Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
ba94897
·
verified ·
1 Parent(s): c65f871

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-38000/config.json +71 -0
  2. checkpoint-38000/generation_config.json +6 -0
  3. checkpoint-38000/model.safetensors +3 -0
  4. checkpoint-38000/optimizer.pt +3 -0
  5. checkpoint-38000/rng_state_0.pth +3 -0
  6. checkpoint-38000/rng_state_1.pth +3 -0
  7. checkpoint-38000/rng_state_10.pth +3 -0
  8. checkpoint-38000/rng_state_11.pth +3 -0
  9. checkpoint-38000/rng_state_12.pth +3 -0
  10. checkpoint-38000/rng_state_13.pth +3 -0
  11. checkpoint-38000/rng_state_14.pth +3 -0
  12. checkpoint-38000/rng_state_15.pth +3 -0
  13. checkpoint-38000/rng_state_2.pth +3 -0
  14. checkpoint-38000/rng_state_3.pth +3 -0
  15. checkpoint-38000/rng_state_4.pth +3 -0
  16. checkpoint-38000/rng_state_5.pth +3 -0
  17. checkpoint-38000/rng_state_6.pth +3 -0
  18. checkpoint-38000/rng_state_7.pth +3 -0
  19. checkpoint-38000/rng_state_8.pth +3 -0
  20. checkpoint-38000/rng_state_9.pth +3 -0
  21. checkpoint-38000/scheduler.pt +3 -0
  22. checkpoint-38000/trainer_state.json +0 -0
  23. checkpoint-38000/training_args.bin +3 -0
checkpoint-38000/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-38000/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-38000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3497568eacdc14eae44b1fb881ed8d002d8b99d82043bea0196fe286505737af
3
+ size 4714837766
checkpoint-38000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e4da8d4f2905b2155d9b862c4801c2f38fdf740ae4b7a5b785b2cc156e6ef5d
3
+ size 6630552226
checkpoint-38000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ea3f89f3a7f187cccac0c3573fbd49020ff6f57e365ac73035a419588410d5e
3
+ size 15920
checkpoint-38000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9ef9209ed7310da63a5492ea8e748381ecd8bf0ff7bcf2c30837ffafeeaf4b2
3
+ size 15920
checkpoint-38000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e56fb50413d6edd348d1e725d66358a58308c244bf8814b213fd8c2a9070e4f
3
+ size 15933
checkpoint-38000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6466c56e67b2c61a20fe716ebe998f294ee2d4af0dedb0b3ae3c81f0eaaceb5c
3
+ size 15933
checkpoint-38000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:476978eb522cfb64fef4b0f5308cac060d144142c51513a82f35109927345b08
3
+ size 15933
checkpoint-38000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b85c4f20e1203b7bcd65e9e1b3894508125fa86ced44ce5f902f91ff6eb00c36
3
+ size 15933
checkpoint-38000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f610b923d3480279d4af80873dc0b5548d493381b259b171b5e45310a49f860
3
+ size 15933
checkpoint-38000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7949c285151310dc46079c221e58e7d189f49b4a841dd8eb4dc2dd1cafd26483
3
+ size 15933
checkpoint-38000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a142f7ad36b7dfcd7b03acb35954fdada10be4e2a661b70605a2e54a18c8c71
3
+ size 15920
checkpoint-38000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a942b78051118229c2f8dc57e464d89c240cc396232757b0eef159bfd7320f3
3
+ size 15920
checkpoint-38000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81639d2a26094c9d13e58fe2003701259cb71f2d4a35470209ea82b8bb05f64c
3
+ size 15920
checkpoint-38000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1cebfaa379739cebd1105e3f3bd160db84a362ce890bc701c9f976e242475a0
3
+ size 15920
checkpoint-38000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7d3b4da471cd5290f4f57d71e19940153918c856c6fb6d0005ba8597c244ff7
3
+ size 15920
checkpoint-38000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1020c0afd6c18067f4451cc1ae7fe91331a7d96c003800a628eb37524961ad0
3
+ size 15920
checkpoint-38000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0706ada50d1e79aa91cb1618366155f720ed5e7cbb7871e1ccec99900903260b
3
+ size 15920
checkpoint-38000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9d666160610ebadbdc3dce7aefdd9464d0079b18d087c3346efec5281b788e3
3
+ size 15920
checkpoint-38000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8365b7b873509a6c7cf9b5cc77ed98a004c4d46913094aa05bc9f136ea89c62
3
+ size 1064
checkpoint-38000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-38000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0b1cea90d0b060e093f84804350c592f2c6b7d416059c6f26b38d91f21c3d5e
3
+ size 5752