nvan15 commited on
Commit
27ff7e3
·
verified ·
1 Parent(s): f750623

Batch upload part 7

Browse files
Files changed (50) hide show
  1. nl_tasks/exps/run_ex21/ft2/adapter_config.json +18 -0
  2. nl_tasks/exps/run_ex22/ft/adapter_config.json +18 -0
  3. nl_tasks/exps/run_ex22/ft/special_tokens_map.json +24 -0
  4. nl_tasks/exps/run_ex22/ft/tokenizer.json +0 -0
  5. nl_tasks/exps/run_ex22/ft/tokenizer.model +3 -0
  6. nl_tasks/exps/run_ex22/ft/tokenizer_config.json +43 -0
  7. nl_tasks/exps/run_ex22/ft2/adapter_config.json +18 -0
  8. nl_tasks/exps/run_ex22/ft2/adapter_model.bin +3 -0
  9. nl_tasks/exps/run_ex22/trainer_state.json +743 -0
  10. nl_tasks/exps/run_ex23/ft/adapter_config.json +18 -0
  11. nl_tasks/exps/run_ex23/ft/special_tokens_map.json +24 -0
  12. nl_tasks/exps/run_ex23/ft/tokenizer.json +0 -0
  13. nl_tasks/exps/run_ex23/ft/tokenizer.model +3 -0
  14. nl_tasks/exps/run_ex23/ft/tokenizer_config.json +43 -0
  15. nl_tasks/exps/run_ex23/ft2/adapter_config.json +18 -0
  16. nl_tasks/exps/run_ex23/ft2/adapter_model.bin +3 -0
  17. nl_tasks/exps/run_ex23/trainer_state.json +1093 -0
  18. nl_tasks/exps/run_ex24/ft/adapter_config.json +18 -0
  19. nl_tasks/exps/run_ex24/ft/special_tokens_map.json +24 -0
  20. nl_tasks/exps/run_ex24/ft/tokenizer.json +0 -0
  21. nl_tasks/exps/run_ex24/ft/tokenizer.model +3 -0
  22. nl_tasks/exps/run_ex24/ft/tokenizer_config.json +43 -0
  23. nl_tasks/exps/run_ex24/ft2/adapter_config.json +18 -0
  24. nl_tasks/exps/run_ex24/ft2/adapter_model.bin +3 -0
  25. nl_tasks/exps/run_ex24/trainer_state.json +1093 -0
  26. nl_tasks/exps/run_ex25/ft/adapter_config.json +18 -0
  27. nl_tasks/exps/run_ex25/ft/special_tokens_map.json +24 -0
  28. nl_tasks/exps/run_ex25/ft/tokenizer.json +0 -0
  29. nl_tasks/exps/run_ex25/ft/tokenizer.model +3 -0
  30. nl_tasks/exps/run_ex25/ft/tokenizer_config.json +43 -0
  31. nl_tasks/exps/run_ex25/ft2/adapter_config.json +18 -0
  32. nl_tasks/exps/run_ex25/ft2/adapter_model.bin +3 -0
  33. nl_tasks/exps/run_ex25/trainer_state.json +1093 -0
  34. nl_tasks/exps/run_ex26/ft/adapter_config.json +18 -0
  35. nl_tasks/exps/run_ex26/ft/special_tokens_map.json +24 -0
  36. nl_tasks/exps/run_ex26/ft/tokenizer.json +0 -0
  37. nl_tasks/exps/run_ex26/ft/tokenizer.model +3 -0
  38. nl_tasks/exps/run_ex26/ft/tokenizer_config.json +43 -0
  39. nl_tasks/exps/run_ex26/ft2/adapter_config.json +18 -0
  40. nl_tasks/exps/run_ex26/ft2/adapter_model.bin +3 -0
  41. nl_tasks/exps/run_ex26/trainer_state.json +1093 -0
  42. nl_tasks/exps/run_ex27/ft/adapter_config.json +18 -0
  43. nl_tasks/exps/run_ex27/ft/special_tokens_map.json +24 -0
  44. nl_tasks/exps/run_ex27/ft/tokenizer.json +0 -0
  45. nl_tasks/exps/run_ex27/ft/tokenizer.model +3 -0
  46. nl_tasks/exps/run_ex27/ft/tokenizer_config.json +43 -0
  47. nl_tasks/exps/run_ex27/ft2/adapter_config.json +18 -0
  48. nl_tasks/exps/run_ex27/ft2/adapter_model.bin +3 -0
  49. nl_tasks/exps/run_ex27/trainer_state.json +1093 -0
  50. nl_tasks/exps/run_ex28/trainer_state.json +1093 -0
nl_tasks/exps/run_ex21/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex22/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex22/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex22/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex22/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex22/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex22/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex22/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92f573cb0b4609d2cb0f402d7e70f2c7fdbe010e9626324a83ade3e5a4836266
3
+ size 33602915
nl_tasks/exps/run_ex22/trainer_state.json ADDED
@@ -0,0 +1,743 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
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": 0.02,
14
+ "grad_norm": 0.5767698884010315,
15
+ "learning_rate": 9.6e-05,
16
+ "loss": 0.8383,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.22841285169124603,
22
+ "learning_rate": 0.00019600000000000002,
23
+ "loss": 0.4641,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.20203162729740143,
29
+ "learning_rate": 0.000296,
30
+ "loss": 0.4022,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.15483643114566803,
36
+ "learning_rate": 0.00039600000000000003,
37
+ "loss": 0.3678,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.2592383027076721,
43
+ "learning_rate": 0.000496,
44
+ "loss": 0.3497,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.1872025728225708,
50
+ "learning_rate": 0.000596,
51
+ "loss": 0.3412,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.18230123817920685,
57
+ "learning_rate": 0.000696,
58
+ "loss": 0.335,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.21839694678783417,
64
+ "learning_rate": 0.000796,
65
+ "loss": 0.3342,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.1940021961927414,
71
+ "learning_rate": 0.000896,
72
+ "loss": 0.3304,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.1906810700893402,
78
+ "learning_rate": 0.000996,
79
+ "loss": 0.3266,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.26560282707214355,
85
+ "learning_rate": 0.0009997192908557321,
86
+ "loss": 0.3267,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.24125456809997559,
92
+ "learning_rate": 0.000998830238119205,
93
+ "loss": 0.336,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.197306290268898,
99
+ "learning_rate": 0.000997333437576437,
100
+ "loss": 0.3246,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.22942069172859192,
106
+ "learning_rate": 0.0009952307128483257,
107
+ "loss": 0.3277,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.28770017623901367,
113
+ "learning_rate": 0.0009925246257810518,
114
+ "loss": 0.3185,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.30775460600852966,
120
+ "learning_rate": 0.0009892184733248665,
121
+ "loss": 0.3169,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.19444003701210022,
127
+ "learning_rate": 0.0009853162835172637,
128
+ "loss": 0.3039,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.23387335240840912,
134
+ "learning_rate": 0.0009808228105754376,
135
+ "loss": 0.3074,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.2246161848306656,
141
+ "learning_rate": 0.0009757435291040016,
142
+ "loss": 0.3023,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.23986800014972687,
148
+ "learning_rate": 0.0009700846274250251,
149
+ "loss": 0.3,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.1957164704799652,
155
+ "learning_rate": 0.000963853000038517,
156
+ "loss": 0.2987,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.24343261122703552,
162
+ "learning_rate": 0.0009570562392225395,
163
+ "loss": 0.2847,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.2263241708278656,
169
+ "learning_rate": 0.0009497026257831855,
170
+ "loss": 0.3051,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.22311192750930786,
176
+ "learning_rate": 0.0009418011189656941,
177
+ "loss": 0.2891,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.21272866427898407,
183
+ "learning_rate": 0.0009333613455389882,
184
+ "loss": 0.303,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.2171148806810379,
190
+ "learning_rate": 0.000924393588066941,
191
+ "loss": 0.2924,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.20269569754600525,
197
+ "learning_rate": 0.0009149087723806549,
198
+ "loss": 0.2799,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.22589229047298431,
204
+ "learning_rate": 0.0009049184542670199,
205
+ "loss": 0.2809,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.23199684917926788,
211
+ "learning_rate": 0.0008944348053897671,
212
+ "loss": 0.2872,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.20177872478961945,
218
+ "learning_rate": 0.0008834705984601709,
219
+ "loss": 0.2834,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.23552103340625763,
225
+ "learning_rate": 0.0008720391916754683,
226
+ "loss": 0.2713,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.23510734736919403,
232
+ "learning_rate": 0.0008601545124439535,
233
+ "loss": 0.2763,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.2040044516324997,
239
+ "learning_rate": 0.0008478310404165754,
240
+ "loss": 0.2682,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.23244167864322662,
246
+ "learning_rate": 0.0008350837898457143,
247
+ "loss": 0.2702,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.16786731779575348,
253
+ "learning_rate": 0.0008219282912926269,
254
+ "loss": 0.2776,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.18036706745624542,
260
+ "learning_rate": 0.0008083805727058513,
261
+ "loss": 0.2702,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.1846761405467987,
267
+ "learning_rate": 0.0007944571398936193,
268
+ "loss": 0.2675,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.18422862887382507,
274
+ "learning_rate": 0.0007801749564140723,
275
+ "loss": 0.2638,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.17170637845993042,
281
+ "learning_rate": 0.0007655514229077783,
282
+ "loss": 0.261,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.22848133742809296,
288
+ "learning_rate": 0.0007506043558977322,
289
+ "loss": 0.2679,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.18595680594444275,
295
+ "learning_rate": 0.0007353519660826664,
296
+ "loss": 0.2659,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.19739358127117157,
302
+ "learning_rate": 0.00071981283615012,
303
+ "loss": 0.2561,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.18625503778457642,
309
+ "learning_rate": 0.0007040058981362964,
310
+ "loss": 0.2521,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.20161661505699158,
316
+ "learning_rate": 0.0006879504103602934,
317
+ "loss": 0.2518,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.18369227647781372,
323
+ "learning_rate": 0.0006716659339608077,
324
+ "loss": 0.2626,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.15661346912384033,
330
+ "learning_rate": 0.0006551723090639006,
331
+ "loss": 0.2535,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.19566485285758972,
337
+ "learning_rate": 0.0006384896306108612,
338
+ "loss": 0.256,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.1498168408870697,
344
+ "learning_rate": 0.0006216382238756146,
345
+ "loss": 0.2557,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.18538439273834229,
351
+ "learning_rate": 0.0006046386197015076,
352
+ "loss": 0.2586,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.17976002395153046,
358
+ "learning_rate": 0.0005875115294876381,
359
+ "loss": 0.2464,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.18410831689834595,
365
+ "learning_rate": 0.0005702778199552054,
366
+ "loss": 0.2234,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.14994341135025024,
372
+ "learning_rate": 0.000552958487724626,
373
+ "loss": 0.2305,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.16480422019958496,
379
+ "learning_rate": 0.0005355746337343836,
380
+ "loss": 0.2271,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.16974572837352753,
386
+ "learning_rate": 0.0005181474375327879,
387
+ "loss": 0.2311,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.16218312084674835,
393
+ "learning_rate": 0.0005006981314739573,
394
+ "loss": 0.2351,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.1601516455411911,
400
+ "learning_rate": 0.00048324797484946424,
401
+ "loss": 0.2329,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.15736624598503113,
407
+ "learning_rate": 0.0004658182279871657,
408
+ "loss": 0.2245,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.16985422372817993,
414
+ "learning_rate": 0.00044843012634876645,
415
+ "loss": 0.2305,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.15030717849731445,
421
+ "learning_rate": 0.000431104854657681,
422
+ "loss": 0.2274,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.1833723485469818,
428
+ "learning_rate": 0.0004138635210887117,
429
+ "loss": 0.2292,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.1704292595386505,
435
+ "learning_rate": 0.0003967271315509884,
436
+ "loss": 0.2317,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.15471942722797394,
442
+ "learning_rate": 0.0003797165640955041,
443
+ "loss": 0.2234,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.14952746033668518,
449
+ "learning_rate": 0.0003628525434784268,
450
+ "loss": 0.2214,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.1543506234884262,
456
+ "learning_rate": 0.0003461556159111748,
457
+ "loss": 0.2159,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.1592603474855423,
463
+ "learning_rate": 0.0003296461240280242,
464
+ "loss": 0.2173,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.16056233644485474,
470
+ "learning_rate": 0.00031334418210174266,
471
+ "loss": 0.2123,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.14286746084690094,
477
+ "learning_rate": 0.0002972696515374455,
478
+ "loss": 0.2152,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.16939015686511993,
484
+ "learning_rate": 0.00028144211667453366,
485
+ "loss": 0.2145,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.15761587023735046,
491
+ "learning_rate": 0.00026588086092619277,
492
+ "loss": 0.2153,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.17639687657356262,
498
+ "learning_rate": 0.00025060484328552466,
499
+ "loss": 0.2156,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.15203754603862762,
505
+ "learning_rate": 0.00023563267522693415,
506
+ "loss": 0.2165,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.1759003847837448,
512
+ "learning_rate": 0.0002209825980309151,
513
+ "loss": 0.2028,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.15083126723766327,
519
+ "learning_rate": 0.00020667246055985938,
520
+ "loss": 0.2097,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.19524288177490234,
526
+ "learning_rate": 0.00019271969751196778,
527
+ "loss": 0.2146,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.1552341729402542,
533
+ "learning_rate": 0.00017914130817975592,
534
+ "loss": 0.2146,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.17365647852420807,
540
+ "learning_rate": 0.00016595383573903412,
541
+ "loss": 0.2116,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.17446695268154144,
547
+ "learning_rate": 0.0001531733470935976,
548
+ "loss": 0.2101,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.15145274996757507,
554
+ "learning_rate": 0.00014081541330017704,
555
+ "loss": 0.2001,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.16965870559215546,
561
+ "learning_rate": 0.00012889509059750602,
562
+ "loss": 0.2146,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.1585461050271988,
568
+ "learning_rate": 0.00011742690206261292,
569
+ "loss": 0.2126,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.15166988968849182,
575
+ "learning_rate": 0.0001064248199166884,
576
+ "loss": 0.1989,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.16476476192474365,
582
+ "learning_rate": 9.590224850208645e-05,
583
+ "loss": 0.208,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.13906484842300415,
589
+ "learning_rate": 8.587200795119792e-05,
590
+ "loss": 0.2127,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.15508808195590973,
596
+ "learning_rate": 7.634631856709389e-05,
597
+ "loss": 0.2087,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.1457161009311676,
603
+ "learning_rate": 6.733678593496901e-05,
604
+ "loss": 0.2074,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.1518443375825882,
610
+ "learning_rate": 5.885438678252342e-05,
611
+ "loss": 0.1996,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.1431756168603897,
617
+ "learning_rate": 5.0909455606510726e-05,
618
+ "loss": 0.2175,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.15276211500167847,
624
+ "learning_rate": 4.3511672081746386e-05,
625
+ "loss": 0.2072,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.15774431824684143,
631
+ "learning_rate": 3.667004926791395e-05,
632
+ "loss": 0.2004,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.1650596261024475,
638
+ "learning_rate": 3.0392922628540875e-05,
639
+ "loss": 0.209,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.14472806453704834,
645
+ "learning_rate": 2.468793987551998e-05,
646
+ "loss": 0.2014,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.16512498259544373,
652
+ "learning_rate": 1.9562051651550784e-05,
653
+ "loss": 0.2066,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.16627739369869232,
659
+ "learning_rate": 1.5021503061851349e-05,
660
+ "loss": 0.2082,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.1565823256969452,
666
+ "learning_rate": 1.1071826065460589e-05,
667
+ "loss": 0.2,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.17333842813968658,
673
+ "learning_rate": 7.717832735397334e-06,
674
+ "loss": 0.2243,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.15922032296657562,
680
+ "learning_rate": 4.963609395891299e-06,
681
+ "loss": 0.2091,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.1890290379524231,
687
+ "learning_rate": 2.81251164382601e-06,
688
+ "loss": 0.2019,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.16079087555408478,
694
+ "learning_rate": 1.267160260461253e-06,
695
+ "loss": 0.2036,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.1514529585838318,
701
+ "learning_rate": 3.2943801841439634e-07,
702
+ "loss": 0.2098,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.14568041265010834,
708
+ "learning_rate": 4.873877924582715e-10,
709
+ "loss": 0.2068,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.0,
714
+ "step": 2500,
715
+ "total_flos": 1.62588235137024e+18,
716
+ "train_loss": 0.2613293670654297,
717
+ "train_runtime": 2323.6228,
718
+ "train_samples_per_second": 34.429,
719
+ "train_steps_per_second": 1.076
720
+ }
721
+ ],
722
+ "logging_steps": 25,
723
+ "max_steps": 2500,
724
+ "num_input_tokens_seen": 0,
725
+ "num_train_epochs": 2,
726
+ "save_steps": 0,
727
+ "stateful_callbacks": {
728
+ "TrainerControl": {
729
+ "args": {
730
+ "should_epoch_stop": false,
731
+ "should_evaluate": false,
732
+ "should_log": false,
733
+ "should_save": true,
734
+ "should_training_stop": true
735
+ },
736
+ "attributes": {}
737
+ }
738
+ },
739
+ "total_flos": 1.62588235137024e+18,
740
+ "train_batch_size": 32,
741
+ "trial_name": null,
742
+ "trial_params": null
743
+ }
nl_tasks/exps/run_ex23/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex23/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex23/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex23/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex23/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex23/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex23/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fe184c2f0054501fd2277b1721d9d9e59e0db117f73a18649bff587c9e5ef2c
3
+ size 33602915
nl_tasks/exps/run_ex23/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.48084303736686707,
15
+ "learning_rate": 6.4e-05,
16
+ "loss": 0.8758,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.2218693643808365,
22
+ "learning_rate": 0.00013066666666666665,
23
+ "loss": 0.5086,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.1952614188194275,
29
+ "learning_rate": 0.00019733333333333335,
30
+ "loss": 0.4161,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.17632220685482025,
36
+ "learning_rate": 0.000264,
37
+ "loss": 0.3766,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.19628162682056427,
43
+ "learning_rate": 0.00033066666666666666,
44
+ "loss": 0.3566,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.2977882921695709,
50
+ "learning_rate": 0.0003973333333333333,
51
+ "loss": 0.3472,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.17526467144489288,
57
+ "learning_rate": 0.00046400000000000006,
58
+ "loss": 0.3401,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.21100583672523499,
64
+ "learning_rate": 0.0005306666666666666,
65
+ "loss": 0.3369,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.1849939078092575,
71
+ "learning_rate": 0.0005973333333333334,
72
+ "loss": 0.3319,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.1735353320837021,
78
+ "learning_rate": 0.0006640000000000001,
79
+ "loss": 0.3244,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.1939871460199356,
85
+ "learning_rate": 0.0007306666666666667,
86
+ "loss": 0.3236,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.20389701426029205,
92
+ "learning_rate": 0.0007973333333333334,
93
+ "loss": 0.3336,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.18185636401176453,
99
+ "learning_rate": 0.000864,
100
+ "loss": 0.3241,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.22774292528629303,
106
+ "learning_rate": 0.0009306666666666667,
107
+ "loss": 0.3284,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.24220959842205048,
113
+ "learning_rate": 0.0009973333333333334,
114
+ "loss": 0.3239,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.2494906187057495,
120
+ "learning_rate": 0.0009998752338940611,
121
+ "loss": 0.3234,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.20020228624343872,
127
+ "learning_rate": 0.0009994799931474475,
128
+ "loss": 0.311,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.21591724455356598,
134
+ "learning_rate": 0.0009988142751731796,
135
+ "loss": 0.3148,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.22068051993846893,
141
+ "learning_rate": 0.0009978784404692847,
142
+ "loss": 0.3071,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.24510782957077026,
148
+ "learning_rate": 0.0009966729958067638,
149
+ "loss": 0.3075,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.2199760228395462,
155
+ "learning_rate": 0.0009951985939551673,
156
+ "loss": 0.3064,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.237228661775589,
162
+ "learning_rate": 0.0009934560333291077,
163
+ "loss": 0.2927,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.24607574939727783,
169
+ "learning_rate": 0.0009914462575559045,
170
+ "loss": 0.3116,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.23642563819885254,
176
+ "learning_rate": 0.000989170354964594,
177
+ "loss": 0.2974,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.22067102789878845,
183
+ "learning_rate": 0.0009866295579965781,
184
+ "loss": 0.3096,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.2489147186279297,
190
+ "learning_rate": 0.000983825242538238,
191
+ "loss": 0.3007,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.20756591856479645,
197
+ "learning_rate": 0.0009807589271758651,
198
+ "loss": 0.2873,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.2658780515193939,
204
+ "learning_rate": 0.0009774322723733215,
205
+ "loss": 0.2903,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.2340981364250183,
211
+ "learning_rate": 0.0009738470795728685,
212
+ "loss": 0.2951,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.23202940821647644,
218
+ "learning_rate": 0.000970005290219654,
219
+ "loss": 0.2894,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.22148098051548004,
225
+ "learning_rate": 0.0009659089847103863,
226
+ "loss": 0.2785,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.23035770654678345,
232
+ "learning_rate": 0.0009615603812667617,
233
+ "loss": 0.2833,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.20710399746894836,
239
+ "learning_rate": 0.0009569618347342592,
240
+ "loss": 0.274,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.21183089911937714,
246
+ "learning_rate": 0.0009521158353069494,
247
+ "loss": 0.2765,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.19836094975471497,
253
+ "learning_rate": 0.0009470250071790107,
254
+ "loss": 0.2846,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.19088484346866608,
260
+ "learning_rate": 0.0009416921071236822,
261
+ "loss": 0.2768,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.20932093262672424,
267
+ "learning_rate": 0.0009361200230004219,
268
+ "loss": 0.275,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.1993676722049713,
274
+ "learning_rate": 0.0009303117721910801,
275
+ "loss": 0.2717,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.19074216485023499,
281
+ "learning_rate": 0.0009242704999659339,
282
+ "loss": 0.2681,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.3126646876335144,
288
+ "learning_rate": 0.0009179994777804677,
289
+ "loss": 0.2749,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.21068663895130157,
295
+ "learning_rate": 0.0009115021015038226,
296
+ "loss": 0.2737,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.22428499162197113,
302
+ "learning_rate": 0.0009047818895798731,
303
+ "loss": 0.2631,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.2032303512096405,
309
+ "learning_rate": 0.0008978424811219277,
310
+ "loss": 0.2573,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.17513667047023773,
316
+ "learning_rate": 0.0008906876339420851,
317
+ "loss": 0.2591,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.19266444444656372,
323
+ "learning_rate": 0.0008833212225163124,
324
+ "loss": 0.2688,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.1570984125137329,
330
+ "learning_rate": 0.0008757472358863481,
331
+ "loss": 0.2592,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.21406503021717072,
337
+ "learning_rate": 0.0008679697754995655,
338
+ "loss": 0.2634,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.17088258266448975,
344
+ "learning_rate": 0.0008599930529879669,
345
+ "loss": 0.2606,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.18541710078716278,
351
+ "learning_rate": 0.0008518213878875102,
352
+ "loss": 0.2643,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.24920210242271423,
358
+ "learning_rate": 0.0008434592052990044,
359
+ "loss": 0.2529,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.1960878223180771,
365
+ "learning_rate": 0.0008349110334918391,
366
+ "loss": 0.2325,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.15122298896312714,
372
+ "learning_rate": 0.0008261815014518466,
373
+ "loss": 0.2407,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.16863854229450226,
379
+ "learning_rate": 0.000817275336374625,
380
+ "loss": 0.236,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.18454797565937042,
386
+ "learning_rate": 0.0008081973611056783,
387
+ "loss": 0.2401,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.14638011157512665,
393
+ "learning_rate": 0.0007989524915287594,
394
+ "loss": 0.2445,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.23536516726016998,
400
+ "learning_rate": 0.000789545733903834,
401
+ "loss": 0.2427,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.1823291927576065,
407
+ "learning_rate": 0.0007799821821561011,
408
+ "loss": 0.2342,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.16896045207977295,
414
+ "learning_rate": 0.0007702670151175434,
415
+ "loss": 0.2422,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.14874465763568878,
421
+ "learning_rate": 0.0007604054937224989,
422
+ "loss": 0.2367,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.17434418201446533,
428
+ "learning_rate": 0.00075040295815877,
429
+ "loss": 0.2394,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.1558532416820526,
435
+ "learning_rate": 0.0007402648249758203,
436
+ "loss": 0.2439,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.14951705932617188,
442
+ "learning_rate": 0.0007299965841516163,
443
+ "loss": 0.2324,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.17535166442394257,
449
+ "learning_rate": 0.0007196037961197066,
450
+ "loss": 0.2328,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.14109523594379425,
456
+ "learning_rate": 0.0007090920887581507,
457
+ "loss": 0.2274,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.16802997887134552,
463
+ "learning_rate": 0.0006984671543419228,
464
+ "loss": 0.2278,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.16117294132709503,
470
+ "learning_rate": 0.0006877347464604445,
471
+ "loss": 0.2217,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.13962742686271667,
477
+ "learning_rate": 0.0006769006769019148,
478
+ "loss": 0.2262,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.1441992223262787,
484
+ "learning_rate": 0.0006659708125061241,
485
+ "loss": 0.2255,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.14966872334480286,
491
+ "learning_rate": 0.0006549510719874577,
492
+ "loss": 0.2267,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.1714620292186737,
498
+ "learning_rate": 0.0006438474227298065,
499
+ "loss": 0.2267,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.14379504323005676,
505
+ "learning_rate": 0.0006326658775551235,
506
+ "loss": 0.2296,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.16740331053733826,
512
+ "learning_rate": 0.0006214124914673755,
513
+ "loss": 0.2148,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.13457053899765015,
519
+ "learning_rate": 0.0006100933583736508,
520
+ "loss": 0.2218,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.191158264875412,
526
+ "learning_rate": 0.0005987146077842015,
527
+ "loss": 0.2249,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.14882108569145203,
533
+ "learning_rate": 0.000587282401493205,
534
+ "loss": 0.2261,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.16314193606376648,
540
+ "learning_rate": 0.0005758029302420446,
541
+ "loss": 0.2235,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.14912760257720947,
547
+ "learning_rate": 0.0005642824103669125,
548
+ "loss": 0.222,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.15002597868442535,
554
+ "learning_rate": 0.000552727080432556,
555
+ "loss": 0.2117,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.1659402698278427,
561
+ "learning_rate": 0.0005411431978539828,
562
+ "loss": 0.2265,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.14878462255001068,
568
+ "learning_rate": 0.0005295370355079614,
569
+ "loss": 0.2254,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.13876402378082275,
575
+ "learning_rate": 0.0005179148783361473,
576
+ "loss": 0.2101,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.13744057714939117,
582
+ "learning_rate": 0.0005062830199416764,
583
+ "loss": 0.219,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.13792552053928375,
589
+ "learning_rate": 0.0004946477591810677,
590
+ "loss": 0.2243,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.1432466059923172,
596
+ "learning_rate": 0.00048301539675328197,
597
+ "loss": 0.2211,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.14318040013313293,
603
+ "learning_rate": 0.000471392231787781,
604
+ "loss": 0.2197,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.14931629598140717,
610
+ "learning_rate": 0.0004597845584334386,
611
+ "loss": 0.2107,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.12764166295528412,
617
+ "learning_rate": 0.00044819866245014836,
618
+ "loss": 0.228,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.13491638004779816,
624
+ "learning_rate": 0.0004366408178049728,
625
+ "loss": 0.2184,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.14066922664642334,
631
+ "learning_rate": 0.0004251172832746799,
632
+ "loss": 0.2109,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.13541120290756226,
638
+ "learning_rate": 0.00041363429905650543,
639
+ "loss": 0.2197,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.1387101262807846,
645
+ "learning_rate": 0.00040219808338897594,
646
+ "loss": 0.2126,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.1422133594751358,
652
+ "learning_rate": 0.0003908148291846224,
653
+ "loss": 0.2156,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.13704848289489746,
659
+ "learning_rate": 0.00037949070067640927,
660
+ "loss": 0.2182,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.1567024290561676,
666
+ "learning_rate": 0.00036823183007969373,
667
+ "loss": 0.2096,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.16875536739826202,
673
+ "learning_rate": 0.00035704431427152237,
674
+ "loss": 0.2333,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.15320144593715668,
680
+ "learning_rate": 0.00034593421148906525,
681
+ "loss": 0.2176,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.16132648289203644,
687
+ "learning_rate": 0.0003349075380489731,
688
+ "loss": 0.2105,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.1457502841949463,
694
+ "learning_rate": 0.0003239702650894364,
695
+ "loss": 0.2116,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.13578177988529205,
701
+ "learning_rate": 0.00031312831533670773,
702
+ "loss": 0.2171,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.1279817819595337,
708
+ "learning_rate": 0.0003023875598978419,
709
+ "loss": 0.2128,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.1550670564174652,
715
+ "learning_rate": 0.0002917538150813876,
716
+ "loss": 0.1909,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.1622520536184311,
722
+ "learning_rate": 0.00028123283924775356,
723
+ "loss": 0.1833,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.1415659636259079,
729
+ "learning_rate": 0.00027083032969095505,
730
+ "loss": 0.1915,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.15416212379932404,
736
+ "learning_rate": 0.0002605519195534288,
737
+ "loss": 0.1834,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.14025457203388214,
743
+ "learning_rate": 0.00025040317477558616,
744
+ "loss": 0.1903,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.14042459428310394,
750
+ "learning_rate": 0.00024038959108175928,
751
+ "loss": 0.184,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.13581395149230957,
757
+ "learning_rate": 0.0002305165910041689,
758
+ "loss": 0.1909,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.15179701149463654,
764
+ "learning_rate": 0.00022078952094652704,
765
+ "loss": 0.187,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.1464652419090271,
771
+ "learning_rate": 0.0002112136482888663,
772
+ "loss": 0.1939,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.15152765810489655,
778
+ "learning_rate": 0.00020179415853515908,
779
+ "loss": 0.1894,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.13474231958389282,
785
+ "learning_rate": 0.00019253615250527738,
786
+ "loss": 0.1832,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.1342306286096573,
792
+ "learning_rate": 0.00018344464357280722,
793
+ "loss": 0.1838,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.13246561586856842,
799
+ "learning_rate": 0.00017452455495021963,
800
+ "loss": 0.1942,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.13951082527637482,
806
+ "learning_rate": 0.00016578071702286396,
807
+ "loss": 0.1816,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.13456101715564728,
813
+ "learning_rate": 0.00015721786473322824,
814
+ "loss": 0.185,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.13722851872444153,
820
+ "learning_rate": 0.0001488406350168854,
821
+ "loss": 0.193,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.12870334088802338,
827
+ "learning_rate": 0.00014065356429151033,
828
+ "loss": 0.1819,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.14238068461418152,
834
+ "learning_rate": 0.0001326610860003293,
835
+ "loss": 0.1884,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.13999466598033905,
841
+ "learning_rate": 0.00012486752821133312,
842
+ "loss": 0.1822,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.1360911726951599,
848
+ "learning_rate": 0.00011727711127355117,
849
+ "loss": 0.183,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.14922741055488586,
855
+ "learning_rate": 0.00010989394553165833,
856
+ "loss": 0.1861,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.1330823451280594,
862
+ "learning_rate": 0.00010272202910015082,
863
+ "loss": 0.1885,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.14635007083415985,
869
+ "learning_rate": 9.576524569829692e-05,
870
+ "loss": 0.1871,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.15063054859638214,
876
+ "learning_rate": 8.902736254703348e-05,
877
+ "loss": 0.1822,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.15703986585140228,
883
+ "learning_rate": 8.251202832895066e-05,
884
+ "loss": 0.1895,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.14821434020996094,
890
+ "learning_rate": 7.622277121246512e-05,
891
+ "loss": 0.1825,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.14606979489326477,
897
+ "learning_rate": 7.01629969412545e-05,
898
+ "loss": 0.1865,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.1357213705778122,
904
+ "learning_rate": 6.433598698998766e-05,
905
+ "loss": 0.1827,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.13560768961906433,
911
+ "learning_rate": 5.874489678734812e-05,
912
+ "loss": 0.1791,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.15358999371528625,
918
+ "learning_rate": 5.3392754007313305e-05,
919
+ "loss": 0.1802,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.1459738314151764,
925
+ "learning_rate": 4.828245692961608e-05,
926
+ "loss": 0.1823,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.13794459402561188,
932
+ "learning_rate": 4.341677287027529e-05,
933
+ "loss": 0.1812,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.1369146853685379,
939
+ "learning_rate": 3.8798336683045054e-05,
940
+ "loss": 0.1845,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.1478826403617859,
946
+ "learning_rate": 3.442964933259474e-05,
947
+ "loss": 0.1863,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.13746634125709534,
953
+ "learning_rate": 3.0313076540192996e-05,
954
+ "loss": 0.183,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.1631135642528534,
960
+ "learning_rate": 2.6450847502627883e-05,
961
+ "loss": 0.1944,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.11646343022584915,
967
+ "learning_rate": 2.2845053685056904e-05,
968
+ "loss": 0.185,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.13773666322231293,
974
+ "learning_rate": 1.9497647688442477e-05,
975
+ "loss": 0.1759,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.16493980586528778,
981
+ "learning_rate": 1.6410442192183573e-05,
982
+ "loss": 0.1789,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.14070625603199005,
988
+ "learning_rate": 1.3585108972518078e-05,
989
+ "loss": 0.1804,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.142463818192482,
995
+ "learning_rate": 1.1023177997226297e-05,
996
+ "loss": 0.1834,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.1492580622434616,
1002
+ "learning_rate": 8.72603659712662e-06,
1003
+ "loss": 0.1791,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.13910527527332306,
1009
+ "learning_rate": 6.694928714811255e-06,
1010
+ "loss": 0.1765,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.14240440726280212,
1016
+ "learning_rate": 4.93095423102935e-06,
1017
+ "loss": 0.1841,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.14672136306762695,
1023
+ "learning_rate": 3.435068369082306e-06,
1024
+ "loss": 0.1847,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.16310083866119385,
1030
+ "learning_rate": 2.2080811775535003e-06,
1031
+ "loss": 0.1822,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.14073088765144348,
1037
+ "learning_rate": 1.2506570916523408e-06,
1038
+ "loss": 0.1771,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.14259496331214905,
1044
+ "learning_rate": 5.633145734114664e-07,
1045
+ "loss": 0.1789,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.1306188851594925,
1051
+ "learning_rate": 1.4642583092999705e-07,
1052
+ "loss": 0.1778,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.12538310885429382,
1058
+ "learning_rate": 2.166168162065496e-10,
1059
+ "loss": 0.1828,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.24165712381998697,
1067
+ "train_runtime": 3457.8533,
1068
+ "train_samples_per_second": 34.704,
1069
+ "train_steps_per_second": 1.084
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }
nl_tasks/exps/run_ex24/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex24/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex24/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex24/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex24/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex24/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex24/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9286ac0906fe2d5e4381d3fac1b20723cd01c3f173e47f3c779884b82886b499
3
+ size 33602915
nl_tasks/exps/run_ex24/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.20721325278282166,
15
+ "learning_rate": 0.00064,
16
+ "loss": 0.6429,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.2121923863887787,
22
+ "learning_rate": 0.0013066666666666665,
23
+ "loss": 0.3881,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.33458471298217773,
29
+ "learning_rate": 0.0019733333333333334,
30
+ "loss": 0.3796,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.4793953597545624,
36
+ "learning_rate": 0.00264,
37
+ "loss": 0.3743,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 1.7511122226715088,
43
+ "learning_rate": 0.0033066666666666665,
44
+ "loss": 0.3919,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.4728436768054962,
50
+ "learning_rate": 0.003973333333333333,
51
+ "loss": 0.4166,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.6393271088600159,
57
+ "learning_rate": 0.00464,
58
+ "loss": 0.3913,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.368111252784729,
64
+ "learning_rate": 0.005306666666666666,
65
+ "loss": 0.3718,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.33478349447250366,
71
+ "learning_rate": 0.005973333333333334,
72
+ "loss": 0.3573,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.1816096007823944,
78
+ "learning_rate": 0.00664,
79
+ "loss": 0.3395,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.1968374103307724,
85
+ "learning_rate": 0.007306666666666667,
86
+ "loss": 0.3327,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.22355352342128754,
92
+ "learning_rate": 0.007973333333333334,
93
+ "loss": 0.3403,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.09445228427648544,
99
+ "learning_rate": 0.00864,
100
+ "loss": 0.3281,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.11095510423183441,
106
+ "learning_rate": 0.009306666666666666,
107
+ "loss": 0.3263,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.12437857687473297,
113
+ "learning_rate": 0.009973333333333332,
114
+ "loss": 0.3186,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.12249485403299332,
120
+ "learning_rate": 0.009998752338940611,
121
+ "loss": 0.3178,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.09227700531482697,
127
+ "learning_rate": 0.009994799931474474,
128
+ "loss": 0.3047,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.08722732216119766,
134
+ "learning_rate": 0.009988142751731795,
135
+ "loss": 0.3051,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.0776078999042511,
141
+ "learning_rate": 0.009978784404692846,
142
+ "loss": 0.2982,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.07535282522439957,
148
+ "learning_rate": 0.009966729958067638,
149
+ "loss": 0.2927,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.0663604885339737,
155
+ "learning_rate": 0.009951985939551673,
156
+ "loss": 0.2934,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.0725964903831482,
162
+ "learning_rate": 0.009934560333291076,
163
+ "loss": 0.2794,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.07474663853645325,
169
+ "learning_rate": 0.009914462575559044,
170
+ "loss": 0.2972,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.06644977629184723,
176
+ "learning_rate": 0.009891703549645938,
177
+ "loss": 0.2811,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.055261533707380295,
183
+ "learning_rate": 0.009866295579965781,
184
+ "loss": 0.2946,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.05713541805744171,
190
+ "learning_rate": 0.009838252425382379,
191
+ "loss": 0.2857,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.049348220229148865,
197
+ "learning_rate": 0.00980758927175865,
198
+ "loss": 0.2728,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.08552377671003342,
204
+ "learning_rate": 0.009774322723733216,
205
+ "loss": 0.2747,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.06687381863594055,
211
+ "learning_rate": 0.009738470795728685,
212
+ "loss": 0.2803,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.06733611971139908,
218
+ "learning_rate": 0.00970005290219654,
219
+ "loss": 0.2782,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.0917917937040329,
225
+ "learning_rate": 0.009659089847103862,
226
+ "loss": 0.2664,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.05817510187625885,
232
+ "learning_rate": 0.009615603812667618,
233
+ "loss": 0.273,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.057370614260435104,
239
+ "learning_rate": 0.009569618347342592,
240
+ "loss": 0.265,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.05295673757791519,
246
+ "learning_rate": 0.009521158353069494,
247
+ "loss": 0.2661,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.04886120930314064,
253
+ "learning_rate": 0.009470250071790108,
254
+ "loss": 0.274,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.053231801837682724,
260
+ "learning_rate": 0.009416921071236822,
261
+ "loss": 0.2685,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.056384552270174026,
267
+ "learning_rate": 0.009361200230004219,
268
+ "loss": 0.2658,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.05400422215461731,
274
+ "learning_rate": 0.009303117721910802,
275
+ "loss": 0.2637,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.04524515941739082,
281
+ "learning_rate": 0.009242704999659339,
282
+ "loss": 0.2593,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.06007376313209534,
288
+ "learning_rate": 0.009179994777804677,
289
+ "loss": 0.267,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.05714306980371475,
295
+ "learning_rate": 0.009115021015038227,
296
+ "loss": 0.2662,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.0461350679397583,
302
+ "learning_rate": 0.009047818895798731,
303
+ "loss": 0.2573,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.046251099556684494,
309
+ "learning_rate": 0.008978424811219276,
310
+ "loss": 0.2522,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.051277272403240204,
316
+ "learning_rate": 0.00890687633942085,
317
+ "loss": 0.2537,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.05851563438773155,
323
+ "learning_rate": 0.008833212225163124,
324
+ "loss": 0.2625,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.045936476439237595,
330
+ "learning_rate": 0.008757472358863481,
331
+ "loss": 0.2557,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.047913141548633575,
337
+ "learning_rate": 0.008679697754995655,
338
+ "loss": 0.2603,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.041925642639398575,
344
+ "learning_rate": 0.008599930529879668,
345
+ "loss": 0.2584,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.0477525033056736,
351
+ "learning_rate": 0.008518213878875101,
352
+ "loss": 0.2607,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.04758713021874428,
358
+ "learning_rate": 0.008434592052990044,
359
+ "loss": 0.2498,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.046754106879234314,
365
+ "learning_rate": 0.00834911033491839,
366
+ "loss": 0.226,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.03786053881049156,
372
+ "learning_rate": 0.008261815014518465,
373
+ "loss": 0.2346,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.04742316156625748,
379
+ "learning_rate": 0.00817275336374625,
380
+ "loss": 0.2312,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.04774940013885498,
386
+ "learning_rate": 0.008081973611056783,
387
+ "loss": 0.236,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.035334907472133636,
393
+ "learning_rate": 0.007989524915287595,
394
+ "loss": 0.2396,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.04027274250984192,
400
+ "learning_rate": 0.00789545733903834,
401
+ "loss": 0.2381,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.03744512051343918,
407
+ "learning_rate": 0.00779982182156101,
408
+ "loss": 0.2301,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.06715697795152664,
414
+ "learning_rate": 0.007702670151175435,
415
+ "loss": 0.237,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.0376882441341877,
421
+ "learning_rate": 0.007604054937224989,
422
+ "loss": 0.2345,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.04710308834910393,
428
+ "learning_rate": 0.0075040295815877005,
429
+ "loss": 0.2365,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.03648149594664574,
435
+ "learning_rate": 0.007402648249758204,
436
+ "loss": 0.2401,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.03415144234895706,
442
+ "learning_rate": 0.007299965841516163,
443
+ "loss": 0.2313,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.039939895272254944,
449
+ "learning_rate": 0.007196037961197066,
450
+ "loss": 0.2303,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.0336175300180912,
456
+ "learning_rate": 0.007090920887581507,
457
+ "loss": 0.2253,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.03801807761192322,
463
+ "learning_rate": 0.0069846715434192285,
464
+ "loss": 0.2265,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.03818347677588463,
470
+ "learning_rate": 0.006877347464604445,
471
+ "loss": 0.2201,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.03448287397623062,
477
+ "learning_rate": 0.006769006769019147,
478
+ "loss": 0.2251,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.035138051956892014,
484
+ "learning_rate": 0.006659708125061241,
485
+ "loss": 0.2244,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.03661292791366577,
491
+ "learning_rate": 0.006549510719874577,
492
+ "loss": 0.2252,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.038888540118932724,
498
+ "learning_rate": 0.0064384742272980645,
499
+ "loss": 0.2252,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.03597492724657059,
505
+ "learning_rate": 0.006326658775551235,
506
+ "loss": 0.2276,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.04128337651491165,
512
+ "learning_rate": 0.006214124914673755,
513
+ "loss": 0.2139,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.03295692428946495,
519
+ "learning_rate": 0.006100933583736507,
520
+ "loss": 0.2216,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.04477379098534584,
526
+ "learning_rate": 0.005987146077842015,
527
+ "loss": 0.2247,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.0380830354988575,
533
+ "learning_rate": 0.00587282401493205,
534
+ "loss": 0.2254,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.03974383324384689,
540
+ "learning_rate": 0.005758029302420446,
541
+ "loss": 0.224,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.03354157134890556,
547
+ "learning_rate": 0.005642824103669126,
548
+ "loss": 0.2209,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.03308559209108353,
554
+ "learning_rate": 0.00552727080432556,
555
+ "loss": 0.2112,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.04159720614552498,
561
+ "learning_rate": 0.005411431978539828,
562
+ "loss": 0.226,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.034576594829559326,
568
+ "learning_rate": 0.005295370355079614,
569
+ "loss": 0.2247,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.0362069196999073,
575
+ "learning_rate": 0.005179148783361473,
576
+ "loss": 0.2097,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.04194594547152519,
582
+ "learning_rate": 0.005062830199416763,
583
+ "loss": 0.2194,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.032347146421670914,
589
+ "learning_rate": 0.004946477591810677,
590
+ "loss": 0.2242,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.033743757754564285,
596
+ "learning_rate": 0.00483015396753282,
597
+ "loss": 0.2209,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.03348388895392418,
603
+ "learning_rate": 0.00471392231787781,
604
+ "loss": 0.2188,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.0340593196451664,
610
+ "learning_rate": 0.004597845584334386,
611
+ "loss": 0.2115,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.030615000054240227,
617
+ "learning_rate": 0.004481986624501484,
618
+ "loss": 0.2279,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.033475857228040695,
624
+ "learning_rate": 0.004366408178049728,
625
+ "loss": 0.2186,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.03454643115401268,
631
+ "learning_rate": 0.004251172832746799,
632
+ "loss": 0.2099,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.03211412578821182,
638
+ "learning_rate": 0.004136342990565055,
639
+ "loss": 0.2206,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.03475036099553108,
645
+ "learning_rate": 0.00402198083388976,
646
+ "loss": 0.2119,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.04023045673966408,
652
+ "learning_rate": 0.0039081482918462244,
653
+ "loss": 0.215,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.03013662062585354,
659
+ "learning_rate": 0.003794907006764093,
660
+ "loss": 0.2189,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.03530523553490639,
666
+ "learning_rate": 0.0036823183007969372,
667
+ "loss": 0.2099,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.038400132209062576,
673
+ "learning_rate": 0.003570443142715224,
674
+ "loss": 0.234,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.03556523099541664,
680
+ "learning_rate": 0.0034593421148906523,
681
+ "loss": 0.2187,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.04098186269402504,
687
+ "learning_rate": 0.0033490753804897313,
688
+ "loss": 0.21,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.036452557891607285,
694
+ "learning_rate": 0.0032397026508943637,
695
+ "loss": 0.2118,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.03421608358621597,
701
+ "learning_rate": 0.003131283153367077,
702
+ "loss": 0.2171,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.030221326276659966,
708
+ "learning_rate": 0.003023875598978419,
709
+ "loss": 0.2125,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.035226210951805115,
715
+ "learning_rate": 0.002917538150813876,
716
+ "loss": 0.1886,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.03686540946364403,
722
+ "learning_rate": 0.0028123283924775355,
723
+ "loss": 0.1823,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.03392050787806511,
729
+ "learning_rate": 0.0027083032969095508,
730
+ "loss": 0.1899,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.03575480356812477,
736
+ "learning_rate": 0.002605519195534288,
737
+ "loss": 0.1812,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.031089287251234055,
743
+ "learning_rate": 0.0025040317477558614,
744
+ "loss": 0.1897,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.031031444668769836,
750
+ "learning_rate": 0.002403895910817593,
751
+ "loss": 0.1816,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.03138287365436554,
757
+ "learning_rate": 0.002305165910041689,
758
+ "loss": 0.1891,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.035979412496089935,
764
+ "learning_rate": 0.0022078952094652703,
765
+ "loss": 0.1854,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.03277301788330078,
771
+ "learning_rate": 0.0021121364828886632,
772
+ "loss": 0.1917,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.03231624513864517,
778
+ "learning_rate": 0.002017941585351591,
779
+ "loss": 0.1872,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.03247809037566185,
785
+ "learning_rate": 0.001925361525052774,
786
+ "loss": 0.1817,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.030497411265969276,
792
+ "learning_rate": 0.001834446435728072,
793
+ "loss": 0.1821,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.03446466475725174,
799
+ "learning_rate": 0.0017452455495021962,
800
+ "loss": 0.1925,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.03429936617612839,
806
+ "learning_rate": 0.0016578071702286396,
807
+ "loss": 0.1798,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.030741166323423386,
813
+ "learning_rate": 0.0015721786473322824,
814
+ "loss": 0.1842,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.02960963360965252,
820
+ "learning_rate": 0.0014884063501688538,
821
+ "loss": 0.1913,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.026998884975910187,
827
+ "learning_rate": 0.0014065356429151033,
828
+ "loss": 0.18,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.04103350639343262,
834
+ "learning_rate": 0.0013266108600032927,
835
+ "loss": 0.1867,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.03400329872965813,
841
+ "learning_rate": 0.0012486752821133313,
842
+ "loss": 0.1798,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.03053884394466877,
848
+ "learning_rate": 0.0011727711127355117,
849
+ "loss": 0.181,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.03675390034914017,
855
+ "learning_rate": 0.0010989394553165832,
856
+ "loss": 0.182,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.030522828921675682,
862
+ "learning_rate": 0.0010272202910015082,
863
+ "loss": 0.1862,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.03357331454753876,
869
+ "learning_rate": 0.0009576524569829692,
870
+ "loss": 0.1853,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.037533897906541824,
876
+ "learning_rate": 0.0008902736254703347,
877
+ "loss": 0.1795,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.03461192548274994,
883
+ "learning_rate": 0.0008251202832895066,
884
+ "loss": 0.1873,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.03553595393896103,
890
+ "learning_rate": 0.0007622277121246512,
891
+ "loss": 0.1789,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.03394347429275513,
897
+ "learning_rate": 0.000701629969412545,
898
+ "loss": 0.185,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.0283980555832386,
904
+ "learning_rate": 0.0006433598698998766,
905
+ "loss": 0.1805,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.03025851771235466,
911
+ "learning_rate": 0.0005874489678734812,
912
+ "loss": 0.1762,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.03542542830109596,
918
+ "learning_rate": 0.000533927540073133,
919
+ "loss": 0.1782,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.03365590050816536,
925
+ "learning_rate": 0.0004828245692961608,
926
+ "loss": 0.1795,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.0319739505648613,
932
+ "learning_rate": 0.0004341677287027529,
933
+ "loss": 0.1788,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.030165359377861023,
939
+ "learning_rate": 0.00038798336683045057,
940
+ "loss": 0.1814,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.03518758341670036,
946
+ "learning_rate": 0.0003442964933259474,
947
+ "loss": 0.1836,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.036399878561496735,
953
+ "learning_rate": 0.00030313076540192997,
954
+ "loss": 0.1793,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.04823492839932442,
960
+ "learning_rate": 0.00026450847502627883,
961
+ "loss": 0.1904,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.02590683102607727,
967
+ "learning_rate": 0.00022845053685056904,
968
+ "loss": 0.1816,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.030667992308735847,
974
+ "learning_rate": 0.00019497647688442477,
975
+ "loss": 0.1726,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.036195989698171616,
981
+ "learning_rate": 0.00016410442192183573,
982
+ "loss": 0.1759,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.033511728048324585,
988
+ "learning_rate": 0.0001358510897251808,
989
+ "loss": 0.1769,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.03249663487076759,
995
+ "learning_rate": 0.00011023177997226297,
996
+ "loss": 0.1799,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.035352785140275955,
1002
+ "learning_rate": 8.726036597126619e-05,
1003
+ "loss": 0.176,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.03339748457074165,
1009
+ "learning_rate": 6.694928714811255e-05,
1010
+ "loss": 0.1732,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.03594556823372841,
1016
+ "learning_rate": 4.93095423102935e-05,
1017
+ "loss": 0.1812,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.03444835543632507,
1023
+ "learning_rate": 3.435068369082306e-05,
1024
+ "loss": 0.1807,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.04031510651111603,
1030
+ "learning_rate": 2.2080811775535005e-05,
1031
+ "loss": 0.1779,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.03477194532752037,
1037
+ "learning_rate": 1.2506570916523408e-05,
1038
+ "loss": 0.173,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.03274468705058098,
1044
+ "learning_rate": 5.633145734114664e-06,
1045
+ "loss": 0.1752,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.02915850654244423,
1051
+ "learning_rate": 1.4642583092999705e-06,
1052
+ "loss": 0.1739,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.027722839266061783,
1058
+ "learning_rate": 2.166168162065496e-09,
1059
+ "loss": 0.1796,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.23716249809265136,
1067
+ "train_runtime": 3463.2844,
1068
+ "train_samples_per_second": 34.649,
1069
+ "train_steps_per_second": 1.083
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }
nl_tasks/exps/run_ex25/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex25/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex25/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex25/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex25/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex25/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex25/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e54f96c611761c8f54fb1b97d9d3282a5a1f6499b2f14ce38d4a7c4d7f849202
3
+ size 33602915
nl_tasks/exps/run_ex25/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.2934044301509857,
15
+ "learning_rate": 0.000128,
16
+ "loss": 0.8066,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.20665736496448517,
22
+ "learning_rate": 0.0002613333333333333,
23
+ "loss": 0.4446,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.17562589049339294,
29
+ "learning_rate": 0.0003946666666666667,
30
+ "loss": 0.3942,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.1656724214553833,
36
+ "learning_rate": 0.000528,
37
+ "loss": 0.362,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.17268958687782288,
43
+ "learning_rate": 0.0006613333333333333,
44
+ "loss": 0.3454,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.219743013381958,
50
+ "learning_rate": 0.0007946666666666666,
51
+ "loss": 0.3382,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.1991826742887497,
57
+ "learning_rate": 0.0009280000000000001,
58
+ "loss": 0.3339,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.23559680581092834,
64
+ "learning_rate": 0.0010613333333333332,
65
+ "loss": 0.3343,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.23135684430599213,
71
+ "learning_rate": 0.0011946666666666668,
72
+ "loss": 0.3329,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.24547122418880463,
78
+ "learning_rate": 0.0013280000000000002,
79
+ "loss": 0.3294,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.43551498651504517,
85
+ "learning_rate": 0.0014613333333333334,
86
+ "loss": 0.3412,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.3377213478088379,
92
+ "learning_rate": 0.0015946666666666668,
93
+ "loss": 0.7195,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.24051505327224731,
99
+ "learning_rate": 0.001728,
100
+ "loss": 0.3476,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.25818943977355957,
106
+ "learning_rate": 0.0018613333333333333,
107
+ "loss": 0.3497,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.2813394367694855,
113
+ "learning_rate": 0.0019946666666666667,
114
+ "loss": 0.3444,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.294381320476532,
120
+ "learning_rate": 0.0019997504677881223,
121
+ "loss": 0.3383,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.2145279496908188,
127
+ "learning_rate": 0.001998959986294895,
128
+ "loss": 0.322,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.23977665603160858,
134
+ "learning_rate": 0.001997628550346359,
135
+ "loss": 0.3223,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.19997288286685944,
141
+ "learning_rate": 0.0019957568809385695,
142
+ "loss": 0.3121,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.19473682343959808,
148
+ "learning_rate": 0.0019933459916135275,
149
+ "loss": 0.3085,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.16213826835155487,
155
+ "learning_rate": 0.0019903971879103345,
156
+ "loss": 0.3046,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.2148275524377823,
162
+ "learning_rate": 0.0019869120666582154,
163
+ "loss": 0.289,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.19214589893817902,
169
+ "learning_rate": 0.001982892515111809,
170
+ "loss": 0.3065,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.21339517831802368,
176
+ "learning_rate": 0.001978340709929188,
177
+ "loss": 0.291,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.17655597627162933,
183
+ "learning_rate": 0.0019732591159931563,
184
+ "loss": 0.3024,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.19270546734333038,
190
+ "learning_rate": 0.001967650485076476,
191
+ "loss": 0.2943,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.12036184221506119,
197
+ "learning_rate": 0.0019615178543517303,
198
+ "loss": 0.2795,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.21230359375476837,
204
+ "learning_rate": 0.001954864544746643,
205
+ "loss": 0.2803,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.20738306641578674,
211
+ "learning_rate": 0.001947694159145737,
212
+ "loss": 0.2865,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.16709482669830322,
218
+ "learning_rate": 0.001940010580439308,
219
+ "loss": 0.2818,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.17816157639026642,
225
+ "learning_rate": 0.0019318179694207726,
226
+ "loss": 0.2694,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.5133901834487915,
232
+ "learning_rate": 0.0019231207625335235,
233
+ "loss": 0.2757,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.14037635922431946,
239
+ "learning_rate": 0.0019139236694685185,
240
+ "loss": 0.2692,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.16915568709373474,
246
+ "learning_rate": 0.0019042316706138988,
247
+ "loss": 0.2692,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.1274840533733368,
253
+ "learning_rate": 0.0018940500143580215,
254
+ "loss": 0.2774,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.14953529834747314,
260
+ "learning_rate": 0.0018833842142473643,
261
+ "loss": 0.2709,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.17669950425624847,
267
+ "learning_rate": 0.0018722400460008438,
268
+ "loss": 0.2674,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.137393981218338,
274
+ "learning_rate": 0.0018606235443821603,
275
+ "loss": 0.2647,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.1340445578098297,
281
+ "learning_rate": 0.0018485409999318679,
282
+ "loss": 0.2614,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.14075660705566406,
288
+ "learning_rate": 0.0018359989555609354,
289
+ "loss": 0.2679,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.20850388705730438,
295
+ "learning_rate": 0.0018230042030076452,
296
+ "loss": 0.2681,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.21324606239795685,
302
+ "learning_rate": 0.0018095637791597462,
303
+ "loss": 0.2577,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.13143779337406158,
309
+ "learning_rate": 0.0017956849622438554,
310
+ "loss": 0.2522,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.12891457974910736,
316
+ "learning_rate": 0.0017813752678841701,
317
+ "loss": 0.2538,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.13101314008235931,
323
+ "learning_rate": 0.0017666424450326248,
324
+ "loss": 0.263,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.12617208063602448,
330
+ "learning_rate": 0.0017514944717726961,
331
+ "loss": 0.2556,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.12779447436332703,
337
+ "learning_rate": 0.001735939550999131,
338
+ "loss": 0.2578,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.10568287968635559,
344
+ "learning_rate": 0.0017199861059759337,
345
+ "loss": 0.2586,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.12661688029766083,
351
+ "learning_rate": 0.0017036427757750204,
352
+ "loss": 0.2604,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.12812583148479462,
358
+ "learning_rate": 0.0016869184105980089,
359
+ "loss": 0.2493,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.11161061376333237,
365
+ "learning_rate": 0.0016698220669836782,
366
+ "loss": 0.2259,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.09668291360139847,
372
+ "learning_rate": 0.0016523630029036931,
373
+ "loss": 0.235,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.13240110874176025,
379
+ "learning_rate": 0.00163455067274925,
380
+ "loss": 0.2315,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.11607251316308975,
386
+ "learning_rate": 0.0016163947222113566,
387
+ "loss": 0.2352,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.08608829230070114,
393
+ "learning_rate": 0.0015979049830575189,
394
+ "loss": 0.2397,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.10748381912708282,
400
+ "learning_rate": 0.001579091467807668,
401
+ "loss": 0.238,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.11148440837860107,
407
+ "learning_rate": 0.0015599643643122022,
408
+ "loss": 0.2302,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.10197737067937851,
414
+ "learning_rate": 0.0015405340302350869,
415
+ "loss": 0.2375,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.09502054750919342,
421
+ "learning_rate": 0.0015208109874449978,
422
+ "loss": 0.2336,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.13474340736865997,
428
+ "learning_rate": 0.00150080591631754,
429
+ "loss": 0.2351,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.10017239302396774,
435
+ "learning_rate": 0.0014805296499516406,
436
+ "loss": 0.2395,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.0989546924829483,
442
+ "learning_rate": 0.0014599931683032326,
443
+ "loss": 0.2301,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.09925360232591629,
449
+ "learning_rate": 0.0014392075922394132,
450
+ "loss": 0.2296,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.09363091737031937,
456
+ "learning_rate": 0.0014181841775163014,
457
+ "loss": 0.225,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.09964188933372498,
463
+ "learning_rate": 0.0013969343086838457,
464
+ "loss": 0.2249,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.09907900542020798,
470
+ "learning_rate": 0.001375469492920889,
471
+ "loss": 0.2189,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.0817643478512764,
477
+ "learning_rate": 0.0013538013538038295,
478
+ "loss": 0.2247,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.09736189246177673,
484
+ "learning_rate": 0.0013319416250122482,
485
+ "loss": 0.2231,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.09870969504117966,
491
+ "learning_rate": 0.0013099021439749155,
492
+ "loss": 0.2238,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.12818622589111328,
498
+ "learning_rate": 0.001287694845459613,
499
+ "loss": 0.2251,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.09466996788978577,
505
+ "learning_rate": 0.001265331755110247,
506
+ "loss": 0.2252,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.11681800335645676,
512
+ "learning_rate": 0.001242824982934751,
513
+ "loss": 0.2126,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.08363161236047745,
519
+ "learning_rate": 0.0012201867167473015,
520
+ "loss": 0.2192,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.10942152142524719,
526
+ "learning_rate": 0.001197429215568403,
527
+ "loss": 0.224,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.08799609541893005,
533
+ "learning_rate": 0.00117456480298641,
534
+ "loss": 0.2239,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.10763717442750931,
540
+ "learning_rate": 0.0011516058604840891,
541
+ "loss": 0.2209,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.0964960902929306,
547
+ "learning_rate": 0.001128564820733825,
548
+ "loss": 0.2199,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.08642950654029846,
554
+ "learning_rate": 0.001105454160865112,
555
+ "loss": 0.2101,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.10640878975391388,
561
+ "learning_rate": 0.0010822863957079655,
562
+ "loss": 0.2234,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.09686343371868134,
568
+ "learning_rate": 0.0010590740710159229,
569
+ "loss": 0.2231,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.0885419249534607,
575
+ "learning_rate": 0.0010358297566722945,
576
+ "loss": 0.2086,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.09378661215305328,
582
+ "learning_rate": 0.0010125660398833528,
583
+ "loss": 0.2177,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.0935426726937294,
589
+ "learning_rate": 0.0009892955183621354,
590
+ "loss": 0.2227,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.09251715987920761,
596
+ "learning_rate": 0.0009660307935065639,
597
+ "loss": 0.2185,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.09296725690364838,
603
+ "learning_rate": 0.000942784463575562,
604
+ "loss": 0.2172,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.09848319739103317,
610
+ "learning_rate": 0.0009195691168668773,
611
+ "loss": 0.21,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.07870358228683472,
617
+ "learning_rate": 0.0008963973249002967,
618
+ "loss": 0.2269,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.0861450806260109,
624
+ "learning_rate": 0.0008732816356099456,
625
+ "loss": 0.2176,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.09164830297231674,
631
+ "learning_rate": 0.0008502345665493598,
632
+ "loss": 0.2092,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.08600718528032303,
638
+ "learning_rate": 0.0008272685981130109,
639
+ "loss": 0.219,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.08620688319206238,
645
+ "learning_rate": 0.0008043961667779519,
646
+ "loss": 0.2103,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.09346953779459,
652
+ "learning_rate": 0.0007816296583692448,
653
+ "loss": 0.2135,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.08523987978696823,
659
+ "learning_rate": 0.0007589814013528185,
660
+ "loss": 0.2169,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.3015783429145813,
666
+ "learning_rate": 0.0007364636601593875,
667
+ "loss": 0.2085,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.09597095847129822,
673
+ "learning_rate": 0.0007140886285430447,
674
+ "loss": 0.2321,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.08514101058244705,
680
+ "learning_rate": 0.0006918684229781305,
681
+ "loss": 0.2164,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.1003098413348198,
687
+ "learning_rate": 0.0006698150760979462,
688
+ "loss": 0.2087,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.08927905559539795,
694
+ "learning_rate": 0.0006479405301788728,
695
+ "loss": 0.2099,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.09423171728849411,
701
+ "learning_rate": 0.0006262566306734155,
702
+ "loss": 0.2154,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.08049587905406952,
708
+ "learning_rate": 0.0006047751197956838,
709
+ "loss": 0.2115,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.08967506885528564,
715
+ "learning_rate": 0.0005835076301627752,
716
+ "loss": 0.1896,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.10433576256036758,
722
+ "learning_rate": 0.0005624656784955071,
723
+ "loss": 0.1821,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.087051622569561,
729
+ "learning_rate": 0.0005416606593819101,
730
+ "loss": 0.1907,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.10010000318288803,
736
+ "learning_rate": 0.0005211038391068576,
737
+ "loss": 0.181,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.08194483816623688,
743
+ "learning_rate": 0.0005008063495511723,
744
+ "loss": 0.1898,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.08231155574321747,
750
+ "learning_rate": 0.00048077918216351856,
751
+ "loss": 0.1825,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.08368801325559616,
757
+ "learning_rate": 0.0004610331820083378,
758
+ "loss": 0.1903,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.09819929301738739,
764
+ "learning_rate": 0.00044157904189305407,
765
+ "loss": 0.186,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.09869074076414108,
771
+ "learning_rate": 0.0004224272965777326,
772
+ "loss": 0.1913,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.08644938468933105,
778
+ "learning_rate": 0.00040358831707031816,
779
+ "loss": 0.1877,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.08564961701631546,
785
+ "learning_rate": 0.00038507230501055477,
786
+ "loss": 0.1818,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.08038187026977539,
792
+ "learning_rate": 0.00036688928714561443,
793
+ "loss": 0.1822,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.08794587105512619,
799
+ "learning_rate": 0.00034904910990043927,
800
+ "loss": 0.1922,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.08452560752630234,
806
+ "learning_rate": 0.0003315614340457279,
807
+ "loss": 0.1798,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.08093718439340591,
813
+ "learning_rate": 0.0003144357294664565,
814
+ "loss": 0.1838,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.08113941550254822,
820
+ "learning_rate": 0.0002976812700337708,
821
+ "loss": 0.1912,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.07102052122354507,
827
+ "learning_rate": 0.00028130712858302065,
828
+ "loss": 0.1807,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.10384222865104675,
834
+ "learning_rate": 0.0002653221720006586,
835
+ "loss": 0.187,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.0866876021027565,
841
+ "learning_rate": 0.00024973505642266624,
842
+ "loss": 0.1802,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.08205445855855942,
848
+ "learning_rate": 0.00023455422254710235,
849
+ "loss": 0.1822,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.08937009423971176,
855
+ "learning_rate": 0.00021978789106331665,
856
+ "loss": 0.1834,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.08640176802873611,
862
+ "learning_rate": 0.00020544405820030164,
863
+ "loss": 0.1867,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.08852454274892807,
869
+ "learning_rate": 0.00019153049139659383,
870
+ "loss": 0.1865,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.09717807173728943,
876
+ "learning_rate": 0.00017805472509406696,
877
+ "loss": 0.1803,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.09453548491001129,
883
+ "learning_rate": 0.00016502405665790132,
884
+ "loss": 0.1888,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.08937527984380722,
890
+ "learning_rate": 0.00015244554242493024,
891
+ "loss": 0.1801,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.09483719617128372,
897
+ "learning_rate": 0.000140325993882509,
898
+ "loss": 0.1845,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.07975872606039047,
904
+ "learning_rate": 0.00012867197397997533,
905
+ "loss": 0.1804,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.08552587777376175,
911
+ "learning_rate": 0.00011748979357469624,
912
+ "loss": 0.1769,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.09845510870218277,
918
+ "learning_rate": 0.00010678550801462661,
919
+ "loss": 0.1791,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.08743355423212051,
925
+ "learning_rate": 9.656491385923216e-05,
926
+ "loss": 0.1807,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.08147986233234406,
932
+ "learning_rate": 8.683354574055058e-05,
933
+ "loss": 0.18,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.0853717103600502,
939
+ "learning_rate": 7.759667336609011e-05,
940
+ "loss": 0.1828,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.09387767314910889,
946
+ "learning_rate": 6.885929866518948e-05,
947
+ "loss": 0.1853,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.08295658230781555,
953
+ "learning_rate": 6.062615308038599e-05,
954
+ "loss": 0.1807,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.11373701691627502,
960
+ "learning_rate": 5.290169500525577e-05,
961
+ "loss": 0.192,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.0731569230556488,
967
+ "learning_rate": 4.569010737011381e-05,
968
+ "loss": 0.1833,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.08211272954940796,
974
+ "learning_rate": 3.8995295376884954e-05,
975
+ "loss": 0.1728,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.09878759831190109,
981
+ "learning_rate": 3.2820884384367145e-05,
982
+ "loss": 0.1777,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.09063831716775894,
988
+ "learning_rate": 2.7170217945036156e-05,
989
+ "loss": 0.179,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.0808299332857132,
995
+ "learning_rate": 2.2046355994452594e-05,
996
+ "loss": 0.181,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.09642569720745087,
1002
+ "learning_rate": 1.745207319425324e-05,
1003
+ "loss": 0.177,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.09298644214868546,
1009
+ "learning_rate": 1.338985742962251e-05,
1010
+ "loss": 0.1752,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.08527287095785141,
1016
+ "learning_rate": 9.8619084620587e-06,
1017
+ "loss": 0.1824,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.09812653809785843,
1023
+ "learning_rate": 6.870136738164612e-06,
1024
+ "loss": 0.1835,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.10046068578958511,
1030
+ "learning_rate": 4.416162355107001e-06,
1031
+ "loss": 0.1799,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.08379827439785004,
1037
+ "learning_rate": 2.5013141833046815e-06,
1038
+ "loss": 0.1742,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.10984520614147186,
1044
+ "learning_rate": 1.1266291468229327e-06,
1045
+ "loss": 0.1771,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.07823975384235382,
1051
+ "learning_rate": 2.928516618599941e-07,
1052
+ "loss": 0.1748,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.07705021649599075,
1058
+ "learning_rate": 4.332336324130992e-10,
1059
+ "loss": 0.1822,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.24114867858886718,
1067
+ "train_runtime": 3465.2523,
1068
+ "train_samples_per_second": 34.63,
1069
+ "train_steps_per_second": 1.082
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }
nl_tasks/exps/run_ex26/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex26/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex26/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex26/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex26/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex26/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex26/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93bc9ae21a36df9047de9bf8857c7f1171e5589422f1934b10a15b42ec3323e2
3
+ size 33602915
nl_tasks/exps/run_ex26/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.308439701795578,
15
+ "learning_rate": 0.00032,
16
+ "loss": 0.7082,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.4491778016090393,
22
+ "learning_rate": 0.0006533333333333332,
23
+ "loss": 0.4048,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.205202117562294,
29
+ "learning_rate": 0.0009866666666666667,
30
+ "loss": 0.3748,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.20341382920742035,
36
+ "learning_rate": 0.00132,
37
+ "loss": 0.3515,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.38580241799354553,
43
+ "learning_rate": 0.0016533333333333333,
44
+ "loss": 0.3459,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 1.0493764877319336,
50
+ "learning_rate": 0.0019866666666666665,
51
+ "loss": 0.3969,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.38665246963500977,
57
+ "learning_rate": 0.00232,
58
+ "loss": 0.4444,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.3482854664325714,
64
+ "learning_rate": 0.002653333333333333,
65
+ "loss": 0.3755,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.33611181378364563,
71
+ "learning_rate": 0.002986666666666667,
72
+ "loss": 0.3769,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.2831563949584961,
78
+ "learning_rate": 0.00332,
79
+ "loss": 0.3664,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.48948103189468384,
85
+ "learning_rate": 0.0036533333333333335,
86
+ "loss": 0.4322,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.44456496834754944,
92
+ "learning_rate": 0.003986666666666667,
93
+ "loss": 0.359,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.23869431018829346,
99
+ "learning_rate": 0.00432,
100
+ "loss": 0.3418,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.18535733222961426,
106
+ "learning_rate": 0.004653333333333333,
107
+ "loss": 0.3417,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.18292084336280823,
113
+ "learning_rate": 0.004986666666666666,
114
+ "loss": 0.3265,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.14866161346435547,
120
+ "learning_rate": 0.004999376169470306,
121
+ "loss": 0.3215,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.12140727043151855,
127
+ "learning_rate": 0.004997399965737237,
128
+ "loss": 0.3041,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.12332931160926819,
134
+ "learning_rate": 0.0049940713758658975,
135
+ "loss": 0.3055,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.13055121898651123,
141
+ "learning_rate": 0.004989392202346423,
142
+ "loss": 0.2984,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.12080452591180801,
148
+ "learning_rate": 0.004983364979033819,
149
+ "loss": 0.2944,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.094502754509449,
155
+ "learning_rate": 0.004975992969775836,
156
+ "loss": 0.2931,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.13612881302833557,
162
+ "learning_rate": 0.004967280166645538,
163
+ "loss": 0.2814,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.1155838817358017,
169
+ "learning_rate": 0.004957231287779522,
170
+ "loss": 0.3145,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.1393253207206726,
176
+ "learning_rate": 0.004945851774822969,
177
+ "loss": 0.2847,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.09930860251188278,
183
+ "learning_rate": 0.0049331477899828904,
184
+ "loss": 0.2955,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.08679531514644623,
190
+ "learning_rate": 0.0049191262126911895,
191
+ "loss": 0.2871,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.06685009598731995,
197
+ "learning_rate": 0.004903794635879325,
198
+ "loss": 0.2745,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.09721576422452927,
204
+ "learning_rate": 0.004887161361866608,
205
+ "loss": 0.2748,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.11243274807929993,
211
+ "learning_rate": 0.004869235397864343,
212
+ "loss": 0.2826,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.10606742650270462,
218
+ "learning_rate": 0.00485002645109827,
219
+ "loss": 0.2782,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.09463471174240112,
225
+ "learning_rate": 0.004829544923551931,
226
+ "loss": 0.268,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.08187498897314072,
232
+ "learning_rate": 0.004807801906333809,
233
+ "loss": 0.2718,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.07924027740955353,
239
+ "learning_rate": 0.004784809173671296,
240
+ "loss": 0.2642,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.08812413364648819,
246
+ "learning_rate": 0.004760579176534747,
247
+ "loss": 0.2668,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.06769583374261856,
253
+ "learning_rate": 0.004735125035895054,
254
+ "loss": 0.2744,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.07282879948616028,
260
+ "learning_rate": 0.004708460535618411,
261
+ "loss": 0.2693,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.07845976203680038,
267
+ "learning_rate": 0.0046806001150021094,
268
+ "loss": 0.264,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.07892794162034988,
274
+ "learning_rate": 0.004651558860955401,
275
+ "loss": 0.2638,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.06784877926111221,
281
+ "learning_rate": 0.004621352499829669,
282
+ "loss": 0.2594,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.09675692021846771,
288
+ "learning_rate": 0.0045899973889023385,
289
+ "loss": 0.2654,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.08054903894662857,
295
+ "learning_rate": 0.004557510507519113,
296
+ "loss": 0.2652,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.07471887022256851,
302
+ "learning_rate": 0.004523909447899366,
303
+ "loss": 0.2558,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.07312784343957901,
309
+ "learning_rate": 0.004489212405609638,
310
+ "loss": 0.2532,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.08276063948869705,
316
+ "learning_rate": 0.004453438169710425,
317
+ "loss": 0.2523,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.07462652772665024,
323
+ "learning_rate": 0.004416606112581562,
324
+ "loss": 0.2619,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.06282167881727219,
330
+ "learning_rate": 0.0043787361794317406,
331
+ "loss": 0.2544,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.06843234598636627,
337
+ "learning_rate": 0.004339848877497827,
338
+ "loss": 0.2573,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.05832453444600105,
344
+ "learning_rate": 0.004299965264939834,
345
+ "loss": 0.2563,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.07380504906177521,
351
+ "learning_rate": 0.004259106939437551,
352
+ "loss": 0.2604,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.0749950185418129,
358
+ "learning_rate": 0.004217296026495022,
359
+ "loss": 0.2478,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.0739532932639122,
365
+ "learning_rate": 0.004174555167459195,
366
+ "loss": 0.2262,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.055929724127054214,
372
+ "learning_rate": 0.0041309075072592325,
373
+ "loss": 0.2335,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.06944834440946579,
379
+ "learning_rate": 0.004086376681873125,
380
+ "loss": 0.23,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.06275296211242676,
386
+ "learning_rate": 0.004040986805528391,
387
+ "loss": 0.237,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.052883587777614594,
393
+ "learning_rate": 0.003994762457643798,
394
+ "loss": 0.2398,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.06906875967979431,
400
+ "learning_rate": 0.00394772866951917,
401
+ "loss": 0.2365,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.07487814128398895,
407
+ "learning_rate": 0.003899910910780505,
408
+ "loss": 0.2292,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.060511570423841476,
414
+ "learning_rate": 0.0038513350755877175,
415
+ "loss": 0.2362,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.07363221794366837,
421
+ "learning_rate": 0.0038020274686124947,
422
+ "loss": 0.2331,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.06942912936210632,
428
+ "learning_rate": 0.0037520147907938502,
429
+ "loss": 0.2346,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.05655023828148842,
435
+ "learning_rate": 0.003701324124879102,
436
+ "loss": 0.2382,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.05511761084198952,
442
+ "learning_rate": 0.0036499829207580813,
443
+ "loss": 0.2301,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.05750414729118347,
449
+ "learning_rate": 0.003598018980598533,
450
+ "loss": 0.2287,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.05184759199619293,
456
+ "learning_rate": 0.0035454604437907536,
457
+ "loss": 0.2248,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.05860360339283943,
463
+ "learning_rate": 0.0034923357717096142,
464
+ "loss": 0.2265,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.059097886085510254,
470
+ "learning_rate": 0.0034386737323022225,
471
+ "loss": 0.2201,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.04928320646286011,
477
+ "learning_rate": 0.0033845033845095736,
478
+ "loss": 0.2244,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.05335696414113045,
484
+ "learning_rate": 0.0033298540625306206,
485
+ "loss": 0.2238,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.057368285953998566,
491
+ "learning_rate": 0.0032747553599372886,
492
+ "loss": 0.2242,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.06468666344881058,
498
+ "learning_rate": 0.0032192371136490323,
499
+ "loss": 0.2243,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.057294175028800964,
505
+ "learning_rate": 0.0031633293877756175,
506
+ "loss": 0.2262,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.05811116099357605,
512
+ "learning_rate": 0.0031070624573368774,
513
+ "loss": 0.2128,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.04844126105308533,
519
+ "learning_rate": 0.0030504667918682536,
520
+ "loss": 0.2191,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.06403350085020065,
526
+ "learning_rate": 0.0029935730389210077,
527
+ "loss": 0.2238,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.059780120849609375,
533
+ "learning_rate": 0.002936412007466025,
534
+ "loss": 0.2242,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.06413201242685318,
540
+ "learning_rate": 0.002879014651210223,
541
+ "loss": 0.2215,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.056221649050712585,
547
+ "learning_rate": 0.002821412051834563,
548
+ "loss": 0.2206,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.05171119421720505,
554
+ "learning_rate": 0.00276363540216278,
555
+ "loss": 0.21,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.05907101929187775,
561
+ "learning_rate": 0.002705715989269914,
562
+ "loss": 0.2226,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.055108439177274704,
568
+ "learning_rate": 0.002647685177539807,
569
+ "loss": 0.2226,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.05309450253844261,
575
+ "learning_rate": 0.0025895743916807363,
576
+ "loss": 0.2087,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.05426125228404999,
582
+ "learning_rate": 0.0025314150997083817,
583
+ "loss": 0.2185,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.0457584485411644,
589
+ "learning_rate": 0.0024732387959053386,
590
+ "loss": 0.2227,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.05757022649049759,
596
+ "learning_rate": 0.00241507698376641,
597
+ "loss": 0.2193,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.05329664424061775,
603
+ "learning_rate": 0.002356961158938905,
604
+ "loss": 0.2184,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.05725402012467384,
610
+ "learning_rate": 0.002298922792167193,
611
+ "loss": 0.2099,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.04788336530327797,
617
+ "learning_rate": 0.002240993312250742,
618
+ "loss": 0.226,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.04869900271296501,
624
+ "learning_rate": 0.002183204089024864,
625
+ "loss": 0.2172,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.05592982470989227,
631
+ "learning_rate": 0.0021255864163733994,
632
+ "loss": 0.2084,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.049687620252370834,
638
+ "learning_rate": 0.0020681714952825274,
639
+ "loss": 0.2188,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.04963243752717972,
645
+ "learning_rate": 0.00201099041694488,
646
+ "loss": 0.2109,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.05767963454127312,
652
+ "learning_rate": 0.0019540741459231122,
653
+ "loss": 0.2138,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.04665152728557587,
659
+ "learning_rate": 0.0018974535033820464,
660
+ "loss": 0.2168,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.052285660058259964,
666
+ "learning_rate": 0.0018411591503984686,
667
+ "loss": 0.2088,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.056856900453567505,
673
+ "learning_rate": 0.001785221571357612,
674
+ "loss": 0.232,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.05064110830426216,
680
+ "learning_rate": 0.0017296710574453262,
681
+ "loss": 0.2169,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.056195858865976334,
687
+ "learning_rate": 0.0016745376902448656,
688
+ "loss": 0.2089,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.0578073114156723,
694
+ "learning_rate": 0.0016198513254471819,
695
+ "loss": 0.2105,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.04981910437345505,
701
+ "learning_rate": 0.0015656415766835386,
702
+ "loss": 0.2156,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.0438181608915329,
708
+ "learning_rate": 0.0015119377994892095,
709
+ "loss": 0.2116,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.05480600520968437,
715
+ "learning_rate": 0.001458769075406938,
716
+ "loss": 0.1884,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.05522935092449188,
722
+ "learning_rate": 0.0014061641962387678,
723
+ "loss": 0.1811,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.057728227227926254,
729
+ "learning_rate": 0.0013541516484547754,
730
+ "loss": 0.1887,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.056935157626867294,
736
+ "learning_rate": 0.001302759597767144,
737
+ "loss": 0.1803,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.04977869242429733,
743
+ "learning_rate": 0.0012520158738779307,
744
+ "loss": 0.1889,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.050365712493658066,
750
+ "learning_rate": 0.0012019479554087964,
751
+ "loss": 0.1809,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.04901596158742905,
757
+ "learning_rate": 0.0011525829550208444,
758
+ "loss": 0.1895,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.05424964427947998,
764
+ "learning_rate": 0.0011039476047326351,
765
+ "loss": 0.1852,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.052948519587516785,
771
+ "learning_rate": 0.0010560682414443316,
772
+ "loss": 0.1908,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.05285203084349632,
778
+ "learning_rate": 0.0010089707926757955,
779
+ "loss": 0.1861,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.047558143734931946,
785
+ "learning_rate": 0.000962680762526387,
786
+ "loss": 0.1811,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.050169140100479126,
792
+ "learning_rate": 0.000917223217864036,
793
+ "loss": 0.1824,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.04792075231671333,
799
+ "learning_rate": 0.0008726227747510981,
800
+ "loss": 0.1917,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.048594892024993896,
806
+ "learning_rate": 0.0008289035851143198,
807
+ "loss": 0.1794,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.04949982091784477,
813
+ "learning_rate": 0.0007860893236661412,
814
+ "loss": 0.1831,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.052018143236637115,
820
+ "learning_rate": 0.0007442031750844269,
821
+ "loss": 0.1894,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.04490148648619652,
827
+ "learning_rate": 0.0007032678214575517,
828
+ "loss": 0.1788,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.05236239731311798,
834
+ "learning_rate": 0.0006633054300016464,
835
+ "loss": 0.1864,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.047926224768161774,
841
+ "learning_rate": 0.0006243376410566656,
842
+ "loss": 0.1798,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.05037904903292656,
848
+ "learning_rate": 0.0005863855563677558,
849
+ "loss": 0.181,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.05419327691197395,
855
+ "learning_rate": 0.0005494697276582916,
856
+ "loss": 0.1822,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.05214129015803337,
862
+ "learning_rate": 0.0005136101455007541,
863
+ "loss": 0.1854,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.0533323772251606,
869
+ "learning_rate": 0.0004788262284914846,
870
+ "loss": 0.1855,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.05604487657546997,
876
+ "learning_rate": 0.00044513681273516736,
877
+ "loss": 0.1793,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.05135011672973633,
883
+ "learning_rate": 0.0004125601416447533,
884
+ "loss": 0.1876,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.0539252869784832,
890
+ "learning_rate": 0.0003811138560623256,
891
+ "loss": 0.1782,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.05228358134627342,
897
+ "learning_rate": 0.0003508149847062725,
898
+ "loss": 0.1831,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.046433333307504654,
904
+ "learning_rate": 0.0003216799349499383,
905
+ "loss": 0.1793,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.04742318019270897,
911
+ "learning_rate": 0.0002937244839367406,
912
+ "loss": 0.1761,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.05307656154036522,
918
+ "learning_rate": 0.0002669637700365665,
919
+ "loss": 0.177,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.055305834859609604,
925
+ "learning_rate": 0.0002414122846480804,
926
+ "loss": 0.1793,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.04545231908559799,
932
+ "learning_rate": 0.00021708386435137644,
933
+ "loss": 0.178,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.04912173002958298,
939
+ "learning_rate": 0.00019399168341522528,
940
+ "loss": 0.1804,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.0657554641366005,
946
+ "learning_rate": 0.0001721482466629737,
947
+ "loss": 0.1834,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.047934047877788544,
953
+ "learning_rate": 0.00015156538270096498,
954
+ "loss": 0.1805,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.06532304733991623,
960
+ "learning_rate": 0.00013225423751313942,
961
+ "loss": 0.1908,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.03938894718885422,
967
+ "learning_rate": 0.00011422526842528452,
968
+ "loss": 0.1817,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.046688247472047806,
974
+ "learning_rate": 9.748823844221238e-05,
975
+ "loss": 0.1727,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.05760841444134712,
981
+ "learning_rate": 8.205221096091786e-05,
982
+ "loss": 0.1756,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.0510096400976181,
988
+ "learning_rate": 6.79255448625904e-05,
989
+ "loss": 0.1767,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.049026183784008026,
995
+ "learning_rate": 5.511588998613149e-05,
996
+ "loss": 0.1789,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.05264630913734436,
1002
+ "learning_rate": 4.3630182985633096e-05,
1003
+ "loss": 0.1751,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.05136013776063919,
1009
+ "learning_rate": 3.3474643574056276e-05,
1010
+ "loss": 0.1726,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.049536991864442825,
1016
+ "learning_rate": 2.465477115514675e-05,
1017
+ "loss": 0.1811,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.05471661314368248,
1023
+ "learning_rate": 1.717534184541153e-05,
1024
+ "loss": 0.1807,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.059411708265542984,
1030
+ "learning_rate": 1.1040405887767503e-05,
1031
+ "loss": 0.1778,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.048990510404109955,
1037
+ "learning_rate": 6.253285458261704e-06,
1038
+ "loss": 0.1737,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.053671903908252716,
1044
+ "learning_rate": 2.816572867057332e-06,
1045
+ "loss": 0.1762,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.043852001428604126,
1051
+ "learning_rate": 7.321291546499853e-07,
1052
+ "loss": 0.1743,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.043689530342817307,
1058
+ "learning_rate": 1.083084081032748e-09,
1059
+ "loss": 0.1797,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.23841547063191731,
1067
+ "train_runtime": 3464.8361,
1068
+ "train_samples_per_second": 34.634,
1069
+ "train_steps_per_second": 1.082
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }
nl_tasks/exps/run_ex27/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex27/ft/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
nl_tasks/exps/run_ex27/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/exps/run_ex27/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/exps/run_ex27/ft/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
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": 512,
37
+ "pad_token": "<unk>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
nl_tasks/exps/run_ex27/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/exps/run_ex27/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad569dbd90021db6f9222a57761dec497150748c0f5dbedf597ea06cec130bba
3
+ size 33602915
nl_tasks/exps/run_ex27/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.25781920552253723,
15
+ "learning_rate": 0.000512,
16
+ "loss": 0.6638,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 0.19873183965682983,
22
+ "learning_rate": 0.0010453333333333332,
23
+ "loss": 0.3921,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.22360016405582428,
29
+ "learning_rate": 0.0015786666666666668,
30
+ "loss": 0.3705,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.268097460269928,
36
+ "learning_rate": 0.002112,
37
+ "loss": 0.3635,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.74379962682724,
43
+ "learning_rate": 0.0026453333333333333,
44
+ "loss": 0.3711,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.6203864216804504,
50
+ "learning_rate": 0.0031786666666666664,
51
+ "loss": 0.4046,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 0.435539186000824,
57
+ "learning_rate": 0.0037120000000000005,
58
+ "loss": 0.5032,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.23134496808052063,
64
+ "learning_rate": 0.004245333333333333,
65
+ "loss": 0.3804,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.25716540217399597,
71
+ "learning_rate": 0.004778666666666667,
72
+ "loss": 0.3628,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.17056621611118317,
78
+ "learning_rate": 0.005312000000000001,
79
+ "loss": 0.3466,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.1669658124446869,
85
+ "learning_rate": 0.0058453333333333335,
86
+ "loss": 0.333,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.1642168164253235,
92
+ "learning_rate": 0.006378666666666667,
93
+ "loss": 0.3424,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.12736473977565765,
99
+ "learning_rate": 0.006912,
100
+ "loss": 0.3298,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.128292977809906,
106
+ "learning_rate": 0.007445333333333333,
107
+ "loss": 0.3299,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.12831421196460724,
113
+ "learning_rate": 0.007978666666666667,
114
+ "loss": 0.3203,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.12132629007101059,
120
+ "learning_rate": 0.007999001871152489,
121
+ "loss": 0.318,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.10372071713209152,
127
+ "learning_rate": 0.00799583994517958,
128
+ "loss": 0.3041,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.08704401552677155,
134
+ "learning_rate": 0.007990514201385437,
135
+ "loss": 0.3061,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.09257054328918457,
141
+ "learning_rate": 0.007983027523754278,
142
+ "loss": 0.2988,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.09537217020988464,
148
+ "learning_rate": 0.00797338396645411,
149
+ "loss": 0.2937,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.07897094637155533,
155
+ "learning_rate": 0.007961588751641338,
156
+ "loss": 0.2943,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.0941765084862709,
162
+ "learning_rate": 0.007947648266632861,
163
+ "loss": 0.2815,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.08107373118400574,
169
+ "learning_rate": 0.007931570060447236,
170
+ "loss": 0.2995,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.08375738561153412,
176
+ "learning_rate": 0.007913362839716751,
177
+ "loss": 0.2843,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.07901784777641296,
183
+ "learning_rate": 0.007893036463972625,
184
+ "loss": 0.2987,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.07139074057340622,
190
+ "learning_rate": 0.007870601940305904,
191
+ "loss": 0.2893,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.0510406494140625,
197
+ "learning_rate": 0.007846071417406921,
198
+ "loss": 0.2759,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.06341639906167984,
204
+ "learning_rate": 0.007819458178986572,
205
+ "loss": 0.2768,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.07159160822629929,
211
+ "learning_rate": 0.007790776636582948,
212
+ "loss": 0.2831,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.09310116618871689,
218
+ "learning_rate": 0.007760042321757232,
219
+ "loss": 0.2787,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.06396382302045822,
225
+ "learning_rate": 0.00772727187768309,
226
+ "loss": 0.2665,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.07068858295679092,
232
+ "learning_rate": 0.007692483050134094,
233
+ "loss": 0.2731,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.06336300075054169,
239
+ "learning_rate": 0.007655694677874074,
240
+ "loss": 0.2652,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.06757763773202896,
246
+ "learning_rate": 0.007616926682455595,
247
+ "loss": 0.2659,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.050533879548311234,
253
+ "learning_rate": 0.007576200057432086,
254
+ "loss": 0.2748,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.052018292248249054,
260
+ "learning_rate": 0.007533536856989457,
261
+ "loss": 0.2689,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.06127012521028519,
267
+ "learning_rate": 0.007488960184003375,
268
+ "loss": 0.2666,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.05782606080174446,
274
+ "learning_rate": 0.007442494177528641,
275
+ "loss": 0.2643,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.04911736026406288,
281
+ "learning_rate": 0.007394163999727471,
282
+ "loss": 0.2597,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.07270637154579163,
288
+ "learning_rate": 0.007343995822243742,
289
+ "loss": 0.2688,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.0592631995677948,
295
+ "learning_rate": 0.007292016812030581,
296
+ "loss": 0.2664,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.05052737891674042,
302
+ "learning_rate": 0.007238255116638985,
303
+ "loss": 0.2569,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.05483546853065491,
309
+ "learning_rate": 0.007182739848975422,
310
+ "loss": 0.2545,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.05168973654508591,
316
+ "learning_rate": 0.0071255010715366805,
317
+ "loss": 0.2542,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.056612785905599594,
323
+ "learning_rate": 0.007066569780130499,
324
+ "loss": 0.2633,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.04777904227375984,
330
+ "learning_rate": 0.0070059778870907845,
331
+ "loss": 0.2562,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.052652597427368164,
337
+ "learning_rate": 0.006943758203996524,
338
+ "loss": 0.2589,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.04456433653831482,
344
+ "learning_rate": 0.006879944423903735,
345
+ "loss": 0.2576,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.05583874136209488,
351
+ "learning_rate": 0.006814571103100082,
352
+ "loss": 0.2603,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.05586734786629677,
358
+ "learning_rate": 0.0067476736423920354,
359
+ "loss": 0.2496,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.05297844856977463,
365
+ "learning_rate": 0.006679288267934713,
366
+ "loss": 0.2265,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.04152635857462883,
372
+ "learning_rate": 0.0066094520116147725,
373
+ "loss": 0.235,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.052927907556295395,
379
+ "learning_rate": 0.006538202690997,
380
+ "loss": 0.2311,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.051669735461473465,
386
+ "learning_rate": 0.0064655788888454266,
387
+ "loss": 0.2364,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.040959082543849945,
393
+ "learning_rate": 0.0063916199322300755,
394
+ "loss": 0.2401,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.04207254573702812,
400
+ "learning_rate": 0.006316365871230672,
401
+ "loss": 0.2383,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.048728130757808685,
407
+ "learning_rate": 0.006239857457248809,
408
+ "loss": 0.2308,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.043461356312036514,
414
+ "learning_rate": 0.0061621361209403475,
415
+ "loss": 0.2383,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.04390329867601395,
421
+ "learning_rate": 0.006083243949779991,
422
+ "loss": 0.2349,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.07007128745317459,
428
+ "learning_rate": 0.00600322366527016,
429
+ "loss": 0.2378,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.046169713139534,
435
+ "learning_rate": 0.005922118599806563,
436
+ "loss": 0.2425,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.04116000607609749,
442
+ "learning_rate": 0.00583997267321293,
443
+ "loss": 0.232,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.04283928498625755,
449
+ "learning_rate": 0.005756830368957653,
450
+ "loss": 0.2316,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.04096667468547821,
456
+ "learning_rate": 0.005672736710065206,
457
+ "loss": 0.2268,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.045082151889801025,
463
+ "learning_rate": 0.005587737234735383,
464
+ "loss": 0.2276,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.0428253673017025,
470
+ "learning_rate": 0.005501877971683556,
471
+ "loss": 0.2211,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.037039726972579956,
477
+ "learning_rate": 0.005415205415215318,
478
+ "loss": 0.2269,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.04006600379943848,
484
+ "learning_rate": 0.005327766500048993,
485
+ "loss": 0.2251,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.039758071303367615,
491
+ "learning_rate": 0.005239608575899662,
492
+ "loss": 0.2254,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.04773048684000969,
498
+ "learning_rate": 0.005150779381838452,
499
+ "loss": 0.2259,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.03774907439947128,
505
+ "learning_rate": 0.005061327020440988,
506
+ "loss": 0.2277,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.040254510939121246,
512
+ "learning_rate": 0.004971299931739004,
513
+ "loss": 0.2145,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.03746659308671951,
519
+ "learning_rate": 0.004880746866989206,
520
+ "loss": 0.2214,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.04677619785070419,
526
+ "learning_rate": 0.004789716862273612,
527
+ "loss": 0.2251,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.03931615129113197,
533
+ "learning_rate": 0.00469825921194564,
534
+ "loss": 0.2254,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.05020896717905998,
540
+ "learning_rate": 0.0046064234419363565,
541
+ "loss": 0.2236,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.04491092264652252,
547
+ "learning_rate": 0.0045142592829353,
548
+ "loss": 0.2218,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.03652544692158699,
554
+ "learning_rate": 0.004421816643460448,
555
+ "loss": 0.2115,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.046602796763181686,
561
+ "learning_rate": 0.004329145582831862,
562
+ "loss": 0.2262,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.03735148161649704,
568
+ "learning_rate": 0.0042362962840636915,
569
+ "loss": 0.2255,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.03894055634737015,
575
+ "learning_rate": 0.004143319026689178,
576
+ "loss": 0.2106,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.03984660282731056,
582
+ "learning_rate": 0.004050264159533411,
583
+ "loss": 0.2199,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.03481967747211456,
589
+ "learning_rate": 0.003957182073448542,
590
+ "loss": 0.2248,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.045008812099695206,
596
+ "learning_rate": 0.0038641231740262558,
597
+ "loss": 0.2215,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.03925010934472084,
603
+ "learning_rate": 0.003771137854302248,
604
+ "loss": 0.2198,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.040725499391555786,
610
+ "learning_rate": 0.003678276467467509,
611
+ "loss": 0.2112,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.03515653312206268,
617
+ "learning_rate": 0.003585589299601187,
618
+ "loss": 0.228,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.036875877529382706,
624
+ "learning_rate": 0.0034931265424397824,
625
+ "loss": 0.2197,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.040114812552928925,
631
+ "learning_rate": 0.003400938266197439,
632
+ "loss": 0.2107,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.037918511778116226,
638
+ "learning_rate": 0.0033090743924520435,
639
+ "loss": 0.2205,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.03853275254368782,
645
+ "learning_rate": 0.0032175846671118075,
646
+ "loss": 0.2123,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.038103483617305756,
652
+ "learning_rate": 0.003126518633476979,
653
+ "loss": 0.2161,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.03494865819811821,
659
+ "learning_rate": 0.003035925605411274,
660
+ "loss": 0.2189,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.03664889186620712,
666
+ "learning_rate": 0.00294585464063755,
667
+ "loss": 0.2108,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.04148728772997856,
673
+ "learning_rate": 0.002856354514172179,
674
+ "loss": 0.2338,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.04232242703437805,
680
+ "learning_rate": 0.002767473691912522,
681
+ "loss": 0.2186,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.04087619483470917,
687
+ "learning_rate": 0.002679260304391785,
688
+ "loss": 0.2111,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.03856271877884865,
694
+ "learning_rate": 0.002591762120715491,
695
+ "loss": 0.2123,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.04001619666814804,
701
+ "learning_rate": 0.002505026522693662,
702
+ "loss": 0.2177,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.03274156525731087,
708
+ "learning_rate": 0.002419100479182735,
709
+ "loss": 0.2137,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.04001650959253311,
715
+ "learning_rate": 0.0023340305206511008,
716
+ "loss": 0.1889,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.04055812954902649,
722
+ "learning_rate": 0.0022498627139820285,
723
+ "loss": 0.1836,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.03762733191251755,
729
+ "learning_rate": 0.0021666426375276404,
730
+ "loss": 0.1909,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.039257150143384933,
736
+ "learning_rate": 0.0020844153564274306,
737
+ "loss": 0.1822,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.0384918637573719,
743
+ "learning_rate": 0.0020032253982046892,
744
+ "loss": 0.1897,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.03713955730199814,
750
+ "learning_rate": 0.0019231167286540742,
751
+ "loss": 0.1836,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.034854769706726074,
757
+ "learning_rate": 0.0018441327280333511,
758
+ "loss": 0.1903,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.0406007245182991,
764
+ "learning_rate": 0.0017663161675722163,
765
+ "loss": 0.1867,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.03872333839535713,
771
+ "learning_rate": 0.0016897091863109304,
772
+ "loss": 0.1921,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.036862798035144806,
778
+ "learning_rate": 0.0016143532682812726,
779
+ "loss": 0.1891,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.03593632951378822,
785
+ "learning_rate": 0.001540289220042219,
786
+ "loss": 0.1831,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.034924060106277466,
792
+ "learning_rate": 0.0014675571485824577,
793
+ "loss": 0.1831,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.03330395743250847,
799
+ "learning_rate": 0.001396196439601757,
800
+ "loss": 0.1928,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.033493608236312866,
806
+ "learning_rate": 0.0013262457361829117,
807
+ "loss": 0.1812,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.033618029206991196,
813
+ "learning_rate": 0.001257742917865826,
814
+ "loss": 0.1854,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.03629603609442711,
820
+ "learning_rate": 0.0011907250801350831,
821
+ "loss": 0.191,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.030817655846476555,
827
+ "learning_rate": 0.0011252285143320826,
828
+ "loss": 0.1809,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.039837613701820374,
834
+ "learning_rate": 0.0010612886880026343,
835
+ "loss": 0.1881,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.034760717302560806,
841
+ "learning_rate": 0.000998940225690665,
842
+ "loss": 0.1811,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.03357568383216858,
848
+ "learning_rate": 0.0009382168901884094,
849
+ "loss": 0.1824,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.04034045711159706,
855
+ "learning_rate": 0.0008791515642532666,
856
+ "loss": 0.1842,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.033517543226480484,
862
+ "learning_rate": 0.0008217762328012066,
863
+ "loss": 0.187,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.03797078877687454,
869
+ "learning_rate": 0.0007661219655863753,
870
+ "loss": 0.1866,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.03740476444363594,
876
+ "learning_rate": 0.0007122189003762678,
877
+ "loss": 0.1801,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.04116297513246536,
883
+ "learning_rate": 0.0006600962266316053,
884
+ "loss": 0.1885,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.0396578311920166,
890
+ "learning_rate": 0.000609782169699721,
891
+ "loss": 0.18,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.04095142334699631,
897
+ "learning_rate": 0.000561303975530036,
898
+ "loss": 0.1843,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.035744231194257736,
904
+ "learning_rate": 0.0005146878959199013,
905
+ "loss": 0.1806,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.034600187093019485,
911
+ "learning_rate": 0.000469959174298785,
912
+ "loss": 0.1769,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.04248780012130737,
918
+ "learning_rate": 0.00042714203205850644,
919
+ "loss": 0.1778,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.03829626366496086,
925
+ "learning_rate": 0.00038625965543692865,
926
+ "loss": 0.1809,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.03590141609311104,
932
+ "learning_rate": 0.00034733418296220233,
933
+ "loss": 0.1798,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.04315745085477829,
939
+ "learning_rate": 0.00031038669346436043,
940
+ "loss": 0.1818,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.040417563170194626,
946
+ "learning_rate": 0.0002754371946607579,
947
+ "loss": 0.1848,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.039845407009124756,
953
+ "learning_rate": 0.00024250461232154397,
954
+ "loss": 0.1805,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.04121808707714081,
960
+ "learning_rate": 0.00021160678002102307,
961
+ "loss": 0.1923,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.02966867946088314,
967
+ "learning_rate": 0.00018276042948045523,
968
+ "loss": 0.182,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.032377708703279495,
974
+ "learning_rate": 0.00015598118150753982,
975
+ "loss": 0.1735,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.041362907737493515,
981
+ "learning_rate": 0.00013128353753746858,
982
+ "loss": 0.1779,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.034777916967868805,
988
+ "learning_rate": 0.00010868087178014462,
989
+ "loss": 0.1782,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.037447091192007065,
995
+ "learning_rate": 8.818542397781038e-05,
996
+ "loss": 0.1799,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.03927723690867424,
1002
+ "learning_rate": 6.980829277701295e-05,
1003
+ "loss": 0.1771,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.037519097328186035,
1009
+ "learning_rate": 5.355942971849004e-05,
1010
+ "loss": 0.1731,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.03824010491371155,
1016
+ "learning_rate": 3.94476338482348e-05,
1017
+ "loss": 0.182,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.03913429006934166,
1023
+ "learning_rate": 2.748054695265845e-05,
1024
+ "loss": 0.1824,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.04404062405228615,
1030
+ "learning_rate": 1.7664649420428003e-05,
1031
+ "loss": 0.1789,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.0337517075240612,
1037
+ "learning_rate": 1.0005256733218726e-05,
1038
+ "loss": 0.1739,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.03655869513750076,
1044
+ "learning_rate": 4.506516587291731e-06,
1045
+ "loss": 0.1764,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.032187193632125854,
1051
+ "learning_rate": 1.1714066474399764e-06,
1052
+ "loss": 0.1753,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.03194441273808479,
1058
+ "learning_rate": 1.7329345296523968e-09,
1059
+ "loss": 0.1811,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.238677699025472,
1067
+ "train_runtime": 3462.4523,
1068
+ "train_samples_per_second": 34.658,
1069
+ "train_steps_per_second": 1.083
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }
nl_tasks/exps/run_ex28/trainer_state.json ADDED
@@ -0,0 +1,1093 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3750,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.02,
14
+ "grad_norm": 0.3354637026786804,
15
+ "learning_rate": 0.00128,
16
+ "loss": 0.5949,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04,
21
+ "grad_norm": 24.028608322143555,
22
+ "learning_rate": 0.002613333333333333,
23
+ "loss": 0.3875,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.06,
28
+ "grad_norm": 0.8168126344680786,
29
+ "learning_rate": 0.003946666666666667,
30
+ "loss": 0.4411,
31
+ "step": 75
32
+ },
33
+ {
34
+ "epoch": 0.08,
35
+ "grad_norm": 0.4146294891834259,
36
+ "learning_rate": 0.00528,
37
+ "loss": 0.4422,
38
+ "step": 100
39
+ },
40
+ {
41
+ "epoch": 0.1,
42
+ "grad_norm": 0.3548749089241028,
43
+ "learning_rate": 0.006613333333333333,
44
+ "loss": 0.4081,
45
+ "step": 125
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "grad_norm": 0.46628203988075256,
50
+ "learning_rate": 0.007946666666666666,
51
+ "loss": 0.3749,
52
+ "step": 150
53
+ },
54
+ {
55
+ "epoch": 0.14,
56
+ "grad_norm": 3.237746477127075,
57
+ "learning_rate": 0.00928,
58
+ "loss": 0.3762,
59
+ "step": 175
60
+ },
61
+ {
62
+ "epoch": 0.16,
63
+ "grad_norm": 0.14871637523174286,
64
+ "learning_rate": 0.010613333333333332,
65
+ "loss": 0.3558,
66
+ "step": 200
67
+ },
68
+ {
69
+ "epoch": 0.18,
70
+ "grad_norm": 0.14990335702896118,
71
+ "learning_rate": 0.011946666666666668,
72
+ "loss": 0.3415,
73
+ "step": 225
74
+ },
75
+ {
76
+ "epoch": 0.2,
77
+ "grad_norm": 0.1260153353214264,
78
+ "learning_rate": 0.01328,
79
+ "loss": 0.3336,
80
+ "step": 250
81
+ },
82
+ {
83
+ "epoch": 0.22,
84
+ "grad_norm": 0.12185318768024445,
85
+ "learning_rate": 0.014613333333333334,
86
+ "loss": 0.3263,
87
+ "step": 275
88
+ },
89
+ {
90
+ "epoch": 0.24,
91
+ "grad_norm": 0.12580621242523193,
92
+ "learning_rate": 0.015946666666666668,
93
+ "loss": 0.3427,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 0.26,
98
+ "grad_norm": 0.09728040546178818,
99
+ "learning_rate": 0.01728,
100
+ "loss": 0.3286,
101
+ "step": 325
102
+ },
103
+ {
104
+ "epoch": 0.28,
105
+ "grad_norm": 0.07153791934251785,
106
+ "learning_rate": 0.018613333333333332,
107
+ "loss": 0.3346,
108
+ "step": 350
109
+ },
110
+ {
111
+ "epoch": 0.3,
112
+ "grad_norm": 0.07144256681203842,
113
+ "learning_rate": 0.019946666666666665,
114
+ "loss": 0.3233,
115
+ "step": 375
116
+ },
117
+ {
118
+ "epoch": 0.32,
119
+ "grad_norm": 0.08008267730474472,
120
+ "learning_rate": 0.019997504677881223,
121
+ "loss": 0.3181,
122
+ "step": 400
123
+ },
124
+ {
125
+ "epoch": 0.34,
126
+ "grad_norm": 0.04643591120839119,
127
+ "learning_rate": 0.01998959986294895,
128
+ "loss": 0.304,
129
+ "step": 425
130
+ },
131
+ {
132
+ "epoch": 0.36,
133
+ "grad_norm": 0.05323820933699608,
134
+ "learning_rate": 0.01997628550346359,
135
+ "loss": 0.3053,
136
+ "step": 450
137
+ },
138
+ {
139
+ "epoch": 0.38,
140
+ "grad_norm": 0.049062035977840424,
141
+ "learning_rate": 0.019957568809385692,
142
+ "loss": 0.2969,
143
+ "step": 475
144
+ },
145
+ {
146
+ "epoch": 0.4,
147
+ "grad_norm": 0.06179765611886978,
148
+ "learning_rate": 0.019933459916135276,
149
+ "loss": 0.2954,
150
+ "step": 500
151
+ },
152
+ {
153
+ "epoch": 0.42,
154
+ "grad_norm": 0.04942788556218147,
155
+ "learning_rate": 0.019903971879103345,
156
+ "loss": 0.2957,
157
+ "step": 525
158
+ },
159
+ {
160
+ "epoch": 0.44,
161
+ "grad_norm": 0.05358466878533363,
162
+ "learning_rate": 0.01986912066658215,
163
+ "loss": 0.2813,
164
+ "step": 550
165
+ },
166
+ {
167
+ "epoch": 0.46,
168
+ "grad_norm": 0.046258535236120224,
169
+ "learning_rate": 0.019828925151118088,
170
+ "loss": 0.2995,
171
+ "step": 575
172
+ },
173
+ {
174
+ "epoch": 0.48,
175
+ "grad_norm": 0.04525147005915642,
176
+ "learning_rate": 0.019783407099291876,
177
+ "loss": 0.2839,
178
+ "step": 600
179
+ },
180
+ {
181
+ "epoch": 0.5,
182
+ "grad_norm": 0.04446304962038994,
183
+ "learning_rate": 0.019732591159931562,
184
+ "loss": 0.2961,
185
+ "step": 625
186
+ },
187
+ {
188
+ "epoch": 0.52,
189
+ "grad_norm": 0.04161718487739563,
190
+ "learning_rate": 0.019676504850764758,
191
+ "loss": 0.2866,
192
+ "step": 650
193
+ },
194
+ {
195
+ "epoch": 0.54,
196
+ "grad_norm": 0.03767893463373184,
197
+ "learning_rate": 0.0196151785435173,
198
+ "loss": 0.2761,
199
+ "step": 675
200
+ },
201
+ {
202
+ "epoch": 0.56,
203
+ "grad_norm": 0.04273810610175133,
204
+ "learning_rate": 0.019548645447466432,
205
+ "loss": 0.2767,
206
+ "step": 700
207
+ },
208
+ {
209
+ "epoch": 0.58,
210
+ "grad_norm": 0.0471462719142437,
211
+ "learning_rate": 0.01947694159145737,
212
+ "loss": 0.283,
213
+ "step": 725
214
+ },
215
+ {
216
+ "epoch": 0.6,
217
+ "grad_norm": 0.03773315250873566,
218
+ "learning_rate": 0.01940010580439308,
219
+ "loss": 0.2799,
220
+ "step": 750
221
+ },
222
+ {
223
+ "epoch": 0.62,
224
+ "grad_norm": 0.04052406921982765,
225
+ "learning_rate": 0.019318179694207724,
226
+ "loss": 0.2679,
227
+ "step": 775
228
+ },
229
+ {
230
+ "epoch": 0.64,
231
+ "grad_norm": 0.04028647765517235,
232
+ "learning_rate": 0.019231207625335235,
233
+ "loss": 0.2734,
234
+ "step": 800
235
+ },
236
+ {
237
+ "epoch": 0.66,
238
+ "grad_norm": 0.03707930073142052,
239
+ "learning_rate": 0.019139236694685185,
240
+ "loss": 0.2665,
241
+ "step": 825
242
+ },
243
+ {
244
+ "epoch": 0.68,
245
+ "grad_norm": 0.0416373685002327,
246
+ "learning_rate": 0.019042316706138988,
247
+ "loss": 0.2673,
248
+ "step": 850
249
+ },
250
+ {
251
+ "epoch": 0.7,
252
+ "grad_norm": 0.03203510865569115,
253
+ "learning_rate": 0.018940500143580216,
254
+ "loss": 0.2757,
255
+ "step": 875
256
+ },
257
+ {
258
+ "epoch": 0.72,
259
+ "grad_norm": 0.031195444986224174,
260
+ "learning_rate": 0.018833842142473644,
261
+ "loss": 0.2698,
262
+ "step": 900
263
+ },
264
+ {
265
+ "epoch": 0.74,
266
+ "grad_norm": 0.03926091641187668,
267
+ "learning_rate": 0.018722400460008438,
268
+ "loss": 0.2676,
269
+ "step": 925
270
+ },
271
+ {
272
+ "epoch": 0.76,
273
+ "grad_norm": 0.03447200357913971,
274
+ "learning_rate": 0.018606235443821603,
275
+ "loss": 0.2646,
276
+ "step": 950
277
+ },
278
+ {
279
+ "epoch": 0.78,
280
+ "grad_norm": 0.04425395652651787,
281
+ "learning_rate": 0.018485409999318678,
282
+ "loss": 0.2613,
283
+ "step": 975
284
+ },
285
+ {
286
+ "epoch": 0.8,
287
+ "grad_norm": 0.04347499832510948,
288
+ "learning_rate": 0.018359989555609354,
289
+ "loss": 0.2694,
290
+ "step": 1000
291
+ },
292
+ {
293
+ "epoch": 0.82,
294
+ "grad_norm": 0.039355143904685974,
295
+ "learning_rate": 0.018230042030076454,
296
+ "loss": 0.2672,
297
+ "step": 1025
298
+ },
299
+ {
300
+ "epoch": 0.84,
301
+ "grad_norm": 0.036026645451784134,
302
+ "learning_rate": 0.018095637791597462,
303
+ "loss": 0.2584,
304
+ "step": 1050
305
+ },
306
+ {
307
+ "epoch": 0.86,
308
+ "grad_norm": 0.03510862588882446,
309
+ "learning_rate": 0.017956849622438553,
310
+ "loss": 0.2547,
311
+ "step": 1075
312
+ },
313
+ {
314
+ "epoch": 0.88,
315
+ "grad_norm": 0.033159829676151276,
316
+ "learning_rate": 0.0178137526788417,
317
+ "loss": 0.255,
318
+ "step": 1100
319
+ },
320
+ {
321
+ "epoch": 0.9,
322
+ "grad_norm": 0.03295517340302467,
323
+ "learning_rate": 0.01766642445032625,
324
+ "loss": 0.2646,
325
+ "step": 1125
326
+ },
327
+ {
328
+ "epoch": 0.92,
329
+ "grad_norm": 0.029308080673217773,
330
+ "learning_rate": 0.017514944717726962,
331
+ "loss": 0.2576,
332
+ "step": 1150
333
+ },
334
+ {
335
+ "epoch": 0.94,
336
+ "grad_norm": 0.036943402141332626,
337
+ "learning_rate": 0.01735939550999131,
338
+ "loss": 0.2605,
339
+ "step": 1175
340
+ },
341
+ {
342
+ "epoch": 0.96,
343
+ "grad_norm": 0.02846529334783554,
344
+ "learning_rate": 0.017199861059759337,
345
+ "loss": 0.2594,
346
+ "step": 1200
347
+ },
348
+ {
349
+ "epoch": 0.98,
350
+ "grad_norm": 0.04283558577299118,
351
+ "learning_rate": 0.017036427757750203,
352
+ "loss": 0.2628,
353
+ "step": 1225
354
+ },
355
+ {
356
+ "epoch": 1.0,
357
+ "grad_norm": 0.03367648273706436,
358
+ "learning_rate": 0.016869184105980088,
359
+ "loss": 0.2529,
360
+ "step": 1250
361
+ },
362
+ {
363
+ "epoch": 1.02,
364
+ "grad_norm": 0.03456483781337738,
365
+ "learning_rate": 0.01669822066983678,
366
+ "loss": 0.2288,
367
+ "step": 1275
368
+ },
369
+ {
370
+ "epoch": 1.04,
371
+ "grad_norm": 0.024872206151485443,
372
+ "learning_rate": 0.01652363002903693,
373
+ "loss": 0.2375,
374
+ "step": 1300
375
+ },
376
+ {
377
+ "epoch": 1.06,
378
+ "grad_norm": 0.030464885756373405,
379
+ "learning_rate": 0.0163455067274925,
380
+ "loss": 0.2332,
381
+ "step": 1325
382
+ },
383
+ {
384
+ "epoch": 1.08,
385
+ "grad_norm": 0.031713806092739105,
386
+ "learning_rate": 0.016163947222113565,
387
+ "loss": 0.2398,
388
+ "step": 1350
389
+ },
390
+ {
391
+ "epoch": 1.1,
392
+ "grad_norm": 0.025640636682510376,
393
+ "learning_rate": 0.01597904983057519,
394
+ "loss": 0.2435,
395
+ "step": 1375
396
+ },
397
+ {
398
+ "epoch": 1.12,
399
+ "grad_norm": 0.02801099419593811,
400
+ "learning_rate": 0.01579091467807668,
401
+ "loss": 0.2423,
402
+ "step": 1400
403
+ },
404
+ {
405
+ "epoch": 1.1400000000000001,
406
+ "grad_norm": 0.03283185139298439,
407
+ "learning_rate": 0.01559964364312202,
408
+ "loss": 0.2334,
409
+ "step": 1425
410
+ },
411
+ {
412
+ "epoch": 1.16,
413
+ "grad_norm": 0.0458955243229866,
414
+ "learning_rate": 0.01540534030235087,
415
+ "loss": 0.2426,
416
+ "step": 1450
417
+ },
418
+ {
419
+ "epoch": 1.18,
420
+ "grad_norm": 0.030243318527936935,
421
+ "learning_rate": 0.015208109874449979,
422
+ "loss": 0.2383,
423
+ "step": 1475
424
+ },
425
+ {
426
+ "epoch": 1.2,
427
+ "grad_norm": 0.029557039961218834,
428
+ "learning_rate": 0.015008059163175401,
429
+ "loss": 0.2388,
430
+ "step": 1500
431
+ },
432
+ {
433
+ "epoch": 1.22,
434
+ "grad_norm": 0.028723224997520447,
435
+ "learning_rate": 0.014805296499516408,
436
+ "loss": 0.2434,
437
+ "step": 1525
438
+ },
439
+ {
440
+ "epoch": 1.24,
441
+ "grad_norm": 0.02488754130899906,
442
+ "learning_rate": 0.014599931683032325,
443
+ "loss": 0.2348,
444
+ "step": 1550
445
+ },
446
+ {
447
+ "epoch": 1.26,
448
+ "grad_norm": 0.025774620473384857,
449
+ "learning_rate": 0.014392075922394132,
450
+ "loss": 0.2327,
451
+ "step": 1575
452
+ },
453
+ {
454
+ "epoch": 1.28,
455
+ "grad_norm": 0.027344953268766403,
456
+ "learning_rate": 0.014181841775163014,
457
+ "loss": 0.2298,
458
+ "step": 1600
459
+ },
460
+ {
461
+ "epoch": 1.3,
462
+ "grad_norm": 0.02487209439277649,
463
+ "learning_rate": 0.013969343086838457,
464
+ "loss": 0.2299,
465
+ "step": 1625
466
+ },
467
+ {
468
+ "epoch": 1.32,
469
+ "grad_norm": 0.02767680585384369,
470
+ "learning_rate": 0.01375469492920889,
471
+ "loss": 0.2232,
472
+ "step": 1650
473
+ },
474
+ {
475
+ "epoch": 1.34,
476
+ "grad_norm": 0.023056739941239357,
477
+ "learning_rate": 0.013538013538038294,
478
+ "loss": 0.2287,
479
+ "step": 1675
480
+ },
481
+ {
482
+ "epoch": 1.3599999999999999,
483
+ "grad_norm": 0.027000591158866882,
484
+ "learning_rate": 0.013319416250122482,
485
+ "loss": 0.2288,
486
+ "step": 1700
487
+ },
488
+ {
489
+ "epoch": 1.38,
490
+ "grad_norm": 0.025275854393839836,
491
+ "learning_rate": 0.013099021439749154,
492
+ "loss": 0.2282,
493
+ "step": 1725
494
+ },
495
+ {
496
+ "epoch": 1.4,
497
+ "grad_norm": 0.028009621426463127,
498
+ "learning_rate": 0.012876948454596129,
499
+ "loss": 0.2288,
500
+ "step": 1750
501
+ },
502
+ {
503
+ "epoch": 1.42,
504
+ "grad_norm": 0.023568755015730858,
505
+ "learning_rate": 0.01265331755110247,
506
+ "loss": 0.2302,
507
+ "step": 1775
508
+ },
509
+ {
510
+ "epoch": 1.44,
511
+ "grad_norm": 0.029848851263523102,
512
+ "learning_rate": 0.01242824982934751,
513
+ "loss": 0.2176,
514
+ "step": 1800
515
+ },
516
+ {
517
+ "epoch": 1.46,
518
+ "grad_norm": 0.02378924749791622,
519
+ "learning_rate": 0.012201867167473015,
520
+ "loss": 0.2242,
521
+ "step": 1825
522
+ },
523
+ {
524
+ "epoch": 1.48,
525
+ "grad_norm": 0.030996032059192657,
526
+ "learning_rate": 0.01197429215568403,
527
+ "loss": 0.2275,
528
+ "step": 1850
529
+ },
530
+ {
531
+ "epoch": 1.5,
532
+ "grad_norm": 0.02629832550883293,
533
+ "learning_rate": 0.0117456480298641,
534
+ "loss": 0.2295,
535
+ "step": 1875
536
+ },
537
+ {
538
+ "epoch": 1.52,
539
+ "grad_norm": 0.02670891210436821,
540
+ "learning_rate": 0.011516058604840892,
541
+ "loss": 0.2272,
542
+ "step": 1900
543
+ },
544
+ {
545
+ "epoch": 1.54,
546
+ "grad_norm": 0.027848055586218834,
547
+ "learning_rate": 0.011285648207338251,
548
+ "loss": 0.225,
549
+ "step": 1925
550
+ },
551
+ {
552
+ "epoch": 1.56,
553
+ "grad_norm": 0.025103619322180748,
554
+ "learning_rate": 0.01105454160865112,
555
+ "loss": 0.2148,
556
+ "step": 1950
557
+ },
558
+ {
559
+ "epoch": 1.58,
560
+ "grad_norm": 0.02855285070836544,
561
+ "learning_rate": 0.010822863957079655,
562
+ "loss": 0.2292,
563
+ "step": 1975
564
+ },
565
+ {
566
+ "epoch": 1.6,
567
+ "grad_norm": 0.024564214050769806,
568
+ "learning_rate": 0.010590740710159228,
569
+ "loss": 0.2288,
570
+ "step": 2000
571
+ },
572
+ {
573
+ "epoch": 1.62,
574
+ "grad_norm": 0.029064735397696495,
575
+ "learning_rate": 0.010358297566722945,
576
+ "loss": 0.2128,
577
+ "step": 2025
578
+ },
579
+ {
580
+ "epoch": 1.6400000000000001,
581
+ "grad_norm": 0.024909919127821922,
582
+ "learning_rate": 0.010125660398833527,
583
+ "loss": 0.2235,
584
+ "step": 2050
585
+ },
586
+ {
587
+ "epoch": 1.6600000000000001,
588
+ "grad_norm": 0.022246485576033592,
589
+ "learning_rate": 0.009892955183621354,
590
+ "loss": 0.2283,
591
+ "step": 2075
592
+ },
593
+ {
594
+ "epoch": 1.6800000000000002,
595
+ "grad_norm": 0.024962959811091423,
596
+ "learning_rate": 0.00966030793506564,
597
+ "loss": 0.2243,
598
+ "step": 2100
599
+ },
600
+ {
601
+ "epoch": 1.7,
602
+ "grad_norm": 0.023296242579817772,
603
+ "learning_rate": 0.00942784463575562,
604
+ "loss": 0.2225,
605
+ "step": 2125
606
+ },
607
+ {
608
+ "epoch": 1.72,
609
+ "grad_norm": 0.026053672656416893,
610
+ "learning_rate": 0.009195691168668772,
611
+ "loss": 0.214,
612
+ "step": 2150
613
+ },
614
+ {
615
+ "epoch": 1.74,
616
+ "grad_norm": 0.022729799151420593,
617
+ "learning_rate": 0.008963973249002968,
618
+ "loss": 0.2309,
619
+ "step": 2175
620
+ },
621
+ {
622
+ "epoch": 1.76,
623
+ "grad_norm": 0.023254524916410446,
624
+ "learning_rate": 0.008732816356099456,
625
+ "loss": 0.2217,
626
+ "step": 2200
627
+ },
628
+ {
629
+ "epoch": 1.78,
630
+ "grad_norm": 0.026419617235660553,
631
+ "learning_rate": 0.008502345665493597,
632
+ "loss": 0.215,
633
+ "step": 2225
634
+ },
635
+ {
636
+ "epoch": 1.8,
637
+ "grad_norm": 0.02386222593486309,
638
+ "learning_rate": 0.00827268598113011,
639
+ "loss": 0.225,
640
+ "step": 2250
641
+ },
642
+ {
643
+ "epoch": 1.8199999999999998,
644
+ "grad_norm": 0.023376762866973877,
645
+ "learning_rate": 0.00804396166777952,
646
+ "loss": 0.2156,
647
+ "step": 2275
648
+ },
649
+ {
650
+ "epoch": 1.8399999999999999,
651
+ "grad_norm": 0.025676891207695007,
652
+ "learning_rate": 0.007816296583692449,
653
+ "loss": 0.2188,
654
+ "step": 2300
655
+ },
656
+ {
657
+ "epoch": 1.8599999999999999,
658
+ "grad_norm": 0.021596932783722878,
659
+ "learning_rate": 0.007589814013528186,
660
+ "loss": 0.2217,
661
+ "step": 2325
662
+ },
663
+ {
664
+ "epoch": 1.88,
665
+ "grad_norm": 0.02387806959450245,
666
+ "learning_rate": 0.0073646366015938745,
667
+ "loss": 0.2134,
668
+ "step": 2350
669
+ },
670
+ {
671
+ "epoch": 1.9,
672
+ "grad_norm": 0.02620845101773739,
673
+ "learning_rate": 0.007140886285430448,
674
+ "loss": 0.2373,
675
+ "step": 2375
676
+ },
677
+ {
678
+ "epoch": 1.92,
679
+ "grad_norm": 0.02476939558982849,
680
+ "learning_rate": 0.006918684229781305,
681
+ "loss": 0.2217,
682
+ "step": 2400
683
+ },
684
+ {
685
+ "epoch": 1.94,
686
+ "grad_norm": 0.026036998257040977,
687
+ "learning_rate": 0.0066981507609794625,
688
+ "loss": 0.2138,
689
+ "step": 2425
690
+ },
691
+ {
692
+ "epoch": 1.96,
693
+ "grad_norm": 0.023386968299746513,
694
+ "learning_rate": 0.006479405301788727,
695
+ "loss": 0.2157,
696
+ "step": 2450
697
+ },
698
+ {
699
+ "epoch": 1.98,
700
+ "grad_norm": 0.02384364791214466,
701
+ "learning_rate": 0.006262566306734154,
702
+ "loss": 0.2194,
703
+ "step": 2475
704
+ },
705
+ {
706
+ "epoch": 2.0,
707
+ "grad_norm": 0.02077922783792019,
708
+ "learning_rate": 0.006047751197956838,
709
+ "loss": 0.2168,
710
+ "step": 2500
711
+ },
712
+ {
713
+ "epoch": 2.02,
714
+ "grad_norm": 0.025426547974348068,
715
+ "learning_rate": 0.005835076301627752,
716
+ "loss": 0.1909,
717
+ "step": 2525
718
+ },
719
+ {
720
+ "epoch": 2.04,
721
+ "grad_norm": 0.025073401629924774,
722
+ "learning_rate": 0.005624656784955071,
723
+ "loss": 0.1848,
724
+ "step": 2550
725
+ },
726
+ {
727
+ "epoch": 2.06,
728
+ "grad_norm": 0.026666883379220963,
729
+ "learning_rate": 0.0054166065938191016,
730
+ "loss": 0.1931,
731
+ "step": 2575
732
+ },
733
+ {
734
+ "epoch": 2.08,
735
+ "grad_norm": 0.025660671293735504,
736
+ "learning_rate": 0.005211038391068576,
737
+ "loss": 0.1838,
738
+ "step": 2600
739
+ },
740
+ {
741
+ "epoch": 2.1,
742
+ "grad_norm": 0.02300131320953369,
743
+ "learning_rate": 0.005008063495511723,
744
+ "loss": 0.1932,
745
+ "step": 2625
746
+ },
747
+ {
748
+ "epoch": 2.12,
749
+ "grad_norm": 0.021838432177901268,
750
+ "learning_rate": 0.004807791821635186,
751
+ "loss": 0.1856,
752
+ "step": 2650
753
+ },
754
+ {
755
+ "epoch": 2.14,
756
+ "grad_norm": 0.02283412404358387,
757
+ "learning_rate": 0.004610331820083378,
758
+ "loss": 0.192,
759
+ "step": 2675
760
+ },
761
+ {
762
+ "epoch": 2.16,
763
+ "grad_norm": 0.024570999667048454,
764
+ "learning_rate": 0.004415790418930541,
765
+ "loss": 0.1874,
766
+ "step": 2700
767
+ },
768
+ {
769
+ "epoch": 2.18,
770
+ "grad_norm": 0.02300398424267769,
771
+ "learning_rate": 0.0042242729657773264,
772
+ "loss": 0.1953,
773
+ "step": 2725
774
+ },
775
+ {
776
+ "epoch": 2.2,
777
+ "grad_norm": 0.02675442397594452,
778
+ "learning_rate": 0.004035883170703182,
779
+ "loss": 0.1894,
780
+ "step": 2750
781
+ },
782
+ {
783
+ "epoch": 2.22,
784
+ "grad_norm": 0.023160850629210472,
785
+ "learning_rate": 0.003850723050105548,
786
+ "loss": 0.1844,
787
+ "step": 2775
788
+ },
789
+ {
790
+ "epoch": 2.24,
791
+ "grad_norm": 0.022288281470537186,
792
+ "learning_rate": 0.003668892871456144,
793
+ "loss": 0.1848,
794
+ "step": 2800
795
+ },
796
+ {
797
+ "epoch": 2.26,
798
+ "grad_norm": 0.02427896484732628,
799
+ "learning_rate": 0.0034904910990043924,
800
+ "loss": 0.1943,
801
+ "step": 2825
802
+ },
803
+ {
804
+ "epoch": 2.2800000000000002,
805
+ "grad_norm": 0.027167946100234985,
806
+ "learning_rate": 0.0033156143404572793,
807
+ "loss": 0.1838,
808
+ "step": 2850
809
+ },
810
+ {
811
+ "epoch": 2.3,
812
+ "grad_norm": 0.022306451573967934,
813
+ "learning_rate": 0.003144357294664565,
814
+ "loss": 0.1877,
815
+ "step": 2875
816
+ },
817
+ {
818
+ "epoch": 2.32,
819
+ "grad_norm": 0.02225615084171295,
820
+ "learning_rate": 0.0029768127003377077,
821
+ "loss": 0.1933,
822
+ "step": 2900
823
+ },
824
+ {
825
+ "epoch": 2.34,
826
+ "grad_norm": 0.02079549804329872,
827
+ "learning_rate": 0.0028130712858302066,
828
+ "loss": 0.1821,
829
+ "step": 2925
830
+ },
831
+ {
832
+ "epoch": 2.36,
833
+ "grad_norm": 0.026661023497581482,
834
+ "learning_rate": 0.0026532217200065855,
835
+ "loss": 0.1904,
836
+ "step": 2950
837
+ },
838
+ {
839
+ "epoch": 2.38,
840
+ "grad_norm": 0.02208436280488968,
841
+ "learning_rate": 0.0024973505642266625,
842
+ "loss": 0.1821,
843
+ "step": 2975
844
+ },
845
+ {
846
+ "epoch": 2.4,
847
+ "grad_norm": 0.022112136706709862,
848
+ "learning_rate": 0.0023455422254710233,
849
+ "loss": 0.1835,
850
+ "step": 3000
851
+ },
852
+ {
853
+ "epoch": 2.42,
854
+ "grad_norm": 0.027875550091266632,
855
+ "learning_rate": 0.0021978789106331664,
856
+ "loss": 0.1859,
857
+ "step": 3025
858
+ },
859
+ {
860
+ "epoch": 2.44,
861
+ "grad_norm": 0.021113188937306404,
862
+ "learning_rate": 0.0020544405820030164,
863
+ "loss": 0.1883,
864
+ "step": 3050
865
+ },
866
+ {
867
+ "epoch": 2.46,
868
+ "grad_norm": 0.023992350324988365,
869
+ "learning_rate": 0.0019153049139659384,
870
+ "loss": 0.1877,
871
+ "step": 3075
872
+ },
873
+ {
874
+ "epoch": 2.48,
875
+ "grad_norm": 0.024912724271416664,
876
+ "learning_rate": 0.0017805472509406694,
877
+ "loss": 0.182,
878
+ "step": 3100
879
+ },
880
+ {
881
+ "epoch": 2.5,
882
+ "grad_norm": 0.025338461622595787,
883
+ "learning_rate": 0.0016502405665790132,
884
+ "loss": 0.1901,
885
+ "step": 3125
886
+ },
887
+ {
888
+ "epoch": 2.52,
889
+ "grad_norm": 0.025324387475848198,
890
+ "learning_rate": 0.0015244554242493024,
891
+ "loss": 0.1816,
892
+ "step": 3150
893
+ },
894
+ {
895
+ "epoch": 2.54,
896
+ "grad_norm": 0.025014016777276993,
897
+ "learning_rate": 0.00140325993882509,
898
+ "loss": 0.1859,
899
+ "step": 3175
900
+ },
901
+ {
902
+ "epoch": 2.56,
903
+ "grad_norm": 0.020721474662423134,
904
+ "learning_rate": 0.0012867197397997531,
905
+ "loss": 0.1819,
906
+ "step": 3200
907
+ },
908
+ {
909
+ "epoch": 2.58,
910
+ "grad_norm": 0.022899964824318886,
911
+ "learning_rate": 0.0011748979357469624,
912
+ "loss": 0.1777,
913
+ "step": 3225
914
+ },
915
+ {
916
+ "epoch": 2.6,
917
+ "grad_norm": 0.025203322991728783,
918
+ "learning_rate": 0.001067855080146266,
919
+ "loss": 0.1796,
920
+ "step": 3250
921
+ },
922
+ {
923
+ "epoch": 2.62,
924
+ "grad_norm": 0.02526050992310047,
925
+ "learning_rate": 0.0009656491385923216,
926
+ "loss": 0.1811,
927
+ "step": 3275
928
+ },
929
+ {
930
+ "epoch": 2.64,
931
+ "grad_norm": 0.02366676740348339,
932
+ "learning_rate": 0.0008683354574055058,
933
+ "loss": 0.1805,
934
+ "step": 3300
935
+ },
936
+ {
937
+ "epoch": 2.66,
938
+ "grad_norm": 0.02300945669412613,
939
+ "learning_rate": 0.0007759667336609011,
940
+ "loss": 0.1833,
941
+ "step": 3325
942
+ },
943
+ {
944
+ "epoch": 2.68,
945
+ "grad_norm": 0.022977150976657867,
946
+ "learning_rate": 0.0006885929866518948,
947
+ "loss": 0.1852,
948
+ "step": 3350
949
+ },
950
+ {
951
+ "epoch": 2.7,
952
+ "grad_norm": 0.021491436287760735,
953
+ "learning_rate": 0.0006062615308038599,
954
+ "loss": 0.1818,
955
+ "step": 3375
956
+ },
957
+ {
958
+ "epoch": 2.7199999999999998,
959
+ "grad_norm": 0.028023768216371536,
960
+ "learning_rate": 0.0005290169500525577,
961
+ "loss": 0.1925,
962
+ "step": 3400
963
+ },
964
+ {
965
+ "epoch": 2.74,
966
+ "grad_norm": 0.020721763372421265,
967
+ "learning_rate": 0.0004569010737011381,
968
+ "loss": 0.1822,
969
+ "step": 3425
970
+ },
971
+ {
972
+ "epoch": 2.76,
973
+ "grad_norm": 0.021237611770629883,
974
+ "learning_rate": 0.00038995295376884954,
975
+ "loss": 0.1742,
976
+ "step": 3450
977
+ },
978
+ {
979
+ "epoch": 2.7800000000000002,
980
+ "grad_norm": 0.027307681739330292,
981
+ "learning_rate": 0.00032820884384367145,
982
+ "loss": 0.1784,
983
+ "step": 3475
984
+ },
985
+ {
986
+ "epoch": 2.8,
987
+ "grad_norm": 0.027936382219195366,
988
+ "learning_rate": 0.0002717021794503616,
989
+ "loss": 0.1798,
990
+ "step": 3500
991
+ },
992
+ {
993
+ "epoch": 2.82,
994
+ "grad_norm": 0.022965189069509506,
995
+ "learning_rate": 0.00022046355994452595,
996
+ "loss": 0.1811,
997
+ "step": 3525
998
+ },
999
+ {
1000
+ "epoch": 2.84,
1001
+ "grad_norm": 0.028826095163822174,
1002
+ "learning_rate": 0.00017452073194253238,
1003
+ "loss": 0.1772,
1004
+ "step": 3550
1005
+ },
1006
+ {
1007
+ "epoch": 2.86,
1008
+ "grad_norm": 0.025759048759937286,
1009
+ "learning_rate": 0.0001338985742962251,
1010
+ "loss": 0.1745,
1011
+ "step": 3575
1012
+ },
1013
+ {
1014
+ "epoch": 2.88,
1015
+ "grad_norm": 0.023553024977445602,
1016
+ "learning_rate": 9.8619084620587e-05,
1017
+ "loss": 0.1828,
1018
+ "step": 3600
1019
+ },
1020
+ {
1021
+ "epoch": 2.9,
1022
+ "grad_norm": 0.026909027248620987,
1023
+ "learning_rate": 6.870136738164612e-05,
1024
+ "loss": 0.1831,
1025
+ "step": 3625
1026
+ },
1027
+ {
1028
+ "epoch": 2.92,
1029
+ "grad_norm": 0.027830779552459717,
1030
+ "learning_rate": 4.416162355107001e-05,
1031
+ "loss": 0.1796,
1032
+ "step": 3650
1033
+ },
1034
+ {
1035
+ "epoch": 2.94,
1036
+ "grad_norm": 0.02305370755493641,
1037
+ "learning_rate": 2.5013141833046815e-05,
1038
+ "loss": 0.1748,
1039
+ "step": 3675
1040
+ },
1041
+ {
1042
+ "epoch": 2.96,
1043
+ "grad_norm": 0.02431248314678669,
1044
+ "learning_rate": 1.1266291468229328e-05,
1045
+ "loss": 0.1767,
1046
+ "step": 3700
1047
+ },
1048
+ {
1049
+ "epoch": 2.98,
1050
+ "grad_norm": 0.023445066064596176,
1051
+ "learning_rate": 2.928516618599941e-06,
1052
+ "loss": 0.1756,
1053
+ "step": 3725
1054
+ },
1055
+ {
1056
+ "epoch": 3.0,
1057
+ "grad_norm": 0.020654326304793358,
1058
+ "learning_rate": 4.332336324130992e-09,
1059
+ "loss": 0.1814,
1060
+ "step": 3750
1061
+ },
1062
+ {
1063
+ "epoch": 3.0,
1064
+ "step": 3750,
1065
+ "total_flos": 2.43882352705536e+18,
1066
+ "train_loss": 0.23951186167399088,
1067
+ "train_runtime": 3462.7911,
1068
+ "train_samples_per_second": 34.654,
1069
+ "train_steps_per_second": 1.083
1070
+ }
1071
+ ],
1072
+ "logging_steps": 25,
1073
+ "max_steps": 3750,
1074
+ "num_input_tokens_seen": 0,
1075
+ "num_train_epochs": 3,
1076
+ "save_steps": 0,
1077
+ "stateful_callbacks": {
1078
+ "TrainerControl": {
1079
+ "args": {
1080
+ "should_epoch_stop": false,
1081
+ "should_evaluate": false,
1082
+ "should_log": false,
1083
+ "should_save": true,
1084
+ "should_training_stop": true
1085
+ },
1086
+ "attributes": {}
1087
+ }
1088
+ },
1089
+ "total_flos": 2.43882352705536e+18,
1090
+ "train_batch_size": 32,
1091
+ "trial_name": null,
1092
+ "trial_params": null
1093
+ }