Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
1fde64b
·
verified ·
1 Parent(s): bf134ca

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-51000/config.json +71 -0
  2. checkpoint-51000/generation_config.json +6 -0
  3. checkpoint-51000/model.safetensors +3 -0
  4. checkpoint-51000/optimizer.pt +3 -0
  5. checkpoint-51000/rng_state_0.pth +3 -0
  6. checkpoint-51000/rng_state_1.pth +3 -0
  7. checkpoint-51000/rng_state_10.pth +3 -0
  8. checkpoint-51000/rng_state_11.pth +3 -0
  9. checkpoint-51000/rng_state_12.pth +3 -0
  10. checkpoint-51000/rng_state_13.pth +3 -0
  11. checkpoint-51000/rng_state_14.pth +3 -0
  12. checkpoint-51000/rng_state_15.pth +3 -0
  13. checkpoint-51000/rng_state_2.pth +3 -0
  14. checkpoint-51000/rng_state_3.pth +3 -0
  15. checkpoint-51000/rng_state_4.pth +3 -0
  16. checkpoint-51000/rng_state_5.pth +3 -0
  17. checkpoint-51000/rng_state_6.pth +3 -0
  18. checkpoint-51000/rng_state_7.pth +3 -0
  19. checkpoint-51000/rng_state_8.pth +3 -0
  20. checkpoint-51000/rng_state_9.pth +3 -0
  21. checkpoint-51000/scheduler.pt +3 -0
  22. checkpoint-51000/trainer_state.json +0 -0
  23. checkpoint-51000/training_args.bin +3 -0
checkpoint-51000/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-51000/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-51000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c3935b59860af016a96bfc3ba8fcc5fa239ac30cfd774e4ea3768110f24aeed
3
+ size 4714837766
checkpoint-51000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:658b54a7efb5a2bacd4aeb64c3b2b6233837807451dea8bcffe135a8018af073
3
+ size 6630552226
checkpoint-51000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c9166c80ed89850164efd33f5b1366ec3cebe24af65c629369173a8c994f4b3
3
+ size 15920
checkpoint-51000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:144dcff5261f088f80f27e48f8bb658610d4e10aecd25b3f41bce51c40b3d375
3
+ size 15920
checkpoint-51000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4641de104dea1d828c24e31e69370fa252cf25fb2c8d03444f9c145ff359a4dd
3
+ size 15933
checkpoint-51000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86531734ce6880a165aa5806dd0a4feec0591f08c365049923414c9b83988851
3
+ size 15933
checkpoint-51000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5387d674f079d9c05b404243bc2fafb73d35b76ee790877f991fc506c7f82c2
3
+ size 15933
checkpoint-51000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46ac9dcfdd699e2cc3856a8d11e53466bf6c99a4b88551909684ff26b5af9710
3
+ size 15933
checkpoint-51000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca50f68ea11dd9a13bfc7daf3a93e35ec1940cfd68f45ea9c7c87a8bff3e1379
3
+ size 15933
checkpoint-51000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a75363854b0b5458685a953e60a7ebe0d89a25d761cf6fc3e2bb4d93f91a682
3
+ size 15933
checkpoint-51000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2578f02e49a87f313ce6c8ce7fb12dc524949b85fdb0dfc8601667488822da3c
3
+ size 15920
checkpoint-51000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d8412ab575f2b30bd5aae61e8962f6f4434c425242d3eafb84bb17d6cd8a3c2
3
+ size 15920
checkpoint-51000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a68fda818a5be583ac89b90b888d5524afb24ef43a32d25de3e25d64f5734a8
3
+ size 15920
checkpoint-51000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0086d85ff8b6508291c4b0354172ce396db401fc0d2d157719cd302e4f5b1057
3
+ size 15920
checkpoint-51000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:011f00d588c0d8c6db1634a199b254fbc0770771af29d8a263ed7560951a6e0b
3
+ size 15920
checkpoint-51000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0df36c622209d3a46ecf3eb901046e90c05f5ebb88579861d1c682b8a362f05
3
+ size 15920
checkpoint-51000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28e337f7c159d2c34a8638b8bbb5b8930116c28964b73c63ec4f4a6590646e4c
3
+ size 15920
checkpoint-51000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4c0932cc5bd270ac2f5052056ce1d005cba2070aa25b46c80a8fe7d3be73888
3
+ size 15920
checkpoint-51000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f20076291c49bf98c7f87388b9ff1a63ed3bafab69843ca7e521cb13c1087dc6
3
+ size 1064
checkpoint-51000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-51000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72116f3d21a5966ae8c182f7c70013b565ddb2c8435e9a2bb4d3e1852630876f
3
+ size 5752