Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
b77758b
·
verified ·
1 Parent(s): 773247e

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-20000/config.json +71 -0
  2. checkpoint-20000/generation_config.json +6 -0
  3. checkpoint-20000/model.safetensors +3 -0
  4. checkpoint-20000/optimizer.pt +3 -0
  5. checkpoint-20000/rng_state_0.pth +3 -0
  6. checkpoint-20000/rng_state_1.pth +3 -0
  7. checkpoint-20000/rng_state_10.pth +3 -0
  8. checkpoint-20000/rng_state_11.pth +3 -0
  9. checkpoint-20000/rng_state_12.pth +3 -0
  10. checkpoint-20000/rng_state_13.pth +3 -0
  11. checkpoint-20000/rng_state_14.pth +3 -0
  12. checkpoint-20000/rng_state_15.pth +3 -0
  13. checkpoint-20000/rng_state_2.pth +3 -0
  14. checkpoint-20000/rng_state_3.pth +3 -0
  15. checkpoint-20000/rng_state_4.pth +3 -0
  16. checkpoint-20000/rng_state_5.pth +3 -0
  17. checkpoint-20000/rng_state_6.pth +3 -0
  18. checkpoint-20000/rng_state_7.pth +3 -0
  19. checkpoint-20000/rng_state_8.pth +3 -0
  20. checkpoint-20000/rng_state_9.pth +3 -0
  21. checkpoint-20000/scheduler.pt +3 -0
  22. checkpoint-20000/trainer_state.json +0 -0
  23. checkpoint-20000/training_args.bin +3 -0
checkpoint-20000/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-20000/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-20000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc40531d37b97656178b68687cb2688e8a0e558c04681bd0b67ac354a12da911
3
+ size 4714837766
checkpoint-20000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41d9a6e8a330e35f1c669d1e25771afb42a0c19795667d3ff43c415460bcceda
3
+ size 6630552226
checkpoint-20000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e0b2ffe4af82e31207d4c190b6bec23f2af0e101c1a7fd803c14b080789f9c2
3
+ size 15920
checkpoint-20000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdc9fe98aaa8e47fe1ba32efcd2d823f53487f4c52cc7b60a96416f5fda75497
3
+ size 15920
checkpoint-20000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5066e870737a338c97e9af8d9c12388493e7e36ecbf56231307ce1be19949039
3
+ size 15933
checkpoint-20000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11437550dcc1ae15986f7eea9d1c5b1d21410afa61576add332d14ce167660fa
3
+ size 15933
checkpoint-20000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a38afde1214eb37d2c3102736d09141edf865602dda4950ac69946bf08dac4fd
3
+ size 15933
checkpoint-20000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d70d2cb796ce27b28aac611c60fbcb0dccea6c794b574aabb08428181022df1
3
+ size 15933
checkpoint-20000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99ea7d7b3b62e2458c05461cf77a728d929cb0b02434f12fd0c0cd4946fe0143
3
+ size 15933
checkpoint-20000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa0a772687c2eebd6af19316954ad7d898f6433102e4467e48de9a7313a5528e
3
+ size 15933
checkpoint-20000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c57731706600d387e2f103697009947b5a551281ac6157af1b1337e51dd754b9
3
+ size 15920
checkpoint-20000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8be8852817cc9463c18e5f59fe590a199fc97a898c44b5bcf0092b5ebefb0d7
3
+ size 15920
checkpoint-20000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7b13e893f57761f17c5f64d284dce70c089d825e4ba49fa9686f757df1ae99f
3
+ size 15920
checkpoint-20000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2afdccc87ea2f8979b5a2fd3a77f7bac745c5b7f35299314b6eab0fd5f48e3aa
3
+ size 15920
checkpoint-20000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30625f71b93cf8687c983dd2f599b5a8f096311487c063432a61728b5f2ab12c
3
+ size 15920
checkpoint-20000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33edde482975f7b068783e33885ab45087bfda1f3af0e5a17d458d2d3a71e2e5
3
+ size 15920
checkpoint-20000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31d02caea7493b14dc6eb044ec319557f0f182a41dc4612d4488b7d67cd2d340
3
+ size 15920
checkpoint-20000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:143064f34631bd84bafc83f479e0adf37d8ba2fb315fbd18a760514bdcdf16cb
3
+ size 15920
checkpoint-20000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c84aebd00fa56a3c7d8e7c3e150905f18061ed98c1a033c0c777cbf63241f81
3
+ size 1064
checkpoint-20000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-20000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2232dcd53929ac77cbe2072e0fb6615e745d62f17105ef03a5ce2a3810969b2
3
+ size 5816