rootxhacker commited on
Commit
f527b7a
·
verified ·
1 Parent(s): 0e140f0

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-1000/config.json +30 -0
  2. checkpoint-1000/generation_config.json +7 -0
  3. checkpoint-1000/model.safetensors +3 -0
  4. checkpoint-1000/optimizer.pt +3 -0
  5. checkpoint-1000/rng_state.pth +3 -0
  6. checkpoint-1000/scheduler.pt +3 -0
  7. checkpoint-1000/special_tokens_map.json +24 -0
  8. checkpoint-1000/tokenizer.json +0 -0
  9. checkpoint-1000/tokenizer.model +3 -0
  10. checkpoint-1000/tokenizer_config.json +42 -0
  11. checkpoint-1000/trainer_state.json +934 -0
  12. checkpoint-1000/training_args.bin +3 -0
  13. checkpoint-1500/config.json +30 -0
  14. checkpoint-1500/generation_config.json +7 -0
  15. checkpoint-1500/model.safetensors +3 -0
  16. checkpoint-1500/optimizer.pt +3 -0
  17. checkpoint-1500/rng_state.pth +3 -0
  18. checkpoint-1500/scheduler.pt +3 -0
  19. checkpoint-1500/special_tokens_map.json +24 -0
  20. checkpoint-1500/tokenizer.json +0 -0
  21. checkpoint-1500/tokenizer.model +3 -0
  22. checkpoint-1500/tokenizer_config.json +42 -0
  23. checkpoint-1500/trainer_state.json +1384 -0
  24. checkpoint-1500/training_args.bin +3 -0
  25. checkpoint-2000/config.json +30 -0
  26. checkpoint-2000/generation_config.json +7 -0
  27. checkpoint-2000/model.safetensors +3 -0
  28. checkpoint-2000/optimizer.pt +3 -0
  29. checkpoint-2000/rng_state.pth +3 -0
  30. checkpoint-2000/scheduler.pt +3 -0
  31. checkpoint-2000/special_tokens_map.json +24 -0
  32. checkpoint-2000/tokenizer.json +0 -0
  33. checkpoint-2000/tokenizer.model +3 -0
  34. checkpoint-2000/tokenizer_config.json +42 -0
  35. checkpoint-2000/trainer_state.json +1834 -0
  36. checkpoint-2000/training_args.bin +3 -0
  37. checkpoint-2500/config.json +30 -0
  38. checkpoint-2500/generation_config.json +7 -0
  39. checkpoint-2500/model.safetensors +3 -0
  40. checkpoint-2500/optimizer.pt +3 -0
  41. checkpoint-2500/rng_state.pth +3 -0
  42. checkpoint-2500/scheduler.pt +3 -0
  43. checkpoint-2500/special_tokens_map.json +24 -0
  44. checkpoint-2500/tokenizer.json +0 -0
  45. checkpoint-2500/tokenizer.model +3 -0
  46. checkpoint-2500/tokenizer_config.json +42 -0
  47. checkpoint-2500/trainer_state.json +2284 -0
  48. checkpoint-2500/training_args.bin +3 -0
  49. checkpoint-3000/config.json +30 -0
  50. checkpoint-3000/generation_config.json +7 -0
checkpoint-1000/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 256,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f7395b37e375386f7e020f0feb5871f091865a01cf21af070229210bc865b70
3
+ size 309900448
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3285b9116f849885b7b430d8e5b630a3fa6a4a4197087805d064eac02d1daa9f
3
+ size 619836730
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeecd634c6d2fbef24b299add90a57b4006935a51eedba880504e0df7f872f59
3
+ size 1064
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-1000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,934 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.4912194522903107,
6
+ "eval_steps": 500,
7
+ "global_step": 1000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.004912194522903107,
14
+ "grad_norm": 2.5265462398529053,
15
+ "learning_rate": 1.995577395577396e-05,
16
+ "loss": 5.7783,
17
+ "mean_token_accuracy": 0.17638254687190055,
18
+ "num_tokens": 50745.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.009824389045806214,
23
+ "grad_norm": 1.7381863594055176,
24
+ "learning_rate": 1.9906633906633906e-05,
25
+ "loss": 5.4596,
26
+ "mean_token_accuracy": 0.20633984543383121,
27
+ "num_tokens": 101681.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.01473658356870932,
32
+ "grad_norm": 2.0286130905151367,
33
+ "learning_rate": 1.985749385749386e-05,
34
+ "loss": 5.2069,
35
+ "mean_token_accuracy": 0.2354225989431143,
36
+ "num_tokens": 153620.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.019648778091612428,
41
+ "grad_norm": 2.2828938961029053,
42
+ "learning_rate": 1.980835380835381e-05,
43
+ "loss": 5.0637,
44
+ "mean_token_accuracy": 0.2486753471195698,
45
+ "num_tokens": 205428.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.024560972614515536,
50
+ "grad_norm": 2.247742176055908,
51
+ "learning_rate": 1.975921375921376e-05,
52
+ "loss": 4.7071,
53
+ "mean_token_accuracy": 0.2982919864356518,
54
+ "num_tokens": 260794.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.02947316713741864,
59
+ "grad_norm": 1.4552102088928223,
60
+ "learning_rate": 1.9710073710073712e-05,
61
+ "loss": 4.5951,
62
+ "mean_token_accuracy": 0.3140149611979723,
63
+ "num_tokens": 314216.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.03438536166032175,
68
+ "grad_norm": 2.032104969024658,
69
+ "learning_rate": 1.9660933660933663e-05,
70
+ "loss": 4.3963,
71
+ "mean_token_accuracy": 0.33228743858635423,
72
+ "num_tokens": 365608.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.039297556183224856,
77
+ "grad_norm": 1.5947270393371582,
78
+ "learning_rate": 1.9611793611793614e-05,
79
+ "loss": 4.2578,
80
+ "mean_token_accuracy": 0.3510261565446854,
81
+ "num_tokens": 418509.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.04420975070612796,
86
+ "grad_norm": 1.9174288511276245,
87
+ "learning_rate": 1.9562653562653564e-05,
88
+ "loss": 4.0457,
89
+ "mean_token_accuracy": 0.38185270577669145,
90
+ "num_tokens": 472608.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.04912194522903107,
95
+ "grad_norm": 1.5742830038070679,
96
+ "learning_rate": 1.9513513513513515e-05,
97
+ "loss": 4.0492,
98
+ "mean_token_accuracy": 0.37742224968969823,
99
+ "num_tokens": 521777.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.054034139751934177,
104
+ "grad_norm": 1.5624850988388062,
105
+ "learning_rate": 1.9464373464373466e-05,
106
+ "loss": 3.7053,
107
+ "mean_token_accuracy": 0.41850126534700394,
108
+ "num_tokens": 580161.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.05894633427483728,
113
+ "grad_norm": 2.199150562286377,
114
+ "learning_rate": 1.9415233415233417e-05,
115
+ "loss": 3.6982,
116
+ "mean_token_accuracy": 0.4235368125140667,
117
+ "num_tokens": 627843.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.06385852879774039,
122
+ "grad_norm": 1.8548036813735962,
123
+ "learning_rate": 1.9366093366093367e-05,
124
+ "loss": 3.3414,
125
+ "mean_token_accuracy": 0.4654690891504288,
126
+ "num_tokens": 681443.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.0687707233206435,
131
+ "grad_norm": 2.78169846534729,
132
+ "learning_rate": 1.9316953316953318e-05,
133
+ "loss": 3.2679,
134
+ "mean_token_accuracy": 0.4711888641119003,
135
+ "num_tokens": 733457.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.07368291784354661,
140
+ "grad_norm": 2.4450321197509766,
141
+ "learning_rate": 1.926781326781327e-05,
142
+ "loss": 3.3377,
143
+ "mean_token_accuracy": 0.4799527756869793,
144
+ "num_tokens": 787769.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.07859511236644971,
149
+ "grad_norm": 1.7717996835708618,
150
+ "learning_rate": 1.921867321867322e-05,
151
+ "loss": 2.8853,
152
+ "mean_token_accuracy": 0.541524414718151,
153
+ "num_tokens": 843485.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.08350730688935282,
158
+ "grad_norm": 2.8402254581451416,
159
+ "learning_rate": 1.9169533169533174e-05,
160
+ "loss": 2.862,
161
+ "mean_token_accuracy": 0.5626954190433026,
162
+ "num_tokens": 894003.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.08841950141225592,
167
+ "grad_norm": 1.5486701726913452,
168
+ "learning_rate": 1.912039312039312e-05,
169
+ "loss": 2.7533,
170
+ "mean_token_accuracy": 0.5848119542002678,
171
+ "num_tokens": 949726.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.09333169593515903,
176
+ "grad_norm": 1.3785852193832397,
177
+ "learning_rate": 1.9071253071253072e-05,
178
+ "loss": 2.7602,
179
+ "mean_token_accuracy": 0.5894295200705528,
180
+ "num_tokens": 1000146.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 0.09824389045806214,
185
+ "grad_norm": 1.6824640035629272,
186
+ "learning_rate": 1.9022113022113026e-05,
187
+ "loss": 2.6345,
188
+ "mean_token_accuracy": 0.5961507339030504,
189
+ "num_tokens": 1050142.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 0.10315608498096525,
194
+ "grad_norm": 1.0604822635650635,
195
+ "learning_rate": 1.8972972972972973e-05,
196
+ "loss": 2.4517,
197
+ "mean_token_accuracy": 0.6296438351273537,
198
+ "num_tokens": 1099646.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 0.10806827950386835,
203
+ "grad_norm": 1.9105230569839478,
204
+ "learning_rate": 1.8923832923832924e-05,
205
+ "loss": 2.4563,
206
+ "mean_token_accuracy": 0.6365545220673084,
207
+ "num_tokens": 1148100.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 0.11298047402677146,
212
+ "grad_norm": 1.41056227684021,
213
+ "learning_rate": 1.8874692874692878e-05,
214
+ "loss": 2.4017,
215
+ "mean_token_accuracy": 0.6691348008811474,
216
+ "num_tokens": 1198987.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.11789266854967456,
221
+ "grad_norm": 1.0573596954345703,
222
+ "learning_rate": 1.8825552825552825e-05,
223
+ "loss": 2.3199,
224
+ "mean_token_accuracy": 0.6731609679758549,
225
+ "num_tokens": 1249335.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 0.12280486307257768,
230
+ "grad_norm": 1.6629279851913452,
231
+ "learning_rate": 1.877641277641278e-05,
232
+ "loss": 1.9335,
233
+ "mean_token_accuracy": 0.7327558636665344,
234
+ "num_tokens": 1303157.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 0.12771705759548077,
239
+ "grad_norm": 1.704005479812622,
240
+ "learning_rate": 1.872727272727273e-05,
241
+ "loss": 2.1897,
242
+ "mean_token_accuracy": 0.6974137671291828,
243
+ "num_tokens": 1355455.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 0.1326292521183839,
248
+ "grad_norm": 1.0553226470947266,
249
+ "learning_rate": 1.8678132678132678e-05,
250
+ "loss": 1.9321,
251
+ "mean_token_accuracy": 0.7327657997608185,
252
+ "num_tokens": 1407508.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 0.137541446641287,
257
+ "grad_norm": 1.101886510848999,
258
+ "learning_rate": 1.8628992628992632e-05,
259
+ "loss": 1.9929,
260
+ "mean_token_accuracy": 0.7272698871791363,
261
+ "num_tokens": 1460595.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 0.1424536411641901,
266
+ "grad_norm": 2.005901575088501,
267
+ "learning_rate": 1.8579852579852583e-05,
268
+ "loss": 1.7147,
269
+ "mean_token_accuracy": 0.7604942478239536,
270
+ "num_tokens": 1512647.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 0.14736583568709322,
275
+ "grad_norm": 7.757036209106445,
276
+ "learning_rate": 1.853071253071253e-05,
277
+ "loss": 1.7571,
278
+ "mean_token_accuracy": 0.7603825606405735,
279
+ "num_tokens": 1565516.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 0.1522780302099963,
284
+ "grad_norm": 0.8850768208503723,
285
+ "learning_rate": 1.8481572481572484e-05,
286
+ "loss": 1.5439,
287
+ "mean_token_accuracy": 0.7915346205234528,
288
+ "num_tokens": 1620194.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 0.15719022473289943,
293
+ "grad_norm": 1.2887756824493408,
294
+ "learning_rate": 1.8432432432432435e-05,
295
+ "loss": 1.5419,
296
+ "mean_token_accuracy": 0.7885693073272705,
297
+ "num_tokens": 1672317.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 0.16210241925580254,
302
+ "grad_norm": 2.562720537185669,
303
+ "learning_rate": 1.8383292383292385e-05,
304
+ "loss": 1.5783,
305
+ "mean_token_accuracy": 0.7889993876218796,
306
+ "num_tokens": 1728513.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 0.16701461377870563,
311
+ "grad_norm": 2.907665491104126,
312
+ "learning_rate": 1.8334152334152336e-05,
313
+ "loss": 1.6573,
314
+ "mean_token_accuracy": 0.7675686649978161,
315
+ "num_tokens": 1777759.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 0.17192680830160875,
320
+ "grad_norm": 0.8430731892585754,
321
+ "learning_rate": 1.8285012285012287e-05,
322
+ "loss": 1.664,
323
+ "mean_token_accuracy": 0.7618721649050713,
324
+ "num_tokens": 1833279.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 0.17683900282451184,
329
+ "grad_norm": 1.1680371761322021,
330
+ "learning_rate": 1.8235872235872238e-05,
331
+ "loss": 1.515,
332
+ "mean_token_accuracy": 0.7789990782737732,
333
+ "num_tokens": 1882531.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 0.18175119734741496,
338
+ "grad_norm": 2.009002208709717,
339
+ "learning_rate": 1.818673218673219e-05,
340
+ "loss": 1.4555,
341
+ "mean_token_accuracy": 0.7983467862010002,
342
+ "num_tokens": 1934619.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 0.18666339187031805,
347
+ "grad_norm": 1.3894531726837158,
348
+ "learning_rate": 1.813759213759214e-05,
349
+ "loss": 1.6206,
350
+ "mean_token_accuracy": 0.7681202955543995,
351
+ "num_tokens": 1987231.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 0.19157558639322117,
356
+ "grad_norm": 0.7266082763671875,
357
+ "learning_rate": 1.808845208845209e-05,
358
+ "loss": 1.3022,
359
+ "mean_token_accuracy": 0.818102091550827,
360
+ "num_tokens": 2039985.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 0.1964877809161243,
365
+ "grad_norm": 0.7502278089523315,
366
+ "learning_rate": 1.803931203931204e-05,
367
+ "loss": 1.3902,
368
+ "mean_token_accuracy": 0.792433574795723,
369
+ "num_tokens": 2092178.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 0.20139997543902738,
374
+ "grad_norm": 0.8366674780845642,
375
+ "learning_rate": 1.799017199017199e-05,
376
+ "loss": 1.5385,
377
+ "mean_token_accuracy": 0.7903823673725128,
378
+ "num_tokens": 2151243.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 0.2063121699619305,
383
+ "grad_norm": 0.7726922035217285,
384
+ "learning_rate": 1.7941031941031942e-05,
385
+ "loss": 1.3529,
386
+ "mean_token_accuracy": 0.8017133295536041,
387
+ "num_tokens": 2201072.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 0.2112243644848336,
392
+ "grad_norm": 0.650055468082428,
393
+ "learning_rate": 1.7891891891891893e-05,
394
+ "loss": 1.4312,
395
+ "mean_token_accuracy": 0.7957115054130555,
396
+ "num_tokens": 2256591.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 0.2161365590077367,
401
+ "grad_norm": 0.9430303573608398,
402
+ "learning_rate": 1.7842751842751843e-05,
403
+ "loss": 1.1938,
404
+ "mean_token_accuracy": 0.8211165018379688,
405
+ "num_tokens": 2308048.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 0.22104875353063982,
410
+ "grad_norm": 0.9191327691078186,
411
+ "learning_rate": 1.7793611793611794e-05,
412
+ "loss": 1.1662,
413
+ "mean_token_accuracy": 0.8298586897552014,
414
+ "num_tokens": 2363573.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 0.22596094805354291,
419
+ "grad_norm": 0.7493109703063965,
420
+ "learning_rate": 1.7744471744471745e-05,
421
+ "loss": 1.5908,
422
+ "mean_token_accuracy": 0.7651749603450299,
423
+ "num_tokens": 2413187.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 0.23087314257644603,
428
+ "grad_norm": 0.5522422790527344,
429
+ "learning_rate": 1.7695331695331696e-05,
430
+ "loss": 1.3707,
431
+ "mean_token_accuracy": 0.7843399420380592,
432
+ "num_tokens": 2468041.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 0.23578533709934912,
437
+ "grad_norm": 0.6577270030975342,
438
+ "learning_rate": 1.764619164619165e-05,
439
+ "loss": 1.4679,
440
+ "mean_token_accuracy": 0.7856365032494068,
441
+ "num_tokens": 2517753.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 0.24069753162225224,
446
+ "grad_norm": 0.9699747562408447,
447
+ "learning_rate": 1.7597051597051597e-05,
448
+ "loss": 1.3493,
449
+ "mean_token_accuracy": 0.8016418114304542,
450
+ "num_tokens": 2568072.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 0.24560972614515536,
455
+ "grad_norm": 2.0955657958984375,
456
+ "learning_rate": 1.7547911547911548e-05,
457
+ "loss": 0.9634,
458
+ "mean_token_accuracy": 0.8412673205137253,
459
+ "num_tokens": 2617143.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 0.25052192066805845,
464
+ "grad_norm": 0.7577691078186035,
465
+ "learning_rate": 1.7498771498771502e-05,
466
+ "loss": 1.0889,
467
+ "mean_token_accuracy": 0.8479016363620758,
468
+ "num_tokens": 2669318.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 0.25543411519096154,
473
+ "grad_norm": 0.5891643762588501,
474
+ "learning_rate": 1.744963144963145e-05,
475
+ "loss": 1.1811,
476
+ "mean_token_accuracy": 0.82891755849123,
477
+ "num_tokens": 2723151.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 0.2603463097138647,
482
+ "grad_norm": 0.7815893292427063,
483
+ "learning_rate": 1.74004914004914e-05,
484
+ "loss": 1.3488,
485
+ "mean_token_accuracy": 0.7971963949501515,
486
+ "num_tokens": 2774274.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 0.2652585042367678,
491
+ "grad_norm": 0.365824818611145,
492
+ "learning_rate": 1.7351351351351354e-05,
493
+ "loss": 1.129,
494
+ "mean_token_accuracy": 0.8214888103306294,
495
+ "num_tokens": 2825301.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 0.27017069875967087,
500
+ "grad_norm": 0.5682836174964905,
501
+ "learning_rate": 1.73022113022113e-05,
502
+ "loss": 1.0752,
503
+ "mean_token_accuracy": 0.8513655051589012,
504
+ "num_tokens": 2877767.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 0.275082893282574,
509
+ "grad_norm": 0.9484438300132751,
510
+ "learning_rate": 1.7253071253071256e-05,
511
+ "loss": 1.1606,
512
+ "mean_token_accuracy": 0.8191152907907963,
513
+ "num_tokens": 2929216.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 0.2799950878054771,
518
+ "grad_norm": 0.7097403407096863,
519
+ "learning_rate": 1.7203931203931206e-05,
520
+ "loss": 1.0245,
521
+ "mean_token_accuracy": 0.8594455644488335,
522
+ "num_tokens": 2983287.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 0.2849072823283802,
527
+ "grad_norm": 2.197983741760254,
528
+ "learning_rate": 1.7154791154791157e-05,
529
+ "loss": 1.0116,
530
+ "mean_token_accuracy": 0.8458577767014503,
531
+ "num_tokens": 3036426.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 0.2898194768512833,
536
+ "grad_norm": 0.9750831723213196,
537
+ "learning_rate": 1.7105651105651108e-05,
538
+ "loss": 1.2482,
539
+ "mean_token_accuracy": 0.8097934015095234,
540
+ "num_tokens": 3090471.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 0.29473167137418643,
545
+ "grad_norm": 0.7452868223190308,
546
+ "learning_rate": 1.705651105651106e-05,
547
+ "loss": 0.9643,
548
+ "mean_token_accuracy": 0.8562554508447647,
549
+ "num_tokens": 3143095.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 0.2996438658970895,
554
+ "grad_norm": 0.3684370815753937,
555
+ "learning_rate": 1.700737100737101e-05,
556
+ "loss": 0.8587,
557
+ "mean_token_accuracy": 0.8684959009289741,
558
+ "num_tokens": 3193778.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 0.3045560604199926,
563
+ "grad_norm": 0.4814823567867279,
564
+ "learning_rate": 1.695823095823096e-05,
565
+ "loss": 0.8246,
566
+ "mean_token_accuracy": 0.8646811529994011,
567
+ "num_tokens": 3249776.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 0.30946825494289576,
572
+ "grad_norm": 0.6298699378967285,
573
+ "learning_rate": 1.690909090909091e-05,
574
+ "loss": 1.0043,
575
+ "mean_token_accuracy": 0.849489164352417,
576
+ "num_tokens": 3301301.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 0.31438044946579885,
581
+ "grad_norm": 0.6802086234092712,
582
+ "learning_rate": 1.685995085995086e-05,
583
+ "loss": 0.9932,
584
+ "mean_token_accuracy": 0.8412040494382381,
585
+ "num_tokens": 3351834.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 0.31929264398870194,
590
+ "grad_norm": 0.4441145062446594,
591
+ "learning_rate": 1.6810810810810812e-05,
592
+ "loss": 0.9093,
593
+ "mean_token_accuracy": 0.8644197151064873,
594
+ "num_tokens": 3402630.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 0.3242048385116051,
599
+ "grad_norm": 0.5664909482002258,
600
+ "learning_rate": 1.6761670761670763e-05,
601
+ "loss": 1.0511,
602
+ "mean_token_accuracy": 0.8302886992692947,
603
+ "num_tokens": 3453600.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 0.3291170330345082,
608
+ "grad_norm": 0.898231029510498,
609
+ "learning_rate": 1.6712530712530714e-05,
610
+ "loss": 1.2663,
611
+ "mean_token_accuracy": 0.7889795318245888,
612
+ "num_tokens": 3501154.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 0.33402922755741127,
617
+ "grad_norm": 0.49926525354385376,
618
+ "learning_rate": 1.6663390663390664e-05,
619
+ "loss": 1.0539,
620
+ "mean_token_accuracy": 0.8462485030293465,
621
+ "num_tokens": 3552076.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 0.33894142208031436,
626
+ "grad_norm": 0.5327655673027039,
627
+ "learning_rate": 1.6614250614250615e-05,
628
+ "loss": 1.0757,
629
+ "mean_token_accuracy": 0.8260165579617024,
630
+ "num_tokens": 3602651.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 0.3438536166032175,
635
+ "grad_norm": 0.5295992493629456,
636
+ "learning_rate": 1.6565110565110566e-05,
637
+ "loss": 0.9826,
638
+ "mean_token_accuracy": 0.8445935681462288,
639
+ "num_tokens": 3656310.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 0.3487658111261206,
644
+ "grad_norm": 0.584396243095398,
645
+ "learning_rate": 1.6515970515970517e-05,
646
+ "loss": 0.9771,
647
+ "mean_token_accuracy": 0.8513481065630912,
648
+ "num_tokens": 3712764.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 0.3536780056490237,
653
+ "grad_norm": 0.5786782503128052,
654
+ "learning_rate": 1.6466830466830467e-05,
655
+ "loss": 1.0812,
656
+ "mean_token_accuracy": 0.8179432228207588,
657
+ "num_tokens": 3761995.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 0.35859020017192683,
662
+ "grad_norm": 0.3496871888637543,
663
+ "learning_rate": 1.6417690417690418e-05,
664
+ "loss": 1.0407,
665
+ "mean_token_accuracy": 0.8126879334449768,
666
+ "num_tokens": 3811493.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 0.3635023946948299,
671
+ "grad_norm": 0.5038619041442871,
672
+ "learning_rate": 1.636855036855037e-05,
673
+ "loss": 1.1602,
674
+ "mean_token_accuracy": 0.8141254656016826,
675
+ "num_tokens": 3859376.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 0.368414589217733,
680
+ "grad_norm": 0.4643881320953369,
681
+ "learning_rate": 1.631941031941032e-05,
682
+ "loss": 0.8819,
683
+ "mean_token_accuracy": 0.8545275427401066,
684
+ "num_tokens": 3912052.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 0.3733267837406361,
689
+ "grad_norm": 0.4222355782985687,
690
+ "learning_rate": 1.6270270270270274e-05,
691
+ "loss": 0.8333,
692
+ "mean_token_accuracy": 0.8552621826529503,
693
+ "num_tokens": 3965078.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 0.37823897826353925,
698
+ "grad_norm": 0.4513651728630066,
699
+ "learning_rate": 1.6221130221130224e-05,
700
+ "loss": 1.0386,
701
+ "mean_token_accuracy": 0.8259748943150044,
702
+ "num_tokens": 4015489.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 0.38315117278644234,
707
+ "grad_norm": 0.5292590856552124,
708
+ "learning_rate": 1.6171990171990172e-05,
709
+ "loss": 0.8968,
710
+ "mean_token_accuracy": 0.8551326602697372,
711
+ "num_tokens": 4067831.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 0.38806336730934543,
716
+ "grad_norm": 0.49815845489501953,
717
+ "learning_rate": 1.6122850122850126e-05,
718
+ "loss": 1.0808,
719
+ "mean_token_accuracy": 0.8252614364027977,
720
+ "num_tokens": 4117945.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 0.3929755618322486,
725
+ "grad_norm": 0.339728981256485,
726
+ "learning_rate": 1.6073710073710077e-05,
727
+ "loss": 0.9569,
728
+ "mean_token_accuracy": 0.8419950112700463,
729
+ "num_tokens": 4171138.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 0.39788775635515167,
734
+ "grad_norm": 0.4530418813228607,
735
+ "learning_rate": 1.6024570024570024e-05,
736
+ "loss": 1.1355,
737
+ "mean_token_accuracy": 0.8250428631901741,
738
+ "num_tokens": 4223765.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 0.40279995087805476,
743
+ "grad_norm": 0.5202032923698425,
744
+ "learning_rate": 1.5975429975429978e-05,
745
+ "loss": 0.8306,
746
+ "mean_token_accuracy": 0.865328436344862,
747
+ "num_tokens": 4278056.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 0.4077121454009579,
752
+ "grad_norm": 0.4732300043106079,
753
+ "learning_rate": 1.592628992628993e-05,
754
+ "loss": 1.3396,
755
+ "mean_token_accuracy": 0.783781211823225,
756
+ "num_tokens": 4326704.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 0.412624339923861,
761
+ "grad_norm": 0.49258068203926086,
762
+ "learning_rate": 1.5877149877149876e-05,
763
+ "loss": 0.8462,
764
+ "mean_token_accuracy": 0.8517311483621597,
765
+ "num_tokens": 4378538.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 0.4175365344467641,
770
+ "grad_norm": 0.8045883774757385,
771
+ "learning_rate": 1.582800982800983e-05,
772
+ "loss": 0.9785,
773
+ "mean_token_accuracy": 0.8369067624211312,
774
+ "num_tokens": 4429893.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 0.4224487289696672,
779
+ "grad_norm": 0.361854612827301,
780
+ "learning_rate": 1.577886977886978e-05,
781
+ "loss": 0.936,
782
+ "mean_token_accuracy": 0.8446702107787132,
783
+ "num_tokens": 4481594.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 0.4273609234925703,
788
+ "grad_norm": 0.593479573726654,
789
+ "learning_rate": 1.572972972972973e-05,
790
+ "loss": 0.8245,
791
+ "mean_token_accuracy": 0.8750210002064704,
792
+ "num_tokens": 4534098.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 0.4322731180154734,
797
+ "grad_norm": 0.31826287508010864,
798
+ "learning_rate": 1.5680589680589682e-05,
799
+ "loss": 0.94,
800
+ "mean_token_accuracy": 0.8408755764365197,
801
+ "num_tokens": 4587111.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 0.4371853125383765,
806
+ "grad_norm": 1.4076229333877563,
807
+ "learning_rate": 1.5631449631449633e-05,
808
+ "loss": 1.0176,
809
+ "mean_token_accuracy": 0.8321189761161805,
810
+ "num_tokens": 4640388.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 0.44209750706127965,
815
+ "grad_norm": 0.6677824854850769,
816
+ "learning_rate": 1.5582309582309584e-05,
817
+ "loss": 1.2891,
818
+ "mean_token_accuracy": 0.7943175598978996,
819
+ "num_tokens": 4691468.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 0.44700970158418274,
824
+ "grad_norm": 0.4233896732330322,
825
+ "learning_rate": 1.5533169533169535e-05,
826
+ "loss": 1.1053,
827
+ "mean_token_accuracy": 0.8279901027679444,
828
+ "num_tokens": 4741844.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 0.45192189610708583,
833
+ "grad_norm": 0.35591593384742737,
834
+ "learning_rate": 1.5484029484029485e-05,
835
+ "loss": 0.8832,
836
+ "mean_token_accuracy": 0.8491488620638847,
837
+ "num_tokens": 4793741.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 0.4568340906299889,
842
+ "grad_norm": 0.6583911180496216,
843
+ "learning_rate": 1.5434889434889436e-05,
844
+ "loss": 0.8387,
845
+ "mean_token_accuracy": 0.8648917526006699,
846
+ "num_tokens": 4848466.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 0.46174628515289207,
851
+ "grad_norm": 0.4065934717655182,
852
+ "learning_rate": 1.5385749385749387e-05,
853
+ "loss": 0.9242,
854
+ "mean_token_accuracy": 0.8412024281919003,
855
+ "num_tokens": 4901992.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 0.46665847967579516,
860
+ "grad_norm": 0.3458426594734192,
861
+ "learning_rate": 1.5336609336609338e-05,
862
+ "loss": 0.8545,
863
+ "mean_token_accuracy": 0.855231161415577,
864
+ "num_tokens": 4956964.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 0.47157067419869825,
869
+ "grad_norm": 0.6710091233253479,
870
+ "learning_rate": 1.5287469287469288e-05,
871
+ "loss": 0.9583,
872
+ "mean_token_accuracy": 0.8393168188631535,
873
+ "num_tokens": 5011287.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 0.4764828687216014,
878
+ "grad_norm": 0.3753456771373749,
879
+ "learning_rate": 1.5238329238329239e-05,
880
+ "loss": 0.9738,
881
+ "mean_token_accuracy": 0.8375437319278717,
882
+ "num_tokens": 5061714.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 0.4813950632445045,
887
+ "grad_norm": 1.2011836767196655,
888
+ "learning_rate": 1.5189189189189191e-05,
889
+ "loss": 1.0972,
890
+ "mean_token_accuracy": 0.8070526115596295,
891
+ "num_tokens": 5114645.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 0.4863072577674076,
896
+ "grad_norm": 0.37889617681503296,
897
+ "learning_rate": 1.5140049140049142e-05,
898
+ "loss": 0.9442,
899
+ "mean_token_accuracy": 0.8458199493587018,
900
+ "num_tokens": 5165485.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 0.4912194522903107,
905
+ "grad_norm": 0.6613748669624329,
906
+ "learning_rate": 1.5090909090909091e-05,
907
+ "loss": 1.0196,
908
+ "mean_token_accuracy": 0.8272999472916126,
909
+ "num_tokens": 5216189.0,
910
+ "step": 1000
911
+ }
912
+ ],
913
+ "logging_steps": 10,
914
+ "max_steps": 4070,
915
+ "num_input_tokens_seen": 0,
916
+ "num_train_epochs": 2,
917
+ "save_steps": 500,
918
+ "stateful_callbacks": {
919
+ "TrainerControl": {
920
+ "args": {
921
+ "should_epoch_stop": false,
922
+ "should_evaluate": false,
923
+ "should_log": false,
924
+ "should_save": true,
925
+ "should_training_stop": false
926
+ },
927
+ "attributes": {}
928
+ }
929
+ },
930
+ "total_flos": 2496109039294464.0,
931
+ "train_batch_size": 4,
932
+ "trial_name": null,
933
+ "trial_params": null
934
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f63465acd95be88e019f04b736064426478945f5bc6a4d825ebd1918a32c185
3
+ size 5624
checkpoint-1500/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 256,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-1500/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-1500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea0629dd2abd2e666406c88a39560ec581af3595c16bbb8ff35329466bd63d2d
3
+ size 309900448
checkpoint-1500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0cd50ea568f86477ec67ecdaa77e138a2b039cb8e4ad4c51a9fa85e7f7eeeb1a
3
+ size 619836730
checkpoint-1500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
3
+ size 14244
checkpoint-1500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca361f1f2c04365c2c5cae49e0862ec9ba61ccb6a79f1f4ad7fce3b1d9facd48
3
+ size 1064
checkpoint-1500/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-1500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1500/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1500/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-1500/trainer_state.json ADDED
@@ -0,0 +1,1384 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.736829178435466,
6
+ "eval_steps": 500,
7
+ "global_step": 1500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.004912194522903107,
14
+ "grad_norm": 2.5265462398529053,
15
+ "learning_rate": 1.995577395577396e-05,
16
+ "loss": 5.7783,
17
+ "mean_token_accuracy": 0.17638254687190055,
18
+ "num_tokens": 50745.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.009824389045806214,
23
+ "grad_norm": 1.7381863594055176,
24
+ "learning_rate": 1.9906633906633906e-05,
25
+ "loss": 5.4596,
26
+ "mean_token_accuracy": 0.20633984543383121,
27
+ "num_tokens": 101681.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.01473658356870932,
32
+ "grad_norm": 2.0286130905151367,
33
+ "learning_rate": 1.985749385749386e-05,
34
+ "loss": 5.2069,
35
+ "mean_token_accuracy": 0.2354225989431143,
36
+ "num_tokens": 153620.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.019648778091612428,
41
+ "grad_norm": 2.2828938961029053,
42
+ "learning_rate": 1.980835380835381e-05,
43
+ "loss": 5.0637,
44
+ "mean_token_accuracy": 0.2486753471195698,
45
+ "num_tokens": 205428.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.024560972614515536,
50
+ "grad_norm": 2.247742176055908,
51
+ "learning_rate": 1.975921375921376e-05,
52
+ "loss": 4.7071,
53
+ "mean_token_accuracy": 0.2982919864356518,
54
+ "num_tokens": 260794.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.02947316713741864,
59
+ "grad_norm": 1.4552102088928223,
60
+ "learning_rate": 1.9710073710073712e-05,
61
+ "loss": 4.5951,
62
+ "mean_token_accuracy": 0.3140149611979723,
63
+ "num_tokens": 314216.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.03438536166032175,
68
+ "grad_norm": 2.032104969024658,
69
+ "learning_rate": 1.9660933660933663e-05,
70
+ "loss": 4.3963,
71
+ "mean_token_accuracy": 0.33228743858635423,
72
+ "num_tokens": 365608.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.039297556183224856,
77
+ "grad_norm": 1.5947270393371582,
78
+ "learning_rate": 1.9611793611793614e-05,
79
+ "loss": 4.2578,
80
+ "mean_token_accuracy": 0.3510261565446854,
81
+ "num_tokens": 418509.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.04420975070612796,
86
+ "grad_norm": 1.9174288511276245,
87
+ "learning_rate": 1.9562653562653564e-05,
88
+ "loss": 4.0457,
89
+ "mean_token_accuracy": 0.38185270577669145,
90
+ "num_tokens": 472608.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.04912194522903107,
95
+ "grad_norm": 1.5742830038070679,
96
+ "learning_rate": 1.9513513513513515e-05,
97
+ "loss": 4.0492,
98
+ "mean_token_accuracy": 0.37742224968969823,
99
+ "num_tokens": 521777.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.054034139751934177,
104
+ "grad_norm": 1.5624850988388062,
105
+ "learning_rate": 1.9464373464373466e-05,
106
+ "loss": 3.7053,
107
+ "mean_token_accuracy": 0.41850126534700394,
108
+ "num_tokens": 580161.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.05894633427483728,
113
+ "grad_norm": 2.199150562286377,
114
+ "learning_rate": 1.9415233415233417e-05,
115
+ "loss": 3.6982,
116
+ "mean_token_accuracy": 0.4235368125140667,
117
+ "num_tokens": 627843.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.06385852879774039,
122
+ "grad_norm": 1.8548036813735962,
123
+ "learning_rate": 1.9366093366093367e-05,
124
+ "loss": 3.3414,
125
+ "mean_token_accuracy": 0.4654690891504288,
126
+ "num_tokens": 681443.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.0687707233206435,
131
+ "grad_norm": 2.78169846534729,
132
+ "learning_rate": 1.9316953316953318e-05,
133
+ "loss": 3.2679,
134
+ "mean_token_accuracy": 0.4711888641119003,
135
+ "num_tokens": 733457.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.07368291784354661,
140
+ "grad_norm": 2.4450321197509766,
141
+ "learning_rate": 1.926781326781327e-05,
142
+ "loss": 3.3377,
143
+ "mean_token_accuracy": 0.4799527756869793,
144
+ "num_tokens": 787769.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.07859511236644971,
149
+ "grad_norm": 1.7717996835708618,
150
+ "learning_rate": 1.921867321867322e-05,
151
+ "loss": 2.8853,
152
+ "mean_token_accuracy": 0.541524414718151,
153
+ "num_tokens": 843485.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.08350730688935282,
158
+ "grad_norm": 2.8402254581451416,
159
+ "learning_rate": 1.9169533169533174e-05,
160
+ "loss": 2.862,
161
+ "mean_token_accuracy": 0.5626954190433026,
162
+ "num_tokens": 894003.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.08841950141225592,
167
+ "grad_norm": 1.5486701726913452,
168
+ "learning_rate": 1.912039312039312e-05,
169
+ "loss": 2.7533,
170
+ "mean_token_accuracy": 0.5848119542002678,
171
+ "num_tokens": 949726.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.09333169593515903,
176
+ "grad_norm": 1.3785852193832397,
177
+ "learning_rate": 1.9071253071253072e-05,
178
+ "loss": 2.7602,
179
+ "mean_token_accuracy": 0.5894295200705528,
180
+ "num_tokens": 1000146.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 0.09824389045806214,
185
+ "grad_norm": 1.6824640035629272,
186
+ "learning_rate": 1.9022113022113026e-05,
187
+ "loss": 2.6345,
188
+ "mean_token_accuracy": 0.5961507339030504,
189
+ "num_tokens": 1050142.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 0.10315608498096525,
194
+ "grad_norm": 1.0604822635650635,
195
+ "learning_rate": 1.8972972972972973e-05,
196
+ "loss": 2.4517,
197
+ "mean_token_accuracy": 0.6296438351273537,
198
+ "num_tokens": 1099646.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 0.10806827950386835,
203
+ "grad_norm": 1.9105230569839478,
204
+ "learning_rate": 1.8923832923832924e-05,
205
+ "loss": 2.4563,
206
+ "mean_token_accuracy": 0.6365545220673084,
207
+ "num_tokens": 1148100.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 0.11298047402677146,
212
+ "grad_norm": 1.41056227684021,
213
+ "learning_rate": 1.8874692874692878e-05,
214
+ "loss": 2.4017,
215
+ "mean_token_accuracy": 0.6691348008811474,
216
+ "num_tokens": 1198987.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.11789266854967456,
221
+ "grad_norm": 1.0573596954345703,
222
+ "learning_rate": 1.8825552825552825e-05,
223
+ "loss": 2.3199,
224
+ "mean_token_accuracy": 0.6731609679758549,
225
+ "num_tokens": 1249335.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 0.12280486307257768,
230
+ "grad_norm": 1.6629279851913452,
231
+ "learning_rate": 1.877641277641278e-05,
232
+ "loss": 1.9335,
233
+ "mean_token_accuracy": 0.7327558636665344,
234
+ "num_tokens": 1303157.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 0.12771705759548077,
239
+ "grad_norm": 1.704005479812622,
240
+ "learning_rate": 1.872727272727273e-05,
241
+ "loss": 2.1897,
242
+ "mean_token_accuracy": 0.6974137671291828,
243
+ "num_tokens": 1355455.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 0.1326292521183839,
248
+ "grad_norm": 1.0553226470947266,
249
+ "learning_rate": 1.8678132678132678e-05,
250
+ "loss": 1.9321,
251
+ "mean_token_accuracy": 0.7327657997608185,
252
+ "num_tokens": 1407508.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 0.137541446641287,
257
+ "grad_norm": 1.101886510848999,
258
+ "learning_rate": 1.8628992628992632e-05,
259
+ "loss": 1.9929,
260
+ "mean_token_accuracy": 0.7272698871791363,
261
+ "num_tokens": 1460595.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 0.1424536411641901,
266
+ "grad_norm": 2.005901575088501,
267
+ "learning_rate": 1.8579852579852583e-05,
268
+ "loss": 1.7147,
269
+ "mean_token_accuracy": 0.7604942478239536,
270
+ "num_tokens": 1512647.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 0.14736583568709322,
275
+ "grad_norm": 7.757036209106445,
276
+ "learning_rate": 1.853071253071253e-05,
277
+ "loss": 1.7571,
278
+ "mean_token_accuracy": 0.7603825606405735,
279
+ "num_tokens": 1565516.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 0.1522780302099963,
284
+ "grad_norm": 0.8850768208503723,
285
+ "learning_rate": 1.8481572481572484e-05,
286
+ "loss": 1.5439,
287
+ "mean_token_accuracy": 0.7915346205234528,
288
+ "num_tokens": 1620194.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 0.15719022473289943,
293
+ "grad_norm": 1.2887756824493408,
294
+ "learning_rate": 1.8432432432432435e-05,
295
+ "loss": 1.5419,
296
+ "mean_token_accuracy": 0.7885693073272705,
297
+ "num_tokens": 1672317.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 0.16210241925580254,
302
+ "grad_norm": 2.562720537185669,
303
+ "learning_rate": 1.8383292383292385e-05,
304
+ "loss": 1.5783,
305
+ "mean_token_accuracy": 0.7889993876218796,
306
+ "num_tokens": 1728513.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 0.16701461377870563,
311
+ "grad_norm": 2.907665491104126,
312
+ "learning_rate": 1.8334152334152336e-05,
313
+ "loss": 1.6573,
314
+ "mean_token_accuracy": 0.7675686649978161,
315
+ "num_tokens": 1777759.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 0.17192680830160875,
320
+ "grad_norm": 0.8430731892585754,
321
+ "learning_rate": 1.8285012285012287e-05,
322
+ "loss": 1.664,
323
+ "mean_token_accuracy": 0.7618721649050713,
324
+ "num_tokens": 1833279.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 0.17683900282451184,
329
+ "grad_norm": 1.1680371761322021,
330
+ "learning_rate": 1.8235872235872238e-05,
331
+ "loss": 1.515,
332
+ "mean_token_accuracy": 0.7789990782737732,
333
+ "num_tokens": 1882531.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 0.18175119734741496,
338
+ "grad_norm": 2.009002208709717,
339
+ "learning_rate": 1.818673218673219e-05,
340
+ "loss": 1.4555,
341
+ "mean_token_accuracy": 0.7983467862010002,
342
+ "num_tokens": 1934619.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 0.18666339187031805,
347
+ "grad_norm": 1.3894531726837158,
348
+ "learning_rate": 1.813759213759214e-05,
349
+ "loss": 1.6206,
350
+ "mean_token_accuracy": 0.7681202955543995,
351
+ "num_tokens": 1987231.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 0.19157558639322117,
356
+ "grad_norm": 0.7266082763671875,
357
+ "learning_rate": 1.808845208845209e-05,
358
+ "loss": 1.3022,
359
+ "mean_token_accuracy": 0.818102091550827,
360
+ "num_tokens": 2039985.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 0.1964877809161243,
365
+ "grad_norm": 0.7502278089523315,
366
+ "learning_rate": 1.803931203931204e-05,
367
+ "loss": 1.3902,
368
+ "mean_token_accuracy": 0.792433574795723,
369
+ "num_tokens": 2092178.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 0.20139997543902738,
374
+ "grad_norm": 0.8366674780845642,
375
+ "learning_rate": 1.799017199017199e-05,
376
+ "loss": 1.5385,
377
+ "mean_token_accuracy": 0.7903823673725128,
378
+ "num_tokens": 2151243.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 0.2063121699619305,
383
+ "grad_norm": 0.7726922035217285,
384
+ "learning_rate": 1.7941031941031942e-05,
385
+ "loss": 1.3529,
386
+ "mean_token_accuracy": 0.8017133295536041,
387
+ "num_tokens": 2201072.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 0.2112243644848336,
392
+ "grad_norm": 0.650055468082428,
393
+ "learning_rate": 1.7891891891891893e-05,
394
+ "loss": 1.4312,
395
+ "mean_token_accuracy": 0.7957115054130555,
396
+ "num_tokens": 2256591.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 0.2161365590077367,
401
+ "grad_norm": 0.9430303573608398,
402
+ "learning_rate": 1.7842751842751843e-05,
403
+ "loss": 1.1938,
404
+ "mean_token_accuracy": 0.8211165018379688,
405
+ "num_tokens": 2308048.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 0.22104875353063982,
410
+ "grad_norm": 0.9191327691078186,
411
+ "learning_rate": 1.7793611793611794e-05,
412
+ "loss": 1.1662,
413
+ "mean_token_accuracy": 0.8298586897552014,
414
+ "num_tokens": 2363573.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 0.22596094805354291,
419
+ "grad_norm": 0.7493109703063965,
420
+ "learning_rate": 1.7744471744471745e-05,
421
+ "loss": 1.5908,
422
+ "mean_token_accuracy": 0.7651749603450299,
423
+ "num_tokens": 2413187.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 0.23087314257644603,
428
+ "grad_norm": 0.5522422790527344,
429
+ "learning_rate": 1.7695331695331696e-05,
430
+ "loss": 1.3707,
431
+ "mean_token_accuracy": 0.7843399420380592,
432
+ "num_tokens": 2468041.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 0.23578533709934912,
437
+ "grad_norm": 0.6577270030975342,
438
+ "learning_rate": 1.764619164619165e-05,
439
+ "loss": 1.4679,
440
+ "mean_token_accuracy": 0.7856365032494068,
441
+ "num_tokens": 2517753.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 0.24069753162225224,
446
+ "grad_norm": 0.9699747562408447,
447
+ "learning_rate": 1.7597051597051597e-05,
448
+ "loss": 1.3493,
449
+ "mean_token_accuracy": 0.8016418114304542,
450
+ "num_tokens": 2568072.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 0.24560972614515536,
455
+ "grad_norm": 2.0955657958984375,
456
+ "learning_rate": 1.7547911547911548e-05,
457
+ "loss": 0.9634,
458
+ "mean_token_accuracy": 0.8412673205137253,
459
+ "num_tokens": 2617143.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 0.25052192066805845,
464
+ "grad_norm": 0.7577691078186035,
465
+ "learning_rate": 1.7498771498771502e-05,
466
+ "loss": 1.0889,
467
+ "mean_token_accuracy": 0.8479016363620758,
468
+ "num_tokens": 2669318.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 0.25543411519096154,
473
+ "grad_norm": 0.5891643762588501,
474
+ "learning_rate": 1.744963144963145e-05,
475
+ "loss": 1.1811,
476
+ "mean_token_accuracy": 0.82891755849123,
477
+ "num_tokens": 2723151.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 0.2603463097138647,
482
+ "grad_norm": 0.7815893292427063,
483
+ "learning_rate": 1.74004914004914e-05,
484
+ "loss": 1.3488,
485
+ "mean_token_accuracy": 0.7971963949501515,
486
+ "num_tokens": 2774274.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 0.2652585042367678,
491
+ "grad_norm": 0.365824818611145,
492
+ "learning_rate": 1.7351351351351354e-05,
493
+ "loss": 1.129,
494
+ "mean_token_accuracy": 0.8214888103306294,
495
+ "num_tokens": 2825301.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 0.27017069875967087,
500
+ "grad_norm": 0.5682836174964905,
501
+ "learning_rate": 1.73022113022113e-05,
502
+ "loss": 1.0752,
503
+ "mean_token_accuracy": 0.8513655051589012,
504
+ "num_tokens": 2877767.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 0.275082893282574,
509
+ "grad_norm": 0.9484438300132751,
510
+ "learning_rate": 1.7253071253071256e-05,
511
+ "loss": 1.1606,
512
+ "mean_token_accuracy": 0.8191152907907963,
513
+ "num_tokens": 2929216.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 0.2799950878054771,
518
+ "grad_norm": 0.7097403407096863,
519
+ "learning_rate": 1.7203931203931206e-05,
520
+ "loss": 1.0245,
521
+ "mean_token_accuracy": 0.8594455644488335,
522
+ "num_tokens": 2983287.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 0.2849072823283802,
527
+ "grad_norm": 2.197983741760254,
528
+ "learning_rate": 1.7154791154791157e-05,
529
+ "loss": 1.0116,
530
+ "mean_token_accuracy": 0.8458577767014503,
531
+ "num_tokens": 3036426.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 0.2898194768512833,
536
+ "grad_norm": 0.9750831723213196,
537
+ "learning_rate": 1.7105651105651108e-05,
538
+ "loss": 1.2482,
539
+ "mean_token_accuracy": 0.8097934015095234,
540
+ "num_tokens": 3090471.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 0.29473167137418643,
545
+ "grad_norm": 0.7452868223190308,
546
+ "learning_rate": 1.705651105651106e-05,
547
+ "loss": 0.9643,
548
+ "mean_token_accuracy": 0.8562554508447647,
549
+ "num_tokens": 3143095.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 0.2996438658970895,
554
+ "grad_norm": 0.3684370815753937,
555
+ "learning_rate": 1.700737100737101e-05,
556
+ "loss": 0.8587,
557
+ "mean_token_accuracy": 0.8684959009289741,
558
+ "num_tokens": 3193778.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 0.3045560604199926,
563
+ "grad_norm": 0.4814823567867279,
564
+ "learning_rate": 1.695823095823096e-05,
565
+ "loss": 0.8246,
566
+ "mean_token_accuracy": 0.8646811529994011,
567
+ "num_tokens": 3249776.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 0.30946825494289576,
572
+ "grad_norm": 0.6298699378967285,
573
+ "learning_rate": 1.690909090909091e-05,
574
+ "loss": 1.0043,
575
+ "mean_token_accuracy": 0.849489164352417,
576
+ "num_tokens": 3301301.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 0.31438044946579885,
581
+ "grad_norm": 0.6802086234092712,
582
+ "learning_rate": 1.685995085995086e-05,
583
+ "loss": 0.9932,
584
+ "mean_token_accuracy": 0.8412040494382381,
585
+ "num_tokens": 3351834.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 0.31929264398870194,
590
+ "grad_norm": 0.4441145062446594,
591
+ "learning_rate": 1.6810810810810812e-05,
592
+ "loss": 0.9093,
593
+ "mean_token_accuracy": 0.8644197151064873,
594
+ "num_tokens": 3402630.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 0.3242048385116051,
599
+ "grad_norm": 0.5664909482002258,
600
+ "learning_rate": 1.6761670761670763e-05,
601
+ "loss": 1.0511,
602
+ "mean_token_accuracy": 0.8302886992692947,
603
+ "num_tokens": 3453600.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 0.3291170330345082,
608
+ "grad_norm": 0.898231029510498,
609
+ "learning_rate": 1.6712530712530714e-05,
610
+ "loss": 1.2663,
611
+ "mean_token_accuracy": 0.7889795318245888,
612
+ "num_tokens": 3501154.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 0.33402922755741127,
617
+ "grad_norm": 0.49926525354385376,
618
+ "learning_rate": 1.6663390663390664e-05,
619
+ "loss": 1.0539,
620
+ "mean_token_accuracy": 0.8462485030293465,
621
+ "num_tokens": 3552076.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 0.33894142208031436,
626
+ "grad_norm": 0.5327655673027039,
627
+ "learning_rate": 1.6614250614250615e-05,
628
+ "loss": 1.0757,
629
+ "mean_token_accuracy": 0.8260165579617024,
630
+ "num_tokens": 3602651.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 0.3438536166032175,
635
+ "grad_norm": 0.5295992493629456,
636
+ "learning_rate": 1.6565110565110566e-05,
637
+ "loss": 0.9826,
638
+ "mean_token_accuracy": 0.8445935681462288,
639
+ "num_tokens": 3656310.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 0.3487658111261206,
644
+ "grad_norm": 0.584396243095398,
645
+ "learning_rate": 1.6515970515970517e-05,
646
+ "loss": 0.9771,
647
+ "mean_token_accuracy": 0.8513481065630912,
648
+ "num_tokens": 3712764.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 0.3536780056490237,
653
+ "grad_norm": 0.5786782503128052,
654
+ "learning_rate": 1.6466830466830467e-05,
655
+ "loss": 1.0812,
656
+ "mean_token_accuracy": 0.8179432228207588,
657
+ "num_tokens": 3761995.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 0.35859020017192683,
662
+ "grad_norm": 0.3496871888637543,
663
+ "learning_rate": 1.6417690417690418e-05,
664
+ "loss": 1.0407,
665
+ "mean_token_accuracy": 0.8126879334449768,
666
+ "num_tokens": 3811493.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 0.3635023946948299,
671
+ "grad_norm": 0.5038619041442871,
672
+ "learning_rate": 1.636855036855037e-05,
673
+ "loss": 1.1602,
674
+ "mean_token_accuracy": 0.8141254656016826,
675
+ "num_tokens": 3859376.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 0.368414589217733,
680
+ "grad_norm": 0.4643881320953369,
681
+ "learning_rate": 1.631941031941032e-05,
682
+ "loss": 0.8819,
683
+ "mean_token_accuracy": 0.8545275427401066,
684
+ "num_tokens": 3912052.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 0.3733267837406361,
689
+ "grad_norm": 0.4222355782985687,
690
+ "learning_rate": 1.6270270270270274e-05,
691
+ "loss": 0.8333,
692
+ "mean_token_accuracy": 0.8552621826529503,
693
+ "num_tokens": 3965078.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 0.37823897826353925,
698
+ "grad_norm": 0.4513651728630066,
699
+ "learning_rate": 1.6221130221130224e-05,
700
+ "loss": 1.0386,
701
+ "mean_token_accuracy": 0.8259748943150044,
702
+ "num_tokens": 4015489.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 0.38315117278644234,
707
+ "grad_norm": 0.5292590856552124,
708
+ "learning_rate": 1.6171990171990172e-05,
709
+ "loss": 0.8968,
710
+ "mean_token_accuracy": 0.8551326602697372,
711
+ "num_tokens": 4067831.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 0.38806336730934543,
716
+ "grad_norm": 0.49815845489501953,
717
+ "learning_rate": 1.6122850122850126e-05,
718
+ "loss": 1.0808,
719
+ "mean_token_accuracy": 0.8252614364027977,
720
+ "num_tokens": 4117945.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 0.3929755618322486,
725
+ "grad_norm": 0.339728981256485,
726
+ "learning_rate": 1.6073710073710077e-05,
727
+ "loss": 0.9569,
728
+ "mean_token_accuracy": 0.8419950112700463,
729
+ "num_tokens": 4171138.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 0.39788775635515167,
734
+ "grad_norm": 0.4530418813228607,
735
+ "learning_rate": 1.6024570024570024e-05,
736
+ "loss": 1.1355,
737
+ "mean_token_accuracy": 0.8250428631901741,
738
+ "num_tokens": 4223765.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 0.40279995087805476,
743
+ "grad_norm": 0.5202032923698425,
744
+ "learning_rate": 1.5975429975429978e-05,
745
+ "loss": 0.8306,
746
+ "mean_token_accuracy": 0.865328436344862,
747
+ "num_tokens": 4278056.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 0.4077121454009579,
752
+ "grad_norm": 0.4732300043106079,
753
+ "learning_rate": 1.592628992628993e-05,
754
+ "loss": 1.3396,
755
+ "mean_token_accuracy": 0.783781211823225,
756
+ "num_tokens": 4326704.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 0.412624339923861,
761
+ "grad_norm": 0.49258068203926086,
762
+ "learning_rate": 1.5877149877149876e-05,
763
+ "loss": 0.8462,
764
+ "mean_token_accuracy": 0.8517311483621597,
765
+ "num_tokens": 4378538.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 0.4175365344467641,
770
+ "grad_norm": 0.8045883774757385,
771
+ "learning_rate": 1.582800982800983e-05,
772
+ "loss": 0.9785,
773
+ "mean_token_accuracy": 0.8369067624211312,
774
+ "num_tokens": 4429893.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 0.4224487289696672,
779
+ "grad_norm": 0.361854612827301,
780
+ "learning_rate": 1.577886977886978e-05,
781
+ "loss": 0.936,
782
+ "mean_token_accuracy": 0.8446702107787132,
783
+ "num_tokens": 4481594.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 0.4273609234925703,
788
+ "grad_norm": 0.593479573726654,
789
+ "learning_rate": 1.572972972972973e-05,
790
+ "loss": 0.8245,
791
+ "mean_token_accuracy": 0.8750210002064704,
792
+ "num_tokens": 4534098.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 0.4322731180154734,
797
+ "grad_norm": 0.31826287508010864,
798
+ "learning_rate": 1.5680589680589682e-05,
799
+ "loss": 0.94,
800
+ "mean_token_accuracy": 0.8408755764365197,
801
+ "num_tokens": 4587111.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 0.4371853125383765,
806
+ "grad_norm": 1.4076229333877563,
807
+ "learning_rate": 1.5631449631449633e-05,
808
+ "loss": 1.0176,
809
+ "mean_token_accuracy": 0.8321189761161805,
810
+ "num_tokens": 4640388.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 0.44209750706127965,
815
+ "grad_norm": 0.6677824854850769,
816
+ "learning_rate": 1.5582309582309584e-05,
817
+ "loss": 1.2891,
818
+ "mean_token_accuracy": 0.7943175598978996,
819
+ "num_tokens": 4691468.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 0.44700970158418274,
824
+ "grad_norm": 0.4233896732330322,
825
+ "learning_rate": 1.5533169533169535e-05,
826
+ "loss": 1.1053,
827
+ "mean_token_accuracy": 0.8279901027679444,
828
+ "num_tokens": 4741844.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 0.45192189610708583,
833
+ "grad_norm": 0.35591593384742737,
834
+ "learning_rate": 1.5484029484029485e-05,
835
+ "loss": 0.8832,
836
+ "mean_token_accuracy": 0.8491488620638847,
837
+ "num_tokens": 4793741.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 0.4568340906299889,
842
+ "grad_norm": 0.6583911180496216,
843
+ "learning_rate": 1.5434889434889436e-05,
844
+ "loss": 0.8387,
845
+ "mean_token_accuracy": 0.8648917526006699,
846
+ "num_tokens": 4848466.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 0.46174628515289207,
851
+ "grad_norm": 0.4065934717655182,
852
+ "learning_rate": 1.5385749385749387e-05,
853
+ "loss": 0.9242,
854
+ "mean_token_accuracy": 0.8412024281919003,
855
+ "num_tokens": 4901992.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 0.46665847967579516,
860
+ "grad_norm": 0.3458426594734192,
861
+ "learning_rate": 1.5336609336609338e-05,
862
+ "loss": 0.8545,
863
+ "mean_token_accuracy": 0.855231161415577,
864
+ "num_tokens": 4956964.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 0.47157067419869825,
869
+ "grad_norm": 0.6710091233253479,
870
+ "learning_rate": 1.5287469287469288e-05,
871
+ "loss": 0.9583,
872
+ "mean_token_accuracy": 0.8393168188631535,
873
+ "num_tokens": 5011287.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 0.4764828687216014,
878
+ "grad_norm": 0.3753456771373749,
879
+ "learning_rate": 1.5238329238329239e-05,
880
+ "loss": 0.9738,
881
+ "mean_token_accuracy": 0.8375437319278717,
882
+ "num_tokens": 5061714.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 0.4813950632445045,
887
+ "grad_norm": 1.2011836767196655,
888
+ "learning_rate": 1.5189189189189191e-05,
889
+ "loss": 1.0972,
890
+ "mean_token_accuracy": 0.8070526115596295,
891
+ "num_tokens": 5114645.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 0.4863072577674076,
896
+ "grad_norm": 0.37889617681503296,
897
+ "learning_rate": 1.5140049140049142e-05,
898
+ "loss": 0.9442,
899
+ "mean_token_accuracy": 0.8458199493587018,
900
+ "num_tokens": 5165485.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 0.4912194522903107,
905
+ "grad_norm": 0.6613748669624329,
906
+ "learning_rate": 1.5090909090909091e-05,
907
+ "loss": 1.0196,
908
+ "mean_token_accuracy": 0.8272999472916126,
909
+ "num_tokens": 5216189.0,
910
+ "step": 1000
911
+ },
912
+ {
913
+ "epoch": 0.4961316468132138,
914
+ "grad_norm": 0.7825655937194824,
915
+ "learning_rate": 1.5041769041769044e-05,
916
+ "loss": 1.011,
917
+ "mean_token_accuracy": 0.8262499138712883,
918
+ "num_tokens": 5269248.0,
919
+ "step": 1010
920
+ },
921
+ {
922
+ "epoch": 0.5010438413361169,
923
+ "grad_norm": 0.2106683850288391,
924
+ "learning_rate": 1.4992628992628994e-05,
925
+ "loss": 0.9098,
926
+ "mean_token_accuracy": 0.8512444771826267,
927
+ "num_tokens": 5322524.0,
928
+ "step": 1020
929
+ },
930
+ {
931
+ "epoch": 0.50595603585902,
932
+ "grad_norm": 0.8093978762626648,
933
+ "learning_rate": 1.4943488943488943e-05,
934
+ "loss": 1.2403,
935
+ "mean_token_accuracy": 0.7969906754791737,
936
+ "num_tokens": 5370123.0,
937
+ "step": 1030
938
+ },
939
+ {
940
+ "epoch": 0.5108682303819231,
941
+ "grad_norm": 0.5412647724151611,
942
+ "learning_rate": 1.4894348894348896e-05,
943
+ "loss": 0.8821,
944
+ "mean_token_accuracy": 0.8460625484585762,
945
+ "num_tokens": 5422125.0,
946
+ "step": 1040
947
+ },
948
+ {
949
+ "epoch": 0.5157804249048262,
950
+ "grad_norm": 0.5411800146102905,
951
+ "learning_rate": 1.4845208845208847e-05,
952
+ "loss": 1.0804,
953
+ "mean_token_accuracy": 0.8199978180229663,
954
+ "num_tokens": 5472127.0,
955
+ "step": 1050
956
+ },
957
+ {
958
+ "epoch": 0.5206926194277294,
959
+ "grad_norm": 0.4349406063556671,
960
+ "learning_rate": 1.4796068796068797e-05,
961
+ "loss": 0.7559,
962
+ "mean_token_accuracy": 0.8662853054702282,
963
+ "num_tokens": 5525028.0,
964
+ "step": 1060
965
+ },
966
+ {
967
+ "epoch": 0.5256048139506324,
968
+ "grad_norm": 0.4738285541534424,
969
+ "learning_rate": 1.4746928746928748e-05,
970
+ "loss": 1.1225,
971
+ "mean_token_accuracy": 0.8103010386228562,
972
+ "num_tokens": 5570859.0,
973
+ "step": 1070
974
+ },
975
+ {
976
+ "epoch": 0.5305170084735356,
977
+ "grad_norm": 0.6644457578659058,
978
+ "learning_rate": 1.46977886977887e-05,
979
+ "loss": 0.9334,
980
+ "mean_token_accuracy": 0.8213524237275124,
981
+ "num_tokens": 5618933.0,
982
+ "step": 1080
983
+ },
984
+ {
985
+ "epoch": 0.5354292029964387,
986
+ "grad_norm": 0.3038887083530426,
987
+ "learning_rate": 1.464864864864865e-05,
988
+ "loss": 1.0292,
989
+ "mean_token_accuracy": 0.810487725585699,
990
+ "num_tokens": 5664922.0,
991
+ "step": 1090
992
+ },
993
+ {
994
+ "epoch": 0.5403413975193417,
995
+ "grad_norm": 0.26083576679229736,
996
+ "learning_rate": 1.45995085995086e-05,
997
+ "loss": 0.9084,
998
+ "mean_token_accuracy": 0.836492794752121,
999
+ "num_tokens": 5716671.0,
1000
+ "step": 1100
1001
+ },
1002
+ {
1003
+ "epoch": 0.5452535920422449,
1004
+ "grad_norm": 0.36848321557044983,
1005
+ "learning_rate": 1.4550368550368553e-05,
1006
+ "loss": 1.1544,
1007
+ "mean_token_accuracy": 0.8033395648002625,
1008
+ "num_tokens": 5764576.0,
1009
+ "step": 1110
1010
+ },
1011
+ {
1012
+ "epoch": 0.550165786565148,
1013
+ "grad_norm": 0.3338620662689209,
1014
+ "learning_rate": 1.4501228501228502e-05,
1015
+ "loss": 1.0114,
1016
+ "mean_token_accuracy": 0.8188160583376884,
1017
+ "num_tokens": 5816699.0,
1018
+ "step": 1120
1019
+ },
1020
+ {
1021
+ "epoch": 0.5550779810880511,
1022
+ "grad_norm": 0.3278094232082367,
1023
+ "learning_rate": 1.4452088452088452e-05,
1024
+ "loss": 0.6189,
1025
+ "mean_token_accuracy": 0.8895823881030083,
1026
+ "num_tokens": 5874257.0,
1027
+ "step": 1130
1028
+ },
1029
+ {
1030
+ "epoch": 0.5599901756109542,
1031
+ "grad_norm": 0.4236205220222473,
1032
+ "learning_rate": 1.4402948402948405e-05,
1033
+ "loss": 0.9474,
1034
+ "mean_token_accuracy": 0.8435555145144462,
1035
+ "num_tokens": 5925628.0,
1036
+ "step": 1140
1037
+ },
1038
+ {
1039
+ "epoch": 0.5649023701338572,
1040
+ "grad_norm": 0.21693912148475647,
1041
+ "learning_rate": 1.4353808353808354e-05,
1042
+ "loss": 0.9109,
1043
+ "mean_token_accuracy": 0.842586936801672,
1044
+ "num_tokens": 5977130.0,
1045
+ "step": 1150
1046
+ },
1047
+ {
1048
+ "epoch": 0.5698145646567604,
1049
+ "grad_norm": 0.6288137435913086,
1050
+ "learning_rate": 1.4304668304668306e-05,
1051
+ "loss": 0.7005,
1052
+ "mean_token_accuracy": 0.877720920741558,
1053
+ "num_tokens": 6032405.0,
1054
+ "step": 1160
1055
+ },
1056
+ {
1057
+ "epoch": 0.5747267591796635,
1058
+ "grad_norm": 0.29411551356315613,
1059
+ "learning_rate": 1.4255528255528257e-05,
1060
+ "loss": 0.9163,
1061
+ "mean_token_accuracy": 0.8345251567661762,
1062
+ "num_tokens": 6081769.0,
1063
+ "step": 1170
1064
+ },
1065
+ {
1066
+ "epoch": 0.5796389537025666,
1067
+ "grad_norm": 0.3542836308479309,
1068
+ "learning_rate": 1.420638820638821e-05,
1069
+ "loss": 0.8983,
1070
+ "mean_token_accuracy": 0.8478083908557892,
1071
+ "num_tokens": 6131660.0,
1072
+ "step": 1180
1073
+ },
1074
+ {
1075
+ "epoch": 0.5845511482254697,
1076
+ "grad_norm": 0.6326059103012085,
1077
+ "learning_rate": 1.4157248157248158e-05,
1078
+ "loss": 1.055,
1079
+ "mean_token_accuracy": 0.8133970260620117,
1080
+ "num_tokens": 6177706.0,
1081
+ "step": 1190
1082
+ },
1083
+ {
1084
+ "epoch": 0.5894633427483729,
1085
+ "grad_norm": 0.39950093626976013,
1086
+ "learning_rate": 1.410810810810811e-05,
1087
+ "loss": 0.6446,
1088
+ "mean_token_accuracy": 0.8883207038044929,
1089
+ "num_tokens": 6231530.0,
1090
+ "step": 1200
1091
+ },
1092
+ {
1093
+ "epoch": 0.5943755372712759,
1094
+ "grad_norm": 1.4949785470962524,
1095
+ "learning_rate": 1.4058968058968062e-05,
1096
+ "loss": 0.8527,
1097
+ "mean_token_accuracy": 0.8495218820869923,
1098
+ "num_tokens": 6284936.0,
1099
+ "step": 1210
1100
+ },
1101
+ {
1102
+ "epoch": 0.599287731794179,
1103
+ "grad_norm": 0.5013307332992554,
1104
+ "learning_rate": 1.400982800982801e-05,
1105
+ "loss": 0.9342,
1106
+ "mean_token_accuracy": 0.8461359746754169,
1107
+ "num_tokens": 6338328.0,
1108
+ "step": 1220
1109
+ },
1110
+ {
1111
+ "epoch": 0.6041999263170822,
1112
+ "grad_norm": 0.4373924136161804,
1113
+ "learning_rate": 1.3960687960687961e-05,
1114
+ "loss": 0.7957,
1115
+ "mean_token_accuracy": 0.8590179920196533,
1116
+ "num_tokens": 6391247.0,
1117
+ "step": 1230
1118
+ },
1119
+ {
1120
+ "epoch": 0.6091121208399852,
1121
+ "grad_norm": 0.30794695019721985,
1122
+ "learning_rate": 1.3911547911547914e-05,
1123
+ "loss": 0.9578,
1124
+ "mean_token_accuracy": 0.8326818406581878,
1125
+ "num_tokens": 6444004.0,
1126
+ "step": 1240
1127
+ },
1128
+ {
1129
+ "epoch": 0.6140243153628884,
1130
+ "grad_norm": 0.3918187618255615,
1131
+ "learning_rate": 1.3862407862407863e-05,
1132
+ "loss": 0.8732,
1133
+ "mean_token_accuracy": 0.8658828616142273,
1134
+ "num_tokens": 6495578.0,
1135
+ "step": 1250
1136
+ },
1137
+ {
1138
+ "epoch": 0.6189365098857915,
1139
+ "grad_norm": 0.2752901315689087,
1140
+ "learning_rate": 1.3813267813267814e-05,
1141
+ "loss": 0.9506,
1142
+ "mean_token_accuracy": 0.8301159009337425,
1143
+ "num_tokens": 6545604.0,
1144
+ "step": 1260
1145
+ },
1146
+ {
1147
+ "epoch": 0.6238487044086946,
1148
+ "grad_norm": 0.460627943277359,
1149
+ "learning_rate": 1.3764127764127766e-05,
1150
+ "loss": 1.0007,
1151
+ "mean_token_accuracy": 0.8227074682712555,
1152
+ "num_tokens": 6594381.0,
1153
+ "step": 1270
1154
+ },
1155
+ {
1156
+ "epoch": 0.6287608989315977,
1157
+ "grad_norm": 0.4065866768360138,
1158
+ "learning_rate": 1.3714987714987715e-05,
1159
+ "loss": 0.8056,
1160
+ "mean_token_accuracy": 0.8646926179528236,
1161
+ "num_tokens": 6649190.0,
1162
+ "step": 1280
1163
+ },
1164
+ {
1165
+ "epoch": 0.6336730934545008,
1166
+ "grad_norm": 0.4343368113040924,
1167
+ "learning_rate": 1.3665847665847667e-05,
1168
+ "loss": 0.8879,
1169
+ "mean_token_accuracy": 0.8385777972638607,
1170
+ "num_tokens": 6701394.0,
1171
+ "step": 1290
1172
+ },
1173
+ {
1174
+ "epoch": 0.6385852879774039,
1175
+ "grad_norm": 0.5364151000976562,
1176
+ "learning_rate": 1.3616707616707618e-05,
1177
+ "loss": 0.9419,
1178
+ "mean_token_accuracy": 0.8308121666312218,
1179
+ "num_tokens": 6752270.0,
1180
+ "step": 1300
1181
+ },
1182
+ {
1183
+ "epoch": 0.643497482500307,
1184
+ "grad_norm": 0.3982967734336853,
1185
+ "learning_rate": 1.3567567567567567e-05,
1186
+ "loss": 1.0305,
1187
+ "mean_token_accuracy": 0.8274823799729347,
1188
+ "num_tokens": 6805857.0,
1189
+ "step": 1310
1190
+ },
1191
+ {
1192
+ "epoch": 0.6484096770232102,
1193
+ "grad_norm": 0.7117066383361816,
1194
+ "learning_rate": 1.351842751842752e-05,
1195
+ "loss": 1.0594,
1196
+ "mean_token_accuracy": 0.8103688299655915,
1197
+ "num_tokens": 6852455.0,
1198
+ "step": 1320
1199
+ },
1200
+ {
1201
+ "epoch": 0.6533218715461132,
1202
+ "grad_norm": 0.3083662986755371,
1203
+ "learning_rate": 1.346928746928747e-05,
1204
+ "loss": 1.0412,
1205
+ "mean_token_accuracy": 0.8186712831258773,
1206
+ "num_tokens": 6896950.0,
1207
+ "step": 1330
1208
+ },
1209
+ {
1210
+ "epoch": 0.6582340660690164,
1211
+ "grad_norm": 0.4959128499031067,
1212
+ "learning_rate": 1.342014742014742e-05,
1213
+ "loss": 0.9348,
1214
+ "mean_token_accuracy": 0.8453299880027771,
1215
+ "num_tokens": 6950814.0,
1216
+ "step": 1340
1217
+ },
1218
+ {
1219
+ "epoch": 0.6631462605919194,
1220
+ "grad_norm": 0.8182358145713806,
1221
+ "learning_rate": 1.3371007371007372e-05,
1222
+ "loss": 1.0552,
1223
+ "mean_token_accuracy": 0.8226786606013775,
1224
+ "num_tokens": 7003460.0,
1225
+ "step": 1350
1226
+ },
1227
+ {
1228
+ "epoch": 0.6680584551148225,
1229
+ "grad_norm": 0.43365928530693054,
1230
+ "learning_rate": 1.3321867321867323e-05,
1231
+ "loss": 0.8795,
1232
+ "mean_token_accuracy": 0.8458721756935119,
1233
+ "num_tokens": 7054901.0,
1234
+ "step": 1360
1235
+ },
1236
+ {
1237
+ "epoch": 0.6729706496377257,
1238
+ "grad_norm": 0.2876506745815277,
1239
+ "learning_rate": 1.3272727272727275e-05,
1240
+ "loss": 0.8282,
1241
+ "mean_token_accuracy": 0.8459444969892502,
1242
+ "num_tokens": 7106038.0,
1243
+ "step": 1370
1244
+ },
1245
+ {
1246
+ "epoch": 0.6778828441606287,
1247
+ "grad_norm": 0.4010944068431854,
1248
+ "learning_rate": 1.3223587223587224e-05,
1249
+ "loss": 0.7836,
1250
+ "mean_token_accuracy": 0.8610207386314869,
1251
+ "num_tokens": 7159651.0,
1252
+ "step": 1380
1253
+ },
1254
+ {
1255
+ "epoch": 0.6827950386835319,
1256
+ "grad_norm": 0.19336983561515808,
1257
+ "learning_rate": 1.3174447174447176e-05,
1258
+ "loss": 1.0039,
1259
+ "mean_token_accuracy": 0.8369563437998295,
1260
+ "num_tokens": 7204811.0,
1261
+ "step": 1390
1262
+ },
1263
+ {
1264
+ "epoch": 0.687707233206435,
1265
+ "grad_norm": 0.5056695938110352,
1266
+ "learning_rate": 1.3125307125307127e-05,
1267
+ "loss": 0.6854,
1268
+ "mean_token_accuracy": 0.878055976331234,
1269
+ "num_tokens": 7256836.0,
1270
+ "step": 1400
1271
+ },
1272
+ {
1273
+ "epoch": 0.692619427729338,
1274
+ "grad_norm": 0.3462684750556946,
1275
+ "learning_rate": 1.3076167076167076e-05,
1276
+ "loss": 1.2627,
1277
+ "mean_token_accuracy": 0.7998166553676128,
1278
+ "num_tokens": 7308913.0,
1279
+ "step": 1410
1280
+ },
1281
+ {
1282
+ "epoch": 0.6975316222522412,
1283
+ "grad_norm": 0.2867954671382904,
1284
+ "learning_rate": 1.3027027027027029e-05,
1285
+ "loss": 0.6718,
1286
+ "mean_token_accuracy": 0.8842718601226807,
1287
+ "num_tokens": 7365814.0,
1288
+ "step": 1420
1289
+ },
1290
+ {
1291
+ "epoch": 0.7024438167751443,
1292
+ "grad_norm": 0.4659688472747803,
1293
+ "learning_rate": 1.297788697788698e-05,
1294
+ "loss": 1.1291,
1295
+ "mean_token_accuracy": 0.8083386868238449,
1296
+ "num_tokens": 7419238.0,
1297
+ "step": 1430
1298
+ },
1299
+ {
1300
+ "epoch": 0.7073560112980474,
1301
+ "grad_norm": 0.41312354803085327,
1302
+ "learning_rate": 1.2928746928746928e-05,
1303
+ "loss": 0.8625,
1304
+ "mean_token_accuracy": 0.8470474675297737,
1305
+ "num_tokens": 7469177.0,
1306
+ "step": 1440
1307
+ },
1308
+ {
1309
+ "epoch": 0.7122682058209505,
1310
+ "grad_norm": 0.5044501423835754,
1311
+ "learning_rate": 1.2879606879606881e-05,
1312
+ "loss": 0.8186,
1313
+ "mean_token_accuracy": 0.8533941991627216,
1314
+ "num_tokens": 7520617.0,
1315
+ "step": 1450
1316
+ },
1317
+ {
1318
+ "epoch": 0.7171804003438537,
1319
+ "grad_norm": 0.3712249994277954,
1320
+ "learning_rate": 1.2830466830466832e-05,
1321
+ "loss": 0.912,
1322
+ "mean_token_accuracy": 0.838789189606905,
1323
+ "num_tokens": 7569361.0,
1324
+ "step": 1460
1325
+ },
1326
+ {
1327
+ "epoch": 0.7220925948667567,
1328
+ "grad_norm": 0.5090559720993042,
1329
+ "learning_rate": 1.2781326781326782e-05,
1330
+ "loss": 0.9559,
1331
+ "mean_token_accuracy": 0.8298265844583511,
1332
+ "num_tokens": 7619902.0,
1333
+ "step": 1470
1334
+ },
1335
+ {
1336
+ "epoch": 0.7270047893896598,
1337
+ "grad_norm": 0.40056052803993225,
1338
+ "learning_rate": 1.2732186732186733e-05,
1339
+ "loss": 0.9697,
1340
+ "mean_token_accuracy": 0.8409195326268672,
1341
+ "num_tokens": 7671364.0,
1342
+ "step": 1480
1343
+ },
1344
+ {
1345
+ "epoch": 0.731916983912563,
1346
+ "grad_norm": 0.5502124428749084,
1347
+ "learning_rate": 1.2683046683046685e-05,
1348
+ "loss": 0.974,
1349
+ "mean_token_accuracy": 0.8395135879516602,
1350
+ "num_tokens": 7721501.0,
1351
+ "step": 1490
1352
+ },
1353
+ {
1354
+ "epoch": 0.736829178435466,
1355
+ "grad_norm": 0.49662184715270996,
1356
+ "learning_rate": 1.2633906633906635e-05,
1357
+ "loss": 0.9764,
1358
+ "mean_token_accuracy": 0.8396995335817337,
1359
+ "num_tokens": 7771721.0,
1360
+ "step": 1500
1361
+ }
1362
+ ],
1363
+ "logging_steps": 10,
1364
+ "max_steps": 4070,
1365
+ "num_input_tokens_seen": 0,
1366
+ "num_train_epochs": 2,
1367
+ "save_steps": 500,
1368
+ "stateful_callbacks": {
1369
+ "TrainerControl": {
1370
+ "args": {
1371
+ "should_epoch_stop": false,
1372
+ "should_evaluate": false,
1373
+ "should_log": false,
1374
+ "should_save": true,
1375
+ "should_training_stop": false
1376
+ },
1377
+ "attributes": {}
1378
+ }
1379
+ },
1380
+ "total_flos": 3728779897522176.0,
1381
+ "train_batch_size": 4,
1382
+ "trial_name": null,
1383
+ "trial_params": null
1384
+ }
checkpoint-1500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f63465acd95be88e019f04b736064426478945f5bc6a4d825ebd1918a32c185
3
+ size 5624
checkpoint-2000/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 256,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3824a210c206d5d5491ba736aba3435ebacea1c722f9c74a279b3d583479cab
3
+ size 309900448
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92fa9fcf020f273ef184b82aa11943d9a87a5b2461971520c00913705cbe6e3e
3
+ size 619836730
checkpoint-2000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d9cd6a0487226e5bd30d1846894c82af483733ab4381b75bae9c0745e05d405
3
+ size 14244
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88e61fef9f4f907eb38d9682fc9809a7c039e80531a21b5d3ebdb66b732cd2ff
3
+ size 1064
checkpoint-2000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-2000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-2000/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-2000/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,1834 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.9824389045806214,
6
+ "eval_steps": 500,
7
+ "global_step": 2000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.004912194522903107,
14
+ "grad_norm": 2.5265462398529053,
15
+ "learning_rate": 1.995577395577396e-05,
16
+ "loss": 5.7783,
17
+ "mean_token_accuracy": 0.17638254687190055,
18
+ "num_tokens": 50745.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.009824389045806214,
23
+ "grad_norm": 1.7381863594055176,
24
+ "learning_rate": 1.9906633906633906e-05,
25
+ "loss": 5.4596,
26
+ "mean_token_accuracy": 0.20633984543383121,
27
+ "num_tokens": 101681.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.01473658356870932,
32
+ "grad_norm": 2.0286130905151367,
33
+ "learning_rate": 1.985749385749386e-05,
34
+ "loss": 5.2069,
35
+ "mean_token_accuracy": 0.2354225989431143,
36
+ "num_tokens": 153620.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.019648778091612428,
41
+ "grad_norm": 2.2828938961029053,
42
+ "learning_rate": 1.980835380835381e-05,
43
+ "loss": 5.0637,
44
+ "mean_token_accuracy": 0.2486753471195698,
45
+ "num_tokens": 205428.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.024560972614515536,
50
+ "grad_norm": 2.247742176055908,
51
+ "learning_rate": 1.975921375921376e-05,
52
+ "loss": 4.7071,
53
+ "mean_token_accuracy": 0.2982919864356518,
54
+ "num_tokens": 260794.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.02947316713741864,
59
+ "grad_norm": 1.4552102088928223,
60
+ "learning_rate": 1.9710073710073712e-05,
61
+ "loss": 4.5951,
62
+ "mean_token_accuracy": 0.3140149611979723,
63
+ "num_tokens": 314216.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.03438536166032175,
68
+ "grad_norm": 2.032104969024658,
69
+ "learning_rate": 1.9660933660933663e-05,
70
+ "loss": 4.3963,
71
+ "mean_token_accuracy": 0.33228743858635423,
72
+ "num_tokens": 365608.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.039297556183224856,
77
+ "grad_norm": 1.5947270393371582,
78
+ "learning_rate": 1.9611793611793614e-05,
79
+ "loss": 4.2578,
80
+ "mean_token_accuracy": 0.3510261565446854,
81
+ "num_tokens": 418509.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.04420975070612796,
86
+ "grad_norm": 1.9174288511276245,
87
+ "learning_rate": 1.9562653562653564e-05,
88
+ "loss": 4.0457,
89
+ "mean_token_accuracy": 0.38185270577669145,
90
+ "num_tokens": 472608.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.04912194522903107,
95
+ "grad_norm": 1.5742830038070679,
96
+ "learning_rate": 1.9513513513513515e-05,
97
+ "loss": 4.0492,
98
+ "mean_token_accuracy": 0.37742224968969823,
99
+ "num_tokens": 521777.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.054034139751934177,
104
+ "grad_norm": 1.5624850988388062,
105
+ "learning_rate": 1.9464373464373466e-05,
106
+ "loss": 3.7053,
107
+ "mean_token_accuracy": 0.41850126534700394,
108
+ "num_tokens": 580161.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.05894633427483728,
113
+ "grad_norm": 2.199150562286377,
114
+ "learning_rate": 1.9415233415233417e-05,
115
+ "loss": 3.6982,
116
+ "mean_token_accuracy": 0.4235368125140667,
117
+ "num_tokens": 627843.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.06385852879774039,
122
+ "grad_norm": 1.8548036813735962,
123
+ "learning_rate": 1.9366093366093367e-05,
124
+ "loss": 3.3414,
125
+ "mean_token_accuracy": 0.4654690891504288,
126
+ "num_tokens": 681443.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.0687707233206435,
131
+ "grad_norm": 2.78169846534729,
132
+ "learning_rate": 1.9316953316953318e-05,
133
+ "loss": 3.2679,
134
+ "mean_token_accuracy": 0.4711888641119003,
135
+ "num_tokens": 733457.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.07368291784354661,
140
+ "grad_norm": 2.4450321197509766,
141
+ "learning_rate": 1.926781326781327e-05,
142
+ "loss": 3.3377,
143
+ "mean_token_accuracy": 0.4799527756869793,
144
+ "num_tokens": 787769.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.07859511236644971,
149
+ "grad_norm": 1.7717996835708618,
150
+ "learning_rate": 1.921867321867322e-05,
151
+ "loss": 2.8853,
152
+ "mean_token_accuracy": 0.541524414718151,
153
+ "num_tokens": 843485.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.08350730688935282,
158
+ "grad_norm": 2.8402254581451416,
159
+ "learning_rate": 1.9169533169533174e-05,
160
+ "loss": 2.862,
161
+ "mean_token_accuracy": 0.5626954190433026,
162
+ "num_tokens": 894003.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.08841950141225592,
167
+ "grad_norm": 1.5486701726913452,
168
+ "learning_rate": 1.912039312039312e-05,
169
+ "loss": 2.7533,
170
+ "mean_token_accuracy": 0.5848119542002678,
171
+ "num_tokens": 949726.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.09333169593515903,
176
+ "grad_norm": 1.3785852193832397,
177
+ "learning_rate": 1.9071253071253072e-05,
178
+ "loss": 2.7602,
179
+ "mean_token_accuracy": 0.5894295200705528,
180
+ "num_tokens": 1000146.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 0.09824389045806214,
185
+ "grad_norm": 1.6824640035629272,
186
+ "learning_rate": 1.9022113022113026e-05,
187
+ "loss": 2.6345,
188
+ "mean_token_accuracy": 0.5961507339030504,
189
+ "num_tokens": 1050142.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 0.10315608498096525,
194
+ "grad_norm": 1.0604822635650635,
195
+ "learning_rate": 1.8972972972972973e-05,
196
+ "loss": 2.4517,
197
+ "mean_token_accuracy": 0.6296438351273537,
198
+ "num_tokens": 1099646.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 0.10806827950386835,
203
+ "grad_norm": 1.9105230569839478,
204
+ "learning_rate": 1.8923832923832924e-05,
205
+ "loss": 2.4563,
206
+ "mean_token_accuracy": 0.6365545220673084,
207
+ "num_tokens": 1148100.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 0.11298047402677146,
212
+ "grad_norm": 1.41056227684021,
213
+ "learning_rate": 1.8874692874692878e-05,
214
+ "loss": 2.4017,
215
+ "mean_token_accuracy": 0.6691348008811474,
216
+ "num_tokens": 1198987.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.11789266854967456,
221
+ "grad_norm": 1.0573596954345703,
222
+ "learning_rate": 1.8825552825552825e-05,
223
+ "loss": 2.3199,
224
+ "mean_token_accuracy": 0.6731609679758549,
225
+ "num_tokens": 1249335.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 0.12280486307257768,
230
+ "grad_norm": 1.6629279851913452,
231
+ "learning_rate": 1.877641277641278e-05,
232
+ "loss": 1.9335,
233
+ "mean_token_accuracy": 0.7327558636665344,
234
+ "num_tokens": 1303157.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 0.12771705759548077,
239
+ "grad_norm": 1.704005479812622,
240
+ "learning_rate": 1.872727272727273e-05,
241
+ "loss": 2.1897,
242
+ "mean_token_accuracy": 0.6974137671291828,
243
+ "num_tokens": 1355455.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 0.1326292521183839,
248
+ "grad_norm": 1.0553226470947266,
249
+ "learning_rate": 1.8678132678132678e-05,
250
+ "loss": 1.9321,
251
+ "mean_token_accuracy": 0.7327657997608185,
252
+ "num_tokens": 1407508.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 0.137541446641287,
257
+ "grad_norm": 1.101886510848999,
258
+ "learning_rate": 1.8628992628992632e-05,
259
+ "loss": 1.9929,
260
+ "mean_token_accuracy": 0.7272698871791363,
261
+ "num_tokens": 1460595.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 0.1424536411641901,
266
+ "grad_norm": 2.005901575088501,
267
+ "learning_rate": 1.8579852579852583e-05,
268
+ "loss": 1.7147,
269
+ "mean_token_accuracy": 0.7604942478239536,
270
+ "num_tokens": 1512647.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 0.14736583568709322,
275
+ "grad_norm": 7.757036209106445,
276
+ "learning_rate": 1.853071253071253e-05,
277
+ "loss": 1.7571,
278
+ "mean_token_accuracy": 0.7603825606405735,
279
+ "num_tokens": 1565516.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 0.1522780302099963,
284
+ "grad_norm": 0.8850768208503723,
285
+ "learning_rate": 1.8481572481572484e-05,
286
+ "loss": 1.5439,
287
+ "mean_token_accuracy": 0.7915346205234528,
288
+ "num_tokens": 1620194.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 0.15719022473289943,
293
+ "grad_norm": 1.2887756824493408,
294
+ "learning_rate": 1.8432432432432435e-05,
295
+ "loss": 1.5419,
296
+ "mean_token_accuracy": 0.7885693073272705,
297
+ "num_tokens": 1672317.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 0.16210241925580254,
302
+ "grad_norm": 2.562720537185669,
303
+ "learning_rate": 1.8383292383292385e-05,
304
+ "loss": 1.5783,
305
+ "mean_token_accuracy": 0.7889993876218796,
306
+ "num_tokens": 1728513.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 0.16701461377870563,
311
+ "grad_norm": 2.907665491104126,
312
+ "learning_rate": 1.8334152334152336e-05,
313
+ "loss": 1.6573,
314
+ "mean_token_accuracy": 0.7675686649978161,
315
+ "num_tokens": 1777759.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 0.17192680830160875,
320
+ "grad_norm": 0.8430731892585754,
321
+ "learning_rate": 1.8285012285012287e-05,
322
+ "loss": 1.664,
323
+ "mean_token_accuracy": 0.7618721649050713,
324
+ "num_tokens": 1833279.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 0.17683900282451184,
329
+ "grad_norm": 1.1680371761322021,
330
+ "learning_rate": 1.8235872235872238e-05,
331
+ "loss": 1.515,
332
+ "mean_token_accuracy": 0.7789990782737732,
333
+ "num_tokens": 1882531.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 0.18175119734741496,
338
+ "grad_norm": 2.009002208709717,
339
+ "learning_rate": 1.818673218673219e-05,
340
+ "loss": 1.4555,
341
+ "mean_token_accuracy": 0.7983467862010002,
342
+ "num_tokens": 1934619.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 0.18666339187031805,
347
+ "grad_norm": 1.3894531726837158,
348
+ "learning_rate": 1.813759213759214e-05,
349
+ "loss": 1.6206,
350
+ "mean_token_accuracy": 0.7681202955543995,
351
+ "num_tokens": 1987231.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 0.19157558639322117,
356
+ "grad_norm": 0.7266082763671875,
357
+ "learning_rate": 1.808845208845209e-05,
358
+ "loss": 1.3022,
359
+ "mean_token_accuracy": 0.818102091550827,
360
+ "num_tokens": 2039985.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 0.1964877809161243,
365
+ "grad_norm": 0.7502278089523315,
366
+ "learning_rate": 1.803931203931204e-05,
367
+ "loss": 1.3902,
368
+ "mean_token_accuracy": 0.792433574795723,
369
+ "num_tokens": 2092178.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 0.20139997543902738,
374
+ "grad_norm": 0.8366674780845642,
375
+ "learning_rate": 1.799017199017199e-05,
376
+ "loss": 1.5385,
377
+ "mean_token_accuracy": 0.7903823673725128,
378
+ "num_tokens": 2151243.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 0.2063121699619305,
383
+ "grad_norm": 0.7726922035217285,
384
+ "learning_rate": 1.7941031941031942e-05,
385
+ "loss": 1.3529,
386
+ "mean_token_accuracy": 0.8017133295536041,
387
+ "num_tokens": 2201072.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 0.2112243644848336,
392
+ "grad_norm": 0.650055468082428,
393
+ "learning_rate": 1.7891891891891893e-05,
394
+ "loss": 1.4312,
395
+ "mean_token_accuracy": 0.7957115054130555,
396
+ "num_tokens": 2256591.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 0.2161365590077367,
401
+ "grad_norm": 0.9430303573608398,
402
+ "learning_rate": 1.7842751842751843e-05,
403
+ "loss": 1.1938,
404
+ "mean_token_accuracy": 0.8211165018379688,
405
+ "num_tokens": 2308048.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 0.22104875353063982,
410
+ "grad_norm": 0.9191327691078186,
411
+ "learning_rate": 1.7793611793611794e-05,
412
+ "loss": 1.1662,
413
+ "mean_token_accuracy": 0.8298586897552014,
414
+ "num_tokens": 2363573.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 0.22596094805354291,
419
+ "grad_norm": 0.7493109703063965,
420
+ "learning_rate": 1.7744471744471745e-05,
421
+ "loss": 1.5908,
422
+ "mean_token_accuracy": 0.7651749603450299,
423
+ "num_tokens": 2413187.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 0.23087314257644603,
428
+ "grad_norm": 0.5522422790527344,
429
+ "learning_rate": 1.7695331695331696e-05,
430
+ "loss": 1.3707,
431
+ "mean_token_accuracy": 0.7843399420380592,
432
+ "num_tokens": 2468041.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 0.23578533709934912,
437
+ "grad_norm": 0.6577270030975342,
438
+ "learning_rate": 1.764619164619165e-05,
439
+ "loss": 1.4679,
440
+ "mean_token_accuracy": 0.7856365032494068,
441
+ "num_tokens": 2517753.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 0.24069753162225224,
446
+ "grad_norm": 0.9699747562408447,
447
+ "learning_rate": 1.7597051597051597e-05,
448
+ "loss": 1.3493,
449
+ "mean_token_accuracy": 0.8016418114304542,
450
+ "num_tokens": 2568072.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 0.24560972614515536,
455
+ "grad_norm": 2.0955657958984375,
456
+ "learning_rate": 1.7547911547911548e-05,
457
+ "loss": 0.9634,
458
+ "mean_token_accuracy": 0.8412673205137253,
459
+ "num_tokens": 2617143.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 0.25052192066805845,
464
+ "grad_norm": 0.7577691078186035,
465
+ "learning_rate": 1.7498771498771502e-05,
466
+ "loss": 1.0889,
467
+ "mean_token_accuracy": 0.8479016363620758,
468
+ "num_tokens": 2669318.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 0.25543411519096154,
473
+ "grad_norm": 0.5891643762588501,
474
+ "learning_rate": 1.744963144963145e-05,
475
+ "loss": 1.1811,
476
+ "mean_token_accuracy": 0.82891755849123,
477
+ "num_tokens": 2723151.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 0.2603463097138647,
482
+ "grad_norm": 0.7815893292427063,
483
+ "learning_rate": 1.74004914004914e-05,
484
+ "loss": 1.3488,
485
+ "mean_token_accuracy": 0.7971963949501515,
486
+ "num_tokens": 2774274.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 0.2652585042367678,
491
+ "grad_norm": 0.365824818611145,
492
+ "learning_rate": 1.7351351351351354e-05,
493
+ "loss": 1.129,
494
+ "mean_token_accuracy": 0.8214888103306294,
495
+ "num_tokens": 2825301.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 0.27017069875967087,
500
+ "grad_norm": 0.5682836174964905,
501
+ "learning_rate": 1.73022113022113e-05,
502
+ "loss": 1.0752,
503
+ "mean_token_accuracy": 0.8513655051589012,
504
+ "num_tokens": 2877767.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 0.275082893282574,
509
+ "grad_norm": 0.9484438300132751,
510
+ "learning_rate": 1.7253071253071256e-05,
511
+ "loss": 1.1606,
512
+ "mean_token_accuracy": 0.8191152907907963,
513
+ "num_tokens": 2929216.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 0.2799950878054771,
518
+ "grad_norm": 0.7097403407096863,
519
+ "learning_rate": 1.7203931203931206e-05,
520
+ "loss": 1.0245,
521
+ "mean_token_accuracy": 0.8594455644488335,
522
+ "num_tokens": 2983287.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 0.2849072823283802,
527
+ "grad_norm": 2.197983741760254,
528
+ "learning_rate": 1.7154791154791157e-05,
529
+ "loss": 1.0116,
530
+ "mean_token_accuracy": 0.8458577767014503,
531
+ "num_tokens": 3036426.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 0.2898194768512833,
536
+ "grad_norm": 0.9750831723213196,
537
+ "learning_rate": 1.7105651105651108e-05,
538
+ "loss": 1.2482,
539
+ "mean_token_accuracy": 0.8097934015095234,
540
+ "num_tokens": 3090471.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 0.29473167137418643,
545
+ "grad_norm": 0.7452868223190308,
546
+ "learning_rate": 1.705651105651106e-05,
547
+ "loss": 0.9643,
548
+ "mean_token_accuracy": 0.8562554508447647,
549
+ "num_tokens": 3143095.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 0.2996438658970895,
554
+ "grad_norm": 0.3684370815753937,
555
+ "learning_rate": 1.700737100737101e-05,
556
+ "loss": 0.8587,
557
+ "mean_token_accuracy": 0.8684959009289741,
558
+ "num_tokens": 3193778.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 0.3045560604199926,
563
+ "grad_norm": 0.4814823567867279,
564
+ "learning_rate": 1.695823095823096e-05,
565
+ "loss": 0.8246,
566
+ "mean_token_accuracy": 0.8646811529994011,
567
+ "num_tokens": 3249776.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 0.30946825494289576,
572
+ "grad_norm": 0.6298699378967285,
573
+ "learning_rate": 1.690909090909091e-05,
574
+ "loss": 1.0043,
575
+ "mean_token_accuracy": 0.849489164352417,
576
+ "num_tokens": 3301301.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 0.31438044946579885,
581
+ "grad_norm": 0.6802086234092712,
582
+ "learning_rate": 1.685995085995086e-05,
583
+ "loss": 0.9932,
584
+ "mean_token_accuracy": 0.8412040494382381,
585
+ "num_tokens": 3351834.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 0.31929264398870194,
590
+ "grad_norm": 0.4441145062446594,
591
+ "learning_rate": 1.6810810810810812e-05,
592
+ "loss": 0.9093,
593
+ "mean_token_accuracy": 0.8644197151064873,
594
+ "num_tokens": 3402630.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 0.3242048385116051,
599
+ "grad_norm": 0.5664909482002258,
600
+ "learning_rate": 1.6761670761670763e-05,
601
+ "loss": 1.0511,
602
+ "mean_token_accuracy": 0.8302886992692947,
603
+ "num_tokens": 3453600.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 0.3291170330345082,
608
+ "grad_norm": 0.898231029510498,
609
+ "learning_rate": 1.6712530712530714e-05,
610
+ "loss": 1.2663,
611
+ "mean_token_accuracy": 0.7889795318245888,
612
+ "num_tokens": 3501154.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 0.33402922755741127,
617
+ "grad_norm": 0.49926525354385376,
618
+ "learning_rate": 1.6663390663390664e-05,
619
+ "loss": 1.0539,
620
+ "mean_token_accuracy": 0.8462485030293465,
621
+ "num_tokens": 3552076.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 0.33894142208031436,
626
+ "grad_norm": 0.5327655673027039,
627
+ "learning_rate": 1.6614250614250615e-05,
628
+ "loss": 1.0757,
629
+ "mean_token_accuracy": 0.8260165579617024,
630
+ "num_tokens": 3602651.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 0.3438536166032175,
635
+ "grad_norm": 0.5295992493629456,
636
+ "learning_rate": 1.6565110565110566e-05,
637
+ "loss": 0.9826,
638
+ "mean_token_accuracy": 0.8445935681462288,
639
+ "num_tokens": 3656310.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 0.3487658111261206,
644
+ "grad_norm": 0.584396243095398,
645
+ "learning_rate": 1.6515970515970517e-05,
646
+ "loss": 0.9771,
647
+ "mean_token_accuracy": 0.8513481065630912,
648
+ "num_tokens": 3712764.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 0.3536780056490237,
653
+ "grad_norm": 0.5786782503128052,
654
+ "learning_rate": 1.6466830466830467e-05,
655
+ "loss": 1.0812,
656
+ "mean_token_accuracy": 0.8179432228207588,
657
+ "num_tokens": 3761995.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 0.35859020017192683,
662
+ "grad_norm": 0.3496871888637543,
663
+ "learning_rate": 1.6417690417690418e-05,
664
+ "loss": 1.0407,
665
+ "mean_token_accuracy": 0.8126879334449768,
666
+ "num_tokens": 3811493.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 0.3635023946948299,
671
+ "grad_norm": 0.5038619041442871,
672
+ "learning_rate": 1.636855036855037e-05,
673
+ "loss": 1.1602,
674
+ "mean_token_accuracy": 0.8141254656016826,
675
+ "num_tokens": 3859376.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 0.368414589217733,
680
+ "grad_norm": 0.4643881320953369,
681
+ "learning_rate": 1.631941031941032e-05,
682
+ "loss": 0.8819,
683
+ "mean_token_accuracy": 0.8545275427401066,
684
+ "num_tokens": 3912052.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 0.3733267837406361,
689
+ "grad_norm": 0.4222355782985687,
690
+ "learning_rate": 1.6270270270270274e-05,
691
+ "loss": 0.8333,
692
+ "mean_token_accuracy": 0.8552621826529503,
693
+ "num_tokens": 3965078.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 0.37823897826353925,
698
+ "grad_norm": 0.4513651728630066,
699
+ "learning_rate": 1.6221130221130224e-05,
700
+ "loss": 1.0386,
701
+ "mean_token_accuracy": 0.8259748943150044,
702
+ "num_tokens": 4015489.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 0.38315117278644234,
707
+ "grad_norm": 0.5292590856552124,
708
+ "learning_rate": 1.6171990171990172e-05,
709
+ "loss": 0.8968,
710
+ "mean_token_accuracy": 0.8551326602697372,
711
+ "num_tokens": 4067831.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 0.38806336730934543,
716
+ "grad_norm": 0.49815845489501953,
717
+ "learning_rate": 1.6122850122850126e-05,
718
+ "loss": 1.0808,
719
+ "mean_token_accuracy": 0.8252614364027977,
720
+ "num_tokens": 4117945.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 0.3929755618322486,
725
+ "grad_norm": 0.339728981256485,
726
+ "learning_rate": 1.6073710073710077e-05,
727
+ "loss": 0.9569,
728
+ "mean_token_accuracy": 0.8419950112700463,
729
+ "num_tokens": 4171138.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 0.39788775635515167,
734
+ "grad_norm": 0.4530418813228607,
735
+ "learning_rate": 1.6024570024570024e-05,
736
+ "loss": 1.1355,
737
+ "mean_token_accuracy": 0.8250428631901741,
738
+ "num_tokens": 4223765.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 0.40279995087805476,
743
+ "grad_norm": 0.5202032923698425,
744
+ "learning_rate": 1.5975429975429978e-05,
745
+ "loss": 0.8306,
746
+ "mean_token_accuracy": 0.865328436344862,
747
+ "num_tokens": 4278056.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 0.4077121454009579,
752
+ "grad_norm": 0.4732300043106079,
753
+ "learning_rate": 1.592628992628993e-05,
754
+ "loss": 1.3396,
755
+ "mean_token_accuracy": 0.783781211823225,
756
+ "num_tokens": 4326704.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 0.412624339923861,
761
+ "grad_norm": 0.49258068203926086,
762
+ "learning_rate": 1.5877149877149876e-05,
763
+ "loss": 0.8462,
764
+ "mean_token_accuracy": 0.8517311483621597,
765
+ "num_tokens": 4378538.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 0.4175365344467641,
770
+ "grad_norm": 0.8045883774757385,
771
+ "learning_rate": 1.582800982800983e-05,
772
+ "loss": 0.9785,
773
+ "mean_token_accuracy": 0.8369067624211312,
774
+ "num_tokens": 4429893.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 0.4224487289696672,
779
+ "grad_norm": 0.361854612827301,
780
+ "learning_rate": 1.577886977886978e-05,
781
+ "loss": 0.936,
782
+ "mean_token_accuracy": 0.8446702107787132,
783
+ "num_tokens": 4481594.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 0.4273609234925703,
788
+ "grad_norm": 0.593479573726654,
789
+ "learning_rate": 1.572972972972973e-05,
790
+ "loss": 0.8245,
791
+ "mean_token_accuracy": 0.8750210002064704,
792
+ "num_tokens": 4534098.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 0.4322731180154734,
797
+ "grad_norm": 0.31826287508010864,
798
+ "learning_rate": 1.5680589680589682e-05,
799
+ "loss": 0.94,
800
+ "mean_token_accuracy": 0.8408755764365197,
801
+ "num_tokens": 4587111.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 0.4371853125383765,
806
+ "grad_norm": 1.4076229333877563,
807
+ "learning_rate": 1.5631449631449633e-05,
808
+ "loss": 1.0176,
809
+ "mean_token_accuracy": 0.8321189761161805,
810
+ "num_tokens": 4640388.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 0.44209750706127965,
815
+ "grad_norm": 0.6677824854850769,
816
+ "learning_rate": 1.5582309582309584e-05,
817
+ "loss": 1.2891,
818
+ "mean_token_accuracy": 0.7943175598978996,
819
+ "num_tokens": 4691468.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 0.44700970158418274,
824
+ "grad_norm": 0.4233896732330322,
825
+ "learning_rate": 1.5533169533169535e-05,
826
+ "loss": 1.1053,
827
+ "mean_token_accuracy": 0.8279901027679444,
828
+ "num_tokens": 4741844.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 0.45192189610708583,
833
+ "grad_norm": 0.35591593384742737,
834
+ "learning_rate": 1.5484029484029485e-05,
835
+ "loss": 0.8832,
836
+ "mean_token_accuracy": 0.8491488620638847,
837
+ "num_tokens": 4793741.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 0.4568340906299889,
842
+ "grad_norm": 0.6583911180496216,
843
+ "learning_rate": 1.5434889434889436e-05,
844
+ "loss": 0.8387,
845
+ "mean_token_accuracy": 0.8648917526006699,
846
+ "num_tokens": 4848466.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 0.46174628515289207,
851
+ "grad_norm": 0.4065934717655182,
852
+ "learning_rate": 1.5385749385749387e-05,
853
+ "loss": 0.9242,
854
+ "mean_token_accuracy": 0.8412024281919003,
855
+ "num_tokens": 4901992.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 0.46665847967579516,
860
+ "grad_norm": 0.3458426594734192,
861
+ "learning_rate": 1.5336609336609338e-05,
862
+ "loss": 0.8545,
863
+ "mean_token_accuracy": 0.855231161415577,
864
+ "num_tokens": 4956964.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 0.47157067419869825,
869
+ "grad_norm": 0.6710091233253479,
870
+ "learning_rate": 1.5287469287469288e-05,
871
+ "loss": 0.9583,
872
+ "mean_token_accuracy": 0.8393168188631535,
873
+ "num_tokens": 5011287.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 0.4764828687216014,
878
+ "grad_norm": 0.3753456771373749,
879
+ "learning_rate": 1.5238329238329239e-05,
880
+ "loss": 0.9738,
881
+ "mean_token_accuracy": 0.8375437319278717,
882
+ "num_tokens": 5061714.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 0.4813950632445045,
887
+ "grad_norm": 1.2011836767196655,
888
+ "learning_rate": 1.5189189189189191e-05,
889
+ "loss": 1.0972,
890
+ "mean_token_accuracy": 0.8070526115596295,
891
+ "num_tokens": 5114645.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 0.4863072577674076,
896
+ "grad_norm": 0.37889617681503296,
897
+ "learning_rate": 1.5140049140049142e-05,
898
+ "loss": 0.9442,
899
+ "mean_token_accuracy": 0.8458199493587018,
900
+ "num_tokens": 5165485.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 0.4912194522903107,
905
+ "grad_norm": 0.6613748669624329,
906
+ "learning_rate": 1.5090909090909091e-05,
907
+ "loss": 1.0196,
908
+ "mean_token_accuracy": 0.8272999472916126,
909
+ "num_tokens": 5216189.0,
910
+ "step": 1000
911
+ },
912
+ {
913
+ "epoch": 0.4961316468132138,
914
+ "grad_norm": 0.7825655937194824,
915
+ "learning_rate": 1.5041769041769044e-05,
916
+ "loss": 1.011,
917
+ "mean_token_accuracy": 0.8262499138712883,
918
+ "num_tokens": 5269248.0,
919
+ "step": 1010
920
+ },
921
+ {
922
+ "epoch": 0.5010438413361169,
923
+ "grad_norm": 0.2106683850288391,
924
+ "learning_rate": 1.4992628992628994e-05,
925
+ "loss": 0.9098,
926
+ "mean_token_accuracy": 0.8512444771826267,
927
+ "num_tokens": 5322524.0,
928
+ "step": 1020
929
+ },
930
+ {
931
+ "epoch": 0.50595603585902,
932
+ "grad_norm": 0.8093978762626648,
933
+ "learning_rate": 1.4943488943488943e-05,
934
+ "loss": 1.2403,
935
+ "mean_token_accuracy": 0.7969906754791737,
936
+ "num_tokens": 5370123.0,
937
+ "step": 1030
938
+ },
939
+ {
940
+ "epoch": 0.5108682303819231,
941
+ "grad_norm": 0.5412647724151611,
942
+ "learning_rate": 1.4894348894348896e-05,
943
+ "loss": 0.8821,
944
+ "mean_token_accuracy": 0.8460625484585762,
945
+ "num_tokens": 5422125.0,
946
+ "step": 1040
947
+ },
948
+ {
949
+ "epoch": 0.5157804249048262,
950
+ "grad_norm": 0.5411800146102905,
951
+ "learning_rate": 1.4845208845208847e-05,
952
+ "loss": 1.0804,
953
+ "mean_token_accuracy": 0.8199978180229663,
954
+ "num_tokens": 5472127.0,
955
+ "step": 1050
956
+ },
957
+ {
958
+ "epoch": 0.5206926194277294,
959
+ "grad_norm": 0.4349406063556671,
960
+ "learning_rate": 1.4796068796068797e-05,
961
+ "loss": 0.7559,
962
+ "mean_token_accuracy": 0.8662853054702282,
963
+ "num_tokens": 5525028.0,
964
+ "step": 1060
965
+ },
966
+ {
967
+ "epoch": 0.5256048139506324,
968
+ "grad_norm": 0.4738285541534424,
969
+ "learning_rate": 1.4746928746928748e-05,
970
+ "loss": 1.1225,
971
+ "mean_token_accuracy": 0.8103010386228562,
972
+ "num_tokens": 5570859.0,
973
+ "step": 1070
974
+ },
975
+ {
976
+ "epoch": 0.5305170084735356,
977
+ "grad_norm": 0.6644457578659058,
978
+ "learning_rate": 1.46977886977887e-05,
979
+ "loss": 0.9334,
980
+ "mean_token_accuracy": 0.8213524237275124,
981
+ "num_tokens": 5618933.0,
982
+ "step": 1080
983
+ },
984
+ {
985
+ "epoch": 0.5354292029964387,
986
+ "grad_norm": 0.3038887083530426,
987
+ "learning_rate": 1.464864864864865e-05,
988
+ "loss": 1.0292,
989
+ "mean_token_accuracy": 0.810487725585699,
990
+ "num_tokens": 5664922.0,
991
+ "step": 1090
992
+ },
993
+ {
994
+ "epoch": 0.5403413975193417,
995
+ "grad_norm": 0.26083576679229736,
996
+ "learning_rate": 1.45995085995086e-05,
997
+ "loss": 0.9084,
998
+ "mean_token_accuracy": 0.836492794752121,
999
+ "num_tokens": 5716671.0,
1000
+ "step": 1100
1001
+ },
1002
+ {
1003
+ "epoch": 0.5452535920422449,
1004
+ "grad_norm": 0.36848321557044983,
1005
+ "learning_rate": 1.4550368550368553e-05,
1006
+ "loss": 1.1544,
1007
+ "mean_token_accuracy": 0.8033395648002625,
1008
+ "num_tokens": 5764576.0,
1009
+ "step": 1110
1010
+ },
1011
+ {
1012
+ "epoch": 0.550165786565148,
1013
+ "grad_norm": 0.3338620662689209,
1014
+ "learning_rate": 1.4501228501228502e-05,
1015
+ "loss": 1.0114,
1016
+ "mean_token_accuracy": 0.8188160583376884,
1017
+ "num_tokens": 5816699.0,
1018
+ "step": 1120
1019
+ },
1020
+ {
1021
+ "epoch": 0.5550779810880511,
1022
+ "grad_norm": 0.3278094232082367,
1023
+ "learning_rate": 1.4452088452088452e-05,
1024
+ "loss": 0.6189,
1025
+ "mean_token_accuracy": 0.8895823881030083,
1026
+ "num_tokens": 5874257.0,
1027
+ "step": 1130
1028
+ },
1029
+ {
1030
+ "epoch": 0.5599901756109542,
1031
+ "grad_norm": 0.4236205220222473,
1032
+ "learning_rate": 1.4402948402948405e-05,
1033
+ "loss": 0.9474,
1034
+ "mean_token_accuracy": 0.8435555145144462,
1035
+ "num_tokens": 5925628.0,
1036
+ "step": 1140
1037
+ },
1038
+ {
1039
+ "epoch": 0.5649023701338572,
1040
+ "grad_norm": 0.21693912148475647,
1041
+ "learning_rate": 1.4353808353808354e-05,
1042
+ "loss": 0.9109,
1043
+ "mean_token_accuracy": 0.842586936801672,
1044
+ "num_tokens": 5977130.0,
1045
+ "step": 1150
1046
+ },
1047
+ {
1048
+ "epoch": 0.5698145646567604,
1049
+ "grad_norm": 0.6288137435913086,
1050
+ "learning_rate": 1.4304668304668306e-05,
1051
+ "loss": 0.7005,
1052
+ "mean_token_accuracy": 0.877720920741558,
1053
+ "num_tokens": 6032405.0,
1054
+ "step": 1160
1055
+ },
1056
+ {
1057
+ "epoch": 0.5747267591796635,
1058
+ "grad_norm": 0.29411551356315613,
1059
+ "learning_rate": 1.4255528255528257e-05,
1060
+ "loss": 0.9163,
1061
+ "mean_token_accuracy": 0.8345251567661762,
1062
+ "num_tokens": 6081769.0,
1063
+ "step": 1170
1064
+ },
1065
+ {
1066
+ "epoch": 0.5796389537025666,
1067
+ "grad_norm": 0.3542836308479309,
1068
+ "learning_rate": 1.420638820638821e-05,
1069
+ "loss": 0.8983,
1070
+ "mean_token_accuracy": 0.8478083908557892,
1071
+ "num_tokens": 6131660.0,
1072
+ "step": 1180
1073
+ },
1074
+ {
1075
+ "epoch": 0.5845511482254697,
1076
+ "grad_norm": 0.6326059103012085,
1077
+ "learning_rate": 1.4157248157248158e-05,
1078
+ "loss": 1.055,
1079
+ "mean_token_accuracy": 0.8133970260620117,
1080
+ "num_tokens": 6177706.0,
1081
+ "step": 1190
1082
+ },
1083
+ {
1084
+ "epoch": 0.5894633427483729,
1085
+ "grad_norm": 0.39950093626976013,
1086
+ "learning_rate": 1.410810810810811e-05,
1087
+ "loss": 0.6446,
1088
+ "mean_token_accuracy": 0.8883207038044929,
1089
+ "num_tokens": 6231530.0,
1090
+ "step": 1200
1091
+ },
1092
+ {
1093
+ "epoch": 0.5943755372712759,
1094
+ "grad_norm": 1.4949785470962524,
1095
+ "learning_rate": 1.4058968058968062e-05,
1096
+ "loss": 0.8527,
1097
+ "mean_token_accuracy": 0.8495218820869923,
1098
+ "num_tokens": 6284936.0,
1099
+ "step": 1210
1100
+ },
1101
+ {
1102
+ "epoch": 0.599287731794179,
1103
+ "grad_norm": 0.5013307332992554,
1104
+ "learning_rate": 1.400982800982801e-05,
1105
+ "loss": 0.9342,
1106
+ "mean_token_accuracy": 0.8461359746754169,
1107
+ "num_tokens": 6338328.0,
1108
+ "step": 1220
1109
+ },
1110
+ {
1111
+ "epoch": 0.6041999263170822,
1112
+ "grad_norm": 0.4373924136161804,
1113
+ "learning_rate": 1.3960687960687961e-05,
1114
+ "loss": 0.7957,
1115
+ "mean_token_accuracy": 0.8590179920196533,
1116
+ "num_tokens": 6391247.0,
1117
+ "step": 1230
1118
+ },
1119
+ {
1120
+ "epoch": 0.6091121208399852,
1121
+ "grad_norm": 0.30794695019721985,
1122
+ "learning_rate": 1.3911547911547914e-05,
1123
+ "loss": 0.9578,
1124
+ "mean_token_accuracy": 0.8326818406581878,
1125
+ "num_tokens": 6444004.0,
1126
+ "step": 1240
1127
+ },
1128
+ {
1129
+ "epoch": 0.6140243153628884,
1130
+ "grad_norm": 0.3918187618255615,
1131
+ "learning_rate": 1.3862407862407863e-05,
1132
+ "loss": 0.8732,
1133
+ "mean_token_accuracy": 0.8658828616142273,
1134
+ "num_tokens": 6495578.0,
1135
+ "step": 1250
1136
+ },
1137
+ {
1138
+ "epoch": 0.6189365098857915,
1139
+ "grad_norm": 0.2752901315689087,
1140
+ "learning_rate": 1.3813267813267814e-05,
1141
+ "loss": 0.9506,
1142
+ "mean_token_accuracy": 0.8301159009337425,
1143
+ "num_tokens": 6545604.0,
1144
+ "step": 1260
1145
+ },
1146
+ {
1147
+ "epoch": 0.6238487044086946,
1148
+ "grad_norm": 0.460627943277359,
1149
+ "learning_rate": 1.3764127764127766e-05,
1150
+ "loss": 1.0007,
1151
+ "mean_token_accuracy": 0.8227074682712555,
1152
+ "num_tokens": 6594381.0,
1153
+ "step": 1270
1154
+ },
1155
+ {
1156
+ "epoch": 0.6287608989315977,
1157
+ "grad_norm": 0.4065866768360138,
1158
+ "learning_rate": 1.3714987714987715e-05,
1159
+ "loss": 0.8056,
1160
+ "mean_token_accuracy": 0.8646926179528236,
1161
+ "num_tokens": 6649190.0,
1162
+ "step": 1280
1163
+ },
1164
+ {
1165
+ "epoch": 0.6336730934545008,
1166
+ "grad_norm": 0.4343368113040924,
1167
+ "learning_rate": 1.3665847665847667e-05,
1168
+ "loss": 0.8879,
1169
+ "mean_token_accuracy": 0.8385777972638607,
1170
+ "num_tokens": 6701394.0,
1171
+ "step": 1290
1172
+ },
1173
+ {
1174
+ "epoch": 0.6385852879774039,
1175
+ "grad_norm": 0.5364151000976562,
1176
+ "learning_rate": 1.3616707616707618e-05,
1177
+ "loss": 0.9419,
1178
+ "mean_token_accuracy": 0.8308121666312218,
1179
+ "num_tokens": 6752270.0,
1180
+ "step": 1300
1181
+ },
1182
+ {
1183
+ "epoch": 0.643497482500307,
1184
+ "grad_norm": 0.3982967734336853,
1185
+ "learning_rate": 1.3567567567567567e-05,
1186
+ "loss": 1.0305,
1187
+ "mean_token_accuracy": 0.8274823799729347,
1188
+ "num_tokens": 6805857.0,
1189
+ "step": 1310
1190
+ },
1191
+ {
1192
+ "epoch": 0.6484096770232102,
1193
+ "grad_norm": 0.7117066383361816,
1194
+ "learning_rate": 1.351842751842752e-05,
1195
+ "loss": 1.0594,
1196
+ "mean_token_accuracy": 0.8103688299655915,
1197
+ "num_tokens": 6852455.0,
1198
+ "step": 1320
1199
+ },
1200
+ {
1201
+ "epoch": 0.6533218715461132,
1202
+ "grad_norm": 0.3083662986755371,
1203
+ "learning_rate": 1.346928746928747e-05,
1204
+ "loss": 1.0412,
1205
+ "mean_token_accuracy": 0.8186712831258773,
1206
+ "num_tokens": 6896950.0,
1207
+ "step": 1330
1208
+ },
1209
+ {
1210
+ "epoch": 0.6582340660690164,
1211
+ "grad_norm": 0.4959128499031067,
1212
+ "learning_rate": 1.342014742014742e-05,
1213
+ "loss": 0.9348,
1214
+ "mean_token_accuracy": 0.8453299880027771,
1215
+ "num_tokens": 6950814.0,
1216
+ "step": 1340
1217
+ },
1218
+ {
1219
+ "epoch": 0.6631462605919194,
1220
+ "grad_norm": 0.8182358145713806,
1221
+ "learning_rate": 1.3371007371007372e-05,
1222
+ "loss": 1.0552,
1223
+ "mean_token_accuracy": 0.8226786606013775,
1224
+ "num_tokens": 7003460.0,
1225
+ "step": 1350
1226
+ },
1227
+ {
1228
+ "epoch": 0.6680584551148225,
1229
+ "grad_norm": 0.43365928530693054,
1230
+ "learning_rate": 1.3321867321867323e-05,
1231
+ "loss": 0.8795,
1232
+ "mean_token_accuracy": 0.8458721756935119,
1233
+ "num_tokens": 7054901.0,
1234
+ "step": 1360
1235
+ },
1236
+ {
1237
+ "epoch": 0.6729706496377257,
1238
+ "grad_norm": 0.2876506745815277,
1239
+ "learning_rate": 1.3272727272727275e-05,
1240
+ "loss": 0.8282,
1241
+ "mean_token_accuracy": 0.8459444969892502,
1242
+ "num_tokens": 7106038.0,
1243
+ "step": 1370
1244
+ },
1245
+ {
1246
+ "epoch": 0.6778828441606287,
1247
+ "grad_norm": 0.4010944068431854,
1248
+ "learning_rate": 1.3223587223587224e-05,
1249
+ "loss": 0.7836,
1250
+ "mean_token_accuracy": 0.8610207386314869,
1251
+ "num_tokens": 7159651.0,
1252
+ "step": 1380
1253
+ },
1254
+ {
1255
+ "epoch": 0.6827950386835319,
1256
+ "grad_norm": 0.19336983561515808,
1257
+ "learning_rate": 1.3174447174447176e-05,
1258
+ "loss": 1.0039,
1259
+ "mean_token_accuracy": 0.8369563437998295,
1260
+ "num_tokens": 7204811.0,
1261
+ "step": 1390
1262
+ },
1263
+ {
1264
+ "epoch": 0.687707233206435,
1265
+ "grad_norm": 0.5056695938110352,
1266
+ "learning_rate": 1.3125307125307127e-05,
1267
+ "loss": 0.6854,
1268
+ "mean_token_accuracy": 0.878055976331234,
1269
+ "num_tokens": 7256836.0,
1270
+ "step": 1400
1271
+ },
1272
+ {
1273
+ "epoch": 0.692619427729338,
1274
+ "grad_norm": 0.3462684750556946,
1275
+ "learning_rate": 1.3076167076167076e-05,
1276
+ "loss": 1.2627,
1277
+ "mean_token_accuracy": 0.7998166553676128,
1278
+ "num_tokens": 7308913.0,
1279
+ "step": 1410
1280
+ },
1281
+ {
1282
+ "epoch": 0.6975316222522412,
1283
+ "grad_norm": 0.2867954671382904,
1284
+ "learning_rate": 1.3027027027027029e-05,
1285
+ "loss": 0.6718,
1286
+ "mean_token_accuracy": 0.8842718601226807,
1287
+ "num_tokens": 7365814.0,
1288
+ "step": 1420
1289
+ },
1290
+ {
1291
+ "epoch": 0.7024438167751443,
1292
+ "grad_norm": 0.4659688472747803,
1293
+ "learning_rate": 1.297788697788698e-05,
1294
+ "loss": 1.1291,
1295
+ "mean_token_accuracy": 0.8083386868238449,
1296
+ "num_tokens": 7419238.0,
1297
+ "step": 1430
1298
+ },
1299
+ {
1300
+ "epoch": 0.7073560112980474,
1301
+ "grad_norm": 0.41312354803085327,
1302
+ "learning_rate": 1.2928746928746928e-05,
1303
+ "loss": 0.8625,
1304
+ "mean_token_accuracy": 0.8470474675297737,
1305
+ "num_tokens": 7469177.0,
1306
+ "step": 1440
1307
+ },
1308
+ {
1309
+ "epoch": 0.7122682058209505,
1310
+ "grad_norm": 0.5044501423835754,
1311
+ "learning_rate": 1.2879606879606881e-05,
1312
+ "loss": 0.8186,
1313
+ "mean_token_accuracy": 0.8533941991627216,
1314
+ "num_tokens": 7520617.0,
1315
+ "step": 1450
1316
+ },
1317
+ {
1318
+ "epoch": 0.7171804003438537,
1319
+ "grad_norm": 0.3712249994277954,
1320
+ "learning_rate": 1.2830466830466832e-05,
1321
+ "loss": 0.912,
1322
+ "mean_token_accuracy": 0.838789189606905,
1323
+ "num_tokens": 7569361.0,
1324
+ "step": 1460
1325
+ },
1326
+ {
1327
+ "epoch": 0.7220925948667567,
1328
+ "grad_norm": 0.5090559720993042,
1329
+ "learning_rate": 1.2781326781326782e-05,
1330
+ "loss": 0.9559,
1331
+ "mean_token_accuracy": 0.8298265844583511,
1332
+ "num_tokens": 7619902.0,
1333
+ "step": 1470
1334
+ },
1335
+ {
1336
+ "epoch": 0.7270047893896598,
1337
+ "grad_norm": 0.40056052803993225,
1338
+ "learning_rate": 1.2732186732186733e-05,
1339
+ "loss": 0.9697,
1340
+ "mean_token_accuracy": 0.8409195326268672,
1341
+ "num_tokens": 7671364.0,
1342
+ "step": 1480
1343
+ },
1344
+ {
1345
+ "epoch": 0.731916983912563,
1346
+ "grad_norm": 0.5502124428749084,
1347
+ "learning_rate": 1.2683046683046685e-05,
1348
+ "loss": 0.974,
1349
+ "mean_token_accuracy": 0.8395135879516602,
1350
+ "num_tokens": 7721501.0,
1351
+ "step": 1490
1352
+ },
1353
+ {
1354
+ "epoch": 0.736829178435466,
1355
+ "grad_norm": 0.49662184715270996,
1356
+ "learning_rate": 1.2633906633906635e-05,
1357
+ "loss": 0.9764,
1358
+ "mean_token_accuracy": 0.8396995335817337,
1359
+ "num_tokens": 7771721.0,
1360
+ "step": 1500
1361
+ },
1362
+ {
1363
+ "epoch": 0.7417413729583692,
1364
+ "grad_norm": 0.26574891805648804,
1365
+ "learning_rate": 1.2584766584766585e-05,
1366
+ "loss": 0.8165,
1367
+ "mean_token_accuracy": 0.8401172868907452,
1368
+ "num_tokens": 7822046.0,
1369
+ "step": 1510
1370
+ },
1371
+ {
1372
+ "epoch": 0.7466535674812722,
1373
+ "grad_norm": 0.6338578462600708,
1374
+ "learning_rate": 1.2535626535626538e-05,
1375
+ "loss": 0.9305,
1376
+ "mean_token_accuracy": 0.8319102026522159,
1377
+ "num_tokens": 7874800.0,
1378
+ "step": 1520
1379
+ },
1380
+ {
1381
+ "epoch": 0.7515657620041754,
1382
+ "grad_norm": 0.45722728967666626,
1383
+ "learning_rate": 1.2486486486486487e-05,
1384
+ "loss": 0.817,
1385
+ "mean_token_accuracy": 0.865605479478836,
1386
+ "num_tokens": 7928044.0,
1387
+ "step": 1530
1388
+ },
1389
+ {
1390
+ "epoch": 0.7564779565270785,
1391
+ "grad_norm": 0.2256188988685608,
1392
+ "learning_rate": 1.2437346437346437e-05,
1393
+ "loss": 1.0366,
1394
+ "mean_token_accuracy": 0.8243134960532188,
1395
+ "num_tokens": 7976774.0,
1396
+ "step": 1540
1397
+ },
1398
+ {
1399
+ "epoch": 0.7613901510499815,
1400
+ "grad_norm": 0.3194807469844818,
1401
+ "learning_rate": 1.238820638820639e-05,
1402
+ "loss": 0.8638,
1403
+ "mean_token_accuracy": 0.8425570398569107,
1404
+ "num_tokens": 8027431.0,
1405
+ "step": 1550
1406
+ },
1407
+ {
1408
+ "epoch": 0.7663023455728847,
1409
+ "grad_norm": 0.7870245575904846,
1410
+ "learning_rate": 1.2339066339066339e-05,
1411
+ "loss": 1.1827,
1412
+ "mean_token_accuracy": 0.8022145099937916,
1413
+ "num_tokens": 8076248.0,
1414
+ "step": 1560
1415
+ },
1416
+ {
1417
+ "epoch": 0.7712145400957878,
1418
+ "grad_norm": 1.0986965894699097,
1419
+ "learning_rate": 1.2289926289926291e-05,
1420
+ "loss": 1.0736,
1421
+ "mean_token_accuracy": 0.8118120074272156,
1422
+ "num_tokens": 8128333.0,
1423
+ "step": 1570
1424
+ },
1425
+ {
1426
+ "epoch": 0.7761267346186909,
1427
+ "grad_norm": 0.6505934000015259,
1428
+ "learning_rate": 1.2240786240786242e-05,
1429
+ "loss": 0.957,
1430
+ "mean_token_accuracy": 0.8350290179252624,
1431
+ "num_tokens": 8177196.0,
1432
+ "step": 1580
1433
+ },
1434
+ {
1435
+ "epoch": 0.781038929141594,
1436
+ "grad_norm": 0.35571786761283875,
1437
+ "learning_rate": 1.2191646191646194e-05,
1438
+ "loss": 0.8929,
1439
+ "mean_token_accuracy": 0.8486527316272259,
1440
+ "num_tokens": 8229113.0,
1441
+ "step": 1590
1442
+ },
1443
+ {
1444
+ "epoch": 0.7859511236644972,
1445
+ "grad_norm": 0.4012582004070282,
1446
+ "learning_rate": 1.2142506142506144e-05,
1447
+ "loss": 0.7953,
1448
+ "mean_token_accuracy": 0.8502931199967861,
1449
+ "num_tokens": 8283001.0,
1450
+ "step": 1600
1451
+ },
1452
+ {
1453
+ "epoch": 0.7908633181874002,
1454
+ "grad_norm": 0.25499385595321655,
1455
+ "learning_rate": 1.2093366093366094e-05,
1456
+ "loss": 0.9205,
1457
+ "mean_token_accuracy": 0.8501100361347198,
1458
+ "num_tokens": 8334978.0,
1459
+ "step": 1610
1460
+ },
1461
+ {
1462
+ "epoch": 0.7957755127103033,
1463
+ "grad_norm": 0.5996484756469727,
1464
+ "learning_rate": 1.2044226044226047e-05,
1465
+ "loss": 0.9237,
1466
+ "mean_token_accuracy": 0.8439513966441154,
1467
+ "num_tokens": 8390388.0,
1468
+ "step": 1620
1469
+ },
1470
+ {
1471
+ "epoch": 0.8006877072332065,
1472
+ "grad_norm": 0.4831118583679199,
1473
+ "learning_rate": 1.1995085995085996e-05,
1474
+ "loss": 0.7179,
1475
+ "mean_token_accuracy": 0.8848425835371018,
1476
+ "num_tokens": 8445335.0,
1477
+ "step": 1630
1478
+ },
1479
+ {
1480
+ "epoch": 0.8055999017561095,
1481
+ "grad_norm": 0.3229360580444336,
1482
+ "learning_rate": 1.1945945945945946e-05,
1483
+ "loss": 1.1637,
1484
+ "mean_token_accuracy": 0.8010388404130936,
1485
+ "num_tokens": 8494343.0,
1486
+ "step": 1640
1487
+ },
1488
+ {
1489
+ "epoch": 0.8105120962790127,
1490
+ "grad_norm": 1.2905784845352173,
1491
+ "learning_rate": 1.1896805896805899e-05,
1492
+ "loss": 0.9642,
1493
+ "mean_token_accuracy": 0.829357648640871,
1494
+ "num_tokens": 8542794.0,
1495
+ "step": 1650
1496
+ },
1497
+ {
1498
+ "epoch": 0.8154242908019158,
1499
+ "grad_norm": 0.33358246088027954,
1500
+ "learning_rate": 1.1847665847665848e-05,
1501
+ "loss": 0.8557,
1502
+ "mean_token_accuracy": 0.85107681453228,
1503
+ "num_tokens": 8597837.0,
1504
+ "step": 1660
1505
+ },
1506
+ {
1507
+ "epoch": 0.8203364853248188,
1508
+ "grad_norm": 0.30174124240875244,
1509
+ "learning_rate": 1.17985257985258e-05,
1510
+ "loss": 0.6798,
1511
+ "mean_token_accuracy": 0.8848607018589973,
1512
+ "num_tokens": 8651855.0,
1513
+ "step": 1670
1514
+ },
1515
+ {
1516
+ "epoch": 0.825248679847722,
1517
+ "grad_norm": 0.41789162158966064,
1518
+ "learning_rate": 1.1749385749385751e-05,
1519
+ "loss": 0.7969,
1520
+ "mean_token_accuracy": 0.859107281267643,
1521
+ "num_tokens": 8705043.0,
1522
+ "step": 1680
1523
+ },
1524
+ {
1525
+ "epoch": 0.830160874370625,
1526
+ "grad_norm": 0.5565118193626404,
1527
+ "learning_rate": 1.17002457002457e-05,
1528
+ "loss": 0.8947,
1529
+ "mean_token_accuracy": 0.8513899371027946,
1530
+ "num_tokens": 8753735.0,
1531
+ "step": 1690
1532
+ },
1533
+ {
1534
+ "epoch": 0.8350730688935282,
1535
+ "grad_norm": 0.8451042771339417,
1536
+ "learning_rate": 1.1651105651105653e-05,
1537
+ "loss": 1.1201,
1538
+ "mean_token_accuracy": 0.7937561281025409,
1539
+ "num_tokens": 8799611.0,
1540
+ "step": 1700
1541
+ },
1542
+ {
1543
+ "epoch": 0.8399852634164313,
1544
+ "grad_norm": 0.8886920213699341,
1545
+ "learning_rate": 1.1601965601965603e-05,
1546
+ "loss": 0.8187,
1547
+ "mean_token_accuracy": 0.8588108576834201,
1548
+ "num_tokens": 8852789.0,
1549
+ "step": 1710
1550
+ },
1551
+ {
1552
+ "epoch": 0.8448974579393344,
1553
+ "grad_norm": 0.8036173582077026,
1554
+ "learning_rate": 1.1552825552825552e-05,
1555
+ "loss": 0.8139,
1556
+ "mean_token_accuracy": 0.868566033244133,
1557
+ "num_tokens": 8904446.0,
1558
+ "step": 1720
1559
+ },
1560
+ {
1561
+ "epoch": 0.8498096524622375,
1562
+ "grad_norm": 0.8208529949188232,
1563
+ "learning_rate": 1.1503685503685505e-05,
1564
+ "loss": 0.9869,
1565
+ "mean_token_accuracy": 0.8289718352258205,
1566
+ "num_tokens": 8955674.0,
1567
+ "step": 1730
1568
+ },
1569
+ {
1570
+ "epoch": 0.8547218469851406,
1571
+ "grad_norm": 0.48808610439300537,
1572
+ "learning_rate": 1.1454545454545455e-05,
1573
+ "loss": 0.6682,
1574
+ "mean_token_accuracy": 0.8741697937250137,
1575
+ "num_tokens": 9008895.0,
1576
+ "step": 1740
1577
+ },
1578
+ {
1579
+ "epoch": 0.8596340415080437,
1580
+ "grad_norm": 0.30963972210884094,
1581
+ "learning_rate": 1.1405405405405404e-05,
1582
+ "loss": 0.8816,
1583
+ "mean_token_accuracy": 0.843504398316145,
1584
+ "num_tokens": 9059352.0,
1585
+ "step": 1750
1586
+ },
1587
+ {
1588
+ "epoch": 0.8645462360309468,
1589
+ "grad_norm": 0.2593599855899811,
1590
+ "learning_rate": 1.1356265356265357e-05,
1591
+ "loss": 1.0678,
1592
+ "mean_token_accuracy": 0.8158367194235325,
1593
+ "num_tokens": 9105840.0,
1594
+ "step": 1760
1595
+ },
1596
+ {
1597
+ "epoch": 0.86945843055385,
1598
+ "grad_norm": 0.2662312984466553,
1599
+ "learning_rate": 1.1307125307125308e-05,
1600
+ "loss": 0.988,
1601
+ "mean_token_accuracy": 0.8375108957290649,
1602
+ "num_tokens": 9158645.0,
1603
+ "step": 1770
1604
+ },
1605
+ {
1606
+ "epoch": 0.874370625076753,
1607
+ "grad_norm": 0.2985932528972626,
1608
+ "learning_rate": 1.125798525798526e-05,
1609
+ "loss": 0.7872,
1610
+ "mean_token_accuracy": 0.8592289894819259,
1611
+ "num_tokens": 9205661.0,
1612
+ "step": 1780
1613
+ },
1614
+ {
1615
+ "epoch": 0.8792828195996562,
1616
+ "grad_norm": 0.26380449533462524,
1617
+ "learning_rate": 1.1208845208845209e-05,
1618
+ "loss": 1.1826,
1619
+ "mean_token_accuracy": 0.7935345992445946,
1620
+ "num_tokens": 9251283.0,
1621
+ "step": 1790
1622
+ },
1623
+ {
1624
+ "epoch": 0.8841950141225593,
1625
+ "grad_norm": 0.5365442633628845,
1626
+ "learning_rate": 1.1159705159705162e-05,
1627
+ "loss": 0.9483,
1628
+ "mean_token_accuracy": 0.8279004551470279,
1629
+ "num_tokens": 9298506.0,
1630
+ "step": 1800
1631
+ },
1632
+ {
1633
+ "epoch": 0.8891072086454623,
1634
+ "grad_norm": 0.35537609457969666,
1635
+ "learning_rate": 1.1110565110565112e-05,
1636
+ "loss": 0.6907,
1637
+ "mean_token_accuracy": 0.874860267341137,
1638
+ "num_tokens": 9352301.0,
1639
+ "step": 1810
1640
+ },
1641
+ {
1642
+ "epoch": 0.8940194031683655,
1643
+ "grad_norm": 0.36677488684654236,
1644
+ "learning_rate": 1.1061425061425061e-05,
1645
+ "loss": 0.8506,
1646
+ "mean_token_accuracy": 0.847238763421774,
1647
+ "num_tokens": 9401209.0,
1648
+ "step": 1820
1649
+ },
1650
+ {
1651
+ "epoch": 0.8989315976912686,
1652
+ "grad_norm": 0.35230782628059387,
1653
+ "learning_rate": 1.1012285012285014e-05,
1654
+ "loss": 0.9664,
1655
+ "mean_token_accuracy": 0.8267447024583816,
1656
+ "num_tokens": 9451239.0,
1657
+ "step": 1830
1658
+ },
1659
+ {
1660
+ "epoch": 0.9038437922141717,
1661
+ "grad_norm": 0.5595026612281799,
1662
+ "learning_rate": 1.0963144963144964e-05,
1663
+ "loss": 1.0239,
1664
+ "mean_token_accuracy": 0.8194499731063842,
1665
+ "num_tokens": 9504396.0,
1666
+ "step": 1840
1667
+ },
1668
+ {
1669
+ "epoch": 0.9087559867370748,
1670
+ "grad_norm": 0.38404449820518494,
1671
+ "learning_rate": 1.0914004914004913e-05,
1672
+ "loss": 0.9903,
1673
+ "mean_token_accuracy": 0.8269840359687806,
1674
+ "num_tokens": 9555148.0,
1675
+ "step": 1850
1676
+ },
1677
+ {
1678
+ "epoch": 0.9136681812599778,
1679
+ "grad_norm": 0.41196462512016296,
1680
+ "learning_rate": 1.0864864864864866e-05,
1681
+ "loss": 1.0264,
1682
+ "mean_token_accuracy": 0.8328478842973709,
1683
+ "num_tokens": 9606453.0,
1684
+ "step": 1860
1685
+ },
1686
+ {
1687
+ "epoch": 0.918580375782881,
1688
+ "grad_norm": 0.5028640031814575,
1689
+ "learning_rate": 1.0815724815724817e-05,
1690
+ "loss": 0.7749,
1691
+ "mean_token_accuracy": 0.8660773292183876,
1692
+ "num_tokens": 9658869.0,
1693
+ "step": 1870
1694
+ },
1695
+ {
1696
+ "epoch": 0.9234925703057841,
1697
+ "grad_norm": 0.2295975685119629,
1698
+ "learning_rate": 1.0766584766584767e-05,
1699
+ "loss": 0.8817,
1700
+ "mean_token_accuracy": 0.83365648239851,
1701
+ "num_tokens": 9708994.0,
1702
+ "step": 1880
1703
+ },
1704
+ {
1705
+ "epoch": 0.9284047648286872,
1706
+ "grad_norm": 0.24253033101558685,
1707
+ "learning_rate": 1.0717444717444718e-05,
1708
+ "loss": 0.9226,
1709
+ "mean_token_accuracy": 0.8485260397195816,
1710
+ "num_tokens": 9760081.0,
1711
+ "step": 1890
1712
+ },
1713
+ {
1714
+ "epoch": 0.9333169593515903,
1715
+ "grad_norm": 0.2894159257411957,
1716
+ "learning_rate": 1.066830466830467e-05,
1717
+ "loss": 0.7618,
1718
+ "mean_token_accuracy": 0.8669800266623497,
1719
+ "num_tokens": 9810101.0,
1720
+ "step": 1900
1721
+ },
1722
+ {
1723
+ "epoch": 0.9382291538744935,
1724
+ "grad_norm": 0.36487090587615967,
1725
+ "learning_rate": 1.061916461916462e-05,
1726
+ "loss": 0.8968,
1727
+ "mean_token_accuracy": 0.8443298265337944,
1728
+ "num_tokens": 9861741.0,
1729
+ "step": 1910
1730
+ },
1731
+ {
1732
+ "epoch": 0.9431413483973965,
1733
+ "grad_norm": 0.2829186022281647,
1734
+ "learning_rate": 1.057002457002457e-05,
1735
+ "loss": 0.821,
1736
+ "mean_token_accuracy": 0.8466833367943764,
1737
+ "num_tokens": 9913833.0,
1738
+ "step": 1920
1739
+ },
1740
+ {
1741
+ "epoch": 0.9480535429202996,
1742
+ "grad_norm": 0.6794554591178894,
1743
+ "learning_rate": 1.0520884520884523e-05,
1744
+ "loss": 0.7578,
1745
+ "mean_token_accuracy": 0.8747551962733269,
1746
+ "num_tokens": 9970247.0,
1747
+ "step": 1930
1748
+ },
1749
+ {
1750
+ "epoch": 0.9529657374432028,
1751
+ "grad_norm": 0.6977810263633728,
1752
+ "learning_rate": 1.0471744471744472e-05,
1753
+ "loss": 0.8165,
1754
+ "mean_token_accuracy": 0.8567108184099197,
1755
+ "num_tokens": 10019920.0,
1756
+ "step": 1940
1757
+ },
1758
+ {
1759
+ "epoch": 0.9578779319661058,
1760
+ "grad_norm": 0.504741907119751,
1761
+ "learning_rate": 1.0422604422604422e-05,
1762
+ "loss": 0.9742,
1763
+ "mean_token_accuracy": 0.8300105221569538,
1764
+ "num_tokens": 10066930.0,
1765
+ "step": 1950
1766
+ },
1767
+ {
1768
+ "epoch": 0.962790126489009,
1769
+ "grad_norm": 2.085604190826416,
1770
+ "learning_rate": 1.0373464373464375e-05,
1771
+ "loss": 0.9191,
1772
+ "mean_token_accuracy": 0.8302432119846344,
1773
+ "num_tokens": 10116405.0,
1774
+ "step": 1960
1775
+ },
1776
+ {
1777
+ "epoch": 0.9677023210119121,
1778
+ "grad_norm": 0.5094314813613892,
1779
+ "learning_rate": 1.0324324324324324e-05,
1780
+ "loss": 0.6908,
1781
+ "mean_token_accuracy": 0.8804146230220795,
1782
+ "num_tokens": 10171793.0,
1783
+ "step": 1970
1784
+ },
1785
+ {
1786
+ "epoch": 0.9726145155348151,
1787
+ "grad_norm": 0.7494292259216309,
1788
+ "learning_rate": 1.0275184275184276e-05,
1789
+ "loss": 0.9316,
1790
+ "mean_token_accuracy": 0.8486801907420158,
1791
+ "num_tokens": 10223476.0,
1792
+ "step": 1980
1793
+ },
1794
+ {
1795
+ "epoch": 0.9775267100577183,
1796
+ "grad_norm": 0.5121098756790161,
1797
+ "learning_rate": 1.0226044226044227e-05,
1798
+ "loss": 0.8121,
1799
+ "mean_token_accuracy": 0.8445978663861752,
1800
+ "num_tokens": 10276795.0,
1801
+ "step": 1990
1802
+ },
1803
+ {
1804
+ "epoch": 0.9824389045806214,
1805
+ "grad_norm": 0.22474397718906403,
1806
+ "learning_rate": 1.017690417690418e-05,
1807
+ "loss": 0.7328,
1808
+ "mean_token_accuracy": 0.8750261560082435,
1809
+ "num_tokens": 10327432.0,
1810
+ "step": 2000
1811
+ }
1812
+ ],
1813
+ "logging_steps": 10,
1814
+ "max_steps": 4070,
1815
+ "num_input_tokens_seen": 0,
1816
+ "num_train_epochs": 2,
1817
+ "save_steps": 500,
1818
+ "stateful_callbacks": {
1819
+ "TrainerControl": {
1820
+ "args": {
1821
+ "should_epoch_stop": false,
1822
+ "should_evaluate": false,
1823
+ "should_log": false,
1824
+ "should_save": true,
1825
+ "should_training_stop": false
1826
+ },
1827
+ "attributes": {}
1828
+ }
1829
+ },
1830
+ "total_flos": 4970062051430400.0,
1831
+ "train_batch_size": 4,
1832
+ "trial_name": null,
1833
+ "trial_params": null
1834
+ }
checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f63465acd95be88e019f04b736064426478945f5bc6a4d825ebd1918a32c185
3
+ size 5624
checkpoint-2500/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 256,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-2500/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-2500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7269e6a62d4f8e469e1ac91c3fc87c3622c15acaed0baf41b64c382efff51963
3
+ size 309900448
checkpoint-2500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:524626fa184d64a8deed6d2f16691d94fc9500c69b8bff2b53f8c8c2301bd2ba
3
+ size 619836730
checkpoint-2500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b66e3cc7c452b707ddac5caf0aa17618afb9bc1a0333600a22c4afb353f3165
3
+ size 14244
checkpoint-2500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26c1a0b563555b92f5a3125e4fdd77875692a9192d373c1789d607dee3f6fedd
3
+ size 1064
checkpoint-2500/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-2500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-2500/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-2500/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-2500/trainer_state.json ADDED
@@ -0,0 +1,2284 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.2284170453149945,
6
+ "eval_steps": 500,
7
+ "global_step": 2500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.004912194522903107,
14
+ "grad_norm": 2.5265462398529053,
15
+ "learning_rate": 1.995577395577396e-05,
16
+ "loss": 5.7783,
17
+ "mean_token_accuracy": 0.17638254687190055,
18
+ "num_tokens": 50745.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.009824389045806214,
23
+ "grad_norm": 1.7381863594055176,
24
+ "learning_rate": 1.9906633906633906e-05,
25
+ "loss": 5.4596,
26
+ "mean_token_accuracy": 0.20633984543383121,
27
+ "num_tokens": 101681.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.01473658356870932,
32
+ "grad_norm": 2.0286130905151367,
33
+ "learning_rate": 1.985749385749386e-05,
34
+ "loss": 5.2069,
35
+ "mean_token_accuracy": 0.2354225989431143,
36
+ "num_tokens": 153620.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.019648778091612428,
41
+ "grad_norm": 2.2828938961029053,
42
+ "learning_rate": 1.980835380835381e-05,
43
+ "loss": 5.0637,
44
+ "mean_token_accuracy": 0.2486753471195698,
45
+ "num_tokens": 205428.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.024560972614515536,
50
+ "grad_norm": 2.247742176055908,
51
+ "learning_rate": 1.975921375921376e-05,
52
+ "loss": 4.7071,
53
+ "mean_token_accuracy": 0.2982919864356518,
54
+ "num_tokens": 260794.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.02947316713741864,
59
+ "grad_norm": 1.4552102088928223,
60
+ "learning_rate": 1.9710073710073712e-05,
61
+ "loss": 4.5951,
62
+ "mean_token_accuracy": 0.3140149611979723,
63
+ "num_tokens": 314216.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.03438536166032175,
68
+ "grad_norm": 2.032104969024658,
69
+ "learning_rate": 1.9660933660933663e-05,
70
+ "loss": 4.3963,
71
+ "mean_token_accuracy": 0.33228743858635423,
72
+ "num_tokens": 365608.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.039297556183224856,
77
+ "grad_norm": 1.5947270393371582,
78
+ "learning_rate": 1.9611793611793614e-05,
79
+ "loss": 4.2578,
80
+ "mean_token_accuracy": 0.3510261565446854,
81
+ "num_tokens": 418509.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.04420975070612796,
86
+ "grad_norm": 1.9174288511276245,
87
+ "learning_rate": 1.9562653562653564e-05,
88
+ "loss": 4.0457,
89
+ "mean_token_accuracy": 0.38185270577669145,
90
+ "num_tokens": 472608.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.04912194522903107,
95
+ "grad_norm": 1.5742830038070679,
96
+ "learning_rate": 1.9513513513513515e-05,
97
+ "loss": 4.0492,
98
+ "mean_token_accuracy": 0.37742224968969823,
99
+ "num_tokens": 521777.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.054034139751934177,
104
+ "grad_norm": 1.5624850988388062,
105
+ "learning_rate": 1.9464373464373466e-05,
106
+ "loss": 3.7053,
107
+ "mean_token_accuracy": 0.41850126534700394,
108
+ "num_tokens": 580161.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.05894633427483728,
113
+ "grad_norm": 2.199150562286377,
114
+ "learning_rate": 1.9415233415233417e-05,
115
+ "loss": 3.6982,
116
+ "mean_token_accuracy": 0.4235368125140667,
117
+ "num_tokens": 627843.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.06385852879774039,
122
+ "grad_norm": 1.8548036813735962,
123
+ "learning_rate": 1.9366093366093367e-05,
124
+ "loss": 3.3414,
125
+ "mean_token_accuracy": 0.4654690891504288,
126
+ "num_tokens": 681443.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.0687707233206435,
131
+ "grad_norm": 2.78169846534729,
132
+ "learning_rate": 1.9316953316953318e-05,
133
+ "loss": 3.2679,
134
+ "mean_token_accuracy": 0.4711888641119003,
135
+ "num_tokens": 733457.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.07368291784354661,
140
+ "grad_norm": 2.4450321197509766,
141
+ "learning_rate": 1.926781326781327e-05,
142
+ "loss": 3.3377,
143
+ "mean_token_accuracy": 0.4799527756869793,
144
+ "num_tokens": 787769.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.07859511236644971,
149
+ "grad_norm": 1.7717996835708618,
150
+ "learning_rate": 1.921867321867322e-05,
151
+ "loss": 2.8853,
152
+ "mean_token_accuracy": 0.541524414718151,
153
+ "num_tokens": 843485.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.08350730688935282,
158
+ "grad_norm": 2.8402254581451416,
159
+ "learning_rate": 1.9169533169533174e-05,
160
+ "loss": 2.862,
161
+ "mean_token_accuracy": 0.5626954190433026,
162
+ "num_tokens": 894003.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.08841950141225592,
167
+ "grad_norm": 1.5486701726913452,
168
+ "learning_rate": 1.912039312039312e-05,
169
+ "loss": 2.7533,
170
+ "mean_token_accuracy": 0.5848119542002678,
171
+ "num_tokens": 949726.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.09333169593515903,
176
+ "grad_norm": 1.3785852193832397,
177
+ "learning_rate": 1.9071253071253072e-05,
178
+ "loss": 2.7602,
179
+ "mean_token_accuracy": 0.5894295200705528,
180
+ "num_tokens": 1000146.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 0.09824389045806214,
185
+ "grad_norm": 1.6824640035629272,
186
+ "learning_rate": 1.9022113022113026e-05,
187
+ "loss": 2.6345,
188
+ "mean_token_accuracy": 0.5961507339030504,
189
+ "num_tokens": 1050142.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 0.10315608498096525,
194
+ "grad_norm": 1.0604822635650635,
195
+ "learning_rate": 1.8972972972972973e-05,
196
+ "loss": 2.4517,
197
+ "mean_token_accuracy": 0.6296438351273537,
198
+ "num_tokens": 1099646.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 0.10806827950386835,
203
+ "grad_norm": 1.9105230569839478,
204
+ "learning_rate": 1.8923832923832924e-05,
205
+ "loss": 2.4563,
206
+ "mean_token_accuracy": 0.6365545220673084,
207
+ "num_tokens": 1148100.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 0.11298047402677146,
212
+ "grad_norm": 1.41056227684021,
213
+ "learning_rate": 1.8874692874692878e-05,
214
+ "loss": 2.4017,
215
+ "mean_token_accuracy": 0.6691348008811474,
216
+ "num_tokens": 1198987.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.11789266854967456,
221
+ "grad_norm": 1.0573596954345703,
222
+ "learning_rate": 1.8825552825552825e-05,
223
+ "loss": 2.3199,
224
+ "mean_token_accuracy": 0.6731609679758549,
225
+ "num_tokens": 1249335.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 0.12280486307257768,
230
+ "grad_norm": 1.6629279851913452,
231
+ "learning_rate": 1.877641277641278e-05,
232
+ "loss": 1.9335,
233
+ "mean_token_accuracy": 0.7327558636665344,
234
+ "num_tokens": 1303157.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 0.12771705759548077,
239
+ "grad_norm": 1.704005479812622,
240
+ "learning_rate": 1.872727272727273e-05,
241
+ "loss": 2.1897,
242
+ "mean_token_accuracy": 0.6974137671291828,
243
+ "num_tokens": 1355455.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 0.1326292521183839,
248
+ "grad_norm": 1.0553226470947266,
249
+ "learning_rate": 1.8678132678132678e-05,
250
+ "loss": 1.9321,
251
+ "mean_token_accuracy": 0.7327657997608185,
252
+ "num_tokens": 1407508.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 0.137541446641287,
257
+ "grad_norm": 1.101886510848999,
258
+ "learning_rate": 1.8628992628992632e-05,
259
+ "loss": 1.9929,
260
+ "mean_token_accuracy": 0.7272698871791363,
261
+ "num_tokens": 1460595.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 0.1424536411641901,
266
+ "grad_norm": 2.005901575088501,
267
+ "learning_rate": 1.8579852579852583e-05,
268
+ "loss": 1.7147,
269
+ "mean_token_accuracy": 0.7604942478239536,
270
+ "num_tokens": 1512647.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 0.14736583568709322,
275
+ "grad_norm": 7.757036209106445,
276
+ "learning_rate": 1.853071253071253e-05,
277
+ "loss": 1.7571,
278
+ "mean_token_accuracy": 0.7603825606405735,
279
+ "num_tokens": 1565516.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 0.1522780302099963,
284
+ "grad_norm": 0.8850768208503723,
285
+ "learning_rate": 1.8481572481572484e-05,
286
+ "loss": 1.5439,
287
+ "mean_token_accuracy": 0.7915346205234528,
288
+ "num_tokens": 1620194.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 0.15719022473289943,
293
+ "grad_norm": 1.2887756824493408,
294
+ "learning_rate": 1.8432432432432435e-05,
295
+ "loss": 1.5419,
296
+ "mean_token_accuracy": 0.7885693073272705,
297
+ "num_tokens": 1672317.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 0.16210241925580254,
302
+ "grad_norm": 2.562720537185669,
303
+ "learning_rate": 1.8383292383292385e-05,
304
+ "loss": 1.5783,
305
+ "mean_token_accuracy": 0.7889993876218796,
306
+ "num_tokens": 1728513.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 0.16701461377870563,
311
+ "grad_norm": 2.907665491104126,
312
+ "learning_rate": 1.8334152334152336e-05,
313
+ "loss": 1.6573,
314
+ "mean_token_accuracy": 0.7675686649978161,
315
+ "num_tokens": 1777759.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 0.17192680830160875,
320
+ "grad_norm": 0.8430731892585754,
321
+ "learning_rate": 1.8285012285012287e-05,
322
+ "loss": 1.664,
323
+ "mean_token_accuracy": 0.7618721649050713,
324
+ "num_tokens": 1833279.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 0.17683900282451184,
329
+ "grad_norm": 1.1680371761322021,
330
+ "learning_rate": 1.8235872235872238e-05,
331
+ "loss": 1.515,
332
+ "mean_token_accuracy": 0.7789990782737732,
333
+ "num_tokens": 1882531.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 0.18175119734741496,
338
+ "grad_norm": 2.009002208709717,
339
+ "learning_rate": 1.818673218673219e-05,
340
+ "loss": 1.4555,
341
+ "mean_token_accuracy": 0.7983467862010002,
342
+ "num_tokens": 1934619.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 0.18666339187031805,
347
+ "grad_norm": 1.3894531726837158,
348
+ "learning_rate": 1.813759213759214e-05,
349
+ "loss": 1.6206,
350
+ "mean_token_accuracy": 0.7681202955543995,
351
+ "num_tokens": 1987231.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 0.19157558639322117,
356
+ "grad_norm": 0.7266082763671875,
357
+ "learning_rate": 1.808845208845209e-05,
358
+ "loss": 1.3022,
359
+ "mean_token_accuracy": 0.818102091550827,
360
+ "num_tokens": 2039985.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 0.1964877809161243,
365
+ "grad_norm": 0.7502278089523315,
366
+ "learning_rate": 1.803931203931204e-05,
367
+ "loss": 1.3902,
368
+ "mean_token_accuracy": 0.792433574795723,
369
+ "num_tokens": 2092178.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 0.20139997543902738,
374
+ "grad_norm": 0.8366674780845642,
375
+ "learning_rate": 1.799017199017199e-05,
376
+ "loss": 1.5385,
377
+ "mean_token_accuracy": 0.7903823673725128,
378
+ "num_tokens": 2151243.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 0.2063121699619305,
383
+ "grad_norm": 0.7726922035217285,
384
+ "learning_rate": 1.7941031941031942e-05,
385
+ "loss": 1.3529,
386
+ "mean_token_accuracy": 0.8017133295536041,
387
+ "num_tokens": 2201072.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 0.2112243644848336,
392
+ "grad_norm": 0.650055468082428,
393
+ "learning_rate": 1.7891891891891893e-05,
394
+ "loss": 1.4312,
395
+ "mean_token_accuracy": 0.7957115054130555,
396
+ "num_tokens": 2256591.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 0.2161365590077367,
401
+ "grad_norm": 0.9430303573608398,
402
+ "learning_rate": 1.7842751842751843e-05,
403
+ "loss": 1.1938,
404
+ "mean_token_accuracy": 0.8211165018379688,
405
+ "num_tokens": 2308048.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 0.22104875353063982,
410
+ "grad_norm": 0.9191327691078186,
411
+ "learning_rate": 1.7793611793611794e-05,
412
+ "loss": 1.1662,
413
+ "mean_token_accuracy": 0.8298586897552014,
414
+ "num_tokens": 2363573.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 0.22596094805354291,
419
+ "grad_norm": 0.7493109703063965,
420
+ "learning_rate": 1.7744471744471745e-05,
421
+ "loss": 1.5908,
422
+ "mean_token_accuracy": 0.7651749603450299,
423
+ "num_tokens": 2413187.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 0.23087314257644603,
428
+ "grad_norm": 0.5522422790527344,
429
+ "learning_rate": 1.7695331695331696e-05,
430
+ "loss": 1.3707,
431
+ "mean_token_accuracy": 0.7843399420380592,
432
+ "num_tokens": 2468041.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 0.23578533709934912,
437
+ "grad_norm": 0.6577270030975342,
438
+ "learning_rate": 1.764619164619165e-05,
439
+ "loss": 1.4679,
440
+ "mean_token_accuracy": 0.7856365032494068,
441
+ "num_tokens": 2517753.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 0.24069753162225224,
446
+ "grad_norm": 0.9699747562408447,
447
+ "learning_rate": 1.7597051597051597e-05,
448
+ "loss": 1.3493,
449
+ "mean_token_accuracy": 0.8016418114304542,
450
+ "num_tokens": 2568072.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 0.24560972614515536,
455
+ "grad_norm": 2.0955657958984375,
456
+ "learning_rate": 1.7547911547911548e-05,
457
+ "loss": 0.9634,
458
+ "mean_token_accuracy": 0.8412673205137253,
459
+ "num_tokens": 2617143.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 0.25052192066805845,
464
+ "grad_norm": 0.7577691078186035,
465
+ "learning_rate": 1.7498771498771502e-05,
466
+ "loss": 1.0889,
467
+ "mean_token_accuracy": 0.8479016363620758,
468
+ "num_tokens": 2669318.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 0.25543411519096154,
473
+ "grad_norm": 0.5891643762588501,
474
+ "learning_rate": 1.744963144963145e-05,
475
+ "loss": 1.1811,
476
+ "mean_token_accuracy": 0.82891755849123,
477
+ "num_tokens": 2723151.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 0.2603463097138647,
482
+ "grad_norm": 0.7815893292427063,
483
+ "learning_rate": 1.74004914004914e-05,
484
+ "loss": 1.3488,
485
+ "mean_token_accuracy": 0.7971963949501515,
486
+ "num_tokens": 2774274.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 0.2652585042367678,
491
+ "grad_norm": 0.365824818611145,
492
+ "learning_rate": 1.7351351351351354e-05,
493
+ "loss": 1.129,
494
+ "mean_token_accuracy": 0.8214888103306294,
495
+ "num_tokens": 2825301.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 0.27017069875967087,
500
+ "grad_norm": 0.5682836174964905,
501
+ "learning_rate": 1.73022113022113e-05,
502
+ "loss": 1.0752,
503
+ "mean_token_accuracy": 0.8513655051589012,
504
+ "num_tokens": 2877767.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 0.275082893282574,
509
+ "grad_norm": 0.9484438300132751,
510
+ "learning_rate": 1.7253071253071256e-05,
511
+ "loss": 1.1606,
512
+ "mean_token_accuracy": 0.8191152907907963,
513
+ "num_tokens": 2929216.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 0.2799950878054771,
518
+ "grad_norm": 0.7097403407096863,
519
+ "learning_rate": 1.7203931203931206e-05,
520
+ "loss": 1.0245,
521
+ "mean_token_accuracy": 0.8594455644488335,
522
+ "num_tokens": 2983287.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 0.2849072823283802,
527
+ "grad_norm": 2.197983741760254,
528
+ "learning_rate": 1.7154791154791157e-05,
529
+ "loss": 1.0116,
530
+ "mean_token_accuracy": 0.8458577767014503,
531
+ "num_tokens": 3036426.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 0.2898194768512833,
536
+ "grad_norm": 0.9750831723213196,
537
+ "learning_rate": 1.7105651105651108e-05,
538
+ "loss": 1.2482,
539
+ "mean_token_accuracy": 0.8097934015095234,
540
+ "num_tokens": 3090471.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 0.29473167137418643,
545
+ "grad_norm": 0.7452868223190308,
546
+ "learning_rate": 1.705651105651106e-05,
547
+ "loss": 0.9643,
548
+ "mean_token_accuracy": 0.8562554508447647,
549
+ "num_tokens": 3143095.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 0.2996438658970895,
554
+ "grad_norm": 0.3684370815753937,
555
+ "learning_rate": 1.700737100737101e-05,
556
+ "loss": 0.8587,
557
+ "mean_token_accuracy": 0.8684959009289741,
558
+ "num_tokens": 3193778.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 0.3045560604199926,
563
+ "grad_norm": 0.4814823567867279,
564
+ "learning_rate": 1.695823095823096e-05,
565
+ "loss": 0.8246,
566
+ "mean_token_accuracy": 0.8646811529994011,
567
+ "num_tokens": 3249776.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 0.30946825494289576,
572
+ "grad_norm": 0.6298699378967285,
573
+ "learning_rate": 1.690909090909091e-05,
574
+ "loss": 1.0043,
575
+ "mean_token_accuracy": 0.849489164352417,
576
+ "num_tokens": 3301301.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 0.31438044946579885,
581
+ "grad_norm": 0.6802086234092712,
582
+ "learning_rate": 1.685995085995086e-05,
583
+ "loss": 0.9932,
584
+ "mean_token_accuracy": 0.8412040494382381,
585
+ "num_tokens": 3351834.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 0.31929264398870194,
590
+ "grad_norm": 0.4441145062446594,
591
+ "learning_rate": 1.6810810810810812e-05,
592
+ "loss": 0.9093,
593
+ "mean_token_accuracy": 0.8644197151064873,
594
+ "num_tokens": 3402630.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 0.3242048385116051,
599
+ "grad_norm": 0.5664909482002258,
600
+ "learning_rate": 1.6761670761670763e-05,
601
+ "loss": 1.0511,
602
+ "mean_token_accuracy": 0.8302886992692947,
603
+ "num_tokens": 3453600.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 0.3291170330345082,
608
+ "grad_norm": 0.898231029510498,
609
+ "learning_rate": 1.6712530712530714e-05,
610
+ "loss": 1.2663,
611
+ "mean_token_accuracy": 0.7889795318245888,
612
+ "num_tokens": 3501154.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 0.33402922755741127,
617
+ "grad_norm": 0.49926525354385376,
618
+ "learning_rate": 1.6663390663390664e-05,
619
+ "loss": 1.0539,
620
+ "mean_token_accuracy": 0.8462485030293465,
621
+ "num_tokens": 3552076.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 0.33894142208031436,
626
+ "grad_norm": 0.5327655673027039,
627
+ "learning_rate": 1.6614250614250615e-05,
628
+ "loss": 1.0757,
629
+ "mean_token_accuracy": 0.8260165579617024,
630
+ "num_tokens": 3602651.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 0.3438536166032175,
635
+ "grad_norm": 0.5295992493629456,
636
+ "learning_rate": 1.6565110565110566e-05,
637
+ "loss": 0.9826,
638
+ "mean_token_accuracy": 0.8445935681462288,
639
+ "num_tokens": 3656310.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 0.3487658111261206,
644
+ "grad_norm": 0.584396243095398,
645
+ "learning_rate": 1.6515970515970517e-05,
646
+ "loss": 0.9771,
647
+ "mean_token_accuracy": 0.8513481065630912,
648
+ "num_tokens": 3712764.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 0.3536780056490237,
653
+ "grad_norm": 0.5786782503128052,
654
+ "learning_rate": 1.6466830466830467e-05,
655
+ "loss": 1.0812,
656
+ "mean_token_accuracy": 0.8179432228207588,
657
+ "num_tokens": 3761995.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 0.35859020017192683,
662
+ "grad_norm": 0.3496871888637543,
663
+ "learning_rate": 1.6417690417690418e-05,
664
+ "loss": 1.0407,
665
+ "mean_token_accuracy": 0.8126879334449768,
666
+ "num_tokens": 3811493.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 0.3635023946948299,
671
+ "grad_norm": 0.5038619041442871,
672
+ "learning_rate": 1.636855036855037e-05,
673
+ "loss": 1.1602,
674
+ "mean_token_accuracy": 0.8141254656016826,
675
+ "num_tokens": 3859376.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 0.368414589217733,
680
+ "grad_norm": 0.4643881320953369,
681
+ "learning_rate": 1.631941031941032e-05,
682
+ "loss": 0.8819,
683
+ "mean_token_accuracy": 0.8545275427401066,
684
+ "num_tokens": 3912052.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 0.3733267837406361,
689
+ "grad_norm": 0.4222355782985687,
690
+ "learning_rate": 1.6270270270270274e-05,
691
+ "loss": 0.8333,
692
+ "mean_token_accuracy": 0.8552621826529503,
693
+ "num_tokens": 3965078.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 0.37823897826353925,
698
+ "grad_norm": 0.4513651728630066,
699
+ "learning_rate": 1.6221130221130224e-05,
700
+ "loss": 1.0386,
701
+ "mean_token_accuracy": 0.8259748943150044,
702
+ "num_tokens": 4015489.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 0.38315117278644234,
707
+ "grad_norm": 0.5292590856552124,
708
+ "learning_rate": 1.6171990171990172e-05,
709
+ "loss": 0.8968,
710
+ "mean_token_accuracy": 0.8551326602697372,
711
+ "num_tokens": 4067831.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 0.38806336730934543,
716
+ "grad_norm": 0.49815845489501953,
717
+ "learning_rate": 1.6122850122850126e-05,
718
+ "loss": 1.0808,
719
+ "mean_token_accuracy": 0.8252614364027977,
720
+ "num_tokens": 4117945.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 0.3929755618322486,
725
+ "grad_norm": 0.339728981256485,
726
+ "learning_rate": 1.6073710073710077e-05,
727
+ "loss": 0.9569,
728
+ "mean_token_accuracy": 0.8419950112700463,
729
+ "num_tokens": 4171138.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 0.39788775635515167,
734
+ "grad_norm": 0.4530418813228607,
735
+ "learning_rate": 1.6024570024570024e-05,
736
+ "loss": 1.1355,
737
+ "mean_token_accuracy": 0.8250428631901741,
738
+ "num_tokens": 4223765.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 0.40279995087805476,
743
+ "grad_norm": 0.5202032923698425,
744
+ "learning_rate": 1.5975429975429978e-05,
745
+ "loss": 0.8306,
746
+ "mean_token_accuracy": 0.865328436344862,
747
+ "num_tokens": 4278056.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 0.4077121454009579,
752
+ "grad_norm": 0.4732300043106079,
753
+ "learning_rate": 1.592628992628993e-05,
754
+ "loss": 1.3396,
755
+ "mean_token_accuracy": 0.783781211823225,
756
+ "num_tokens": 4326704.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 0.412624339923861,
761
+ "grad_norm": 0.49258068203926086,
762
+ "learning_rate": 1.5877149877149876e-05,
763
+ "loss": 0.8462,
764
+ "mean_token_accuracy": 0.8517311483621597,
765
+ "num_tokens": 4378538.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 0.4175365344467641,
770
+ "grad_norm": 0.8045883774757385,
771
+ "learning_rate": 1.582800982800983e-05,
772
+ "loss": 0.9785,
773
+ "mean_token_accuracy": 0.8369067624211312,
774
+ "num_tokens": 4429893.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 0.4224487289696672,
779
+ "grad_norm": 0.361854612827301,
780
+ "learning_rate": 1.577886977886978e-05,
781
+ "loss": 0.936,
782
+ "mean_token_accuracy": 0.8446702107787132,
783
+ "num_tokens": 4481594.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 0.4273609234925703,
788
+ "grad_norm": 0.593479573726654,
789
+ "learning_rate": 1.572972972972973e-05,
790
+ "loss": 0.8245,
791
+ "mean_token_accuracy": 0.8750210002064704,
792
+ "num_tokens": 4534098.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 0.4322731180154734,
797
+ "grad_norm": 0.31826287508010864,
798
+ "learning_rate": 1.5680589680589682e-05,
799
+ "loss": 0.94,
800
+ "mean_token_accuracy": 0.8408755764365197,
801
+ "num_tokens": 4587111.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 0.4371853125383765,
806
+ "grad_norm": 1.4076229333877563,
807
+ "learning_rate": 1.5631449631449633e-05,
808
+ "loss": 1.0176,
809
+ "mean_token_accuracy": 0.8321189761161805,
810
+ "num_tokens": 4640388.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 0.44209750706127965,
815
+ "grad_norm": 0.6677824854850769,
816
+ "learning_rate": 1.5582309582309584e-05,
817
+ "loss": 1.2891,
818
+ "mean_token_accuracy": 0.7943175598978996,
819
+ "num_tokens": 4691468.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 0.44700970158418274,
824
+ "grad_norm": 0.4233896732330322,
825
+ "learning_rate": 1.5533169533169535e-05,
826
+ "loss": 1.1053,
827
+ "mean_token_accuracy": 0.8279901027679444,
828
+ "num_tokens": 4741844.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 0.45192189610708583,
833
+ "grad_norm": 0.35591593384742737,
834
+ "learning_rate": 1.5484029484029485e-05,
835
+ "loss": 0.8832,
836
+ "mean_token_accuracy": 0.8491488620638847,
837
+ "num_tokens": 4793741.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 0.4568340906299889,
842
+ "grad_norm": 0.6583911180496216,
843
+ "learning_rate": 1.5434889434889436e-05,
844
+ "loss": 0.8387,
845
+ "mean_token_accuracy": 0.8648917526006699,
846
+ "num_tokens": 4848466.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 0.46174628515289207,
851
+ "grad_norm": 0.4065934717655182,
852
+ "learning_rate": 1.5385749385749387e-05,
853
+ "loss": 0.9242,
854
+ "mean_token_accuracy": 0.8412024281919003,
855
+ "num_tokens": 4901992.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 0.46665847967579516,
860
+ "grad_norm": 0.3458426594734192,
861
+ "learning_rate": 1.5336609336609338e-05,
862
+ "loss": 0.8545,
863
+ "mean_token_accuracy": 0.855231161415577,
864
+ "num_tokens": 4956964.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 0.47157067419869825,
869
+ "grad_norm": 0.6710091233253479,
870
+ "learning_rate": 1.5287469287469288e-05,
871
+ "loss": 0.9583,
872
+ "mean_token_accuracy": 0.8393168188631535,
873
+ "num_tokens": 5011287.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 0.4764828687216014,
878
+ "grad_norm": 0.3753456771373749,
879
+ "learning_rate": 1.5238329238329239e-05,
880
+ "loss": 0.9738,
881
+ "mean_token_accuracy": 0.8375437319278717,
882
+ "num_tokens": 5061714.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 0.4813950632445045,
887
+ "grad_norm": 1.2011836767196655,
888
+ "learning_rate": 1.5189189189189191e-05,
889
+ "loss": 1.0972,
890
+ "mean_token_accuracy": 0.8070526115596295,
891
+ "num_tokens": 5114645.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 0.4863072577674076,
896
+ "grad_norm": 0.37889617681503296,
897
+ "learning_rate": 1.5140049140049142e-05,
898
+ "loss": 0.9442,
899
+ "mean_token_accuracy": 0.8458199493587018,
900
+ "num_tokens": 5165485.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 0.4912194522903107,
905
+ "grad_norm": 0.6613748669624329,
906
+ "learning_rate": 1.5090909090909091e-05,
907
+ "loss": 1.0196,
908
+ "mean_token_accuracy": 0.8272999472916126,
909
+ "num_tokens": 5216189.0,
910
+ "step": 1000
911
+ },
912
+ {
913
+ "epoch": 0.4961316468132138,
914
+ "grad_norm": 0.7825655937194824,
915
+ "learning_rate": 1.5041769041769044e-05,
916
+ "loss": 1.011,
917
+ "mean_token_accuracy": 0.8262499138712883,
918
+ "num_tokens": 5269248.0,
919
+ "step": 1010
920
+ },
921
+ {
922
+ "epoch": 0.5010438413361169,
923
+ "grad_norm": 0.2106683850288391,
924
+ "learning_rate": 1.4992628992628994e-05,
925
+ "loss": 0.9098,
926
+ "mean_token_accuracy": 0.8512444771826267,
927
+ "num_tokens": 5322524.0,
928
+ "step": 1020
929
+ },
930
+ {
931
+ "epoch": 0.50595603585902,
932
+ "grad_norm": 0.8093978762626648,
933
+ "learning_rate": 1.4943488943488943e-05,
934
+ "loss": 1.2403,
935
+ "mean_token_accuracy": 0.7969906754791737,
936
+ "num_tokens": 5370123.0,
937
+ "step": 1030
938
+ },
939
+ {
940
+ "epoch": 0.5108682303819231,
941
+ "grad_norm": 0.5412647724151611,
942
+ "learning_rate": 1.4894348894348896e-05,
943
+ "loss": 0.8821,
944
+ "mean_token_accuracy": 0.8460625484585762,
945
+ "num_tokens": 5422125.0,
946
+ "step": 1040
947
+ },
948
+ {
949
+ "epoch": 0.5157804249048262,
950
+ "grad_norm": 0.5411800146102905,
951
+ "learning_rate": 1.4845208845208847e-05,
952
+ "loss": 1.0804,
953
+ "mean_token_accuracy": 0.8199978180229663,
954
+ "num_tokens": 5472127.0,
955
+ "step": 1050
956
+ },
957
+ {
958
+ "epoch": 0.5206926194277294,
959
+ "grad_norm": 0.4349406063556671,
960
+ "learning_rate": 1.4796068796068797e-05,
961
+ "loss": 0.7559,
962
+ "mean_token_accuracy": 0.8662853054702282,
963
+ "num_tokens": 5525028.0,
964
+ "step": 1060
965
+ },
966
+ {
967
+ "epoch": 0.5256048139506324,
968
+ "grad_norm": 0.4738285541534424,
969
+ "learning_rate": 1.4746928746928748e-05,
970
+ "loss": 1.1225,
971
+ "mean_token_accuracy": 0.8103010386228562,
972
+ "num_tokens": 5570859.0,
973
+ "step": 1070
974
+ },
975
+ {
976
+ "epoch": 0.5305170084735356,
977
+ "grad_norm": 0.6644457578659058,
978
+ "learning_rate": 1.46977886977887e-05,
979
+ "loss": 0.9334,
980
+ "mean_token_accuracy": 0.8213524237275124,
981
+ "num_tokens": 5618933.0,
982
+ "step": 1080
983
+ },
984
+ {
985
+ "epoch": 0.5354292029964387,
986
+ "grad_norm": 0.3038887083530426,
987
+ "learning_rate": 1.464864864864865e-05,
988
+ "loss": 1.0292,
989
+ "mean_token_accuracy": 0.810487725585699,
990
+ "num_tokens": 5664922.0,
991
+ "step": 1090
992
+ },
993
+ {
994
+ "epoch": 0.5403413975193417,
995
+ "grad_norm": 0.26083576679229736,
996
+ "learning_rate": 1.45995085995086e-05,
997
+ "loss": 0.9084,
998
+ "mean_token_accuracy": 0.836492794752121,
999
+ "num_tokens": 5716671.0,
1000
+ "step": 1100
1001
+ },
1002
+ {
1003
+ "epoch": 0.5452535920422449,
1004
+ "grad_norm": 0.36848321557044983,
1005
+ "learning_rate": 1.4550368550368553e-05,
1006
+ "loss": 1.1544,
1007
+ "mean_token_accuracy": 0.8033395648002625,
1008
+ "num_tokens": 5764576.0,
1009
+ "step": 1110
1010
+ },
1011
+ {
1012
+ "epoch": 0.550165786565148,
1013
+ "grad_norm": 0.3338620662689209,
1014
+ "learning_rate": 1.4501228501228502e-05,
1015
+ "loss": 1.0114,
1016
+ "mean_token_accuracy": 0.8188160583376884,
1017
+ "num_tokens": 5816699.0,
1018
+ "step": 1120
1019
+ },
1020
+ {
1021
+ "epoch": 0.5550779810880511,
1022
+ "grad_norm": 0.3278094232082367,
1023
+ "learning_rate": 1.4452088452088452e-05,
1024
+ "loss": 0.6189,
1025
+ "mean_token_accuracy": 0.8895823881030083,
1026
+ "num_tokens": 5874257.0,
1027
+ "step": 1130
1028
+ },
1029
+ {
1030
+ "epoch": 0.5599901756109542,
1031
+ "grad_norm": 0.4236205220222473,
1032
+ "learning_rate": 1.4402948402948405e-05,
1033
+ "loss": 0.9474,
1034
+ "mean_token_accuracy": 0.8435555145144462,
1035
+ "num_tokens": 5925628.0,
1036
+ "step": 1140
1037
+ },
1038
+ {
1039
+ "epoch": 0.5649023701338572,
1040
+ "grad_norm": 0.21693912148475647,
1041
+ "learning_rate": 1.4353808353808354e-05,
1042
+ "loss": 0.9109,
1043
+ "mean_token_accuracy": 0.842586936801672,
1044
+ "num_tokens": 5977130.0,
1045
+ "step": 1150
1046
+ },
1047
+ {
1048
+ "epoch": 0.5698145646567604,
1049
+ "grad_norm": 0.6288137435913086,
1050
+ "learning_rate": 1.4304668304668306e-05,
1051
+ "loss": 0.7005,
1052
+ "mean_token_accuracy": 0.877720920741558,
1053
+ "num_tokens": 6032405.0,
1054
+ "step": 1160
1055
+ },
1056
+ {
1057
+ "epoch": 0.5747267591796635,
1058
+ "grad_norm": 0.29411551356315613,
1059
+ "learning_rate": 1.4255528255528257e-05,
1060
+ "loss": 0.9163,
1061
+ "mean_token_accuracy": 0.8345251567661762,
1062
+ "num_tokens": 6081769.0,
1063
+ "step": 1170
1064
+ },
1065
+ {
1066
+ "epoch": 0.5796389537025666,
1067
+ "grad_norm": 0.3542836308479309,
1068
+ "learning_rate": 1.420638820638821e-05,
1069
+ "loss": 0.8983,
1070
+ "mean_token_accuracy": 0.8478083908557892,
1071
+ "num_tokens": 6131660.0,
1072
+ "step": 1180
1073
+ },
1074
+ {
1075
+ "epoch": 0.5845511482254697,
1076
+ "grad_norm": 0.6326059103012085,
1077
+ "learning_rate": 1.4157248157248158e-05,
1078
+ "loss": 1.055,
1079
+ "mean_token_accuracy": 0.8133970260620117,
1080
+ "num_tokens": 6177706.0,
1081
+ "step": 1190
1082
+ },
1083
+ {
1084
+ "epoch": 0.5894633427483729,
1085
+ "grad_norm": 0.39950093626976013,
1086
+ "learning_rate": 1.410810810810811e-05,
1087
+ "loss": 0.6446,
1088
+ "mean_token_accuracy": 0.8883207038044929,
1089
+ "num_tokens": 6231530.0,
1090
+ "step": 1200
1091
+ },
1092
+ {
1093
+ "epoch": 0.5943755372712759,
1094
+ "grad_norm": 1.4949785470962524,
1095
+ "learning_rate": 1.4058968058968062e-05,
1096
+ "loss": 0.8527,
1097
+ "mean_token_accuracy": 0.8495218820869923,
1098
+ "num_tokens": 6284936.0,
1099
+ "step": 1210
1100
+ },
1101
+ {
1102
+ "epoch": 0.599287731794179,
1103
+ "grad_norm": 0.5013307332992554,
1104
+ "learning_rate": 1.400982800982801e-05,
1105
+ "loss": 0.9342,
1106
+ "mean_token_accuracy": 0.8461359746754169,
1107
+ "num_tokens": 6338328.0,
1108
+ "step": 1220
1109
+ },
1110
+ {
1111
+ "epoch": 0.6041999263170822,
1112
+ "grad_norm": 0.4373924136161804,
1113
+ "learning_rate": 1.3960687960687961e-05,
1114
+ "loss": 0.7957,
1115
+ "mean_token_accuracy": 0.8590179920196533,
1116
+ "num_tokens": 6391247.0,
1117
+ "step": 1230
1118
+ },
1119
+ {
1120
+ "epoch": 0.6091121208399852,
1121
+ "grad_norm": 0.30794695019721985,
1122
+ "learning_rate": 1.3911547911547914e-05,
1123
+ "loss": 0.9578,
1124
+ "mean_token_accuracy": 0.8326818406581878,
1125
+ "num_tokens": 6444004.0,
1126
+ "step": 1240
1127
+ },
1128
+ {
1129
+ "epoch": 0.6140243153628884,
1130
+ "grad_norm": 0.3918187618255615,
1131
+ "learning_rate": 1.3862407862407863e-05,
1132
+ "loss": 0.8732,
1133
+ "mean_token_accuracy": 0.8658828616142273,
1134
+ "num_tokens": 6495578.0,
1135
+ "step": 1250
1136
+ },
1137
+ {
1138
+ "epoch": 0.6189365098857915,
1139
+ "grad_norm": 0.2752901315689087,
1140
+ "learning_rate": 1.3813267813267814e-05,
1141
+ "loss": 0.9506,
1142
+ "mean_token_accuracy": 0.8301159009337425,
1143
+ "num_tokens": 6545604.0,
1144
+ "step": 1260
1145
+ },
1146
+ {
1147
+ "epoch": 0.6238487044086946,
1148
+ "grad_norm": 0.460627943277359,
1149
+ "learning_rate": 1.3764127764127766e-05,
1150
+ "loss": 1.0007,
1151
+ "mean_token_accuracy": 0.8227074682712555,
1152
+ "num_tokens": 6594381.0,
1153
+ "step": 1270
1154
+ },
1155
+ {
1156
+ "epoch": 0.6287608989315977,
1157
+ "grad_norm": 0.4065866768360138,
1158
+ "learning_rate": 1.3714987714987715e-05,
1159
+ "loss": 0.8056,
1160
+ "mean_token_accuracy": 0.8646926179528236,
1161
+ "num_tokens": 6649190.0,
1162
+ "step": 1280
1163
+ },
1164
+ {
1165
+ "epoch": 0.6336730934545008,
1166
+ "grad_norm": 0.4343368113040924,
1167
+ "learning_rate": 1.3665847665847667e-05,
1168
+ "loss": 0.8879,
1169
+ "mean_token_accuracy": 0.8385777972638607,
1170
+ "num_tokens": 6701394.0,
1171
+ "step": 1290
1172
+ },
1173
+ {
1174
+ "epoch": 0.6385852879774039,
1175
+ "grad_norm": 0.5364151000976562,
1176
+ "learning_rate": 1.3616707616707618e-05,
1177
+ "loss": 0.9419,
1178
+ "mean_token_accuracy": 0.8308121666312218,
1179
+ "num_tokens": 6752270.0,
1180
+ "step": 1300
1181
+ },
1182
+ {
1183
+ "epoch": 0.643497482500307,
1184
+ "grad_norm": 0.3982967734336853,
1185
+ "learning_rate": 1.3567567567567567e-05,
1186
+ "loss": 1.0305,
1187
+ "mean_token_accuracy": 0.8274823799729347,
1188
+ "num_tokens": 6805857.0,
1189
+ "step": 1310
1190
+ },
1191
+ {
1192
+ "epoch": 0.6484096770232102,
1193
+ "grad_norm": 0.7117066383361816,
1194
+ "learning_rate": 1.351842751842752e-05,
1195
+ "loss": 1.0594,
1196
+ "mean_token_accuracy": 0.8103688299655915,
1197
+ "num_tokens": 6852455.0,
1198
+ "step": 1320
1199
+ },
1200
+ {
1201
+ "epoch": 0.6533218715461132,
1202
+ "grad_norm": 0.3083662986755371,
1203
+ "learning_rate": 1.346928746928747e-05,
1204
+ "loss": 1.0412,
1205
+ "mean_token_accuracy": 0.8186712831258773,
1206
+ "num_tokens": 6896950.0,
1207
+ "step": 1330
1208
+ },
1209
+ {
1210
+ "epoch": 0.6582340660690164,
1211
+ "grad_norm": 0.4959128499031067,
1212
+ "learning_rate": 1.342014742014742e-05,
1213
+ "loss": 0.9348,
1214
+ "mean_token_accuracy": 0.8453299880027771,
1215
+ "num_tokens": 6950814.0,
1216
+ "step": 1340
1217
+ },
1218
+ {
1219
+ "epoch": 0.6631462605919194,
1220
+ "grad_norm": 0.8182358145713806,
1221
+ "learning_rate": 1.3371007371007372e-05,
1222
+ "loss": 1.0552,
1223
+ "mean_token_accuracy": 0.8226786606013775,
1224
+ "num_tokens": 7003460.0,
1225
+ "step": 1350
1226
+ },
1227
+ {
1228
+ "epoch": 0.6680584551148225,
1229
+ "grad_norm": 0.43365928530693054,
1230
+ "learning_rate": 1.3321867321867323e-05,
1231
+ "loss": 0.8795,
1232
+ "mean_token_accuracy": 0.8458721756935119,
1233
+ "num_tokens": 7054901.0,
1234
+ "step": 1360
1235
+ },
1236
+ {
1237
+ "epoch": 0.6729706496377257,
1238
+ "grad_norm": 0.2876506745815277,
1239
+ "learning_rate": 1.3272727272727275e-05,
1240
+ "loss": 0.8282,
1241
+ "mean_token_accuracy": 0.8459444969892502,
1242
+ "num_tokens": 7106038.0,
1243
+ "step": 1370
1244
+ },
1245
+ {
1246
+ "epoch": 0.6778828441606287,
1247
+ "grad_norm": 0.4010944068431854,
1248
+ "learning_rate": 1.3223587223587224e-05,
1249
+ "loss": 0.7836,
1250
+ "mean_token_accuracy": 0.8610207386314869,
1251
+ "num_tokens": 7159651.0,
1252
+ "step": 1380
1253
+ },
1254
+ {
1255
+ "epoch": 0.6827950386835319,
1256
+ "grad_norm": 0.19336983561515808,
1257
+ "learning_rate": 1.3174447174447176e-05,
1258
+ "loss": 1.0039,
1259
+ "mean_token_accuracy": 0.8369563437998295,
1260
+ "num_tokens": 7204811.0,
1261
+ "step": 1390
1262
+ },
1263
+ {
1264
+ "epoch": 0.687707233206435,
1265
+ "grad_norm": 0.5056695938110352,
1266
+ "learning_rate": 1.3125307125307127e-05,
1267
+ "loss": 0.6854,
1268
+ "mean_token_accuracy": 0.878055976331234,
1269
+ "num_tokens": 7256836.0,
1270
+ "step": 1400
1271
+ },
1272
+ {
1273
+ "epoch": 0.692619427729338,
1274
+ "grad_norm": 0.3462684750556946,
1275
+ "learning_rate": 1.3076167076167076e-05,
1276
+ "loss": 1.2627,
1277
+ "mean_token_accuracy": 0.7998166553676128,
1278
+ "num_tokens": 7308913.0,
1279
+ "step": 1410
1280
+ },
1281
+ {
1282
+ "epoch": 0.6975316222522412,
1283
+ "grad_norm": 0.2867954671382904,
1284
+ "learning_rate": 1.3027027027027029e-05,
1285
+ "loss": 0.6718,
1286
+ "mean_token_accuracy": 0.8842718601226807,
1287
+ "num_tokens": 7365814.0,
1288
+ "step": 1420
1289
+ },
1290
+ {
1291
+ "epoch": 0.7024438167751443,
1292
+ "grad_norm": 0.4659688472747803,
1293
+ "learning_rate": 1.297788697788698e-05,
1294
+ "loss": 1.1291,
1295
+ "mean_token_accuracy": 0.8083386868238449,
1296
+ "num_tokens": 7419238.0,
1297
+ "step": 1430
1298
+ },
1299
+ {
1300
+ "epoch": 0.7073560112980474,
1301
+ "grad_norm": 0.41312354803085327,
1302
+ "learning_rate": 1.2928746928746928e-05,
1303
+ "loss": 0.8625,
1304
+ "mean_token_accuracy": 0.8470474675297737,
1305
+ "num_tokens": 7469177.0,
1306
+ "step": 1440
1307
+ },
1308
+ {
1309
+ "epoch": 0.7122682058209505,
1310
+ "grad_norm": 0.5044501423835754,
1311
+ "learning_rate": 1.2879606879606881e-05,
1312
+ "loss": 0.8186,
1313
+ "mean_token_accuracy": 0.8533941991627216,
1314
+ "num_tokens": 7520617.0,
1315
+ "step": 1450
1316
+ },
1317
+ {
1318
+ "epoch": 0.7171804003438537,
1319
+ "grad_norm": 0.3712249994277954,
1320
+ "learning_rate": 1.2830466830466832e-05,
1321
+ "loss": 0.912,
1322
+ "mean_token_accuracy": 0.838789189606905,
1323
+ "num_tokens": 7569361.0,
1324
+ "step": 1460
1325
+ },
1326
+ {
1327
+ "epoch": 0.7220925948667567,
1328
+ "grad_norm": 0.5090559720993042,
1329
+ "learning_rate": 1.2781326781326782e-05,
1330
+ "loss": 0.9559,
1331
+ "mean_token_accuracy": 0.8298265844583511,
1332
+ "num_tokens": 7619902.0,
1333
+ "step": 1470
1334
+ },
1335
+ {
1336
+ "epoch": 0.7270047893896598,
1337
+ "grad_norm": 0.40056052803993225,
1338
+ "learning_rate": 1.2732186732186733e-05,
1339
+ "loss": 0.9697,
1340
+ "mean_token_accuracy": 0.8409195326268672,
1341
+ "num_tokens": 7671364.0,
1342
+ "step": 1480
1343
+ },
1344
+ {
1345
+ "epoch": 0.731916983912563,
1346
+ "grad_norm": 0.5502124428749084,
1347
+ "learning_rate": 1.2683046683046685e-05,
1348
+ "loss": 0.974,
1349
+ "mean_token_accuracy": 0.8395135879516602,
1350
+ "num_tokens": 7721501.0,
1351
+ "step": 1490
1352
+ },
1353
+ {
1354
+ "epoch": 0.736829178435466,
1355
+ "grad_norm": 0.49662184715270996,
1356
+ "learning_rate": 1.2633906633906635e-05,
1357
+ "loss": 0.9764,
1358
+ "mean_token_accuracy": 0.8396995335817337,
1359
+ "num_tokens": 7771721.0,
1360
+ "step": 1500
1361
+ },
1362
+ {
1363
+ "epoch": 0.7417413729583692,
1364
+ "grad_norm": 0.26574891805648804,
1365
+ "learning_rate": 1.2584766584766585e-05,
1366
+ "loss": 0.8165,
1367
+ "mean_token_accuracy": 0.8401172868907452,
1368
+ "num_tokens": 7822046.0,
1369
+ "step": 1510
1370
+ },
1371
+ {
1372
+ "epoch": 0.7466535674812722,
1373
+ "grad_norm": 0.6338578462600708,
1374
+ "learning_rate": 1.2535626535626538e-05,
1375
+ "loss": 0.9305,
1376
+ "mean_token_accuracy": 0.8319102026522159,
1377
+ "num_tokens": 7874800.0,
1378
+ "step": 1520
1379
+ },
1380
+ {
1381
+ "epoch": 0.7515657620041754,
1382
+ "grad_norm": 0.45722728967666626,
1383
+ "learning_rate": 1.2486486486486487e-05,
1384
+ "loss": 0.817,
1385
+ "mean_token_accuracy": 0.865605479478836,
1386
+ "num_tokens": 7928044.0,
1387
+ "step": 1530
1388
+ },
1389
+ {
1390
+ "epoch": 0.7564779565270785,
1391
+ "grad_norm": 0.2256188988685608,
1392
+ "learning_rate": 1.2437346437346437e-05,
1393
+ "loss": 1.0366,
1394
+ "mean_token_accuracy": 0.8243134960532188,
1395
+ "num_tokens": 7976774.0,
1396
+ "step": 1540
1397
+ },
1398
+ {
1399
+ "epoch": 0.7613901510499815,
1400
+ "grad_norm": 0.3194807469844818,
1401
+ "learning_rate": 1.238820638820639e-05,
1402
+ "loss": 0.8638,
1403
+ "mean_token_accuracy": 0.8425570398569107,
1404
+ "num_tokens": 8027431.0,
1405
+ "step": 1550
1406
+ },
1407
+ {
1408
+ "epoch": 0.7663023455728847,
1409
+ "grad_norm": 0.7870245575904846,
1410
+ "learning_rate": 1.2339066339066339e-05,
1411
+ "loss": 1.1827,
1412
+ "mean_token_accuracy": 0.8022145099937916,
1413
+ "num_tokens": 8076248.0,
1414
+ "step": 1560
1415
+ },
1416
+ {
1417
+ "epoch": 0.7712145400957878,
1418
+ "grad_norm": 1.0986965894699097,
1419
+ "learning_rate": 1.2289926289926291e-05,
1420
+ "loss": 1.0736,
1421
+ "mean_token_accuracy": 0.8118120074272156,
1422
+ "num_tokens": 8128333.0,
1423
+ "step": 1570
1424
+ },
1425
+ {
1426
+ "epoch": 0.7761267346186909,
1427
+ "grad_norm": 0.6505934000015259,
1428
+ "learning_rate": 1.2240786240786242e-05,
1429
+ "loss": 0.957,
1430
+ "mean_token_accuracy": 0.8350290179252624,
1431
+ "num_tokens": 8177196.0,
1432
+ "step": 1580
1433
+ },
1434
+ {
1435
+ "epoch": 0.781038929141594,
1436
+ "grad_norm": 0.35571786761283875,
1437
+ "learning_rate": 1.2191646191646194e-05,
1438
+ "loss": 0.8929,
1439
+ "mean_token_accuracy": 0.8486527316272259,
1440
+ "num_tokens": 8229113.0,
1441
+ "step": 1590
1442
+ },
1443
+ {
1444
+ "epoch": 0.7859511236644972,
1445
+ "grad_norm": 0.4012582004070282,
1446
+ "learning_rate": 1.2142506142506144e-05,
1447
+ "loss": 0.7953,
1448
+ "mean_token_accuracy": 0.8502931199967861,
1449
+ "num_tokens": 8283001.0,
1450
+ "step": 1600
1451
+ },
1452
+ {
1453
+ "epoch": 0.7908633181874002,
1454
+ "grad_norm": 0.25499385595321655,
1455
+ "learning_rate": 1.2093366093366094e-05,
1456
+ "loss": 0.9205,
1457
+ "mean_token_accuracy": 0.8501100361347198,
1458
+ "num_tokens": 8334978.0,
1459
+ "step": 1610
1460
+ },
1461
+ {
1462
+ "epoch": 0.7957755127103033,
1463
+ "grad_norm": 0.5996484756469727,
1464
+ "learning_rate": 1.2044226044226047e-05,
1465
+ "loss": 0.9237,
1466
+ "mean_token_accuracy": 0.8439513966441154,
1467
+ "num_tokens": 8390388.0,
1468
+ "step": 1620
1469
+ },
1470
+ {
1471
+ "epoch": 0.8006877072332065,
1472
+ "grad_norm": 0.4831118583679199,
1473
+ "learning_rate": 1.1995085995085996e-05,
1474
+ "loss": 0.7179,
1475
+ "mean_token_accuracy": 0.8848425835371018,
1476
+ "num_tokens": 8445335.0,
1477
+ "step": 1630
1478
+ },
1479
+ {
1480
+ "epoch": 0.8055999017561095,
1481
+ "grad_norm": 0.3229360580444336,
1482
+ "learning_rate": 1.1945945945945946e-05,
1483
+ "loss": 1.1637,
1484
+ "mean_token_accuracy": 0.8010388404130936,
1485
+ "num_tokens": 8494343.0,
1486
+ "step": 1640
1487
+ },
1488
+ {
1489
+ "epoch": 0.8105120962790127,
1490
+ "grad_norm": 1.2905784845352173,
1491
+ "learning_rate": 1.1896805896805899e-05,
1492
+ "loss": 0.9642,
1493
+ "mean_token_accuracy": 0.829357648640871,
1494
+ "num_tokens": 8542794.0,
1495
+ "step": 1650
1496
+ },
1497
+ {
1498
+ "epoch": 0.8154242908019158,
1499
+ "grad_norm": 0.33358246088027954,
1500
+ "learning_rate": 1.1847665847665848e-05,
1501
+ "loss": 0.8557,
1502
+ "mean_token_accuracy": 0.85107681453228,
1503
+ "num_tokens": 8597837.0,
1504
+ "step": 1660
1505
+ },
1506
+ {
1507
+ "epoch": 0.8203364853248188,
1508
+ "grad_norm": 0.30174124240875244,
1509
+ "learning_rate": 1.17985257985258e-05,
1510
+ "loss": 0.6798,
1511
+ "mean_token_accuracy": 0.8848607018589973,
1512
+ "num_tokens": 8651855.0,
1513
+ "step": 1670
1514
+ },
1515
+ {
1516
+ "epoch": 0.825248679847722,
1517
+ "grad_norm": 0.41789162158966064,
1518
+ "learning_rate": 1.1749385749385751e-05,
1519
+ "loss": 0.7969,
1520
+ "mean_token_accuracy": 0.859107281267643,
1521
+ "num_tokens": 8705043.0,
1522
+ "step": 1680
1523
+ },
1524
+ {
1525
+ "epoch": 0.830160874370625,
1526
+ "grad_norm": 0.5565118193626404,
1527
+ "learning_rate": 1.17002457002457e-05,
1528
+ "loss": 0.8947,
1529
+ "mean_token_accuracy": 0.8513899371027946,
1530
+ "num_tokens": 8753735.0,
1531
+ "step": 1690
1532
+ },
1533
+ {
1534
+ "epoch": 0.8350730688935282,
1535
+ "grad_norm": 0.8451042771339417,
1536
+ "learning_rate": 1.1651105651105653e-05,
1537
+ "loss": 1.1201,
1538
+ "mean_token_accuracy": 0.7937561281025409,
1539
+ "num_tokens": 8799611.0,
1540
+ "step": 1700
1541
+ },
1542
+ {
1543
+ "epoch": 0.8399852634164313,
1544
+ "grad_norm": 0.8886920213699341,
1545
+ "learning_rate": 1.1601965601965603e-05,
1546
+ "loss": 0.8187,
1547
+ "mean_token_accuracy": 0.8588108576834201,
1548
+ "num_tokens": 8852789.0,
1549
+ "step": 1710
1550
+ },
1551
+ {
1552
+ "epoch": 0.8448974579393344,
1553
+ "grad_norm": 0.8036173582077026,
1554
+ "learning_rate": 1.1552825552825552e-05,
1555
+ "loss": 0.8139,
1556
+ "mean_token_accuracy": 0.868566033244133,
1557
+ "num_tokens": 8904446.0,
1558
+ "step": 1720
1559
+ },
1560
+ {
1561
+ "epoch": 0.8498096524622375,
1562
+ "grad_norm": 0.8208529949188232,
1563
+ "learning_rate": 1.1503685503685505e-05,
1564
+ "loss": 0.9869,
1565
+ "mean_token_accuracy": 0.8289718352258205,
1566
+ "num_tokens": 8955674.0,
1567
+ "step": 1730
1568
+ },
1569
+ {
1570
+ "epoch": 0.8547218469851406,
1571
+ "grad_norm": 0.48808610439300537,
1572
+ "learning_rate": 1.1454545454545455e-05,
1573
+ "loss": 0.6682,
1574
+ "mean_token_accuracy": 0.8741697937250137,
1575
+ "num_tokens": 9008895.0,
1576
+ "step": 1740
1577
+ },
1578
+ {
1579
+ "epoch": 0.8596340415080437,
1580
+ "grad_norm": 0.30963972210884094,
1581
+ "learning_rate": 1.1405405405405404e-05,
1582
+ "loss": 0.8816,
1583
+ "mean_token_accuracy": 0.843504398316145,
1584
+ "num_tokens": 9059352.0,
1585
+ "step": 1750
1586
+ },
1587
+ {
1588
+ "epoch": 0.8645462360309468,
1589
+ "grad_norm": 0.2593599855899811,
1590
+ "learning_rate": 1.1356265356265357e-05,
1591
+ "loss": 1.0678,
1592
+ "mean_token_accuracy": 0.8158367194235325,
1593
+ "num_tokens": 9105840.0,
1594
+ "step": 1760
1595
+ },
1596
+ {
1597
+ "epoch": 0.86945843055385,
1598
+ "grad_norm": 0.2662312984466553,
1599
+ "learning_rate": 1.1307125307125308e-05,
1600
+ "loss": 0.988,
1601
+ "mean_token_accuracy": 0.8375108957290649,
1602
+ "num_tokens": 9158645.0,
1603
+ "step": 1770
1604
+ },
1605
+ {
1606
+ "epoch": 0.874370625076753,
1607
+ "grad_norm": 0.2985932528972626,
1608
+ "learning_rate": 1.125798525798526e-05,
1609
+ "loss": 0.7872,
1610
+ "mean_token_accuracy": 0.8592289894819259,
1611
+ "num_tokens": 9205661.0,
1612
+ "step": 1780
1613
+ },
1614
+ {
1615
+ "epoch": 0.8792828195996562,
1616
+ "grad_norm": 0.26380449533462524,
1617
+ "learning_rate": 1.1208845208845209e-05,
1618
+ "loss": 1.1826,
1619
+ "mean_token_accuracy": 0.7935345992445946,
1620
+ "num_tokens": 9251283.0,
1621
+ "step": 1790
1622
+ },
1623
+ {
1624
+ "epoch": 0.8841950141225593,
1625
+ "grad_norm": 0.5365442633628845,
1626
+ "learning_rate": 1.1159705159705162e-05,
1627
+ "loss": 0.9483,
1628
+ "mean_token_accuracy": 0.8279004551470279,
1629
+ "num_tokens": 9298506.0,
1630
+ "step": 1800
1631
+ },
1632
+ {
1633
+ "epoch": 0.8891072086454623,
1634
+ "grad_norm": 0.35537609457969666,
1635
+ "learning_rate": 1.1110565110565112e-05,
1636
+ "loss": 0.6907,
1637
+ "mean_token_accuracy": 0.874860267341137,
1638
+ "num_tokens": 9352301.0,
1639
+ "step": 1810
1640
+ },
1641
+ {
1642
+ "epoch": 0.8940194031683655,
1643
+ "grad_norm": 0.36677488684654236,
1644
+ "learning_rate": 1.1061425061425061e-05,
1645
+ "loss": 0.8506,
1646
+ "mean_token_accuracy": 0.847238763421774,
1647
+ "num_tokens": 9401209.0,
1648
+ "step": 1820
1649
+ },
1650
+ {
1651
+ "epoch": 0.8989315976912686,
1652
+ "grad_norm": 0.35230782628059387,
1653
+ "learning_rate": 1.1012285012285014e-05,
1654
+ "loss": 0.9664,
1655
+ "mean_token_accuracy": 0.8267447024583816,
1656
+ "num_tokens": 9451239.0,
1657
+ "step": 1830
1658
+ },
1659
+ {
1660
+ "epoch": 0.9038437922141717,
1661
+ "grad_norm": 0.5595026612281799,
1662
+ "learning_rate": 1.0963144963144964e-05,
1663
+ "loss": 1.0239,
1664
+ "mean_token_accuracy": 0.8194499731063842,
1665
+ "num_tokens": 9504396.0,
1666
+ "step": 1840
1667
+ },
1668
+ {
1669
+ "epoch": 0.9087559867370748,
1670
+ "grad_norm": 0.38404449820518494,
1671
+ "learning_rate": 1.0914004914004913e-05,
1672
+ "loss": 0.9903,
1673
+ "mean_token_accuracy": 0.8269840359687806,
1674
+ "num_tokens": 9555148.0,
1675
+ "step": 1850
1676
+ },
1677
+ {
1678
+ "epoch": 0.9136681812599778,
1679
+ "grad_norm": 0.41196462512016296,
1680
+ "learning_rate": 1.0864864864864866e-05,
1681
+ "loss": 1.0264,
1682
+ "mean_token_accuracy": 0.8328478842973709,
1683
+ "num_tokens": 9606453.0,
1684
+ "step": 1860
1685
+ },
1686
+ {
1687
+ "epoch": 0.918580375782881,
1688
+ "grad_norm": 0.5028640031814575,
1689
+ "learning_rate": 1.0815724815724817e-05,
1690
+ "loss": 0.7749,
1691
+ "mean_token_accuracy": 0.8660773292183876,
1692
+ "num_tokens": 9658869.0,
1693
+ "step": 1870
1694
+ },
1695
+ {
1696
+ "epoch": 0.9234925703057841,
1697
+ "grad_norm": 0.2295975685119629,
1698
+ "learning_rate": 1.0766584766584767e-05,
1699
+ "loss": 0.8817,
1700
+ "mean_token_accuracy": 0.83365648239851,
1701
+ "num_tokens": 9708994.0,
1702
+ "step": 1880
1703
+ },
1704
+ {
1705
+ "epoch": 0.9284047648286872,
1706
+ "grad_norm": 0.24253033101558685,
1707
+ "learning_rate": 1.0717444717444718e-05,
1708
+ "loss": 0.9226,
1709
+ "mean_token_accuracy": 0.8485260397195816,
1710
+ "num_tokens": 9760081.0,
1711
+ "step": 1890
1712
+ },
1713
+ {
1714
+ "epoch": 0.9333169593515903,
1715
+ "grad_norm": 0.2894159257411957,
1716
+ "learning_rate": 1.066830466830467e-05,
1717
+ "loss": 0.7618,
1718
+ "mean_token_accuracy": 0.8669800266623497,
1719
+ "num_tokens": 9810101.0,
1720
+ "step": 1900
1721
+ },
1722
+ {
1723
+ "epoch": 0.9382291538744935,
1724
+ "grad_norm": 0.36487090587615967,
1725
+ "learning_rate": 1.061916461916462e-05,
1726
+ "loss": 0.8968,
1727
+ "mean_token_accuracy": 0.8443298265337944,
1728
+ "num_tokens": 9861741.0,
1729
+ "step": 1910
1730
+ },
1731
+ {
1732
+ "epoch": 0.9431413483973965,
1733
+ "grad_norm": 0.2829186022281647,
1734
+ "learning_rate": 1.057002457002457e-05,
1735
+ "loss": 0.821,
1736
+ "mean_token_accuracy": 0.8466833367943764,
1737
+ "num_tokens": 9913833.0,
1738
+ "step": 1920
1739
+ },
1740
+ {
1741
+ "epoch": 0.9480535429202996,
1742
+ "grad_norm": 0.6794554591178894,
1743
+ "learning_rate": 1.0520884520884523e-05,
1744
+ "loss": 0.7578,
1745
+ "mean_token_accuracy": 0.8747551962733269,
1746
+ "num_tokens": 9970247.0,
1747
+ "step": 1930
1748
+ },
1749
+ {
1750
+ "epoch": 0.9529657374432028,
1751
+ "grad_norm": 0.6977810263633728,
1752
+ "learning_rate": 1.0471744471744472e-05,
1753
+ "loss": 0.8165,
1754
+ "mean_token_accuracy": 0.8567108184099197,
1755
+ "num_tokens": 10019920.0,
1756
+ "step": 1940
1757
+ },
1758
+ {
1759
+ "epoch": 0.9578779319661058,
1760
+ "grad_norm": 0.504741907119751,
1761
+ "learning_rate": 1.0422604422604422e-05,
1762
+ "loss": 0.9742,
1763
+ "mean_token_accuracy": 0.8300105221569538,
1764
+ "num_tokens": 10066930.0,
1765
+ "step": 1950
1766
+ },
1767
+ {
1768
+ "epoch": 0.962790126489009,
1769
+ "grad_norm": 2.085604190826416,
1770
+ "learning_rate": 1.0373464373464375e-05,
1771
+ "loss": 0.9191,
1772
+ "mean_token_accuracy": 0.8302432119846344,
1773
+ "num_tokens": 10116405.0,
1774
+ "step": 1960
1775
+ },
1776
+ {
1777
+ "epoch": 0.9677023210119121,
1778
+ "grad_norm": 0.5094314813613892,
1779
+ "learning_rate": 1.0324324324324324e-05,
1780
+ "loss": 0.6908,
1781
+ "mean_token_accuracy": 0.8804146230220795,
1782
+ "num_tokens": 10171793.0,
1783
+ "step": 1970
1784
+ },
1785
+ {
1786
+ "epoch": 0.9726145155348151,
1787
+ "grad_norm": 0.7494292259216309,
1788
+ "learning_rate": 1.0275184275184276e-05,
1789
+ "loss": 0.9316,
1790
+ "mean_token_accuracy": 0.8486801907420158,
1791
+ "num_tokens": 10223476.0,
1792
+ "step": 1980
1793
+ },
1794
+ {
1795
+ "epoch": 0.9775267100577183,
1796
+ "grad_norm": 0.5121098756790161,
1797
+ "learning_rate": 1.0226044226044227e-05,
1798
+ "loss": 0.8121,
1799
+ "mean_token_accuracy": 0.8445978663861752,
1800
+ "num_tokens": 10276795.0,
1801
+ "step": 1990
1802
+ },
1803
+ {
1804
+ "epoch": 0.9824389045806214,
1805
+ "grad_norm": 0.22474397718906403,
1806
+ "learning_rate": 1.017690417690418e-05,
1807
+ "loss": 0.7328,
1808
+ "mean_token_accuracy": 0.8750261560082435,
1809
+ "num_tokens": 10327432.0,
1810
+ "step": 2000
1811
+ },
1812
+ {
1813
+ "epoch": 0.9873510991035245,
1814
+ "grad_norm": 0.6674897074699402,
1815
+ "learning_rate": 1.0127764127764129e-05,
1816
+ "loss": 0.9154,
1817
+ "mean_token_accuracy": 0.8323157578706741,
1818
+ "num_tokens": 10376006.0,
1819
+ "step": 2010
1820
+ },
1821
+ {
1822
+ "epoch": 0.9922632936264276,
1823
+ "grad_norm": 0.30504295229911804,
1824
+ "learning_rate": 1.007862407862408e-05,
1825
+ "loss": 0.6402,
1826
+ "mean_token_accuracy": 0.891994646191597,
1827
+ "num_tokens": 10431051.0,
1828
+ "step": 2020
1829
+ },
1830
+ {
1831
+ "epoch": 0.9971754881493308,
1832
+ "grad_norm": 0.39028123021125793,
1833
+ "learning_rate": 1.0029484029484032e-05,
1834
+ "loss": 1.0563,
1835
+ "mean_token_accuracy": 0.8198968544602394,
1836
+ "num_tokens": 10482628.0,
1837
+ "step": 2030
1838
+ },
1839
+ {
1840
+ "epoch": 1.0024560972614516,
1841
+ "grad_norm": 0.4253966212272644,
1842
+ "learning_rate": 9.980343980343982e-06,
1843
+ "loss": 1.1883,
1844
+ "mean_token_accuracy": 0.8364348382484622,
1845
+ "num_tokens": 10540047.0,
1846
+ "step": 2040
1847
+ },
1848
+ {
1849
+ "epoch": 1.0073682917843547,
1850
+ "grad_norm": 0.3172409236431122,
1851
+ "learning_rate": 9.931203931203932e-06,
1852
+ "loss": 0.7889,
1853
+ "mean_token_accuracy": 0.8587934225797653,
1854
+ "num_tokens": 10594209.0,
1855
+ "step": 2050
1856
+ },
1857
+ {
1858
+ "epoch": 1.0122804863072579,
1859
+ "grad_norm": 0.2890489101409912,
1860
+ "learning_rate": 9.882063882063882e-06,
1861
+ "loss": 0.948,
1862
+ "mean_token_accuracy": 0.8293856546282768,
1863
+ "num_tokens": 10639797.0,
1864
+ "step": 2060
1865
+ },
1866
+ {
1867
+ "epoch": 1.0171926808301608,
1868
+ "grad_norm": 0.2283569574356079,
1869
+ "learning_rate": 9.832923832923835e-06,
1870
+ "loss": 0.8061,
1871
+ "mean_token_accuracy": 0.8454312592744827,
1872
+ "num_tokens": 10691672.0,
1873
+ "step": 2070
1874
+ },
1875
+ {
1876
+ "epoch": 1.022104875353064,
1877
+ "grad_norm": 0.53035968542099,
1878
+ "learning_rate": 9.783783783783785e-06,
1879
+ "loss": 0.9801,
1880
+ "mean_token_accuracy": 0.8315344356000424,
1881
+ "num_tokens": 10745561.0,
1882
+ "step": 2080
1883
+ },
1884
+ {
1885
+ "epoch": 1.027017069875967,
1886
+ "grad_norm": 0.18628983199596405,
1887
+ "learning_rate": 9.734643734643734e-06,
1888
+ "loss": 0.8569,
1889
+ "mean_token_accuracy": 0.8437845915555954,
1890
+ "num_tokens": 10798814.0,
1891
+ "step": 2090
1892
+ },
1893
+ {
1894
+ "epoch": 1.0319292643988702,
1895
+ "grad_norm": 0.33779898285865784,
1896
+ "learning_rate": 9.685503685503687e-06,
1897
+ "loss": 0.7244,
1898
+ "mean_token_accuracy": 0.871407438069582,
1899
+ "num_tokens": 10850368.0,
1900
+ "step": 2100
1901
+ },
1902
+ {
1903
+ "epoch": 1.0368414589217734,
1904
+ "grad_norm": 0.3872124254703522,
1905
+ "learning_rate": 9.636363636363638e-06,
1906
+ "loss": 0.7797,
1907
+ "mean_token_accuracy": 0.85552798807621,
1908
+ "num_tokens": 10904570.0,
1909
+ "step": 2110
1910
+ },
1911
+ {
1912
+ "epoch": 1.0417536534446765,
1913
+ "grad_norm": 0.3307070732116699,
1914
+ "learning_rate": 9.587223587223588e-06,
1915
+ "loss": 0.71,
1916
+ "mean_token_accuracy": 0.8858561724424362,
1917
+ "num_tokens": 10958970.0,
1918
+ "step": 2120
1919
+ },
1920
+ {
1921
+ "epoch": 1.0466658479675794,
1922
+ "grad_norm": 0.2884276211261749,
1923
+ "learning_rate": 9.538083538083539e-06,
1924
+ "loss": 0.8769,
1925
+ "mean_token_accuracy": 0.8532223105430603,
1926
+ "num_tokens": 11014110.0,
1927
+ "step": 2130
1928
+ },
1929
+ {
1930
+ "epoch": 1.0515780424904826,
1931
+ "grad_norm": 0.3521365225315094,
1932
+ "learning_rate": 9.48894348894349e-06,
1933
+ "loss": 0.7054,
1934
+ "mean_token_accuracy": 0.8789783351123333,
1935
+ "num_tokens": 11069318.0,
1936
+ "step": 2140
1937
+ },
1938
+ {
1939
+ "epoch": 1.0564902370133857,
1940
+ "grad_norm": 0.7055663466453552,
1941
+ "learning_rate": 9.43980343980344e-06,
1942
+ "loss": 0.9756,
1943
+ "mean_token_accuracy": 0.8204010255634785,
1944
+ "num_tokens": 11117893.0,
1945
+ "step": 2150
1946
+ },
1947
+ {
1948
+ "epoch": 1.0614024315362889,
1949
+ "grad_norm": 0.3963095247745514,
1950
+ "learning_rate": 9.390663390663391e-06,
1951
+ "loss": 0.837,
1952
+ "mean_token_accuracy": 0.8513067305088043,
1953
+ "num_tokens": 11172734.0,
1954
+ "step": 2160
1955
+ },
1956
+ {
1957
+ "epoch": 1.066314626059192,
1958
+ "grad_norm": 0.567969024181366,
1959
+ "learning_rate": 9.341523341523342e-06,
1960
+ "loss": 0.983,
1961
+ "mean_token_accuracy": 0.8270547963678837,
1962
+ "num_tokens": 11223656.0,
1963
+ "step": 2170
1964
+ },
1965
+ {
1966
+ "epoch": 1.071226820582095,
1967
+ "grad_norm": 0.47293904423713684,
1968
+ "learning_rate": 9.292383292383293e-06,
1969
+ "loss": 0.862,
1970
+ "mean_token_accuracy": 0.8456276826560497,
1971
+ "num_tokens": 11274037.0,
1972
+ "step": 2180
1973
+ },
1974
+ {
1975
+ "epoch": 1.076139015104998,
1976
+ "grad_norm": 0.2276432365179062,
1977
+ "learning_rate": 9.243243243243243e-06,
1978
+ "loss": 1.0856,
1979
+ "mean_token_accuracy": 0.8078479148447514,
1980
+ "num_tokens": 11324027.0,
1981
+ "step": 2190
1982
+ },
1983
+ {
1984
+ "epoch": 1.0810512096279012,
1985
+ "grad_norm": 0.5022515058517456,
1986
+ "learning_rate": 9.194103194103194e-06,
1987
+ "loss": 0.7426,
1988
+ "mean_token_accuracy": 0.8734906181693077,
1989
+ "num_tokens": 11374912.0,
1990
+ "step": 2200
1991
+ },
1992
+ {
1993
+ "epoch": 1.0859634041508044,
1994
+ "grad_norm": 0.33764153718948364,
1995
+ "learning_rate": 9.144963144963147e-06,
1996
+ "loss": 0.9392,
1997
+ "mean_token_accuracy": 0.840320710837841,
1998
+ "num_tokens": 11429664.0,
1999
+ "step": 2210
2000
+ },
2001
+ {
2002
+ "epoch": 1.0908755986737075,
2003
+ "grad_norm": 0.4857289791107178,
2004
+ "learning_rate": 9.095823095823096e-06,
2005
+ "loss": 0.7073,
2006
+ "mean_token_accuracy": 0.8782361626625061,
2007
+ "num_tokens": 11480264.0,
2008
+ "step": 2220
2009
+ },
2010
+ {
2011
+ "epoch": 1.0957877931966107,
2012
+ "grad_norm": 0.2599787712097168,
2013
+ "learning_rate": 9.046683046683048e-06,
2014
+ "loss": 0.6804,
2015
+ "mean_token_accuracy": 0.8698379978537559,
2016
+ "num_tokens": 11530942.0,
2017
+ "step": 2230
2018
+ },
2019
+ {
2020
+ "epoch": 1.1006999877195136,
2021
+ "grad_norm": 0.5495312213897705,
2022
+ "learning_rate": 8.997542997542999e-06,
2023
+ "loss": 0.9495,
2024
+ "mean_token_accuracy": 0.8306553058326245,
2025
+ "num_tokens": 11582088.0,
2026
+ "step": 2240
2027
+ },
2028
+ {
2029
+ "epoch": 1.1056121822424168,
2030
+ "grad_norm": 0.3552323579788208,
2031
+ "learning_rate": 8.94840294840295e-06,
2032
+ "loss": 0.944,
2033
+ "mean_token_accuracy": 0.8352946802973747,
2034
+ "num_tokens": 11635417.0,
2035
+ "step": 2250
2036
+ },
2037
+ {
2038
+ "epoch": 1.11052437676532,
2039
+ "grad_norm": 0.4491831958293915,
2040
+ "learning_rate": 8.8992628992629e-06,
2041
+ "loss": 0.8545,
2042
+ "mean_token_accuracy": 0.8397120088338852,
2043
+ "num_tokens": 11686180.0,
2044
+ "step": 2260
2045
+ },
2046
+ {
2047
+ "epoch": 1.115436571288223,
2048
+ "grad_norm": 0.319167822599411,
2049
+ "learning_rate": 8.850122850122851e-06,
2050
+ "loss": 0.8211,
2051
+ "mean_token_accuracy": 0.8514452539384365,
2052
+ "num_tokens": 11740948.0,
2053
+ "step": 2270
2054
+ },
2055
+ {
2056
+ "epoch": 1.1203487658111262,
2057
+ "grad_norm": 0.32985174655914307,
2058
+ "learning_rate": 8.800982800982802e-06,
2059
+ "loss": 0.7957,
2060
+ "mean_token_accuracy": 0.8484233103692531,
2061
+ "num_tokens": 11793790.0,
2062
+ "step": 2280
2063
+ },
2064
+ {
2065
+ "epoch": 1.1252609603340291,
2066
+ "grad_norm": 0.24254177510738373,
2067
+ "learning_rate": 8.751842751842752e-06,
2068
+ "loss": 1.1364,
2069
+ "mean_token_accuracy": 0.8023369990289211,
2070
+ "num_tokens": 11843673.0,
2071
+ "step": 2290
2072
+ },
2073
+ {
2074
+ "epoch": 1.1301731548569323,
2075
+ "grad_norm": 0.3639405071735382,
2076
+ "learning_rate": 8.702702702702703e-06,
2077
+ "loss": 0.7676,
2078
+ "mean_token_accuracy": 0.8642957217991352,
2079
+ "num_tokens": 11900152.0,
2080
+ "step": 2300
2081
+ },
2082
+ {
2083
+ "epoch": 1.1350853493798354,
2084
+ "grad_norm": 0.3733905255794525,
2085
+ "learning_rate": 8.653562653562654e-06,
2086
+ "loss": 0.7797,
2087
+ "mean_token_accuracy": 0.8661677822470665,
2088
+ "num_tokens": 11954897.0,
2089
+ "step": 2310
2090
+ },
2091
+ {
2092
+ "epoch": 1.1399975439027386,
2093
+ "grad_norm": 0.6224343776702881,
2094
+ "learning_rate": 8.604422604422605e-06,
2095
+ "loss": 0.7757,
2096
+ "mean_token_accuracy": 0.863005454838276,
2097
+ "num_tokens": 12008625.0,
2098
+ "step": 2320
2099
+ },
2100
+ {
2101
+ "epoch": 1.1449097384256417,
2102
+ "grad_norm": 0.6588796973228455,
2103
+ "learning_rate": 8.555282555282555e-06,
2104
+ "loss": 0.7962,
2105
+ "mean_token_accuracy": 0.8574208118021488,
2106
+ "num_tokens": 12059951.0,
2107
+ "step": 2330
2108
+ },
2109
+ {
2110
+ "epoch": 1.1498219329485448,
2111
+ "grad_norm": 0.35367152094841003,
2112
+ "learning_rate": 8.506142506142508e-06,
2113
+ "loss": 0.7097,
2114
+ "mean_token_accuracy": 0.8655361987650394,
2115
+ "num_tokens": 12110175.0,
2116
+ "step": 2340
2117
+ },
2118
+ {
2119
+ "epoch": 1.154734127471448,
2120
+ "grad_norm": 0.2823013663291931,
2121
+ "learning_rate": 8.457002457002459e-06,
2122
+ "loss": 0.7687,
2123
+ "mean_token_accuracy": 0.8638794094324111,
2124
+ "num_tokens": 12163112.0,
2125
+ "step": 2350
2126
+ },
2127
+ {
2128
+ "epoch": 1.159646321994351,
2129
+ "grad_norm": 0.24300046265125275,
2130
+ "learning_rate": 8.407862407862408e-06,
2131
+ "loss": 0.7411,
2132
+ "mean_token_accuracy": 0.8563850894570351,
2133
+ "num_tokens": 12213345.0,
2134
+ "step": 2360
2135
+ },
2136
+ {
2137
+ "epoch": 1.164558516517254,
2138
+ "grad_norm": 0.5503495335578918,
2139
+ "learning_rate": 8.35872235872236e-06,
2140
+ "loss": 0.8215,
2141
+ "mean_token_accuracy": 0.8549113497138023,
2142
+ "num_tokens": 12264518.0,
2143
+ "step": 2370
2144
+ },
2145
+ {
2146
+ "epoch": 1.1694707110401572,
2147
+ "grad_norm": 1.1128441095352173,
2148
+ "learning_rate": 8.30958230958231e-06,
2149
+ "loss": 0.8864,
2150
+ "mean_token_accuracy": 0.8373782582581043,
2151
+ "num_tokens": 12315487.0,
2152
+ "step": 2380
2153
+ },
2154
+ {
2155
+ "epoch": 1.1743829055630604,
2156
+ "grad_norm": 0.3037548363208771,
2157
+ "learning_rate": 8.260442260442261e-06,
2158
+ "loss": 1.0085,
2159
+ "mean_token_accuracy": 0.829988357424736,
2160
+ "num_tokens": 12369302.0,
2161
+ "step": 2390
2162
+ },
2163
+ {
2164
+ "epoch": 1.1792951000859635,
2165
+ "grad_norm": 0.5334797501564026,
2166
+ "learning_rate": 8.211302211302212e-06,
2167
+ "loss": 0.9124,
2168
+ "mean_token_accuracy": 0.8381602048873902,
2169
+ "num_tokens": 12418500.0,
2170
+ "step": 2400
2171
+ },
2172
+ {
2173
+ "epoch": 1.1842072946088664,
2174
+ "grad_norm": 0.29138582944869995,
2175
+ "learning_rate": 8.162162162162163e-06,
2176
+ "loss": 0.7768,
2177
+ "mean_token_accuracy": 0.8604146704077721,
2178
+ "num_tokens": 12469596.0,
2179
+ "step": 2410
2180
+ },
2181
+ {
2182
+ "epoch": 1.1891194891317696,
2183
+ "grad_norm": 0.3019094467163086,
2184
+ "learning_rate": 8.113022113022114e-06,
2185
+ "loss": 0.92,
2186
+ "mean_token_accuracy": 0.839411947131157,
2187
+ "num_tokens": 12521960.0,
2188
+ "step": 2420
2189
+ },
2190
+ {
2191
+ "epoch": 1.1940316836546727,
2192
+ "grad_norm": 0.37577369809150696,
2193
+ "learning_rate": 8.063882063882064e-06,
2194
+ "loss": 0.7576,
2195
+ "mean_token_accuracy": 0.8685958325862885,
2196
+ "num_tokens": 12572182.0,
2197
+ "step": 2430
2198
+ },
2199
+ {
2200
+ "epoch": 1.1989438781775759,
2201
+ "grad_norm": 1.5805540084838867,
2202
+ "learning_rate": 8.014742014742015e-06,
2203
+ "loss": 0.8623,
2204
+ "mean_token_accuracy": 0.8443202301859856,
2205
+ "num_tokens": 12623257.0,
2206
+ "step": 2440
2207
+ },
2208
+ {
2209
+ "epoch": 1.203856072700479,
2210
+ "grad_norm": 0.5873517394065857,
2211
+ "learning_rate": 7.965601965601968e-06,
2212
+ "loss": 0.8229,
2213
+ "mean_token_accuracy": 0.8458457998931408,
2214
+ "num_tokens": 12673313.0,
2215
+ "step": 2450
2216
+ },
2217
+ {
2218
+ "epoch": 1.2087682672233822,
2219
+ "grad_norm": 0.4337753355503082,
2220
+ "learning_rate": 7.916461916461917e-06,
2221
+ "loss": 0.7783,
2222
+ "mean_token_accuracy": 0.8670792311429978,
2223
+ "num_tokens": 12726249.0,
2224
+ "step": 2460
2225
+ },
2226
+ {
2227
+ "epoch": 1.213680461746285,
2228
+ "grad_norm": 0.2580586075782776,
2229
+ "learning_rate": 7.867321867321867e-06,
2230
+ "loss": 0.8532,
2231
+ "mean_token_accuracy": 0.8490586526691913,
2232
+ "num_tokens": 12779593.0,
2233
+ "step": 2470
2234
+ },
2235
+ {
2236
+ "epoch": 1.2185926562691882,
2237
+ "grad_norm": 0.6018544435501099,
2238
+ "learning_rate": 7.81818181818182e-06,
2239
+ "loss": 0.9073,
2240
+ "mean_token_accuracy": 0.8452610522508621,
2241
+ "num_tokens": 12829244.0,
2242
+ "step": 2480
2243
+ },
2244
+ {
2245
+ "epoch": 1.2235048507920914,
2246
+ "grad_norm": 0.42789220809936523,
2247
+ "learning_rate": 7.76904176904177e-06,
2248
+ "loss": 0.8733,
2249
+ "mean_token_accuracy": 0.8365338027477265,
2250
+ "num_tokens": 12878770.0,
2251
+ "step": 2490
2252
+ },
2253
+ {
2254
+ "epoch": 1.2284170453149945,
2255
+ "grad_norm": 0.34881478548049927,
2256
+ "learning_rate": 7.71990171990172e-06,
2257
+ "loss": 0.7468,
2258
+ "mean_token_accuracy": 0.863375435769558,
2259
+ "num_tokens": 12931644.0,
2260
+ "step": 2500
2261
+ }
2262
+ ],
2263
+ "logging_steps": 10,
2264
+ "max_steps": 4070,
2265
+ "num_input_tokens_seen": 0,
2266
+ "num_train_epochs": 2,
2267
+ "save_steps": 500,
2268
+ "stateful_callbacks": {
2269
+ "TrainerControl": {
2270
+ "args": {
2271
+ "should_epoch_stop": false,
2272
+ "should_evaluate": false,
2273
+ "should_log": false,
2274
+ "should_save": true,
2275
+ "should_training_stop": false
2276
+ },
2277
+ "attributes": {}
2278
+ }
2279
+ },
2280
+ "total_flos": 6216342387720192.0,
2281
+ "train_batch_size": 4,
2282
+ "trial_name": null,
2283
+ "trial_params": null
2284
+ }
checkpoint-2500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f63465acd95be88e019f04b736064426478945f5bc6a4d825ebd1918a32c185
3
+ size 5624
checkpoint-3000/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 256,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-3000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }