Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
bf134ca
·
verified ·
1 Parent(s): a703944

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-50000/config.json +71 -0
  2. checkpoint-50000/generation_config.json +6 -0
  3. checkpoint-50000/model.safetensors +3 -0
  4. checkpoint-50000/optimizer.pt +3 -0
  5. checkpoint-50000/rng_state_0.pth +3 -0
  6. checkpoint-50000/rng_state_1.pth +3 -0
  7. checkpoint-50000/rng_state_10.pth +3 -0
  8. checkpoint-50000/rng_state_11.pth +3 -0
  9. checkpoint-50000/rng_state_12.pth +3 -0
  10. checkpoint-50000/rng_state_13.pth +3 -0
  11. checkpoint-50000/rng_state_14.pth +3 -0
  12. checkpoint-50000/rng_state_15.pth +3 -0
  13. checkpoint-50000/rng_state_2.pth +3 -0
  14. checkpoint-50000/rng_state_3.pth +3 -0
  15. checkpoint-50000/rng_state_4.pth +3 -0
  16. checkpoint-50000/rng_state_5.pth +3 -0
  17. checkpoint-50000/rng_state_6.pth +3 -0
  18. checkpoint-50000/rng_state_7.pth +3 -0
  19. checkpoint-50000/rng_state_8.pth +3 -0
  20. checkpoint-50000/rng_state_9.pth +3 -0
  21. checkpoint-50000/scheduler.pt +3 -0
  22. checkpoint-50000/trainer_state.json +0 -0
  23. checkpoint-50000/training_args.bin +3 -0
checkpoint-50000/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-50000/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-50000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54253f4b466fd35fb317458dc323e75e99ba0e98f3ca6884a1f7e8ea5b8c9cfd
3
+ size 4714837766
checkpoint-50000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e84d9c0b3e3eab842f6c51eb27accb7d95edd536c0108d79f8252f78dc0366cb
3
+ size 6630552226
checkpoint-50000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c552806574a2a6d69eea7052a62197d2f5111a8b6b4984b0d9fc1c8d4e70ae43
3
+ size 15920
checkpoint-50000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ae2fabd74e2a1b850eb3805e425bc15e5b892adb5c8736587c7e74037685588
3
+ size 15920
checkpoint-50000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2daa958b35a1fb158cde059ac36822fb882f10a0c1dd1752fb3d670e898997a2
3
+ size 15933
checkpoint-50000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b79c0d186c4032c59bfd75dd008346e3f6c2e08425498b23756d1036866276c
3
+ size 15933
checkpoint-50000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ebb381225a6d0da1e902c36ab191986a780cc848178fb7578015f2db235b700
3
+ size 15933
checkpoint-50000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9ea7ef719977f0a5e6990a80a6058ff4c4ee7d5eae21509e1d9f3dfdbbfdc88
3
+ size 15933
checkpoint-50000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:278446b0132aede9d969786c4c697bcae789e7c5de9a6be1edf2d84ced20c5d1
3
+ size 15933
checkpoint-50000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31ad659897cb564c1faf3382d8867f9c94c30c314e18c3a3104ca66564d09afa
3
+ size 15933
checkpoint-50000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26d692ac60081191b3561a85011c0d9b87f956a715ced7306e1d98e38a5a6179
3
+ size 15920
checkpoint-50000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf271921066e7098beddcdeca56440af3dcd2c54f51fce751c1433f4830796bc
3
+ size 15920
checkpoint-50000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61e8e46ebac65fa8de674d9333cd0098cff15919bb87a90a637e58d63476f658
3
+ size 15920
checkpoint-50000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26abe06f57a34b0698fb823f31df63519430dbb3072714e9c74311d1e8357c70
3
+ size 15920
checkpoint-50000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c18cd97180ee67c52039e7afc9b81bf6897c34ce8b75abb1383701dc144ca8a4
3
+ size 15920
checkpoint-50000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c76cdc70e66fab1a4da4fd0f532e99d5b6599c533b45a49f8bc28dfa815fa0a5
3
+ size 15920
checkpoint-50000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98907f5de609c5c795ecd7572d8b3847b2d2a2f4ae129044d98be4ef372f99a9
3
+ size 15920
checkpoint-50000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b165a4381fdff66d378e8f6c6749118e36405e29861076d5efa25473c04e7cc0
3
+ size 15920
checkpoint-50000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e8c62b290c5c06f78a386c6d256c09d0705fd53312a2e572aff46a3515e1dbb
3
+ size 1064
checkpoint-50000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-50000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eea0180cbf06ab70adffdd577f6c76a5a9ddd87755d7906568097e894a9692dd
3
+ size 5816