Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
773247e
·
verified ·
1 Parent(s): 067e647

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-16000/config.json +71 -0
  2. checkpoint-16000/generation_config.json +6 -0
  3. checkpoint-16000/model.safetensors +3 -0
  4. checkpoint-16000/optimizer.pt +3 -0
  5. checkpoint-16000/rng_state_0.pth +3 -0
  6. checkpoint-16000/rng_state_1.pth +3 -0
  7. checkpoint-16000/rng_state_10.pth +3 -0
  8. checkpoint-16000/rng_state_11.pth +3 -0
  9. checkpoint-16000/rng_state_12.pth +3 -0
  10. checkpoint-16000/rng_state_13.pth +3 -0
  11. checkpoint-16000/rng_state_14.pth +3 -0
  12. checkpoint-16000/rng_state_15.pth +3 -0
  13. checkpoint-16000/rng_state_2.pth +3 -0
  14. checkpoint-16000/rng_state_3.pth +3 -0
  15. checkpoint-16000/rng_state_4.pth +3 -0
  16. checkpoint-16000/rng_state_5.pth +3 -0
  17. checkpoint-16000/rng_state_6.pth +3 -0
  18. checkpoint-16000/rng_state_7.pth +3 -0
  19. checkpoint-16000/rng_state_8.pth +3 -0
  20. checkpoint-16000/rng_state_9.pth +3 -0
  21. checkpoint-16000/scheduler.pt +3 -0
  22. checkpoint-16000/trainer_state.json +0 -0
  23. checkpoint-16000/training_args.bin +3 -0
checkpoint-16000/config.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "ar_steps": 1,
3
+ "architectures": [
4
+ "DiffVLMBaseline"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "eos_token_id": 151645,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 1536,
11
+ "image_token_id": 151655,
12
+ "img_cross_attention_dim": 2048,
13
+ "img_diffuser_depth": 12,
14
+ "img_ffn_dim_multiplier": null,
15
+ "img_hidden_size": 1536,
16
+ "img_multiple_of": 256,
17
+ "img_norm_eps": 1e-05,
18
+ "img_num_attention_heads": 12,
19
+ "img_num_kv_heads": 12,
20
+ "img_qk_norm": true,
21
+ "in_channels": 32,
22
+ "initializer_range": 0.02,
23
+ "inject_img_diffuser": false,
24
+ "input_size": 32,
25
+ "intermediate_size": 8960,
26
+ "lora_alpha": 128,
27
+ "lora_bias": "none",
28
+ "lora_dropout": 0.05,
29
+ "lora_enable": false,
30
+ "lora_r": 64,
31
+ "max_position_embeddings": 32768,
32
+ "max_window_layers": 28,
33
+ "model_type": "qwen2_vl",
34
+ "non_linearity": 1,
35
+ "norm_elementwise_affine": true,
36
+ "num_attention_heads": 12,
37
+ "num_hidden_layers": 28,
38
+ "num_key_value_heads": 2,
39
+ "patch_size": 2,
40
+ "rms_norm_eps": 1e-06,
41
+ "rope_scaling": {
42
+ "mrope_section": [
43
+ 16,
44
+ 24,
45
+ 24
46
+ ],
47
+ "rope_type": "default",
48
+ "type": "default"
49
+ },
50
+ "rope_theta": 1000000.0,
51
+ "sample_size": 128,
52
+ "sampling_steps": 28,
53
+ "sliding_window": null,
54
+ "tie_word_embeddings": true,
55
+ "torch_dtype": "bfloat16",
56
+ "transformers_version": "4.47.0",
57
+ "use_cache": true,
58
+ "use_sliding_window": false,
59
+ "vae_path": "mit-han-lab/dc-ae-f32c32-sana-1.1-diffusers",
60
+ "video_token_id": 151656,
61
+ "vision_config": {
62
+ "hidden_size": 1536,
63
+ "in_chans": 3,
64
+ "model_type": "qwen2_vl",
65
+ "spatial_patch_size": 14
66
+ },
67
+ "vision_end_token_id": 151653,
68
+ "vision_start_token_id": 151652,
69
+ "vision_token_id": 151654,
70
+ "vocab_size": 151936
71
+ }
checkpoint-16000/generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "4.47.0"
6
+ }
checkpoint-16000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d993c97db170679fe359b9e10531b83001a0866d17b83bd228d29a6189f746b7
3
+ size 4714837766
checkpoint-16000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2a37ffcc52e502dbcb362f133f45adf86e00832e8c9cf06ecb663f01e7b0849
3
+ size 6630552226
checkpoint-16000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c6d23820cc3dd8f2a5e1813122727f41f66650cff09c4469f30f879f53e15646
3
+ size 15920
checkpoint-16000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21ac41c279a085486b3d05a776528640991527c0d9c4e9af6aa064e44a4aa3eb
3
+ size 15920
checkpoint-16000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3d4fe70514110ff108041c14eec1777faf70218f986d5508cd963482ff554a4
3
+ size 15933
checkpoint-16000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c0c6cf2b2c8d7fcedb7f955d9528d05ad9f3c7637729bf60c825a9eb09a485f
3
+ size 15933
checkpoint-16000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7199a0fb1b8c4989f15247eee59e7040af81aed1de7413fdd7a4b3f0ba76f973
3
+ size 15933
checkpoint-16000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48b2d57d75dd346e8b260260d9cdeb3efe051816184bdaba00512d42e70dffb5
3
+ size 15933
checkpoint-16000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46f9f550f3be5535fbc20918d3dee8b91721dfe9b67a30c0dcf8c1267361f5cc
3
+ size 15933
checkpoint-16000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f065d04041222deb21260746264dfdcab788beb8a94b1b10c427f083b8ae4a4a
3
+ size 15933
checkpoint-16000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d1c01fb3adbe081d9156eeffd5bf8a4d07ef274a2e12096366f5c49a9ba933e
3
+ size 15920
checkpoint-16000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d5b96d941361728e5e645e78bc03aacde7838b823c517e02cd2cf60e4f6a172
3
+ size 15920
checkpoint-16000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c464e35b42316feb32717fa907e47a0ab10493ef64595e70cde62dc6fb8a1348
3
+ size 15920
checkpoint-16000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:924915d8156fe374767ce87656b49d1c9c5e91ea7b508f39006f0fe72f1490d8
3
+ size 15920
checkpoint-16000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67f553ed7935d99380598e589e660e75342fb5b50871d3092acecb716882818d
3
+ size 15920
checkpoint-16000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3ad28eb1ff594c905592204b8a39a7922300070872639e7d354180a623acd63
3
+ size 15920
checkpoint-16000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93f9202baf70a44f2f248418814da5e231679999a213f27d6ec9aa6f8db90d1d
3
+ size 15920
checkpoint-16000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:240ff591e452babcbd4e7b0819172c055f614240329373413ef695e45277b36a
3
+ size 15920
checkpoint-16000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a732e5f966a55205323923ecac94a5a27b49eee48a9398e6b7e1587016d48fa
3
+ size 1064
checkpoint-16000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-16000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17f6a15498b31314f433e76921ed6584e744dfc81fed67f2a3c726a9bcd54124
3
+ size 5752