {},
40
+ "model_max_length": 2048,
41
+ "pad_token": "<UNK>",
42
+ "padding_side": "right",
43
+ "tokenizer_class": "CustomWordTokenizer",
44
+ "unk_token": "<UNK>"
45
+ }
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5590429419efacef817dd9d7ca7f9bd980df988cc93038b53fb0597fda3f6e49
3
+ size 21006299
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcdf880640b0f1f0dd22f89b8dc6b4044f277625466d23e0f6cb69077a5681ed
3
+ size 6353
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-120000/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
trabank_vl_dino_pretrain_resize_pure_s142/checkpoint-130000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "LlavaGPTForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 1,
10
+ "detect_loss": false,
11
+ "embd_pdrop": 0.1,
12
+ "eos_token_id": 2,
13
+ "freeze_mm_mlp_adapter": false,
14
+ "hidden_act": "silu",
15
+ "image_aspect_ratio": "pad",
16
+ "initializer_range": 0.02,
17
+ "intermediate_size": 3072,
18
+ "is_decoder": true,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "mm_projector_lr": null,
21
+ "mm_use_im_patch_token": false,
22
+ "mm_use_im_start_end": false,
23
+ "mm_vision_tower": "dino",
24
+ "model_type": "gpt2",
25
+ "n_embd": 768,
26
+ "n_head": 12,
27
+ "n_inner": null,
28
+ "n_layer": 12,
29
+ "n_positions": 1024,
30
+ "num_key_value_heads": 12,
31
+ "reorder_and_upcast_attn": false,
32
+ "resid_pdrop": 0.1,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_scaling": null,
35
+ "rope_theta": 10000.0,
36
+ "scale_attn_by_inverse_layer_idx": false,
37
+ "scale_attn_weights": true,
38
+ "summary_activation": null,
39
+ "summary_first_dropout": 0.1,
40
+ "summary_proj_to_labels": true,
41
+ "summary_type": "cls_index",
42
+ "summary_use_proj": true,
43
+ "tie_word_embeddings": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "tokenizer_model_max_length": 2048,
46
+ "tokenizer_padding_side": "right",
47
+ "torch_dtype": "float32",
48
+ "transformers_version": "4.48.1",
49
+ "tune_mm_mlp_adapter": false,
50
+ "use_cache": false,
51
+ "vision_tower_type": "dino",
52
+ "vocab_size": 25005
53
+ }