Hugging Face's logo

umd-vt-nyu
/
JH_multinode_dc-ae-f32c32-in-1.0-diffusers-768_patch-1_group-7_5e4_residual_attn

Safetensors
Model card Files
xet
Community
jiuhai commited on
Commit
a7b1dce
·
verified ·
1 Parent(s): 8832c74

Upload folder using huggingface_hub

Browse files
Files changed (31) hide show
  1. .gitattributes +2 -0
  2. checkpoint-69000/config.json +79 -0
  3. checkpoint-69000/generation_config.json +6 -0
  4. checkpoint-69000/model.safetensors +3 -0
  5. checkpoint-69000/optimizer.pt +3 -0
  6. checkpoint-69000/rng_state_0.pth +3 -0
  7. checkpoint-69000/rng_state_1.pth +3 -0
  8. checkpoint-69000/rng_state_2.pth +3 -0
  9. checkpoint-69000/rng_state_3.pth +3 -0
  10. checkpoint-69000/rng_state_4.pth +3 -0
  11. checkpoint-69000/rng_state_5.pth +3 -0
  12. checkpoint-69000/rng_state_6.pth +3 -0
  13. checkpoint-69000/rng_state_7.pth +3 -0
  14. checkpoint-69000/scheduler.pt +3 -0
  15. checkpoint-69000/trainer_state.json +3 -0
  16. checkpoint-69000/training_args.bin +3 -0
  17. checkpoint-70000/config.json +79 -0
  18. checkpoint-70000/generation_config.json +6 -0
  19. checkpoint-70000/model.safetensors +3 -0
  20. checkpoint-70000/optimizer.pt +3 -0
  21. checkpoint-70000/rng_state_0.pth +3 -0
  22. checkpoint-70000/rng_state_1.pth +3 -0
  23. checkpoint-70000/rng_state_2.pth +3 -0
  24. checkpoint-70000/rng_state_3.pth +3 -0
  25. checkpoint-70000/rng_state_4.pth +3 -0
  26. checkpoint-70000/rng_state_5.pth +3 -0
  27. checkpoint-70000/rng_state_6.pth +3 -0
  28. checkpoint-70000/rng_state_7.pth +3 -0
  29. checkpoint-70000/scheduler.pt +3 -0
  30. checkpoint-70000/trainer_state.json +3 -0
  31. checkpoint-70000/training_args.bin +3 -0
.gitattributes CHANGED
@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ checkpoint-69000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
37
+ checkpoint-70000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
checkpoint-69000/config.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "ar_steps": 1,
3
+ "architectures": [
4
+ "DiffVLMDiffusion"
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": true,
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-69000/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-69000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a51098b2afe15322b51a087c29e4142c3f531af74c9a759a907a16db5e826ca
3
+ size 4411759432
checkpoint-69000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:24675695673184387e659aa29b14ae3bc71119b1dff813f1645134edb9fef65a
3
+ size 6332050591
checkpoint-69000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f30b7c403509b30632e039ae25bb9aa8784550e0febe7eef397dcf8a9c5e532f
3
+ size 16389
checkpoint-69000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:600df7043caf939ba93ae63e6f0580a9b2697eec9b708c9dac5c8e4657d916af
3
+ size 16389
checkpoint-69000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdfe51d99e59680159316b464eb59dda2c6571b7dfaacd9b060373d15cc68424
3
+ size 16389
checkpoint-69000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e104b38be1cd912e0b308c28d799a442e60e83f9b9370aa67657469d0b98bc2f
3
+ size 16389
checkpoint-69000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e36a7d2b174fddf7713cc2ed6f9c59a5d6f8fad107bd0269559b3d873a5079b1
3
+ size 16389
checkpoint-69000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd286dcadb22f741195e0cb7b30039549cdb1b80749a92555956db0706ccd749
3
+ size 16389
checkpoint-69000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b624071485156193b50227104aca9900007008f46841743fbc594e039667a4bf
3
+ size 16389
checkpoint-69000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13132b06a74066f1377b8ad6d1d86df6a2c3fdcef3967d9262ff93479afa954e
3
+ size 16389
checkpoint-69000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bf03bcb5656b2caa588d6b903a817612191ecda78f273b678407f754182d9085
3
+ size 1465
checkpoint-69000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48684a4abd41c72508d8870345e9f43052313b1d7e9fb6a208ec967313c70404
3
+ size 10548442
checkpoint-69000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcb4f54b41da409c59d9fc2031521301bbcee641932d2290d8486c77ee9eee69
3
+ size 6481
checkpoint-70000/config.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "ar_steps": 1,
3
+ "architectures": [
4
+ "DiffVLMDiffusion"
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": true,
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-70000/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-70000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9dd9da55db7a37b0cdd1912821771b0528539dd6675979d40400cd6100c9657
3
+ size 4411759432
checkpoint-70000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d8e928de8660fc871060c3868ecb8b71f0091ccbfd483e2462da7cc6a645fcbc
3
+ size 6332050591
checkpoint-70000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11e9893a9957c9a22e1ff2ba125c376aac7e3556592a7dad3aca1c022ea99fdf
3
+ size 16389
checkpoint-70000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef8fdd3c8badee7f3a2317525aa97fa7474f2c2af9da1114c93393351135a841
3
+ size 16389
checkpoint-70000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:292e7f428885d66c2fc69c28fbdc2bf6c59d78ce919695b8e3fce8d4beba66fe
3
+ size 16389
checkpoint-70000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15ca920aa4312e2b2eecb8f94b5fca2b1f7cc403d335670300dd6fac530912c3
3
+ size 16389
checkpoint-70000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f14446cb3890fc2ce2fc67f5f53f47bd1734fc97c17f52148bd03f6e3f5dd56
3
+ size 16389
checkpoint-70000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2ba3b60f5ea48c837a4a744db76af1c607f761985f2fea1657156351e511902
3
+ size 16389
checkpoint-70000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:372f74ced183d666d94c75ec68b6b63f73fc1323391b0df68a4151aeeab439bd
3
+ size 16389
checkpoint-70000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1e21f31b182ede37cb8143b698f3e74b7b6e8709a5a8644d6c82bc8e9ccc2fe
3
+ size 16389
checkpoint-70000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fd35a1510d72e0e987cb9bab23577efc4b23d73e0c52ffd0f0ef476e1dde9e0
3
+ size 1465
checkpoint-70000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cac720538bc22756005720bf69b31d39c90b6e3113cd39bded94c006c05637d7
3
+ size 10702097
checkpoint-70000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcb4f54b41da409c59d9fc2031521301bbcee641932d2290d8486c77ee9eee69
3
+ size 6481