Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
a703944
·
verified ·
1 Parent(s): bd90745

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-46000/config.json +71 -0
  2. checkpoint-46000/generation_config.json +6 -0
  3. checkpoint-46000/model.safetensors +3 -0
  4. checkpoint-46000/optimizer.pt +3 -0
  5. checkpoint-46000/rng_state_0.pth +3 -0
  6. checkpoint-46000/rng_state_1.pth +3 -0
  7. checkpoint-46000/rng_state_10.pth +3 -0
  8. checkpoint-46000/rng_state_11.pth +3 -0
  9. checkpoint-46000/rng_state_12.pth +3 -0
  10. checkpoint-46000/rng_state_13.pth +3 -0
  11. checkpoint-46000/rng_state_14.pth +3 -0
  12. checkpoint-46000/rng_state_15.pth +3 -0
  13. checkpoint-46000/rng_state_2.pth +3 -0
  14. checkpoint-46000/rng_state_3.pth +3 -0
  15. checkpoint-46000/rng_state_4.pth +3 -0
  16. checkpoint-46000/rng_state_5.pth +3 -0
  17. checkpoint-46000/rng_state_6.pth +3 -0
  18. checkpoint-46000/rng_state_7.pth +3 -0
  19. checkpoint-46000/rng_state_8.pth +3 -0
  20. checkpoint-46000/rng_state_9.pth +3 -0
  21. checkpoint-46000/scheduler.pt +3 -0
  22. checkpoint-46000/trainer_state.json +0 -0
  23. checkpoint-46000/training_args.bin +3 -0
checkpoint-46000/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-46000/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-46000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ae76177662c808e6132694d50e692d1eea937154dc82b25624a2b5c22086cb8
3
+ size 4714837766
checkpoint-46000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc5a2877275fd09a88e229ff01f6155f32f457f73eb270efa230e20db776fcaa
3
+ size 6630552226
checkpoint-46000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11dc8d55e75a2824b8bec1ff3daf868d67d0350b1e6d4d97285879de0c3c590b
3
+ size 15920
checkpoint-46000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88eb2ec85fa905fd0afde9a7e52a2afb16acfe76b3d7a302032dc6e1aedaba5c
3
+ size 15920
checkpoint-46000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea983fc0a397ddad928610ae69a9cf5d0838acc392dc0517a3bde37728aa9432
3
+ size 15933
checkpoint-46000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eedac35a746bbfd7d1fad43d6a318504d492fe534bbfaf69591243f65856c9a3
3
+ size 15933
checkpoint-46000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:115da6fdb54c84e0900836594e7e942660aa71c38dcfc2105fc7a9ec03829eae
3
+ size 15933
checkpoint-46000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81d66e8bb126aac81159415f59342b0511734b053a23c3a2af7fe73a15cf9ec1
3
+ size 15933
checkpoint-46000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2c9c17c6e3ed8a524976c94cf2bd33c5b70297d8a7af52837219782bc8e872a
3
+ size 15933
checkpoint-46000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9a99c7591e0fa90b49e0fa547e1493992e984a5b3438d745796ab348f6c238c
3
+ size 15933
checkpoint-46000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fca120cbf996ac86190aa051685301596d3c188ccf7d7533a59dbd508555759
3
+ size 15920
checkpoint-46000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a6eb2d540c6c1525e486047785d96ddb59fb34550cb3a428f87403c8adc5228
3
+ size 15920
checkpoint-46000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e304755dc4d34675c799367dd315fe5194e00188b4b128b60e52e041d8b93d0
3
+ size 15920
checkpoint-46000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3f84a7e549917610a1178ead6a1d2888e96dbef6dedd7e8e2ceda3eb21a2911
3
+ size 15920
checkpoint-46000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e781afe0465ab57a29a74d3efd152868bf875bd4f5fad1ebf0716d7ea6688ab9
3
+ size 15920
checkpoint-46000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fac9c46e9b910bea45a4a841adb8e8871aa8414d9108986343f0f1c10e19ba1
3
+ size 15920
checkpoint-46000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3734ff3ffb95d2e154f3c937d57320fcaeafc5348c5426e7ae243cf80160a40f
3
+ size 15920
checkpoint-46000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4da8bc96631d2b927d6a152c832e14af076f9a4238a202cd9585a906e8079cb0
3
+ size 15920
checkpoint-46000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e35e100fa868a377dadb4102242efc6e92d49f341c89c254a42fa82cfac10fd
3
+ size 1064
checkpoint-46000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-46000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74790764509fc5128285c160c457610d95c905b4033cc9562615d93d82f9e270
3
+ size 5752