Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
c65f871
·
verified ·
1 Parent(s): ff59ed7

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-30000/config.json +71 -0
  2. checkpoint-30000/generation_config.json +6 -0
  3. checkpoint-30000/model.safetensors +3 -0
  4. checkpoint-30000/optimizer.pt +3 -0
  5. checkpoint-30000/rng_state_0.pth +3 -0
  6. checkpoint-30000/rng_state_1.pth +3 -0
  7. checkpoint-30000/rng_state_10.pth +3 -0
  8. checkpoint-30000/rng_state_11.pth +3 -0
  9. checkpoint-30000/rng_state_12.pth +3 -0
  10. checkpoint-30000/rng_state_13.pth +3 -0
  11. checkpoint-30000/rng_state_14.pth +3 -0
  12. checkpoint-30000/rng_state_15.pth +3 -0
  13. checkpoint-30000/rng_state_2.pth +3 -0
  14. checkpoint-30000/rng_state_3.pth +3 -0
  15. checkpoint-30000/rng_state_4.pth +3 -0
  16. checkpoint-30000/rng_state_5.pth +3 -0
  17. checkpoint-30000/rng_state_6.pth +3 -0
  18. checkpoint-30000/rng_state_7.pth +3 -0
  19. checkpoint-30000/rng_state_8.pth +3 -0
  20. checkpoint-30000/rng_state_9.pth +3 -0
  21. checkpoint-30000/scheduler.pt +3 -0
  22. checkpoint-30000/trainer_state.json +0 -0
  23. checkpoint-30000/training_args.bin +3 -0
checkpoint-30000/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-30000/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-30000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4a70ed59b8e81be832aa40223fd87ec6756b7c37d4f8129d49e07053b97f05b
3
+ size 4714837766
checkpoint-30000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3658daff9a53a97350be1d452b80083f1803de302f2d529ff2477a93b95d800d
3
+ size 6630552226
checkpoint-30000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a224833d0efc431124a85f98661a3371915a9484eba6d7abe921bb72ae6304e1
3
+ size 15920
checkpoint-30000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1fc2a74a48912177c6d6f83caedbf265fa752e96f5e892ece41525006ab6354
3
+ size 15920
checkpoint-30000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab65e832a0e2ec42d51ba8d3c013bc0cf35ceede1a77af82159e30af3a23daa3
3
+ size 15933
checkpoint-30000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ea118a23a0e90ed1f9a24bd0ff8e66cfabbc7aa924e70cc35f15af3833c8403
3
+ size 15933
checkpoint-30000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d2c9ac9abb5e7407c9fa4cfb5b0b6de11a5c320dac8e4a3cf1fc365530c0240
3
+ size 15933
checkpoint-30000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04a21c59e6623fc1e3226584b4d625a8a4fad87e2d89a81b6142dabfa3261ab1
3
+ size 15933
checkpoint-30000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:001a1c6f2f97337a21ff3831f3972230747d06e1619cb606b9d8982d659adf51
3
+ size 15933
checkpoint-30000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0f9aede8370c9e6dcd78d24f8481898a31ba1d4a7ed14347382f8f5f2752304
3
+ size 15933
checkpoint-30000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:606d3285599bc0c2c711b6e90477b21740293c4ddddb7f07ccf6b188bf769b29
3
+ size 15920
checkpoint-30000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34d03de163df1c1d1ee091163ace1974a3f24a0069bedcc33d9ca30463023ed9
3
+ size 15920
checkpoint-30000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e27464a16f847ca66ba161dae5647c9c3eef3ad2d2069f5d61c6866d029152b
3
+ size 15920
checkpoint-30000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:346ce441e66f79aea87551721ccd68b02497ed53f118045e15ad3f9e5225d076
3
+ size 15920
checkpoint-30000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38a0045609b823739c52bc9f442eb59e4a5ee9a9bd34dca215cf3f90b745348f
3
+ size 15920
checkpoint-30000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:866f4d9a5088bdaa8ffdd8f8280efd5cf912841a4f413dee1ac1e2b0f5fb6a12
3
+ size 15920
checkpoint-30000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8a052e0eb9bebdbfc49759cf1970cc138c908ede58f488716e19110ea1dafd0
3
+ size 15920
checkpoint-30000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b57eefc068af86787af7af1f6d808898377f4bd5085aab14cce9fc2c654f3892
3
+ size 15920
checkpoint-30000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10251bcfbd6cb1085d5e9df908000ba794cb50dfd74fc6b90c9196b38832fa8f
3
+ size 1064
checkpoint-30000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-30000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9ccbc63118a76166e3d90d7e936cd0e63939e809a0cb5a03c3c42a6c6d02f01
3
+ size 5752