Hugging Face's logo

shizhediao2
/
SZ_multinode_vlm_dc-vae_baseline

Safetensors
Model card Files Community
shizhediao2 commited on
Commit
bd90745
·
verified ·
1 Parent(s): ba94897

Upload folder using huggingface_hub

Browse files
Files changed (23) hide show
  1. checkpoint-40000/config.json +71 -0
  2. checkpoint-40000/generation_config.json +6 -0
  3. checkpoint-40000/model.safetensors +3 -0
  4. checkpoint-40000/optimizer.pt +3 -0
  5. checkpoint-40000/rng_state_0.pth +3 -0
  6. checkpoint-40000/rng_state_1.pth +3 -0
  7. checkpoint-40000/rng_state_10.pth +3 -0
  8. checkpoint-40000/rng_state_11.pth +3 -0
  9. checkpoint-40000/rng_state_12.pth +3 -0
  10. checkpoint-40000/rng_state_13.pth +3 -0
  11. checkpoint-40000/rng_state_14.pth +3 -0
  12. checkpoint-40000/rng_state_15.pth +3 -0
  13. checkpoint-40000/rng_state_2.pth +3 -0
  14. checkpoint-40000/rng_state_3.pth +3 -0
  15. checkpoint-40000/rng_state_4.pth +3 -0
  16. checkpoint-40000/rng_state_5.pth +3 -0
  17. checkpoint-40000/rng_state_6.pth +3 -0
  18. checkpoint-40000/rng_state_7.pth +3 -0
  19. checkpoint-40000/rng_state_8.pth +3 -0
  20. checkpoint-40000/rng_state_9.pth +3 -0
  21. checkpoint-40000/scheduler.pt +3 -0
  22. checkpoint-40000/trainer_state.json +0 -0
  23. checkpoint-40000/training_args.bin +3 -0
checkpoint-40000/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-40000/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-40000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0438742f8d83304d2bb8059bc39ed1d0ddba669b34e790520675055fa66fb268
3
+ size 4714837766
checkpoint-40000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e09b82c7c04c49e2d3408d63d800bfbbeda993e5d17a5e97f0bccdb65baa6897
3
+ size 6630552226
checkpoint-40000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15043b5f04713d06572b34bfbeff08312b5ef34da129f396262acb33384945fc
3
+ size 15920
checkpoint-40000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3ed75c5f1ecbbd035652131eee36015240b2f318d53b3e893b2dbf0d3adda1e
3
+ size 15920
checkpoint-40000/rng_state_10.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6142630fea7223d72afc0e17365bcf68b134979c97e48734c7521e4221aa027
3
+ size 15933
checkpoint-40000/rng_state_11.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:515a6ca48a5e8843e13943b88fb55b5599a5b6bed9e3610fe9420a9ed8bb6d5b
3
+ size 15933
checkpoint-40000/rng_state_12.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cdf2266b1e9f05777e436bd856c0b1b03798b898ef188a9c7f94f0e6bc612da
3
+ size 15933
checkpoint-40000/rng_state_13.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4921e408cb482611834b29d169130971b0c7433533782f028cb4efb9806ecc89
3
+ size 15933
checkpoint-40000/rng_state_14.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:434e30bb48737271f9bd072845b0d8b765aeee7fbc6f70691db2d1a6b1b08548
3
+ size 15933
checkpoint-40000/rng_state_15.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c76070ce35687a7468823df34b4bc8a644e9effb391c0c08b518f5df45089fb4
3
+ size 15933
checkpoint-40000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9863f0cc615a31aaea751fa92490bbd26f31946700ad92d810eb0757e07d3c42
3
+ size 15920
checkpoint-40000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20b606dcfb194b299e79ab7a7264ab4fa49c18e23756cb1f031641e4862363aa
3
+ size 15920
checkpoint-40000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29b169ad1a152ba5fff751a20bd71543b3da7d3399887b766e931afdaeb7aeb1
3
+ size 15920
checkpoint-40000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:53f0fc473e41277060745323e2fe885d6fc165a6cac5679c9dc7b687188d884a
3
+ size 15920
checkpoint-40000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bbe5bc40d2cf1126acc7a38614b06fd880481ced59e214bed48223daed768082
3
+ size 15920
checkpoint-40000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:820ade75f74a789820cb72bbb44b51ba12d8209f271429b8825fea859aa25e16
3
+ size 15920
checkpoint-40000/rng_state_8.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03bdc2873d1be8aea6f46cd031a883e19ed65961e75450b60b9141130aeb532d
3
+ size 15920
checkpoint-40000/rng_state_9.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c9966cb16d9fd577a4decbc08103b5c520b2a6227356cc00ac61a4ef8f369a8
3
+ size 15920
checkpoint-40000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a6ddc8cb2af1ec91280ca0f9f919ebd0c98af2adfa8a2cbfd6da4e14464b7e1
3
+ size 1064
checkpoint-40000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-40000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:397907809f8382306eaffeb49f75d1d127f0e72eab7f6218d0992988f025e8ee
3
+ size 5816