Hugging Face's logo

zhiyang1
/
SZ_multinode_dc-ae-f32c32-1.0-768_baseline

Safetensors
Model card Files
xet
Community
zhiyang1 commited on
Commit
1a08a94
·
verified ·
1 Parent(s): b123b27

Upload folder using huggingface_hub

Browse files
Files changed (24) hide show
  1. .gitattributes +1 -0
  2. checkpoint-111000/config.json +79 -0
  3. checkpoint-111000/generation_config.json +6 -0
  4. checkpoint-111000/model.safetensors +3 -0
  5. checkpoint-111000/optimizer.pt +3 -0
  6. checkpoint-111000/rng_state_0.pth +3 -0
  7. checkpoint-111000/rng_state_1.pth +3 -0
  8. checkpoint-111000/rng_state_10.pth +3 -0
  9. checkpoint-111000/rng_state_11.pth +3 -0
  10. checkpoint-111000/rng_state_12.pth +3 -0
  11. checkpoint-111000/rng_state_13.pth +3 -0
  12. checkpoint-111000/rng_state_14.pth +3 -0
  13. checkpoint-111000/rng_state_15.pth +3 -0
  14. checkpoint-111000/rng_state_2.pth +3 -0
  15. checkpoint-111000/rng_state_3.pth +3 -0
  16. checkpoint-111000/rng_state_4.pth +3 -0
  17. checkpoint-111000/rng_state_5.pth +3 -0
  18. checkpoint-111000/rng_state_6.pth +3 -0
  19. checkpoint-111000/rng_state_7.pth +3 -0
  20. checkpoint-111000/rng_state_8.pth +3 -0
  21. checkpoint-111000/rng_state_9.pth +3 -0
  22. checkpoint-111000/scheduler.pt +3 -0
  23. checkpoint-111000/trainer_state.json +3 -0
  24. checkpoint-111000/training_args.bin +3 -0
.gitattributes CHANGED
@@ -49,3 +49,4 @@ checkpoint-99000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
49
  checkpoint-100000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
50
  checkpoint-109000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
51
  checkpoint-110000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
 
 
49
  checkpoint-100000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
50
  checkpoint-109000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
51
  checkpoint-110000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
52
+ checkpoint-111000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
checkpoint-111000/config.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "ar_steps": 1,
3
+ "architectures": [
4
+ "DiffVLMBaseline"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "condition_layer": -1,
9
+ "eos_token_id": 151645,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 1536,
12
+ "image_token_id": 151655,
13
+ "img_cross_attention_dim": 2048,
14
+ "img_diffuser_depth": 2,
15
+ "img_ffn_dim_multiplier": null,
16
+ "img_hidden_size": 1536,
17
+ "img_multiple_of": 256,
18
+ "img_norm_eps": 1e-05,
19
+ "img_num_attention_heads": 12,
20
+ "img_num_kv_heads": 12,
21
+ "img_qk_norm": true,
22
+ "in_channels": 32,
23
+ "initializer_range": 0.02,
24
+ "inject_img_diffuser": false,
25
+ "input_size": 32,
26
+ "intermediate_size": 8960,
27
+ "layer_group_size": 7,
28
+ "layerwise_start_idx": 0,
29
+ "lora_alpha": 128,
30
+ "lora_bias": "none",
31
+ "lora_dropout": 0.05,
32
+ "lora_enable": false,
33
+ "lora_r": 64,
34
+ "max_position_embeddings": 32768,
35
+ "max_window_layers": 28,
36
+ "model_type": "qwen2_vl",
37
+ "non_linearity": 1,
38
+ "norm_elementwise_affine": true,
39
+ "num_attention_heads": 12,
40
+ "num_hidden_layers": 28,
41
+ "num_key_value_heads": 2,
42
+ "patch_size": 1,
43
+ "repa_coeff": 0.1,
44
+ "repa_layers": "2",
45
+ "repa_shared": false,
46
+ "rms_norm_eps": 1e-06,
47
+ "rope_scaling": {
48
+ "mrope_section": [
49
+ 16,
50
+ 24,
51
+ 24
52
+ ],
53
+ "rope_type": "default",
54
+ "type": "default"
55
+ },
56
+ "rope_theta": 1000000.0,
57
+ "sample_size": 128,
58
+ "sampling_steps": 28,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": true,
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.47.0",
63
+ "use_cache": true,
64
+ "use_repa": false,
65
+ "use_residual_attn": false,
66
+ "use_sliding_window": false,
67
+ "vae_path": "mit-han-lab/dc-ae-f32c32-in-1.0-diffusers",
68
+ "video_token_id": 151656,
69
+ "vision_config": {
70
+ "hidden_size": 1536,
71
+ "in_chans": 3,
72
+ "model_type": "qwen2_vl",
73
+ "spatial_patch_size": 14
74
+ },
75
+ "vision_end_token_id": 151653,
76
+ "vision_start_token_id": 151652,
77
+ "vision_token_id": 151654,
78
+ "vocab_size": 151936
79
+ }
checkpoint-111000/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-111000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e50dc18c93ec49ba81a554f48627ed6309e6f9d27a7ab54fff9ff359718e8d14
3
+ size 4410723984
checkpoint-111000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58273090bf6bb4a6191e60a880a883edf61b116153d609360d8e43fcf2e5be72
3
+ size 6330255386
checkpoint-111000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e3b5c37dd5842b1bd1baa9e096bfa041f07726ccc85945db406b60e4bf42332
3
+ size 15984
checkpoint-111000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b099e871d32ae5eb2a20d9cc052c96fd50e35e8ce3f738dda2ef6263f1864896
3
+ size 15984
checkpoint-111000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f76f818067e90c44bec6fe849e294c3efbcab69b2944444f890811cbc8175c8f
3
+ size 15997
checkpoint-111000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e99425fffbeb68713767512666335a9855a89210ce9fd30cc837f44ac3682fb3
3
+ size 15997
checkpoint-111000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d861fcb8579d39ecffee92729d846dddcc7dfe0b4fda742c6cacdf9b07d627e
3
+ size 15997
checkpoint-111000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3a5a789b917c7794fc8af03d9ac4966ba795c348f709fa9bdf9f92714c88a7a
3
+ size 15997
checkpoint-111000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:001ed1f9083f31635a6d836062a32ee5b4d44ad7a5871eb9a92c31d2a45f6553
3
+ size 15997
checkpoint-111000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:172f0b7083b674b49c9550bfe9ec0f2e22f59b489e0ec5010c4576bd93db4f56
3
+ size 15997
checkpoint-111000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7c4da642b395ef65ff78cfcbef7efcaac6b9a34c616603ca6f52275960b61c0
3
+ size 15984
checkpoint-111000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9b89c9e39f3f4388fda9c35998745702156aa38f866cb4bbe11cae145fdcd8b
3
+ size 15984
checkpoint-111000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5407ab30c23e56623081891a4fd9da63c01ed853bb9db6090f1939be9c8ae8d9
3
+ size 15984
checkpoint-111000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8eed85a2e032a4834fb14e10e8a388ba337dfbfe87ba607cc73fce0940c8ef7
3
+ size 15984
checkpoint-111000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:477674d53656fa39ded5ac836aae181b530b2f6560bbef40ec56e834014f0a40
3
+ size 15984
checkpoint-111000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57b3133a23af61fae7a5a4cd8bc26aee2968dee07c667a85e7d2a585443aa547
3
+ size 15984
checkpoint-111000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8a91ff6c7038269a03c9a4ae60f3a9ea018c7c524d783106fa840b729ae8dfa
3
+ size 15984
checkpoint-111000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a47e6ff8da69ffdacae43c04a14dc212d710a895ae47a311100f88140aae78b
3
+ size 15984
checkpoint-111000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a3238adb2d237db5599a7c85b9eb35563f6986a53883944c20b5608780c3de1
3
+ size 1064
checkpoint-111000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f1eabfb58be728e7276684f8350d94dd0e64b9e2c07ac2ca1fe69c56bf950f5
3
+ size 16911716
checkpoint-111000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:faa574cf79cde84154ebb4333b5a22fc9874feef55b7962133f628398fcd6589
3
+ size 5944