Hugging Face's logo

zhiyang1
/
vlm_dc-vae-f32c32-sana-1.1-768_layerwise-0_group-7

Safetensors
Model card Files
xet
Community
zhiyang1 commited on
Commit
3582052
·
verified ·
1 Parent(s): 12930cc

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. checkpoint-10000/config.json +80 -0
  2. checkpoint-10000/generation_config.json +6 -0
  3. checkpoint-10000/model-00001-of-00002.safetensors +3 -0
  4. checkpoint-10000/model-00002-of-00002.safetensors +3 -0
  5. checkpoint-10000/model.safetensors.index.json +0 -0
  6. checkpoint-10000/optimizer.pt +3 -0
  7. checkpoint-10000/rng_state_0.pth +3 -0
  8. checkpoint-10000/rng_state_1.pth +3 -0
  9. checkpoint-10000/rng_state_2.pth +3 -0
  10. checkpoint-10000/rng_state_3.pth +3 -0
  11. checkpoint-10000/rng_state_4.pth +3 -0
  12. checkpoint-10000/rng_state_5.pth +3 -0
  13. checkpoint-10000/rng_state_6.pth +3 -0
  14. checkpoint-10000/rng_state_7.pth +3 -0
  15. checkpoint-10000/scheduler.pt +3 -0
  16. checkpoint-10000/trainer_state.json +0 -0
  17. checkpoint-10000/training_args.bin +3 -0
  18. checkpoint-9250/config.json +80 -0
  19. checkpoint-9250/generation_config.json +6 -0
  20. checkpoint-9250/model-00001-of-00002.safetensors +3 -0
  21. checkpoint-9250/model-00002-of-00002.safetensors +3 -0
  22. checkpoint-9250/model.safetensors.index.json +0 -0
  23. checkpoint-9250/optimizer.pt +3 -0
  24. checkpoint-9250/rng_state_0.pth +3 -0
  25. checkpoint-9250/rng_state_1.pth +3 -0
  26. checkpoint-9250/rng_state_2.pth +3 -0
  27. checkpoint-9250/rng_state_3.pth +3 -0
  28. checkpoint-9250/rng_state_4.pth +3 -0
  29. checkpoint-9250/rng_state_5.pth +3 -0
  30. checkpoint-9250/rng_state_6.pth +3 -0
  31. checkpoint-9250/rng_state_7.pth +3 -0
  32. checkpoint-9250/scheduler.pt +3 -0
  33. checkpoint-9250/trainer_state.json +0 -0
  34. checkpoint-9250/training_args.bin +3 -0
  35. checkpoint-9500/config.json +80 -0
  36. checkpoint-9500/generation_config.json +6 -0
  37. checkpoint-9500/model-00001-of-00002.safetensors +3 -0
  38. checkpoint-9500/model-00002-of-00002.safetensors +3 -0
  39. checkpoint-9500/model.safetensors.index.json +0 -0
  40. checkpoint-9500/optimizer.pt +3 -0
  41. checkpoint-9500/rng_state_0.pth +3 -0
  42. checkpoint-9500/rng_state_1.pth +3 -0
  43. checkpoint-9500/rng_state_2.pth +3 -0
  44. checkpoint-9500/rng_state_3.pth +3 -0
  45. checkpoint-9500/rng_state_4.pth +3 -0
  46. checkpoint-9500/rng_state_5.pth +3 -0
  47. checkpoint-9500/rng_state_6.pth +3 -0
  48. checkpoint-9500/rng_state_7.pth +3 -0
  49. checkpoint-9500/scheduler.pt +3 -0
  50. checkpoint-9500/trainer_state.json +0 -0
checkpoint-10000/config.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "checkpoints/vlm_dc-vae-f32c32-sana-1.1_layerwise-0_group-7/checkpoint-9250",
3
+ "ar_steps": 1,
4
+ "architectures": [
5
+ "DiffVLMDiffusion"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 151643,
9
+ "condition_layer": -1,
10
+ "eos_token_id": 151645,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1536,
13
+ "image_token_id": 151655,
14
+ "img_cross_attention_dim": 2048,
15
+ "img_diffuser_depth": 6,
16
+ "img_ffn_dim_multiplier": null,
17
+ "img_hidden_size": 1536,
18
+ "img_multiple_of": 256,
19
+ "img_norm_eps": 1e-05,
20
+ "img_num_attention_heads": 12,
21
+ "img_num_kv_heads": 12,
22
+ "img_qk_norm": true,
23
+ "in_channels": 32,
24
+ "initializer_range": 0.02,
25
+ "inject_img_diffuser": false,
26
+ "input_size": 32,
27
+ "intermediate_size": 8960,
28
+ "layer_group_size": 7,
29
+ "layerwise_start_idx": 0,
30
+ "lora_alpha": 256,
31
+ "lora_bias": "none",
32
+ "lora_dropout": 0.05,
33
+ "lora_enable": false,
34
+ "lora_r": 128,
35
+ "max_position_embeddings": 32768,
36
+ "max_window_layers": 28,
37
+ "model_type": "qwen2_vl",
38
+ "non_linearity": 1,
39
+ "norm_elementwise_affine": true,
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 28,
42
+ "num_key_value_heads": 2,
43
+ "patch_size": 2,
44
+ "repa_coeff": 0.1,
45
+ "repa_layers": "2",
46
+ "repa_shared": false,
47
+ "rms_norm_eps": 1e-06,
48
+ "rope_scaling": {
49
+ "mrope_section": [
50
+ 16,
51
+ 24,
52
+ 24
53
+ ],
54
+ "rope_type": "default",
55
+ "type": "default"
56
+ },
57
+ "rope_theta": 1000000.0,
58
+ "sample_size": 128,
59
+ "sampling_steps": 28,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "torch_dtype": "float32",
63
+ "transformers_version": "4.47.0",
64
+ "use_cache": true,
65
+ "use_repa": false,
66
+ "use_residual_attn": false,
67
+ "use_sliding_window": false,
68
+ "vae_path": "mit-han-lab/dc-ae-f32c32-sana-1.1-diffusers",
69
+ "video_token_id": 151656,
70
+ "vision_config": {
71
+ "hidden_size": 1536,
72
+ "in_chans": 3,
73
+ "model_type": "qwen2_vl",
74
+ "spatial_patch_size": 14
75
+ },
76
+ "vision_end_token_id": 151653,
77
+ "vision_start_token_id": 151652,
78
+ "vision_token_id": 151654,
79
+ "vocab_size": 151936
80
+ }
checkpoint-10000/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-10000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89f21f90bc523b73cdf3a7ba5a0d46ba708b881833462021e9016920499b54ec
3
+ size 4998598816
checkpoint-10000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e8afd357e3018d21f3ff9af1e218a6a120787a468edfab59d679e9d96a5c555
3
+ size 4990560652
checkpoint-10000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18f8e9a11b8d0f6bec1300daed7ebdb91852500bf45cc1150d035fea35a9bfc3
3
+ size 15084326534
checkpoint-10000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:356296cf368727b0f14f082cd6d973d3746f68721e1ff72d34851f878f705455
3
+ size 15984
checkpoint-10000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6507554a7218dc304b1d5b77aa78afd492b4ce9f82845bd0c14acdc898133fbf
3
+ size 15984
checkpoint-10000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa9622fc69e28136febe7fc4bda8096f9be9ef00f6fc3ed53e550c7b9bf24e75
3
+ size 15984
checkpoint-10000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e6256cbebb72b61aec386e2001c99eca50cd64e3caabed87abbaecab473bfe3
3
+ size 15984
checkpoint-10000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3b910495639901762f951a102243becc7b7c22c9ecf9ac605b3dbaad22daf19
3
+ size 15984
checkpoint-10000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1e54c58e07ae08f4525f416fcf26574303a9933be1edf113de699e25b9e3f03
3
+ size 15984
checkpoint-10000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cc49d27e834b3f570286bef728f6f2fd05bb41e813fe9f2c8cd0474eb34e174
3
+ size 15984
checkpoint-10000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f7c069e906f94302cd351091969433b01cefe19edc1587235be42f669aa2103
3
+ size 15984
checkpoint-10000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8cecba43d7459ec5ce9f816c2e7f8c9a6c0974f0e469d7bde6bc33484cab04b5
3
+ size 1064
checkpoint-10000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06816c37733f99d23f044cefd981b2f404a72ddf40fa59f794154596b842fa95
3
+ size 6072
checkpoint-9250/config.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "checkpoints/vlm_dc-vae-f32c32-sana-1.1_layerwise-0_group-7/checkpoint-9250",
3
+ "ar_steps": 1,
4
+ "architectures": [
5
+ "DiffVLMDiffusion"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 151643,
9
+ "condition_layer": -1,
10
+ "eos_token_id": 151645,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1536,
13
+ "image_token_id": 151655,
14
+ "img_cross_attention_dim": 2048,
15
+ "img_diffuser_depth": 6,
16
+ "img_ffn_dim_multiplier": null,
17
+ "img_hidden_size": 1536,
18
+ "img_multiple_of": 256,
19
+ "img_norm_eps": 1e-05,
20
+ "img_num_attention_heads": 12,
21
+ "img_num_kv_heads": 12,
22
+ "img_qk_norm": true,
23
+ "in_channels": 32,
24
+ "initializer_range": 0.02,
25
+ "inject_img_diffuser": false,
26
+ "input_size": 32,
27
+ "intermediate_size": 8960,
28
+ "layer_group_size": 7,
29
+ "layerwise_start_idx": 0,
30
+ "lora_alpha": 256,
31
+ "lora_bias": "none",
32
+ "lora_dropout": 0.05,
33
+ "lora_enable": false,
34
+ "lora_r": 128,
35
+ "max_position_embeddings": 32768,
36
+ "max_window_layers": 28,
37
+ "model_type": "qwen2_vl",
38
+ "non_linearity": 1,
39
+ "norm_elementwise_affine": true,
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 28,
42
+ "num_key_value_heads": 2,
43
+ "patch_size": 2,
44
+ "repa_coeff": 0.1,
45
+ "repa_layers": "2",
46
+ "repa_shared": false,
47
+ "rms_norm_eps": 1e-06,
48
+ "rope_scaling": {
49
+ "mrope_section": [
50
+ 16,
51
+ 24,
52
+ 24
53
+ ],
54
+ "rope_type": "default",
55
+ "type": "default"
56
+ },
57
+ "rope_theta": 1000000.0,
58
+ "sample_size": 128,
59
+ "sampling_steps": 28,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "torch_dtype": "float32",
63
+ "transformers_version": "4.47.0",
64
+ "use_cache": true,
65
+ "use_repa": false,
66
+ "use_residual_attn": false,
67
+ "use_sliding_window": false,
68
+ "vae_path": "mit-han-lab/dc-ae-f32c32-sana-1.1-diffusers",
69
+ "video_token_id": 151656,
70
+ "vision_config": {
71
+ "hidden_size": 1536,
72
+ "in_chans": 3,
73
+ "model_type": "qwen2_vl",
74
+ "spatial_patch_size": 14
75
+ },
76
+ "vision_end_token_id": 151653,
77
+ "vision_start_token_id": 151652,
78
+ "vision_token_id": 151654,
79
+ "vocab_size": 151936
80
+ }
checkpoint-9250/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-9250/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6d1cbeeefbc8b3d0c08c879cd937dd13a38dffc753d81d8733af02c7f8f260c
3
+ size 4998598816
checkpoint-9250/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39517e88777c7799bf502fcb9668738b511e64ad8fcb899c8472c030637fa531
3
+ size 4990560652
checkpoint-9250/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-9250/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14855296a297fc3cb819a8f1e26fc414f9a01a3d78619c1aaed7173dd7afc238
3
+ size 15084326534
checkpoint-9250/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f15e5ecf2215416cccfb5acf8a5866062dda90abc7d426d4092d9a6ea5701def
3
+ size 15984
checkpoint-9250/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5109a82f4afe291929d6f2f266869e9bc00730c2c499bb040e4ab93e1d5619b7
3
+ size 15984
checkpoint-9250/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38c8fe603eebd1ec48f2340fdffa828296306012f9ae455b491957b35c0e0a5e
3
+ size 15984
checkpoint-9250/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c38efab8f4a1112161f5f95bbefe72f91e2090ed864dcb16b055f0492b4a164
3
+ size 15984
checkpoint-9250/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d831da0f86a0ccb4250a89e2e4cfce06afc7e18e80371373af6370805271d56
3
+ size 15984
checkpoint-9250/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35c020fc505e1a396c96c70e74cebae3796c8f3779874b3643dbc2af655bd5ef
3
+ size 15984
checkpoint-9250/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d90b2d8bff52e46d645fb76cb28243b3ad5922bb8e85430722888010afe993ac
3
+ size 15984
checkpoint-9250/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4444f1ac90b52bff550eeee9bbf3ee70f9daa59e5b6a608866b1b1c58aa2fe09
3
+ size 15984
checkpoint-9250/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88560f892e51e219f2e49d8136a9bbce7f33b04ca2ed8085600760b3201dbd9c
3
+ size 1064
checkpoint-9250/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-9250/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06816c37733f99d23f044cefd981b2f404a72ddf40fa59f794154596b842fa95
3
+ size 6072
checkpoint-9500/config.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "checkpoints/vlm_dc-vae-f32c32-sana-1.1_layerwise-0_group-7/checkpoint-9250",
3
+ "ar_steps": 1,
4
+ "architectures": [
5
+ "DiffVLMDiffusion"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 151643,
9
+ "condition_layer": -1,
10
+ "eos_token_id": 151645,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1536,
13
+ "image_token_id": 151655,
14
+ "img_cross_attention_dim": 2048,
15
+ "img_diffuser_depth": 6,
16
+ "img_ffn_dim_multiplier": null,
17
+ "img_hidden_size": 1536,
18
+ "img_multiple_of": 256,
19
+ "img_norm_eps": 1e-05,
20
+ "img_num_attention_heads": 12,
21
+ "img_num_kv_heads": 12,
22
+ "img_qk_norm": true,
23
+ "in_channels": 32,
24
+ "initializer_range": 0.02,
25
+ "inject_img_diffuser": false,
26
+ "input_size": 32,
27
+ "intermediate_size": 8960,
28
+ "layer_group_size": 7,
29
+ "layerwise_start_idx": 0,
30
+ "lora_alpha": 256,
31
+ "lora_bias": "none",
32
+ "lora_dropout": 0.05,
33
+ "lora_enable": false,
34
+ "lora_r": 128,
35
+ "max_position_embeddings": 32768,
36
+ "max_window_layers": 28,
37
+ "model_type": "qwen2_vl",
38
+ "non_linearity": 1,
39
+ "norm_elementwise_affine": true,
40
+ "num_attention_heads": 12,
41
+ "num_hidden_layers": 28,
42
+ "num_key_value_heads": 2,
43
+ "patch_size": 2,
44
+ "repa_coeff": 0.1,
45
+ "repa_layers": "2",
46
+ "repa_shared": false,
47
+ "rms_norm_eps": 1e-06,
48
+ "rope_scaling": {
49
+ "mrope_section": [
50
+ 16,
51
+ 24,
52
+ 24
53
+ ],
54
+ "rope_type": "default",
55
+ "type": "default"
56
+ },
57
+ "rope_theta": 1000000.0,
58
+ "sample_size": 128,
59
+ "sampling_steps": 28,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "torch_dtype": "float32",
63
+ "transformers_version": "4.47.0",
64
+ "use_cache": true,
65
+ "use_repa": false,
66
+ "use_residual_attn": false,
67
+ "use_sliding_window": false,
68
+ "vae_path": "mit-han-lab/dc-ae-f32c32-sana-1.1-diffusers",
69
+ "video_token_id": 151656,
70
+ "vision_config": {
71
+ "hidden_size": 1536,
72
+ "in_chans": 3,
73
+ "model_type": "qwen2_vl",
74
+ "spatial_patch_size": 14
75
+ },
76
+ "vision_end_token_id": 151653,
77
+ "vision_start_token_id": 151652,
78
+ "vision_token_id": 151654,
79
+ "vocab_size": 151936
80
+ }
checkpoint-9500/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-9500/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:027d78d8ebd5a248c07aa08795be021166ca35cae006930ce7a90250222ae329
3
+ size 4998598816
checkpoint-9500/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad3d6cc77aa0d87ef09990eae4a583cf5992972c92ba1585efe3005a8f84fb25
3
+ size 4990560652
checkpoint-9500/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-9500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffcfe91e3666c374656bf917b8ad4b7ac0da6db13c73d8227f967cead0379e24
3
+ size 15084326534
checkpoint-9500/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4beb399a9aad04e710e18e8622267c919c8a5c94153adef8d6d164aad7f35fe4
3
+ size 15984
checkpoint-9500/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:324837057ced6db286a3516b2a9602e89768f490e46b80d1c76a33b909dcb7da
3
+ size 15984
checkpoint-9500/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fc68175f400c4dea4fa448867394a3058a01ffa73f7484cf5878e8566d8ffbc
3
+ size 15984
checkpoint-9500/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89c00677737944df40c8ff65a78c8683ee5840398d33b26331b8ffae34329a66
3
+ size 15984
checkpoint-9500/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4334aeb339331630c029df65f02b5f93c703add9ea0d60e353d6019e12f0e82
3
+ size 15984
checkpoint-9500/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9485699ba1f8731cde5cfd77a6ce95b8f8317a85bb09a3ba2489b4067501492d
3
+ size 15984
checkpoint-9500/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe104ee345b7463caaac129168eb06b2b53a8623037bf23c4cd2aa8de52ca6f0
3
+ size 15984
checkpoint-9500/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30ec7c0eb2cf7359b7a47c2eb263f7cc67dc1fc25de0de1b4a6e5b7b679cac4c
3
+ size 15984
checkpoint-9500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:24954beb59590793d5fc0ddbe7d1a6127ce1ae2ee15502114a025e447199d147
3
+ size 1064
checkpoint-9500/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff