Hugging Face's logo

xiulinyang
/
IT-GPT2small

PyTorch
gpt2
Model card Files Community
xiulinyang commited on
Commit
871a4c9
·
1 Parent(s): db2ded8

Adding model checkpoints and config files

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-0/config.json +38 -0
  2. checkpoint-0/pytorch_model.bin +3 -0
  3. checkpoint-0/special_tokens_map.json +1 -0
  4. checkpoint-0/tokenizer_config.json +1 -0
  5. checkpoint-0/training_args.bin +3 -0
  6. checkpoint-100/config.json +38 -0
  7. checkpoint-100/optimizer.pt +3 -0
  8. checkpoint-100/pytorch_model.bin +3 -0
  9. checkpoint-100/rng_state.pth +3 -0
  10. checkpoint-100/scaler.pt +3 -0
  11. checkpoint-100/scheduler.pt +3 -0
  12. checkpoint-100/special_tokens_map.json +1 -0
  13. checkpoint-100/tokenizer_config.json +1 -0
  14. checkpoint-100/trainer_state.json +34 -0
  15. checkpoint-100/training_args.bin +3 -0
  16. checkpoint-1000/config.json +38 -0
  17. checkpoint-1000/optimizer.pt +3 -0
  18. checkpoint-1000/pytorch_model.bin +3 -0
  19. checkpoint-1000/rng_state.pth +3 -0
  20. checkpoint-1000/scaler.pt +3 -0
  21. checkpoint-1000/scheduler.pt +3 -0
  22. checkpoint-1000/special_tokens_map.json +1 -0
  23. checkpoint-1000/tokenizer_config.json +1 -0
  24. checkpoint-1000/trainer_state.json +158 -0
  25. checkpoint-1000/training_args.bin +3 -0
  26. checkpoint-1100/config.json +38 -0
  27. checkpoint-1100/optimizer.pt +3 -0
  28. checkpoint-1100/pytorch_model.bin +3 -0
  29. checkpoint-1100/rng_state.pth +3 -0
  30. checkpoint-1100/scaler.pt +3 -0
  31. checkpoint-1100/scheduler.pt +3 -0
  32. checkpoint-1100/special_tokens_map.json +1 -0
  33. checkpoint-1100/tokenizer_config.json +1 -0
  34. checkpoint-1100/trainer_state.json +170 -0
  35. checkpoint-1100/training_args.bin +3 -0
  36. checkpoint-1200/config.json +38 -0
  37. checkpoint-1200/optimizer.pt +3 -0
  38. checkpoint-1200/pytorch_model.bin +3 -0
  39. checkpoint-1200/rng_state.pth +3 -0
  40. checkpoint-1200/scaler.pt +3 -0
  41. checkpoint-1200/scheduler.pt +3 -0
  42. checkpoint-1200/special_tokens_map.json +1 -0
  43. checkpoint-1200/tokenizer_config.json +1 -0
  44. checkpoint-1200/trainer_state.json +182 -0
  45. checkpoint-1200/training_args.bin +3 -0
  46. checkpoint-200/config.json +38 -0
  47. checkpoint-200/optimizer.pt +3 -0
  48. checkpoint-200/pytorch_model.bin +3 -0
  49. checkpoint-200/rng_state.pth +3 -0
  50. checkpoint-200/scaler.pt +3 -0
checkpoint-0/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-0/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44c96477af25ae287f5e32a1c813cbd7036827f82e2676555469fb8c12a1fd61
3
+ size 509616233
checkpoint-0/special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
checkpoint-0/tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"tokenizer_class": "PassthroughTokenizer"}
checkpoint-0/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12da6ea2deede4ec107c28645fefc375a84716d4fc71b25b6ed00ccbeb6d9e20
3
+ size 3183
checkpoint-100/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-100/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:355071beb02f3c4e565795692f79955892f215fb03b3d27eac40d7d4ed71ff7a
3
+ size 994043249
checkpoint-100/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:16f9c853ee33281effb85d9f0a6f42489908f310e6a86bd283f6755fac454787
3
+ size 509616233
checkpoint-100/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:068d34d67bdc1c1aac6bb85691449b1054542ae892410d252bfb25f420c330ac
3
+ size 14567
checkpoint-100/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13a3423b2fe42f204bc8fe2c666ff379f9fd753a0f13613064a5e71e86b519e8
3
+ size 559
checkpoint-100/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75e9b9d31d11c624d89b0c04ad496adf4b5addd3e703848d2583972c703e8da6
3
+ size 623
checkpoint-100/special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
checkpoint-100/tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"tokenizer_class": "PassthroughTokenizer"}
checkpoint-100/trainer_state.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0108333333333333,
5
+ "global_step": 100,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.9999999999999996e-06,
13
+ "loss": 10.853,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 1.02,
18
+ "learning_rate": 0.00025,
19
+ "loss": 7.9559,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 3.01,
24
+ "learning_rate": 0.0005,
25
+ "loss": 6.0245,
26
+ "step": 100
27
+ }
28
+ ],
29
+ "max_steps": 1200,
30
+ "num_train_epochs": 9223372036854775807,
31
+ "total_flos": 2.7439844032512e+16,
32
+ "trial_name": null,
33
+ "trial_params": null
34
+ }
checkpoint-100/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12da6ea2deede4ec107c28645fefc375a84716d4fc71b25b6ed00ccbeb6d9e20
3
+ size 3183
checkpoint-1000/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54e50b3ff99ba1ee355c4014e6602a67a1c1f57ab1f7845ed65ed7bf0c0a82d2
3
+ size 994043441
checkpoint-1000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c78bb0d361574122212359010c238b724be9600f401ee403db850024442da4e
3
+ size 509616233
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49df57de2444cdaa40f7e56e8bdceea54f8837a48418c0d5d16a5de4133e6d0e
3
+ size 14567
checkpoint-1000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f810fc7b695697c440d8985f6042b4ba23a9e1027604c265718b518ca29f1b2b
3
+ size 559
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0691206f4bd9ca409d6e7104087a4e0eb05df8f8f555a400f6ecc532edba52d8
3
+ size 623
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"tokenizer_class": "PassthroughTokenizer"}
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,158 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 34.01166666666666,
5
+ "global_step": 1000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.9999999999999996e-06,
13
+ "loss": 10.853,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 1.02,
18
+ "learning_rate": 0.00025,
19
+ "loss": 7.9559,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 3.01,
24
+ "learning_rate": 0.0005,
25
+ "loss": 6.0245,
26
+ "step": 100
27
+ },
28
+ {
29
+ "epoch": 5.0,
30
+ "learning_rate": 0.0005833333333333333,
31
+ "loss": 5.486,
32
+ "step": 150
33
+ },
34
+ {
35
+ "epoch": 6.02,
36
+ "learning_rate": 0.0005555555555555556,
37
+ "loss": 4.9323,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 8.02,
42
+ "learning_rate": 0.0005277777777777777,
43
+ "loss": 4.7897,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 10.01,
48
+ "learning_rate": 0.0005,
49
+ "loss": 4.6548,
50
+ "step": 300
51
+ },
52
+ {
53
+ "epoch": 12.0,
54
+ "learning_rate": 0.00047222222222222224,
55
+ "loss": 4.5472,
56
+ "step": 350
57
+ },
58
+ {
59
+ "epoch": 13.02,
60
+ "learning_rate": 0.00044444444444444436,
61
+ "loss": 4.3496,
62
+ "step": 400
63
+ },
64
+ {
65
+ "epoch": 15.01,
66
+ "learning_rate": 0.00041666666666666664,
67
+ "loss": 4.2665,
68
+ "step": 450
69
+ },
70
+ {
71
+ "epoch": 17.01,
72
+ "learning_rate": 0.00038888888888888887,
73
+ "loss": 4.098,
74
+ "step": 500
75
+ },
76
+ {
77
+ "epoch": 18.02,
78
+ "learning_rate": 0.0003611111111111111,
79
+ "loss": 3.8954,
80
+ "step": 550
81
+ },
82
+ {
83
+ "epoch": 20.02,
84
+ "learning_rate": 0.0003333333333333333,
85
+ "loss": 3.855,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 22.01,
90
+ "learning_rate": 0.00030555555555555555,
91
+ "loss": 3.7599,
92
+ "step": 650
93
+ },
94
+ {
95
+ "epoch": 24.0,
96
+ "learning_rate": 0.0002777777777777778,
97
+ "loss": 3.676,
98
+ "step": 700
99
+ },
100
+ {
101
+ "epoch": 25.02,
102
+ "learning_rate": 0.00025,
103
+ "loss": 3.5421,
104
+ "step": 750
105
+ },
106
+ {
107
+ "epoch": 27.01,
108
+ "learning_rate": 0.00022222222222222218,
109
+ "loss": 3.5348,
110
+ "step": 800
111
+ },
112
+ {
113
+ "epoch": 29.01,
114
+ "learning_rate": 0.00019444444444444443,
115
+ "loss": 3.4742,
116
+ "step": 850
117
+ },
118
+ {
119
+ "epoch": 31.0,
120
+ "learning_rate": 0.00016666666666666666,
121
+ "loss": 3.416,
122
+ "step": 900
123
+ },
124
+ {
125
+ "epoch": 32.02,
126
+ "learning_rate": 0.0001388888888888889,
127
+ "loss": 3.3145,
128
+ "step": 950
129
+ },
130
+ {
131
+ "epoch": 34.01,
132
+ "learning_rate": 0.00011111111111111109,
133
+ "loss": 3.3272,
134
+ "step": 1000
135
+ },
136
+ {
137
+ "epoch": 34.01,
138
+ "eval_loss": 4.220868110656738,
139
+ "eval_runtime": 2.0433,
140
+ "eval_samples_per_second": 50.898,
141
+ "eval_steps_per_second": 3.426,
142
+ "step": 1000
143
+ },
144
+ {
145
+ "epoch": 34.01,
146
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_loss": 4.220868110656738,
147
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_ppl": 68.0925705251311,
148
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_runtime": 2.0433,
149
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_samples_per_second": 50.898,
150
+ "step": 1000
151
+ }
152
+ ],
153
+ "max_steps": 1200,
154
+ "num_train_epochs": 9223372036854775807,
155
+ "total_flos": 2.75309826932736e+17,
156
+ "trial_name": null,
157
+ "trial_params": null
158
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12da6ea2deede4ec107c28645fefc375a84716d4fc71b25b6ed00ccbeb6d9e20
3
+ size 3183
checkpoint-1100/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-1100/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ceea9ef5f74f5c63bf8bd96a6f62b1a3384e0fb4dbd845269b5f0ffb2608bd82
3
+ size 994043441
checkpoint-1100/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:474b5e94bb163c6ba0de1f297b6ac63d5295590867942ea88ea2d4dfc93d1a3d
3
+ size 509616233
checkpoint-1100/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3814c2584e33b4cf278e617bbadfb38b9200160065ec44f370d5475be439f251
3
+ size 14567
checkpoint-1100/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9fb16c30b686aa43e110b0d33f9d46bf3127b7124542ca8dc34831233d4675a0
3
+ size 559
checkpoint-1100/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f1b7713e4bb40428f29080b7d08d4a52f779ac863737861e4724292b2cf6c59
3
+ size 623
checkpoint-1100/special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
checkpoint-1100/tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"tokenizer_class": "PassthroughTokenizer"}
checkpoint-1100/trainer_state.json ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 37.0225,
5
+ "global_step": 1100,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.9999999999999996e-06,
13
+ "loss": 10.853,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 1.02,
18
+ "learning_rate": 0.00025,
19
+ "loss": 7.9559,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 3.01,
24
+ "learning_rate": 0.0005,
25
+ "loss": 6.0245,
26
+ "step": 100
27
+ },
28
+ {
29
+ "epoch": 5.0,
30
+ "learning_rate": 0.0005833333333333333,
31
+ "loss": 5.486,
32
+ "step": 150
33
+ },
34
+ {
35
+ "epoch": 6.02,
36
+ "learning_rate": 0.0005555555555555556,
37
+ "loss": 4.9323,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 8.02,
42
+ "learning_rate": 0.0005277777777777777,
43
+ "loss": 4.7897,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 10.01,
48
+ "learning_rate": 0.0005,
49
+ "loss": 4.6548,
50
+ "step": 300
51
+ },
52
+ {
53
+ "epoch": 12.0,
54
+ "learning_rate": 0.00047222222222222224,
55
+ "loss": 4.5472,
56
+ "step": 350
57
+ },
58
+ {
59
+ "epoch": 13.02,
60
+ "learning_rate": 0.00044444444444444436,
61
+ "loss": 4.3496,
62
+ "step": 400
63
+ },
64
+ {
65
+ "epoch": 15.01,
66
+ "learning_rate": 0.00041666666666666664,
67
+ "loss": 4.2665,
68
+ "step": 450
69
+ },
70
+ {
71
+ "epoch": 17.01,
72
+ "learning_rate": 0.00038888888888888887,
73
+ "loss": 4.098,
74
+ "step": 500
75
+ },
76
+ {
77
+ "epoch": 18.02,
78
+ "learning_rate": 0.0003611111111111111,
79
+ "loss": 3.8954,
80
+ "step": 550
81
+ },
82
+ {
83
+ "epoch": 20.02,
84
+ "learning_rate": 0.0003333333333333333,
85
+ "loss": 3.855,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 22.01,
90
+ "learning_rate": 0.00030555555555555555,
91
+ "loss": 3.7599,
92
+ "step": 650
93
+ },
94
+ {
95
+ "epoch": 24.0,
96
+ "learning_rate": 0.0002777777777777778,
97
+ "loss": 3.676,
98
+ "step": 700
99
+ },
100
+ {
101
+ "epoch": 25.02,
102
+ "learning_rate": 0.00025,
103
+ "loss": 3.5421,
104
+ "step": 750
105
+ },
106
+ {
107
+ "epoch": 27.01,
108
+ "learning_rate": 0.00022222222222222218,
109
+ "loss": 3.5348,
110
+ "step": 800
111
+ },
112
+ {
113
+ "epoch": 29.01,
114
+ "learning_rate": 0.00019444444444444443,
115
+ "loss": 3.4742,
116
+ "step": 850
117
+ },
118
+ {
119
+ "epoch": 31.0,
120
+ "learning_rate": 0.00016666666666666666,
121
+ "loss": 3.416,
122
+ "step": 900
123
+ },
124
+ {
125
+ "epoch": 32.02,
126
+ "learning_rate": 0.0001388888888888889,
127
+ "loss": 3.3145,
128
+ "step": 950
129
+ },
130
+ {
131
+ "epoch": 34.01,
132
+ "learning_rate": 0.00011111111111111109,
133
+ "loss": 3.3272,
134
+ "step": 1000
135
+ },
136
+ {
137
+ "epoch": 34.01,
138
+ "eval_loss": 4.220868110656738,
139
+ "eval_runtime": 2.0433,
140
+ "eval_samples_per_second": 50.898,
141
+ "eval_steps_per_second": 3.426,
142
+ "step": 1000
143
+ },
144
+ {
145
+ "epoch": 34.01,
146
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_loss": 4.220868110656738,
147
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_ppl": 68.0925705251311,
148
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_runtime": 2.0433,
149
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_samples_per_second": 50.898,
150
+ "step": 1000
151
+ },
152
+ {
153
+ "epoch": 36.01,
154
+ "learning_rate": 8.333333333333333e-05,
155
+ "loss": 3.2877,
156
+ "step": 1050
157
+ },
158
+ {
159
+ "epoch": 37.02,
160
+ "learning_rate": 5.5555555555555545e-05,
161
+ "loss": 3.2001,
162
+ "step": 1100
163
+ }
164
+ ],
165
+ "max_steps": 1200,
166
+ "num_train_epochs": 9223372036854775807,
167
+ "total_flos": 3.02749670965248e+17,
168
+ "trial_name": null,
169
+ "trial_params": null
170
+ }
checkpoint-1100/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12da6ea2deede4ec107c28645fefc375a84716d4fc71b25b6ed00ccbeb6d9e20
3
+ size 3183
checkpoint-1200/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-1200/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d3ac84ed26d30d54994c36be23585ea727e9c98a6d82c553ccd436446d225fb
3
+ size 994043441
checkpoint-1200/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87dbfb54c4aec70eae93f9769bb03aad7188d09c164e9285b8325b201ac8bedb
3
+ size 509616233
checkpoint-1200/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d6b438b53f236eea2e485171b6a5fae822e286d59f6254f913d1b3a717c0c13
3
+ size 14567
checkpoint-1200/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:19c7277eaca0850ae3e9b6790b3d002d820169cce0671185e672c28c8ae8e056
3
+ size 559
checkpoint-1200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:935a8fb09a6e9698d9894853b05e181b3f56098deaaecddde08e55f06bf000c4
3
+ size 623
checkpoint-1200/special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
checkpoint-1200/tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"tokenizer_class": "PassthroughTokenizer"}
checkpoint-1200/trainer_state.json ADDED
@@ -0,0 +1,182 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 41.009166666666665,
5
+ "global_step": 1200,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.9999999999999996e-06,
13
+ "loss": 10.853,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 1.02,
18
+ "learning_rate": 0.00025,
19
+ "loss": 7.9559,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 3.01,
24
+ "learning_rate": 0.0005,
25
+ "loss": 6.0245,
26
+ "step": 100
27
+ },
28
+ {
29
+ "epoch": 5.0,
30
+ "learning_rate": 0.0005833333333333333,
31
+ "loss": 5.486,
32
+ "step": 150
33
+ },
34
+ {
35
+ "epoch": 6.02,
36
+ "learning_rate": 0.0005555555555555556,
37
+ "loss": 4.9323,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 8.02,
42
+ "learning_rate": 0.0005277777777777777,
43
+ "loss": 4.7897,
44
+ "step": 250
45
+ },
46
+ {
47
+ "epoch": 10.01,
48
+ "learning_rate": 0.0005,
49
+ "loss": 4.6548,
50
+ "step": 300
51
+ },
52
+ {
53
+ "epoch": 12.0,
54
+ "learning_rate": 0.00047222222222222224,
55
+ "loss": 4.5472,
56
+ "step": 350
57
+ },
58
+ {
59
+ "epoch": 13.02,
60
+ "learning_rate": 0.00044444444444444436,
61
+ "loss": 4.3496,
62
+ "step": 400
63
+ },
64
+ {
65
+ "epoch": 15.01,
66
+ "learning_rate": 0.00041666666666666664,
67
+ "loss": 4.2665,
68
+ "step": 450
69
+ },
70
+ {
71
+ "epoch": 17.01,
72
+ "learning_rate": 0.00038888888888888887,
73
+ "loss": 4.098,
74
+ "step": 500
75
+ },
76
+ {
77
+ "epoch": 18.02,
78
+ "learning_rate": 0.0003611111111111111,
79
+ "loss": 3.8954,
80
+ "step": 550
81
+ },
82
+ {
83
+ "epoch": 20.02,
84
+ "learning_rate": 0.0003333333333333333,
85
+ "loss": 3.855,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 22.01,
90
+ "learning_rate": 0.00030555555555555555,
91
+ "loss": 3.7599,
92
+ "step": 650
93
+ },
94
+ {
95
+ "epoch": 24.0,
96
+ "learning_rate": 0.0002777777777777778,
97
+ "loss": 3.676,
98
+ "step": 700
99
+ },
100
+ {
101
+ "epoch": 25.02,
102
+ "learning_rate": 0.00025,
103
+ "loss": 3.5421,
104
+ "step": 750
105
+ },
106
+ {
107
+ "epoch": 27.01,
108
+ "learning_rate": 0.00022222222222222218,
109
+ "loss": 3.5348,
110
+ "step": 800
111
+ },
112
+ {
113
+ "epoch": 29.01,
114
+ "learning_rate": 0.00019444444444444443,
115
+ "loss": 3.4742,
116
+ "step": 850
117
+ },
118
+ {
119
+ "epoch": 31.0,
120
+ "learning_rate": 0.00016666666666666666,
121
+ "loss": 3.416,
122
+ "step": 900
123
+ },
124
+ {
125
+ "epoch": 32.02,
126
+ "learning_rate": 0.0001388888888888889,
127
+ "loss": 3.3145,
128
+ "step": 950
129
+ },
130
+ {
131
+ "epoch": 34.01,
132
+ "learning_rate": 0.00011111111111111109,
133
+ "loss": 3.3272,
134
+ "step": 1000
135
+ },
136
+ {
137
+ "epoch": 34.01,
138
+ "eval_loss": 4.220868110656738,
139
+ "eval_runtime": 2.0433,
140
+ "eval_samples_per_second": 50.898,
141
+ "eval_steps_per_second": 3.426,
142
+ "step": 1000
143
+ },
144
+ {
145
+ "epoch": 34.01,
146
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_loss": 4.220868110656738,
147
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_ppl": 68.0925705251311,
148
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_runtime": 2.0433,
149
+ "eval_/local/xiulyang/mission-impossible-language-models/training/babylm_dataset.py_samples_per_second": 50.898,
150
+ "step": 1000
151
+ },
152
+ {
153
+ "epoch": 36.01,
154
+ "learning_rate": 8.333333333333333e-05,
155
+ "loss": 3.2877,
156
+ "step": 1050
157
+ },
158
+ {
159
+ "epoch": 37.02,
160
+ "learning_rate": 5.5555555555555545e-05,
161
+ "loss": 3.2001,
162
+ "step": 1100
163
+ },
164
+ {
165
+ "epoch": 39.02,
166
+ "learning_rate": 2.7777777777777772e-05,
167
+ "loss": 3.2282,
168
+ "step": 1150
169
+ },
170
+ {
171
+ "epoch": 41.01,
172
+ "learning_rate": 0.0,
173
+ "loss": 3.2067,
174
+ "step": 1200
175
+ }
176
+ ],
177
+ "max_steps": 1200,
178
+ "num_train_epochs": 9223372036854775807,
179
+ "total_flos": 3.30417361649664e+17,
180
+ "trial_name": null,
181
+ "trial_params": null
182
+ }
checkpoint-1200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12da6ea2deede4ec107c28645fefc375a84716d4fc71b25b6ed00ccbeb6d9e20
3
+ size 3183
checkpoint-200/config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 2,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_ctx": 1024,
14
+ "n_embd": 768,
15
+ "n_head": 12,
16
+ "n_inner": null,
17
+ "n_layer": 12,
18
+ "n_positions": 1024,
19
+ "reorder_and_upcast_attn": true,
20
+ "resid_pdrop": 0.1,
21
+ "scale_attn_by_inverse_layer_idx": true,
22
+ "scale_attn_weights": true,
23
+ "summary_activation": null,
24
+ "summary_first_dropout": 0.2,
25
+ "summary_proj_to_labels": true,
26
+ "summary_type": "cls_index",
27
+ "summary_use_proj": true,
28
+ "task_specific_params": {
29
+ "text-generation": {
30
+ "do_sample": true,
31
+ "max_length": 1024
32
+ }
33
+ },
34
+ "torch_dtype": "float32",
35
+ "transformers_version": "4.18.0",
36
+ "use_cache": false,
37
+ "vocab_size": 50003
38
+ }
checkpoint-200/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f07f6a27c36610d44f781de8905bef8f2170a7983f666eead01372732bf0d19
3
+ size 994043249
checkpoint-200/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:070142bcca059646f67e0699baf3f6552b889a910bc144bd1a51d08bce6b7e15
3
+ size 509616233
checkpoint-200/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:530aa95d70d69e3a7abffb799fab6fe0f8fba2df37e95d14fa9ecb833aeea60d
3
+ size 14567
checkpoint-200/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb6982c29cd162f49aeb531674acf574eccd46a8f556bec596040d7c3b95200a
3
+ size 559