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
12793e2
·
verified ·
1 Parent(s): 64c334c

Upload folder using huggingface_hub

Browse files
Files changed (16) hide show
  1. .gitattributes +1 -0
  2. checkpoint-80000/config.json +79 -0
  3. checkpoint-80000/generation_config.json +6 -0
  4. checkpoint-80000/model.safetensors +3 -0
  5. checkpoint-80000/optimizer.pt +3 -0
  6. checkpoint-80000/rng_state_0.pth +3 -0
  7. checkpoint-80000/rng_state_1.pth +3 -0
  8. checkpoint-80000/rng_state_2.pth +3 -0
  9. checkpoint-80000/rng_state_3.pth +3 -0
  10. checkpoint-80000/rng_state_4.pth +3 -0
  11. checkpoint-80000/rng_state_5.pth +3 -0
  12. checkpoint-80000/rng_state_6.pth +3 -0
  13. checkpoint-80000/rng_state_7.pth +3 -0
  14. checkpoint-80000/scheduler.pt +3 -0
  15. checkpoint-80000/trainer_state.json +3 -0
  16. checkpoint-80000/training_args.bin +3 -0
.gitattributes CHANGED
@@ -37,3 +37,4 @@ 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
38
  checkpoint-78000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
39
  checkpoint-79000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
 
 
37
  checkpoint-70000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
38
  checkpoint-78000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
39
  checkpoint-79000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
40
+ checkpoint-80000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
checkpoint-80000/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-80000/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-80000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e29ede61e536a56118eba4a2e5e6356cf5ea717dc1f2646b73f4361fe8ff8ee
3
+ size 4411759432
checkpoint-80000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab6d669d65e8b352928192574c13ffa2f53d61fdb8c5b56c4cca459d9ec2cb15
3
+ size 6332050591
checkpoint-80000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0873176611166d90fcafab9b88fb4f2f389740830db39c73716471990adb37a0
3
+ size 16389
checkpoint-80000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e91c7f42fa2c50f51edbd0b6f525a44aad8478c0f653a3892714f9730fe7c78e
3
+ size 16389
checkpoint-80000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df4bd11c7ff1298d5a6c230233c51a28823c175990c8aa767a2c4abe300f3f79
3
+ size 16389
checkpoint-80000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:917848ee6c17ca5a4294f1183b40ac85264c67e585e8edbc7caa110c265ad06d
3
+ size 16389
checkpoint-80000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:042faa81047b043cff4ab98885e184587e7d6b53efdf0225de69847a34d3df98
3
+ size 16389
checkpoint-80000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3f92dc58e52ec9624b5970db24cf136f3474d6cece3545906442c8223a78926a
3
+ size 16389
checkpoint-80000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a16968b5f0c9250f624dcd744fcb4cbfbeecc635f5451a7bce1db2eda4cf5688
3
+ size 16389
checkpoint-80000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34b1e0f7974758b7f1f0f9af41ca66f7f86527b0a3d94979c345e5ea816a23be
3
+ size 16389
checkpoint-80000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bca687f27583cd581f1ddf5ce79cb5ce13c3bccfb4de45f9871eff232089c7f1
3
+ size 1465
checkpoint-80000/trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae2fffbce8fe1b01e8d6601d1ab2ca7df447b950a8540c4de1840ffbc1125524
3
+ size 12239665
checkpoint-80000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a25e8a3121ed011208a5213fb67a4353f4749ecf77a8f84ef0ced2d2a4cf1eb4
3
+ size 6481