Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
ff59ed7
·
verified ·
1 Parent(s): b77758b

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-21000/config.json +71 -0
  2. checkpoint-21000/generation_config.json +6 -0
  3. checkpoint-21000/model.safetensors +3 -0
  4. checkpoint-21000/optimizer.pt +3 -0
  5. checkpoint-21000/rng_state_0.pth +3 -0
  6. checkpoint-21000/rng_state_1.pth +3 -0
  7. checkpoint-21000/rng_state_10.pth +3 -0
  8. checkpoint-21000/rng_state_11.pth +3 -0
  9. checkpoint-21000/rng_state_12.pth +3 -0
  10. checkpoint-21000/rng_state_13.pth +3 -0
  11. checkpoint-21000/rng_state_14.pth +3 -0
  12. checkpoint-21000/rng_state_15.pth +3 -0
  13. checkpoint-21000/rng_state_2.pth +3 -0
  14. checkpoint-21000/rng_state_3.pth +3 -0
  15. checkpoint-21000/rng_state_4.pth +3 -0
  16. checkpoint-21000/rng_state_5.pth +3 -0
  17. checkpoint-21000/rng_state_6.pth +3 -0
  18. checkpoint-21000/rng_state_7.pth +3 -0
  19. checkpoint-21000/rng_state_8.pth +3 -0
  20. checkpoint-21000/rng_state_9.pth +3 -0
  21. checkpoint-21000/scheduler.pt +3 -0
  22. checkpoint-21000/trainer_state.json +0 -0
  23. checkpoint-21000/training_args.bin +3 -0
checkpoint-21000/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-21000/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-21000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fd3319f46b7b86e8826304bde2359b265411f9bdbcc75ab10bce1ccc7d8a49f
3
+ size 4714837766
checkpoint-21000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e326ea08f5a6fa896dcf5b05f5df2f9c7f41d6b78b528a2d1d05a05420db55e1
3
+ size 6630552226
checkpoint-21000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:045f52acde1b13f98a16241e17c182da8f2e2648103f53e2f0f7774291a4b33f
3
+ size 15920
checkpoint-21000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f0fb5fc3df5d5288e0891be58c8bafc255d805e0c70a3d3a16407acc864971b
3
+ size 15920
checkpoint-21000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0cd076598313a39ccd4da074d2f28a737008ff3ea060de82d22059f1ba453faa
3
+ size 15933
checkpoint-21000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d32a2acc33b616c408efe91ed0d08efb382fb13d46b2b66ac804cc1e1c122a89
3
+ size 15933
checkpoint-21000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b5cb28cef2fc4a5fc3c01d82939d866cd9d4b5a25e108a0278e03c963eb22cc
3
+ size 15933
checkpoint-21000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58ed011644289379132859e01fd063dc1bcd3c9ef2cc4a3f8691581e57c7c5f2
3
+ size 15933
checkpoint-21000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab7829f41f478ae6809fdd22ebfed7f6df37f5eef6e384589d934b5e5aad863c
3
+ size 15933
checkpoint-21000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd589de1118cec6fb0f87cc366e456e227b78fe1d58a769a76fddd4d7d26dfee
3
+ size 15933
checkpoint-21000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85421e47054e908dad8a7dfe9e1dc9656ed7d9d26b18fe1f57a2b788632fd0d0
3
+ size 15920
checkpoint-21000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08d9251cc113796e49d121ef3cdece1aeea79ae41775c2bc643016d210c928cd
3
+ size 15920
checkpoint-21000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:079e4867de917c6fe278d8b6ea225b5042556bc44cf2fb57f89f64ef0046f1d5
3
+ size 15920
checkpoint-21000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45fd8192e765faec1e146ac1d31d09e9971cdb4a1c6547746d7a5133a94f27ad
3
+ size 15920
checkpoint-21000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6d75f66884a0576ed835b487e738fe85448ff9fab539142a3fee68624035702
3
+ size 15920
checkpoint-21000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5015a778efbad5caf801eadcebf345c362f0bb1187bacf9faab6f0e543d67ca1
3
+ size 15920
checkpoint-21000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f0519ab3e5643027efe5892c3be52cb61f6c9e922ab85287595cfc81b3ae428
3
+ size 15920
checkpoint-21000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2549b238ef9403712be9cf0955232d161b7002191a0bd99a0f6fb576e1fc540d
3
+ size 15920
checkpoint-21000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6b734c60d44f7f7d7d3df2545aa761c4a2a5282ff383954a512b99d6d0848da
3
+ size 1064
checkpoint-21000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-21000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97c76bc78923da33f72ea5edf873ffc16a05b9bc5073b56fadd70dd6bb34ee29
3
+ size 5752