trapoom555 commited on
Commit
7b65800
1 Parent(s): e1c74ee

Upload checkpoints

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .DS_Store +0 -0
  2. adapter_config.json +29 -0
  3. adapter_model.safetensors +3 -0
  4. checkpoints/.DS_Store +0 -0
  5. checkpoints/checkpoint-1000/.DS_Store +0 -0
  6. checkpoints/checkpoint-1000/adapter_config.json +29 -0
  7. checkpoints/checkpoint-1000/adapter_model.safetensors +3 -0
  8. checkpoints/checkpoint-1000/optimizer.pt +3 -0
  9. checkpoints/checkpoint-1000/rng_state_0.pth +3 -0
  10. checkpoints/checkpoint-1000/rng_state_1.pth +3 -0
  11. checkpoints/checkpoint-1000/rng_state_2.pth +3 -0
  12. checkpoints/checkpoint-1000/rng_state_3.pth +3 -0
  13. checkpoints/checkpoint-1000/scheduler.pt +3 -0
  14. checkpoints/checkpoint-1000/trainer_state.json +721 -0
  15. checkpoints/checkpoint-1000/training_args.bin +3 -0
  16. checkpoints/checkpoint-1500/.DS_Store +0 -0
  17. checkpoints/checkpoint-1500/adapter_config.json +29 -0
  18. checkpoints/checkpoint-1500/adapter_model.safetensors +3 -0
  19. checkpoints/checkpoint-1500/optimizer.pt +3 -0
  20. checkpoints/checkpoint-1500/rng_state_0.pth +3 -0
  21. checkpoints/checkpoint-1500/rng_state_1.pth +3 -0
  22. checkpoints/checkpoint-1500/rng_state_2.pth +3 -0
  23. checkpoints/checkpoint-1500/rng_state_3.pth +3 -0
  24. checkpoints/checkpoint-1500/scheduler.pt +3 -0
  25. checkpoints/checkpoint-1500/trainer_state.json +1071 -0
  26. checkpoints/checkpoint-1500/training_args.bin +3 -0
  27. checkpoints/checkpoint-2000/.DS_Store +0 -0
  28. checkpoints/checkpoint-2000/adapter_config.json +29 -0
  29. checkpoints/checkpoint-2000/adapter_model.safetensors +3 -0
  30. checkpoints/checkpoint-2000/optimizer.pt +3 -0
  31. checkpoints/checkpoint-2000/rng_state_0.pth +3 -0
  32. checkpoints/checkpoint-2000/rng_state_1.pth +3 -0
  33. checkpoints/checkpoint-2000/rng_state_2.pth +3 -0
  34. checkpoints/checkpoint-2000/rng_state_3.pth +3 -0
  35. checkpoints/checkpoint-2000/scheduler.pt +3 -0
  36. checkpoints/checkpoint-2000/trainer_state.json +1421 -0
  37. checkpoints/checkpoint-2000/training_args.bin +3 -0
  38. checkpoints/checkpoint-2500/.DS_Store +0 -0
  39. checkpoints/checkpoint-2500/adapter_config.json +29 -0
  40. checkpoints/checkpoint-2500/adapter_model.safetensors +3 -0
  41. checkpoints/checkpoint-2500/optimizer.pt +3 -0
  42. checkpoints/checkpoint-2500/rng_state_0.pth +3 -0
  43. checkpoints/checkpoint-2500/rng_state_1.pth +3 -0
  44. checkpoints/checkpoint-2500/rng_state_2.pth +3 -0
  45. checkpoints/checkpoint-2500/rng_state_3.pth +3 -0
  46. checkpoints/checkpoint-2500/scheduler.pt +3 -0
  47. checkpoints/checkpoint-2500/trainer_state.json +1771 -0
  48. checkpoints/checkpoint-2500/training_args.bin +3 -0
  49. checkpoints/checkpoint-3000/.DS_Store +0 -0
  50. checkpoints/checkpoint-3000/adapter_config.json +29 -0
.DS_Store ADDED
Binary file (6.15 kB). View file
 
adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e566c61bed3b05cde1ebb3a325febfce0fafe3d3f3e648acb48bf8bd33cedb20
3
+ size 5919456
checkpoints/.DS_Store ADDED
Binary file (14.3 kB). View file
 
checkpoints/checkpoint-1000/.DS_Store ADDED
Binary file (6.15 kB). View file
 
checkpoints/checkpoint-1000/adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }
checkpoints/checkpoint-1000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7021d0f53b2af54bc0fb2b36f002603368f00cc5e37cf8ffa4d35e14e850cfc
3
+ size 5919456
checkpoints/checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cae149eaee16a95e002bf921dbb3bd8869b631f62813911b3680edf18d7ff723
3
+ size 11930938
checkpoints/checkpoint-1000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:287014c23e0c4e9613e974e969516cd0fc0bdce46daf1c6f1c0b66e3eb091e50
3
+ size 15024
checkpoints/checkpoint-1000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ab87bd92d10c436f5b79e618401a8481e12e99083829692c2fc0de84edcad99
3
+ size 15024
checkpoints/checkpoint-1000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8553566dff78f7c07a5a3e5517a0a640c6ef80a0e95eb328fed4c566945f6fd0
3
+ size 15024
checkpoints/checkpoint-1000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60c899ef713ad40446331266b18149afa90e55eed1210839243a923ea8aa772d
3
+ size 15024
checkpoints/checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:feeecf82b24e0ddf3c8d8285a678fe39a1184c1c961cf677b5ac8d36409a9a05
3
+ size 1064
checkpoints/checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,721 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.1015950421619425,
5
+ "eval_steps": 500,
6
+ "global_step": 1000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001015950421619425,
13
+ "grad_norm": 17.625,
14
+ "learning_rate": 5e-06,
15
+ "loss": 3.4264,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.00203190084323885,
20
+ "grad_norm": 12.5625,
21
+ "learning_rate": 1e-05,
22
+ "loss": 3.432,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.003047851264858275,
27
+ "grad_norm": 14.0625,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 3.23,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0040638016864777,
34
+ "grad_norm": 12.4375,
35
+ "learning_rate": 2e-05,
36
+ "loss": 2.9762,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.005079752108097125,
41
+ "grad_norm": 10.0625,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 2.6173,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.00609570252971655,
48
+ "grad_norm": 10.1875,
49
+ "learning_rate": 3e-05,
50
+ "loss": 2.2004,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.007111652951335975,
55
+ "grad_norm": 7.03125,
56
+ "learning_rate": 3.5e-05,
57
+ "loss": 1.4176,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0081276033729554,
62
+ "grad_norm": 4.375,
63
+ "learning_rate": 4e-05,
64
+ "loss": 1.0122,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.009143553794574825,
69
+ "grad_norm": 6.5625,
70
+ "learning_rate": 4.5e-05,
71
+ "loss": 0.9116,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01015950421619425,
76
+ "grad_norm": 5.28125,
77
+ "learning_rate": 5e-05,
78
+ "loss": 0.6832,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.011175454637813675,
83
+ "grad_norm": 5.5,
84
+ "learning_rate": 4.9999870035728426e-05,
85
+ "loss": 0.7355,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0121914050594331,
90
+ "grad_norm": 5.1875,
91
+ "learning_rate": 4.9999480144264944e-05,
92
+ "loss": 0.6673,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.013207355481052525,
97
+ "grad_norm": 4.5,
98
+ "learning_rate": 4.9998830329663314e-05,
99
+ "loss": 0.6792,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.01422330590267195,
104
+ "grad_norm": 3.9375,
105
+ "learning_rate": 4.9997920598679756e-05,
106
+ "loss": 0.6207,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.015239256324291375,
111
+ "grad_norm": 3.15625,
112
+ "learning_rate": 4.999675096077286e-05,
113
+ "loss": 0.483,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.0162552067459108,
118
+ "grad_norm": 5.28125,
119
+ "learning_rate": 4.999532142810354e-05,
120
+ "loss": 0.5319,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.017271157167530225,
125
+ "grad_norm": 4.59375,
126
+ "learning_rate": 4.999363201553483e-05,
127
+ "loss": 0.6052,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.01828710758914965,
132
+ "grad_norm": 5.03125,
133
+ "learning_rate": 4.9991682740631794e-05,
134
+ "loss": 0.4258,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.019303058010769075,
139
+ "grad_norm": 3.859375,
140
+ "learning_rate": 4.998947362366133e-05,
141
+ "loss": 0.4309,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.0203190084323885,
146
+ "grad_norm": 3.328125,
147
+ "learning_rate": 4.998700468759193e-05,
148
+ "loss": 0.3957,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.021334958854007924,
153
+ "grad_norm": 4.9375,
154
+ "learning_rate": 4.9984275958093475e-05,
155
+ "loss": 0.4777,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02235090927562735,
160
+ "grad_norm": 4.78125,
161
+ "learning_rate": 4.998128746353695e-05,
162
+ "loss": 0.3549,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.023366859697246774,
167
+ "grad_norm": 4.0625,
168
+ "learning_rate": 4.997803923499417e-05,
169
+ "loss": 0.4447,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.0243828101188662,
174
+ "grad_norm": 6.375,
175
+ "learning_rate": 4.99745313062374e-05,
176
+ "loss": 0.3808,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.025398760540485624,
181
+ "grad_norm": 3.59375,
182
+ "learning_rate": 4.99707637137391e-05,
183
+ "loss": 0.3827,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.02641471096210505,
188
+ "grad_norm": 3.015625,
189
+ "learning_rate": 4.996673649667145e-05,
190
+ "loss": 0.3694,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.027430661383724474,
195
+ "grad_norm": 2.296875,
196
+ "learning_rate": 4.9962449696906e-05,
197
+ "loss": 0.3586,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.0284466118053439,
202
+ "grad_norm": 4.125,
203
+ "learning_rate": 4.9957903359013214e-05,
204
+ "loss": 0.3832,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.029462562226963324,
209
+ "grad_norm": 3.296875,
210
+ "learning_rate": 4.995309753026201e-05,
211
+ "loss": 0.328,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.03047851264858275,
216
+ "grad_norm": 4.5,
217
+ "learning_rate": 4.994803226061927e-05,
218
+ "loss": 0.3667,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.03149446307020217,
223
+ "grad_norm": 4.3125,
224
+ "learning_rate": 4.994270760274933e-05,
225
+ "loss": 0.3811,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.0325104134918216,
230
+ "grad_norm": 3.421875,
231
+ "learning_rate": 4.99371236120134e-05,
232
+ "loss": 0.3065,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.03352636391344102,
237
+ "grad_norm": 4.6875,
238
+ "learning_rate": 4.993128034646902e-05,
239
+ "loss": 0.4177,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.03454231433506045,
244
+ "grad_norm": 3.046875,
245
+ "learning_rate": 4.992517786686947e-05,
246
+ "loss": 0.33,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.03555826475667987,
251
+ "grad_norm": 1.8828125,
252
+ "learning_rate": 4.9918816236663077e-05,
253
+ "loss": 0.3287,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.0365742151782993,
258
+ "grad_norm": 3.8125,
259
+ "learning_rate": 4.991219552199262e-05,
260
+ "loss": 0.2934,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.03759016559991872,
265
+ "grad_norm": 4.28125,
266
+ "learning_rate": 4.99053157916946e-05,
267
+ "loss": 0.3176,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.03860611602153815,
272
+ "grad_norm": 2.609375,
273
+ "learning_rate": 4.989817711729856e-05,
274
+ "loss": 0.3318,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.03962206644315757,
279
+ "grad_norm": 2.375,
280
+ "learning_rate": 4.98907795730263e-05,
281
+ "loss": 0.3234,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.040638016864777,
286
+ "grad_norm": 4.46875,
287
+ "learning_rate": 4.988312323579114e-05,
288
+ "loss": 0.267,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.04165396728639642,
293
+ "grad_norm": 3.75,
294
+ "learning_rate": 4.98752081851971e-05,
295
+ "loss": 0.3081,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.04266991770801585,
300
+ "grad_norm": 2.203125,
301
+ "learning_rate": 4.986703450353809e-05,
302
+ "loss": 0.2917,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.04368586812963527,
307
+ "grad_norm": 1.6015625,
308
+ "learning_rate": 4.985860227579703e-05,
309
+ "loss": 0.2805,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0447018185512547,
314
+ "grad_norm": 3.140625,
315
+ "learning_rate": 4.984991158964499e-05,
316
+ "loss": 0.3534,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.04571776897287412,
321
+ "grad_norm": 3.296875,
322
+ "learning_rate": 4.9840962535440265e-05,
323
+ "loss": 0.335,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.04673371939449355,
328
+ "grad_norm": 3.25,
329
+ "learning_rate": 4.983175520622744e-05,
330
+ "loss": 0.2544,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.04774966981611297,
335
+ "grad_norm": 2.25,
336
+ "learning_rate": 4.982228969773642e-05,
337
+ "loss": 0.3449,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.0487656202377324,
342
+ "grad_norm": 4.9375,
343
+ "learning_rate": 4.9812566108381435e-05,
344
+ "loss": 0.2964,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.04978157065935182,
349
+ "grad_norm": 1.5703125,
350
+ "learning_rate": 4.9802584539260035e-05,
351
+ "loss": 0.2799,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.05079752108097125,
356
+ "grad_norm": 2.828125,
357
+ "learning_rate": 4.979234509415199e-05,
358
+ "loss": 0.3231,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.05181347150259067,
363
+ "grad_norm": 2.9375,
364
+ "learning_rate": 4.978184787951828e-05,
365
+ "loss": 0.2943,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.0528294219242101,
370
+ "grad_norm": 2.34375,
371
+ "learning_rate": 4.977109300449992e-05,
372
+ "loss": 0.2705,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.05384537234582952,
377
+ "grad_norm": 3.140625,
378
+ "learning_rate": 4.9760080580916876e-05,
379
+ "loss": 0.2998,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.05486132276744895,
384
+ "grad_norm": 3.5625,
385
+ "learning_rate": 4.974881072326688e-05,
386
+ "loss": 0.2595,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.05587727318906837,
391
+ "grad_norm": 4.25,
392
+ "learning_rate": 4.9737283548724236e-05,
393
+ "loss": 0.2803,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.0568932236106878,
398
+ "grad_norm": 4.0625,
399
+ "learning_rate": 4.97254991771386e-05,
400
+ "loss": 0.3511,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.05790917403230722,
405
+ "grad_norm": 2.515625,
406
+ "learning_rate": 4.971345773103377e-05,
407
+ "loss": 0.312,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.05892512445392665,
412
+ "grad_norm": 3.21875,
413
+ "learning_rate": 4.9701159335606365e-05,
414
+ "loss": 0.2482,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.05994107487554607,
419
+ "grad_norm": 5.5,
420
+ "learning_rate": 4.968860411872454e-05,
421
+ "loss": 0.2537,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.0609570252971655,
426
+ "grad_norm": 3.546875,
427
+ "learning_rate": 4.967579221092666e-05,
428
+ "loss": 0.3125,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.06197297571878492,
433
+ "grad_norm": 2.984375,
434
+ "learning_rate": 4.966272374541996e-05,
435
+ "loss": 0.2354,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.06298892614040434,
440
+ "grad_norm": 3.6875,
441
+ "learning_rate": 4.964939885807912e-05,
442
+ "loss": 0.3213,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.06400487656202378,
447
+ "grad_norm": 2.140625,
448
+ "learning_rate": 4.9635817687444876e-05,
449
+ "loss": 0.3003,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.0650208269836432,
454
+ "grad_norm": 3.484375,
455
+ "learning_rate": 4.962198037472259e-05,
456
+ "loss": 0.2996,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.06603677740526262,
461
+ "grad_norm": 3.21875,
462
+ "learning_rate": 4.9607887063780776e-05,
463
+ "loss": 0.2257,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.06705272782688204,
468
+ "grad_norm": 5.375,
469
+ "learning_rate": 4.9593537901149564e-05,
470
+ "loss": 0.223,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.06806867824850148,
475
+ "grad_norm": 4.1875,
476
+ "learning_rate": 4.957893303601924e-05,
477
+ "loss": 0.3407,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.0690846286701209,
482
+ "grad_norm": 3.328125,
483
+ "learning_rate": 4.956407262023866e-05,
484
+ "loss": 0.2589,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.07010057909174032,
489
+ "grad_norm": 2.953125,
490
+ "learning_rate": 4.954895680831367e-05,
491
+ "loss": 0.2949,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.07111652951335974,
496
+ "grad_norm": 4.0625,
497
+ "learning_rate": 4.9533585757405506e-05,
498
+ "loss": 0.2995,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.07213247993497918,
503
+ "grad_norm": 4.625,
504
+ "learning_rate": 4.951795962732917e-05,
505
+ "loss": 0.2894,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.0731484303565986,
510
+ "grad_norm": 3.0,
511
+ "learning_rate": 4.9502078580551755e-05,
512
+ "loss": 0.3082,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.07416438077821802,
517
+ "grad_norm": 3.65625,
518
+ "learning_rate": 4.9485942782190734e-05,
519
+ "loss": 0.2308,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.07518033119983744,
524
+ "grad_norm": 4.78125,
525
+ "learning_rate": 4.9469552400012306e-05,
526
+ "loss": 0.2272,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.07619628162145688,
531
+ "grad_norm": 4.25,
532
+ "learning_rate": 4.94529076044296e-05,
533
+ "loss": 0.2701,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.0772122320430763,
538
+ "grad_norm": 3.140625,
539
+ "learning_rate": 4.94360085685009e-05,
540
+ "loss": 0.2686,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.07822818246469572,
545
+ "grad_norm": 0.765625,
546
+ "learning_rate": 4.9418855467927894e-05,
547
+ "loss": 0.2051,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.07924413288631514,
552
+ "grad_norm": 1.796875,
553
+ "learning_rate": 4.940144848105379e-05,
554
+ "loss": 0.2267,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.08026008330793458,
559
+ "grad_norm": 4.5625,
560
+ "learning_rate": 4.93837877888615e-05,
561
+ "loss": 0.2597,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.081276033729554,
566
+ "grad_norm": 3.03125,
567
+ "learning_rate": 4.9365873574971745e-05,
568
+ "loss": 0.3701,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.08229198415117342,
573
+ "grad_norm": 4.5625,
574
+ "learning_rate": 4.9347706025641136e-05,
575
+ "loss": 0.2559,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.08330793457279284,
580
+ "grad_norm": 3.90625,
581
+ "learning_rate": 4.9329285329760275e-05,
582
+ "loss": 0.2799,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.08432388499441228,
587
+ "grad_norm": 3.140625,
588
+ "learning_rate": 4.9310611678851735e-05,
589
+ "loss": 0.2866,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.0853398354160317,
594
+ "grad_norm": 2.46875,
595
+ "learning_rate": 4.929168526706811e-05,
596
+ "loss": 0.3105,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.08635578583765112,
601
+ "grad_norm": 13.625,
602
+ "learning_rate": 4.927250629119e-05,
603
+ "loss": 0.2454,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.08737173625927054,
608
+ "grad_norm": 3.921875,
609
+ "learning_rate": 4.9253074950623925e-05,
610
+ "loss": 0.2424,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.08838768668088998,
615
+ "grad_norm": 2.90625,
616
+ "learning_rate": 4.9233391447400286e-05,
617
+ "loss": 0.2481,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.0894036371025094,
622
+ "grad_norm": 2.96875,
623
+ "learning_rate": 4.921345598617125e-05,
624
+ "loss": 0.2231,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.09041958752412882,
629
+ "grad_norm": 5.375,
630
+ "learning_rate": 4.9193268774208654e-05,
631
+ "loss": 0.3447,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.09143553794574824,
636
+ "grad_norm": 2.0,
637
+ "learning_rate": 4.9172830021401785e-05,
638
+ "loss": 0.229,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.09245148836736768,
643
+ "grad_norm": 3.1875,
644
+ "learning_rate": 4.9152139940255245e-05,
645
+ "loss": 0.2122,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.0934674387889871,
650
+ "grad_norm": 3.40625,
651
+ "learning_rate": 4.913119874588677e-05,
652
+ "loss": 0.2386,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.09448338921060652,
657
+ "grad_norm": 1.4609375,
658
+ "learning_rate": 4.911000665602489e-05,
659
+ "loss": 0.1944,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.09549933963222594,
664
+ "grad_norm": 5.0625,
665
+ "learning_rate": 4.9088563891006786e-05,
666
+ "loss": 0.2038,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.09651529005384538,
671
+ "grad_norm": 4.53125,
672
+ "learning_rate": 4.906687067377592e-05,
673
+ "loss": 0.3122,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.0975312404754648,
678
+ "grad_norm": 2.84375,
679
+ "learning_rate": 4.904492722987976e-05,
680
+ "loss": 0.3157,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.09854719089708422,
685
+ "grad_norm": 2.171875,
686
+ "learning_rate": 4.902273378746738e-05,
687
+ "loss": 0.3077,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.09956314131870364,
692
+ "grad_norm": 2.84375,
693
+ "learning_rate": 4.9000290577287165e-05,
694
+ "loss": 0.2756,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.10057909174032308,
699
+ "grad_norm": 0.99609375,
700
+ "learning_rate": 4.897759783268434e-05,
701
+ "loss": 0.2915,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.1015950421619425,
706
+ "grad_norm": 3.53125,
707
+ "learning_rate": 4.895465578959859e-05,
708
+ "loss": 0.2052,
709
+ "step": 1000
710
+ }
711
+ ],
712
+ "logging_steps": 10,
713
+ "max_steps": 9843,
714
+ "num_input_tokens_seen": 0,
715
+ "num_train_epochs": 1,
716
+ "save_steps": 500,
717
+ "total_flos": 0.0,
718
+ "train_batch_size": 7,
719
+ "trial_name": null,
720
+ "trial_params": null
721
+ }
checkpoints/checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
3
+ size 4984
checkpoints/checkpoint-1500/.DS_Store ADDED
Binary file (6.15 kB). View file
 
checkpoints/checkpoint-1500/adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }
checkpoints/checkpoint-1500/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0429f41a7679282d393919c7b40c076ed2bde721f9f95c9c14f1f57fa0b63f6
3
+ size 5919456
checkpoints/checkpoint-1500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1cffc930b18a6507977b7928c70647d37429193d760799a24e4aa466c787fd3
3
+ size 11930938
checkpoints/checkpoint-1500/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af2583d0490e9c0c8a5832ae63a5d486d9951078b0ec9594ee0125a1807e7528
3
+ size 15024
checkpoints/checkpoint-1500/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af9fd6e5a8f29754daaa4f1a3c57f904108d748a165ca0e1ad16571d90e39fa3
3
+ size 15024
checkpoints/checkpoint-1500/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd55da1784f3e2dba7244fb29d3bbc59fbefd6b4bb1357a4ded5822c60485304
3
+ size 15024
checkpoints/checkpoint-1500/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b10bab8fcd02c56978a924d6d0be691a36e321523009f8caa015318d52823f2
3
+ size 15024
checkpoints/checkpoint-1500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68a6cef91a2e578166a3e870fa8585312c20b665553a57976a77b1e7d2ca0ef3
3
+ size 1064
checkpoints/checkpoint-1500/trainer_state.json ADDED
@@ -0,0 +1,1071 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.15239256324291375,
5
+ "eval_steps": 500,
6
+ "global_step": 1500,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001015950421619425,
13
+ "grad_norm": 17.625,
14
+ "learning_rate": 5e-06,
15
+ "loss": 3.4264,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.00203190084323885,
20
+ "grad_norm": 12.5625,
21
+ "learning_rate": 1e-05,
22
+ "loss": 3.432,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.003047851264858275,
27
+ "grad_norm": 14.0625,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 3.23,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0040638016864777,
34
+ "grad_norm": 12.4375,
35
+ "learning_rate": 2e-05,
36
+ "loss": 2.9762,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.005079752108097125,
41
+ "grad_norm": 10.0625,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 2.6173,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.00609570252971655,
48
+ "grad_norm": 10.1875,
49
+ "learning_rate": 3e-05,
50
+ "loss": 2.2004,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.007111652951335975,
55
+ "grad_norm": 7.03125,
56
+ "learning_rate": 3.5e-05,
57
+ "loss": 1.4176,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0081276033729554,
62
+ "grad_norm": 4.375,
63
+ "learning_rate": 4e-05,
64
+ "loss": 1.0122,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.009143553794574825,
69
+ "grad_norm": 6.5625,
70
+ "learning_rate": 4.5e-05,
71
+ "loss": 0.9116,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01015950421619425,
76
+ "grad_norm": 5.28125,
77
+ "learning_rate": 5e-05,
78
+ "loss": 0.6832,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.011175454637813675,
83
+ "grad_norm": 5.5,
84
+ "learning_rate": 4.9999870035728426e-05,
85
+ "loss": 0.7355,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0121914050594331,
90
+ "grad_norm": 5.1875,
91
+ "learning_rate": 4.9999480144264944e-05,
92
+ "loss": 0.6673,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.013207355481052525,
97
+ "grad_norm": 4.5,
98
+ "learning_rate": 4.9998830329663314e-05,
99
+ "loss": 0.6792,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.01422330590267195,
104
+ "grad_norm": 3.9375,
105
+ "learning_rate": 4.9997920598679756e-05,
106
+ "loss": 0.6207,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.015239256324291375,
111
+ "grad_norm": 3.15625,
112
+ "learning_rate": 4.999675096077286e-05,
113
+ "loss": 0.483,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.0162552067459108,
118
+ "grad_norm": 5.28125,
119
+ "learning_rate": 4.999532142810354e-05,
120
+ "loss": 0.5319,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.017271157167530225,
125
+ "grad_norm": 4.59375,
126
+ "learning_rate": 4.999363201553483e-05,
127
+ "loss": 0.6052,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.01828710758914965,
132
+ "grad_norm": 5.03125,
133
+ "learning_rate": 4.9991682740631794e-05,
134
+ "loss": 0.4258,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.019303058010769075,
139
+ "grad_norm": 3.859375,
140
+ "learning_rate": 4.998947362366133e-05,
141
+ "loss": 0.4309,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.0203190084323885,
146
+ "grad_norm": 3.328125,
147
+ "learning_rate": 4.998700468759193e-05,
148
+ "loss": 0.3957,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.021334958854007924,
153
+ "grad_norm": 4.9375,
154
+ "learning_rate": 4.9984275958093475e-05,
155
+ "loss": 0.4777,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02235090927562735,
160
+ "grad_norm": 4.78125,
161
+ "learning_rate": 4.998128746353695e-05,
162
+ "loss": 0.3549,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.023366859697246774,
167
+ "grad_norm": 4.0625,
168
+ "learning_rate": 4.997803923499417e-05,
169
+ "loss": 0.4447,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.0243828101188662,
174
+ "grad_norm": 6.375,
175
+ "learning_rate": 4.99745313062374e-05,
176
+ "loss": 0.3808,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.025398760540485624,
181
+ "grad_norm": 3.59375,
182
+ "learning_rate": 4.99707637137391e-05,
183
+ "loss": 0.3827,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.02641471096210505,
188
+ "grad_norm": 3.015625,
189
+ "learning_rate": 4.996673649667145e-05,
190
+ "loss": 0.3694,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.027430661383724474,
195
+ "grad_norm": 2.296875,
196
+ "learning_rate": 4.9962449696906e-05,
197
+ "loss": 0.3586,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.0284466118053439,
202
+ "grad_norm": 4.125,
203
+ "learning_rate": 4.9957903359013214e-05,
204
+ "loss": 0.3832,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.029462562226963324,
209
+ "grad_norm": 3.296875,
210
+ "learning_rate": 4.995309753026201e-05,
211
+ "loss": 0.328,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.03047851264858275,
216
+ "grad_norm": 4.5,
217
+ "learning_rate": 4.994803226061927e-05,
218
+ "loss": 0.3667,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.03149446307020217,
223
+ "grad_norm": 4.3125,
224
+ "learning_rate": 4.994270760274933e-05,
225
+ "loss": 0.3811,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.0325104134918216,
230
+ "grad_norm": 3.421875,
231
+ "learning_rate": 4.99371236120134e-05,
232
+ "loss": 0.3065,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.03352636391344102,
237
+ "grad_norm": 4.6875,
238
+ "learning_rate": 4.993128034646902e-05,
239
+ "loss": 0.4177,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.03454231433506045,
244
+ "grad_norm": 3.046875,
245
+ "learning_rate": 4.992517786686947e-05,
246
+ "loss": 0.33,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.03555826475667987,
251
+ "grad_norm": 1.8828125,
252
+ "learning_rate": 4.9918816236663077e-05,
253
+ "loss": 0.3287,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.0365742151782993,
258
+ "grad_norm": 3.8125,
259
+ "learning_rate": 4.991219552199262e-05,
260
+ "loss": 0.2934,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.03759016559991872,
265
+ "grad_norm": 4.28125,
266
+ "learning_rate": 4.99053157916946e-05,
267
+ "loss": 0.3176,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.03860611602153815,
272
+ "grad_norm": 2.609375,
273
+ "learning_rate": 4.989817711729856e-05,
274
+ "loss": 0.3318,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.03962206644315757,
279
+ "grad_norm": 2.375,
280
+ "learning_rate": 4.98907795730263e-05,
281
+ "loss": 0.3234,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.040638016864777,
286
+ "grad_norm": 4.46875,
287
+ "learning_rate": 4.988312323579114e-05,
288
+ "loss": 0.267,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.04165396728639642,
293
+ "grad_norm": 3.75,
294
+ "learning_rate": 4.98752081851971e-05,
295
+ "loss": 0.3081,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.04266991770801585,
300
+ "grad_norm": 2.203125,
301
+ "learning_rate": 4.986703450353809e-05,
302
+ "loss": 0.2917,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.04368586812963527,
307
+ "grad_norm": 1.6015625,
308
+ "learning_rate": 4.985860227579703e-05,
309
+ "loss": 0.2805,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0447018185512547,
314
+ "grad_norm": 3.140625,
315
+ "learning_rate": 4.984991158964499e-05,
316
+ "loss": 0.3534,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.04571776897287412,
321
+ "grad_norm": 3.296875,
322
+ "learning_rate": 4.9840962535440265e-05,
323
+ "loss": 0.335,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.04673371939449355,
328
+ "grad_norm": 3.25,
329
+ "learning_rate": 4.983175520622744e-05,
330
+ "loss": 0.2544,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.04774966981611297,
335
+ "grad_norm": 2.25,
336
+ "learning_rate": 4.982228969773642e-05,
337
+ "loss": 0.3449,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.0487656202377324,
342
+ "grad_norm": 4.9375,
343
+ "learning_rate": 4.9812566108381435e-05,
344
+ "loss": 0.2964,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.04978157065935182,
349
+ "grad_norm": 1.5703125,
350
+ "learning_rate": 4.9802584539260035e-05,
351
+ "loss": 0.2799,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.05079752108097125,
356
+ "grad_norm": 2.828125,
357
+ "learning_rate": 4.979234509415199e-05,
358
+ "loss": 0.3231,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.05181347150259067,
363
+ "grad_norm": 2.9375,
364
+ "learning_rate": 4.978184787951828e-05,
365
+ "loss": 0.2943,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.0528294219242101,
370
+ "grad_norm": 2.34375,
371
+ "learning_rate": 4.977109300449992e-05,
372
+ "loss": 0.2705,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.05384537234582952,
377
+ "grad_norm": 3.140625,
378
+ "learning_rate": 4.9760080580916876e-05,
379
+ "loss": 0.2998,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.05486132276744895,
384
+ "grad_norm": 3.5625,
385
+ "learning_rate": 4.974881072326688e-05,
386
+ "loss": 0.2595,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.05587727318906837,
391
+ "grad_norm": 4.25,
392
+ "learning_rate": 4.9737283548724236e-05,
393
+ "loss": 0.2803,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.0568932236106878,
398
+ "grad_norm": 4.0625,
399
+ "learning_rate": 4.97254991771386e-05,
400
+ "loss": 0.3511,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.05790917403230722,
405
+ "grad_norm": 2.515625,
406
+ "learning_rate": 4.971345773103377e-05,
407
+ "loss": 0.312,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.05892512445392665,
412
+ "grad_norm": 3.21875,
413
+ "learning_rate": 4.9701159335606365e-05,
414
+ "loss": 0.2482,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.05994107487554607,
419
+ "grad_norm": 5.5,
420
+ "learning_rate": 4.968860411872454e-05,
421
+ "loss": 0.2537,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.0609570252971655,
426
+ "grad_norm": 3.546875,
427
+ "learning_rate": 4.967579221092666e-05,
428
+ "loss": 0.3125,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.06197297571878492,
433
+ "grad_norm": 2.984375,
434
+ "learning_rate": 4.966272374541996e-05,
435
+ "loss": 0.2354,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.06298892614040434,
440
+ "grad_norm": 3.6875,
441
+ "learning_rate": 4.964939885807912e-05,
442
+ "loss": 0.3213,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.06400487656202378,
447
+ "grad_norm": 2.140625,
448
+ "learning_rate": 4.9635817687444876e-05,
449
+ "loss": 0.3003,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.0650208269836432,
454
+ "grad_norm": 3.484375,
455
+ "learning_rate": 4.962198037472259e-05,
456
+ "loss": 0.2996,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.06603677740526262,
461
+ "grad_norm": 3.21875,
462
+ "learning_rate": 4.9607887063780776e-05,
463
+ "loss": 0.2257,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.06705272782688204,
468
+ "grad_norm": 5.375,
469
+ "learning_rate": 4.9593537901149564e-05,
470
+ "loss": 0.223,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.06806867824850148,
475
+ "grad_norm": 4.1875,
476
+ "learning_rate": 4.957893303601924e-05,
477
+ "loss": 0.3407,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.0690846286701209,
482
+ "grad_norm": 3.328125,
483
+ "learning_rate": 4.956407262023866e-05,
484
+ "loss": 0.2589,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.07010057909174032,
489
+ "grad_norm": 2.953125,
490
+ "learning_rate": 4.954895680831367e-05,
491
+ "loss": 0.2949,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.07111652951335974,
496
+ "grad_norm": 4.0625,
497
+ "learning_rate": 4.9533585757405506e-05,
498
+ "loss": 0.2995,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.07213247993497918,
503
+ "grad_norm": 4.625,
504
+ "learning_rate": 4.951795962732917e-05,
505
+ "loss": 0.2894,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.0731484303565986,
510
+ "grad_norm": 3.0,
511
+ "learning_rate": 4.9502078580551755e-05,
512
+ "loss": 0.3082,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.07416438077821802,
517
+ "grad_norm": 3.65625,
518
+ "learning_rate": 4.9485942782190734e-05,
519
+ "loss": 0.2308,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.07518033119983744,
524
+ "grad_norm": 4.78125,
525
+ "learning_rate": 4.9469552400012306e-05,
526
+ "loss": 0.2272,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.07619628162145688,
531
+ "grad_norm": 4.25,
532
+ "learning_rate": 4.94529076044296e-05,
533
+ "loss": 0.2701,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.0772122320430763,
538
+ "grad_norm": 3.140625,
539
+ "learning_rate": 4.94360085685009e-05,
540
+ "loss": 0.2686,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.07822818246469572,
545
+ "grad_norm": 0.765625,
546
+ "learning_rate": 4.9418855467927894e-05,
547
+ "loss": 0.2051,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.07924413288631514,
552
+ "grad_norm": 1.796875,
553
+ "learning_rate": 4.940144848105379e-05,
554
+ "loss": 0.2267,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.08026008330793458,
559
+ "grad_norm": 4.5625,
560
+ "learning_rate": 4.93837877888615e-05,
561
+ "loss": 0.2597,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.081276033729554,
566
+ "grad_norm": 3.03125,
567
+ "learning_rate": 4.9365873574971745e-05,
568
+ "loss": 0.3701,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.08229198415117342,
573
+ "grad_norm": 4.5625,
574
+ "learning_rate": 4.9347706025641136e-05,
575
+ "loss": 0.2559,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.08330793457279284,
580
+ "grad_norm": 3.90625,
581
+ "learning_rate": 4.9329285329760275e-05,
582
+ "loss": 0.2799,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.08432388499441228,
587
+ "grad_norm": 3.140625,
588
+ "learning_rate": 4.9310611678851735e-05,
589
+ "loss": 0.2866,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.0853398354160317,
594
+ "grad_norm": 2.46875,
595
+ "learning_rate": 4.929168526706811e-05,
596
+ "loss": 0.3105,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.08635578583765112,
601
+ "grad_norm": 13.625,
602
+ "learning_rate": 4.927250629119e-05,
603
+ "loss": 0.2454,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.08737173625927054,
608
+ "grad_norm": 3.921875,
609
+ "learning_rate": 4.9253074950623925e-05,
610
+ "loss": 0.2424,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.08838768668088998,
615
+ "grad_norm": 2.90625,
616
+ "learning_rate": 4.9233391447400286e-05,
617
+ "loss": 0.2481,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.0894036371025094,
622
+ "grad_norm": 2.96875,
623
+ "learning_rate": 4.921345598617125e-05,
624
+ "loss": 0.2231,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.09041958752412882,
629
+ "grad_norm": 5.375,
630
+ "learning_rate": 4.9193268774208654e-05,
631
+ "loss": 0.3447,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.09143553794574824,
636
+ "grad_norm": 2.0,
637
+ "learning_rate": 4.9172830021401785e-05,
638
+ "loss": 0.229,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.09245148836736768,
643
+ "grad_norm": 3.1875,
644
+ "learning_rate": 4.9152139940255245e-05,
645
+ "loss": 0.2122,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.0934674387889871,
650
+ "grad_norm": 3.40625,
651
+ "learning_rate": 4.913119874588677e-05,
652
+ "loss": 0.2386,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.09448338921060652,
657
+ "grad_norm": 1.4609375,
658
+ "learning_rate": 4.911000665602489e-05,
659
+ "loss": 0.1944,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.09549933963222594,
664
+ "grad_norm": 5.0625,
665
+ "learning_rate": 4.9088563891006786e-05,
666
+ "loss": 0.2038,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.09651529005384538,
671
+ "grad_norm": 4.53125,
672
+ "learning_rate": 4.906687067377592e-05,
673
+ "loss": 0.3122,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.0975312404754648,
678
+ "grad_norm": 2.84375,
679
+ "learning_rate": 4.904492722987976e-05,
680
+ "loss": 0.3157,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.09854719089708422,
685
+ "grad_norm": 2.171875,
686
+ "learning_rate": 4.902273378746738e-05,
687
+ "loss": 0.3077,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.09956314131870364,
692
+ "grad_norm": 2.84375,
693
+ "learning_rate": 4.9000290577287165e-05,
694
+ "loss": 0.2756,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.10057909174032308,
699
+ "grad_norm": 0.99609375,
700
+ "learning_rate": 4.897759783268434e-05,
701
+ "loss": 0.2915,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.1015950421619425,
706
+ "grad_norm": 3.53125,
707
+ "learning_rate": 4.895465578959859e-05,
708
+ "loss": 0.2052,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.10261099258356192,
713
+ "grad_norm": 4.0,
714
+ "learning_rate": 4.893146468656159e-05,
715
+ "loss": 0.2499,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.10362694300518134,
720
+ "grad_norm": 1.65625,
721
+ "learning_rate": 4.890802476469452e-05,
722
+ "loss": 0.278,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.10464289342680078,
727
+ "grad_norm": 3.625,
728
+ "learning_rate": 4.888433626770558e-05,
729
+ "loss": 0.2143,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.1056588438484202,
734
+ "grad_norm": 5.0625,
735
+ "learning_rate": 4.886039944188741e-05,
736
+ "loss": 0.2878,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.10667479427003962,
741
+ "grad_norm": 4.5,
742
+ "learning_rate": 4.883621453611461e-05,
743
+ "loss": 0.2744,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.10769074469165904,
748
+ "grad_norm": 4.5625,
749
+ "learning_rate": 4.881178180184106e-05,
750
+ "loss": 0.2734,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.10870669511327848,
755
+ "grad_norm": 3.125,
756
+ "learning_rate": 4.878710149309735e-05,
757
+ "loss": 0.3574,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.1097226455348979,
762
+ "grad_norm": 3.0625,
763
+ "learning_rate": 4.876217386648816e-05,
764
+ "loss": 0.2625,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.11073859595651732,
769
+ "grad_norm": 4.0625,
770
+ "learning_rate": 4.873699918118955e-05,
771
+ "loss": 0.2437,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.11175454637813674,
776
+ "grad_norm": 1.59375,
777
+ "learning_rate": 4.87115776989463e-05,
778
+ "loss": 0.2051,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.11277049679975618,
783
+ "grad_norm": 4.375,
784
+ "learning_rate": 4.8685909684069153e-05,
785
+ "loss": 0.1727,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.1137864472213756,
790
+ "grad_norm": 2.28125,
791
+ "learning_rate": 4.865999540343211e-05,
792
+ "loss": 0.2256,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.11480239764299502,
797
+ "grad_norm": 2.265625,
798
+ "learning_rate": 4.86338351264696e-05,
799
+ "loss": 0.3529,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.11581834806461444,
804
+ "grad_norm": 2.34375,
805
+ "learning_rate": 4.8607429125173754e-05,
806
+ "loss": 0.2113,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.11683429848623388,
811
+ "grad_norm": 0.7578125,
812
+ "learning_rate": 4.858077767409149e-05,
813
+ "loss": 0.2759,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.1178502489078533,
818
+ "grad_norm": 3.640625,
819
+ "learning_rate": 4.855388105032174e-05,
820
+ "loss": 0.2482,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.11886619932947272,
825
+ "grad_norm": 3.5,
826
+ "learning_rate": 4.852673953351249e-05,
827
+ "loss": 0.1865,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.11988214975109214,
832
+ "grad_norm": 3.75,
833
+ "learning_rate": 4.849935340585796e-05,
834
+ "loss": 0.2659,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.12089810017271158,
839
+ "grad_norm": 3.375,
840
+ "learning_rate": 4.8471722952095586e-05,
841
+ "loss": 0.1506,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.121914050594331,
846
+ "grad_norm": 3.34375,
847
+ "learning_rate": 4.844384845950312e-05,
848
+ "loss": 0.307,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.12293000101595042,
853
+ "grad_norm": 1.578125,
854
+ "learning_rate": 4.841573021789561e-05,
855
+ "loss": 0.1952,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.12394595143756984,
860
+ "grad_norm": 1.2890625,
861
+ "learning_rate": 4.838736851962239e-05,
862
+ "loss": 0.1779,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.12496190185918928,
867
+ "grad_norm": 1.265625,
868
+ "learning_rate": 4.835876365956408e-05,
869
+ "loss": 0.1235,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.12597785228080868,
874
+ "grad_norm": 1.9609375,
875
+ "learning_rate": 4.8329915935129436e-05,
876
+ "loss": 0.1876,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.12699380270242813,
881
+ "grad_norm": 1.6328125,
882
+ "learning_rate": 4.830082564625235e-05,
883
+ "loss": 0.2188,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.12800975312404755,
888
+ "grad_norm": 3.96875,
889
+ "learning_rate": 4.8271493095388684e-05,
890
+ "loss": 0.2622,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.12902570354566698,
895
+ "grad_norm": 3.765625,
896
+ "learning_rate": 4.824191858751312e-05,
897
+ "loss": 0.2724,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.1300416539672864,
902
+ "grad_norm": 5.59375,
903
+ "learning_rate": 4.821210243011601e-05,
904
+ "loss": 0.2413,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.13105760438890582,
909
+ "grad_norm": 3.34375,
910
+ "learning_rate": 4.818204493320016e-05,
911
+ "loss": 0.2618,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.13207355481052524,
916
+ "grad_norm": 2.78125,
917
+ "learning_rate": 4.8151746409277634e-05,
918
+ "loss": 0.2295,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.13308950523214466,
923
+ "grad_norm": 3.1875,
924
+ "learning_rate": 4.8121207173366484e-05,
925
+ "loss": 0.2733,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.13410545565376408,
930
+ "grad_norm": 2.28125,
931
+ "learning_rate": 4.809042754298746e-05,
932
+ "loss": 0.2311,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.13512140607538353,
937
+ "grad_norm": 2.171875,
938
+ "learning_rate": 4.805940783816075e-05,
939
+ "loss": 0.2059,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.13613735649700295,
944
+ "grad_norm": 2.796875,
945
+ "learning_rate": 4.8028148381402625e-05,
946
+ "loss": 0.2102,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.13715330691862238,
951
+ "grad_norm": 2.96875,
952
+ "learning_rate": 4.7996649497722084e-05,
953
+ "loss": 0.2708,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.1381692573402418,
958
+ "grad_norm": 2.4375,
959
+ "learning_rate": 4.7964911514617485e-05,
960
+ "loss": 0.2429,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.13918520776186122,
965
+ "grad_norm": 5.8125,
966
+ "learning_rate": 4.793293476207312e-05,
967
+ "loss": 0.2725,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.14020115818348064,
972
+ "grad_norm": 2.40625,
973
+ "learning_rate": 4.790071957255585e-05,
974
+ "loss": 0.2098,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.14121710860510006,
979
+ "grad_norm": 4.25,
980
+ "learning_rate": 4.786826628101154e-05,
981
+ "loss": 0.2101,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.14223305902671948,
986
+ "grad_norm": 2.578125,
987
+ "learning_rate": 4.783557522486167e-05,
988
+ "loss": 0.2624,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.14324900944833893,
993
+ "grad_norm": 3.125,
994
+ "learning_rate": 4.780264674399978e-05,
995
+ "loss": 0.2518,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.14426495986995835,
1000
+ "grad_norm": 3.671875,
1001
+ "learning_rate": 4.7769481180787966e-05,
1002
+ "loss": 0.3112,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.14528091029157778,
1007
+ "grad_norm": 3.984375,
1008
+ "learning_rate": 4.773607888005327e-05,
1009
+ "loss": 0.2747,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1462968607131972,
1014
+ "grad_norm": 3.234375,
1015
+ "learning_rate": 4.770244018908416e-05,
1016
+ "loss": 0.1572,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.14731281113481662,
1021
+ "grad_norm": 4.09375,
1022
+ "learning_rate": 4.766856545762687e-05,
1023
+ "loss": 0.2148,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.14832876155643604,
1028
+ "grad_norm": 1.6875,
1029
+ "learning_rate": 4.763445503788178e-05,
1030
+ "loss": 0.2531,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.14934471197805546,
1035
+ "grad_norm": 2.375,
1036
+ "learning_rate": 4.760010928449976e-05,
1037
+ "loss": 0.199,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.15036066239967488,
1042
+ "grad_norm": 4.6875,
1043
+ "learning_rate": 4.7565528554578485e-05,
1044
+ "loss": 0.2366,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.15137661282129433,
1049
+ "grad_norm": 5.4375,
1050
+ "learning_rate": 4.75307132076587e-05,
1051
+ "loss": 0.1862,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.15239256324291375,
1056
+ "grad_norm": 2.484375,
1057
+ "learning_rate": 4.749566360572049e-05,
1058
+ "loss": 0.2143,
1059
+ "step": 1500
1060
+ }
1061
+ ],
1062
+ "logging_steps": 10,
1063
+ "max_steps": 9843,
1064
+ "num_input_tokens_seen": 0,
1065
+ "num_train_epochs": 1,
1066
+ "save_steps": 500,
1067
+ "total_flos": 0.0,
1068
+ "train_batch_size": 7,
1069
+ "trial_name": null,
1070
+ "trial_params": null
1071
+ }
checkpoints/checkpoint-1500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
3
+ size 4984
checkpoints/checkpoint-2000/.DS_Store ADDED
Binary file (6.15 kB). View file
 
checkpoints/checkpoint-2000/adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }
checkpoints/checkpoint-2000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f511e364452e06b28c03299a544dc2d5fd730fdfd45007cd4ec530e30144139
3
+ size 5919456
checkpoints/checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4121e055d593f91209d1cfca077444bb962a8848054f778e04d22d7e269a91e3
3
+ size 11930938
checkpoints/checkpoint-2000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7814979e8476866bd135a55e64d375b840c3c1436aa60c3d69ece3f0a10c3408
3
+ size 15024
checkpoints/checkpoint-2000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4b8a9f5b33e3bc1f4b31217176bbabc65ace6c56a7bf77b1b7153dc062ba709
3
+ size 15024
checkpoints/checkpoint-2000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70445c98497e5f1d7bea8f44a93bc0211d0226177b834ac2949fe24e3b538d05
3
+ size 15024
checkpoints/checkpoint-2000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c244dce3fafef8abbf503718fb81319ff12831928a8845134b78a845e0c6e14
3
+ size 15024
checkpoints/checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9cf822c73733a63e3bebbb2802abc0acedf3d9423d9f25c6bb103f490864c06
3
+ size 1064
checkpoints/checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,1421 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.203190084323885,
5
+ "eval_steps": 500,
6
+ "global_step": 2000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001015950421619425,
13
+ "grad_norm": 17.625,
14
+ "learning_rate": 5e-06,
15
+ "loss": 3.4264,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.00203190084323885,
20
+ "grad_norm": 12.5625,
21
+ "learning_rate": 1e-05,
22
+ "loss": 3.432,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.003047851264858275,
27
+ "grad_norm": 14.0625,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 3.23,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0040638016864777,
34
+ "grad_norm": 12.4375,
35
+ "learning_rate": 2e-05,
36
+ "loss": 2.9762,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.005079752108097125,
41
+ "grad_norm": 10.0625,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 2.6173,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.00609570252971655,
48
+ "grad_norm": 10.1875,
49
+ "learning_rate": 3e-05,
50
+ "loss": 2.2004,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.007111652951335975,
55
+ "grad_norm": 7.03125,
56
+ "learning_rate": 3.5e-05,
57
+ "loss": 1.4176,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0081276033729554,
62
+ "grad_norm": 4.375,
63
+ "learning_rate": 4e-05,
64
+ "loss": 1.0122,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.009143553794574825,
69
+ "grad_norm": 6.5625,
70
+ "learning_rate": 4.5e-05,
71
+ "loss": 0.9116,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01015950421619425,
76
+ "grad_norm": 5.28125,
77
+ "learning_rate": 5e-05,
78
+ "loss": 0.6832,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.011175454637813675,
83
+ "grad_norm": 5.5,
84
+ "learning_rate": 4.9999870035728426e-05,
85
+ "loss": 0.7355,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0121914050594331,
90
+ "grad_norm": 5.1875,
91
+ "learning_rate": 4.9999480144264944e-05,
92
+ "loss": 0.6673,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.013207355481052525,
97
+ "grad_norm": 4.5,
98
+ "learning_rate": 4.9998830329663314e-05,
99
+ "loss": 0.6792,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.01422330590267195,
104
+ "grad_norm": 3.9375,
105
+ "learning_rate": 4.9997920598679756e-05,
106
+ "loss": 0.6207,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.015239256324291375,
111
+ "grad_norm": 3.15625,
112
+ "learning_rate": 4.999675096077286e-05,
113
+ "loss": 0.483,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.0162552067459108,
118
+ "grad_norm": 5.28125,
119
+ "learning_rate": 4.999532142810354e-05,
120
+ "loss": 0.5319,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.017271157167530225,
125
+ "grad_norm": 4.59375,
126
+ "learning_rate": 4.999363201553483e-05,
127
+ "loss": 0.6052,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.01828710758914965,
132
+ "grad_norm": 5.03125,
133
+ "learning_rate": 4.9991682740631794e-05,
134
+ "loss": 0.4258,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.019303058010769075,
139
+ "grad_norm": 3.859375,
140
+ "learning_rate": 4.998947362366133e-05,
141
+ "loss": 0.4309,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.0203190084323885,
146
+ "grad_norm": 3.328125,
147
+ "learning_rate": 4.998700468759193e-05,
148
+ "loss": 0.3957,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.021334958854007924,
153
+ "grad_norm": 4.9375,
154
+ "learning_rate": 4.9984275958093475e-05,
155
+ "loss": 0.4777,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02235090927562735,
160
+ "grad_norm": 4.78125,
161
+ "learning_rate": 4.998128746353695e-05,
162
+ "loss": 0.3549,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.023366859697246774,
167
+ "grad_norm": 4.0625,
168
+ "learning_rate": 4.997803923499417e-05,
169
+ "loss": 0.4447,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.0243828101188662,
174
+ "grad_norm": 6.375,
175
+ "learning_rate": 4.99745313062374e-05,
176
+ "loss": 0.3808,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.025398760540485624,
181
+ "grad_norm": 3.59375,
182
+ "learning_rate": 4.99707637137391e-05,
183
+ "loss": 0.3827,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.02641471096210505,
188
+ "grad_norm": 3.015625,
189
+ "learning_rate": 4.996673649667145e-05,
190
+ "loss": 0.3694,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.027430661383724474,
195
+ "grad_norm": 2.296875,
196
+ "learning_rate": 4.9962449696906e-05,
197
+ "loss": 0.3586,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.0284466118053439,
202
+ "grad_norm": 4.125,
203
+ "learning_rate": 4.9957903359013214e-05,
204
+ "loss": 0.3832,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.029462562226963324,
209
+ "grad_norm": 3.296875,
210
+ "learning_rate": 4.995309753026201e-05,
211
+ "loss": 0.328,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.03047851264858275,
216
+ "grad_norm": 4.5,
217
+ "learning_rate": 4.994803226061927e-05,
218
+ "loss": 0.3667,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.03149446307020217,
223
+ "grad_norm": 4.3125,
224
+ "learning_rate": 4.994270760274933e-05,
225
+ "loss": 0.3811,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.0325104134918216,
230
+ "grad_norm": 3.421875,
231
+ "learning_rate": 4.99371236120134e-05,
232
+ "loss": 0.3065,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.03352636391344102,
237
+ "grad_norm": 4.6875,
238
+ "learning_rate": 4.993128034646902e-05,
239
+ "loss": 0.4177,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.03454231433506045,
244
+ "grad_norm": 3.046875,
245
+ "learning_rate": 4.992517786686947e-05,
246
+ "loss": 0.33,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.03555826475667987,
251
+ "grad_norm": 1.8828125,
252
+ "learning_rate": 4.9918816236663077e-05,
253
+ "loss": 0.3287,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.0365742151782993,
258
+ "grad_norm": 3.8125,
259
+ "learning_rate": 4.991219552199262e-05,
260
+ "loss": 0.2934,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.03759016559991872,
265
+ "grad_norm": 4.28125,
266
+ "learning_rate": 4.99053157916946e-05,
267
+ "loss": 0.3176,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.03860611602153815,
272
+ "grad_norm": 2.609375,
273
+ "learning_rate": 4.989817711729856e-05,
274
+ "loss": 0.3318,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.03962206644315757,
279
+ "grad_norm": 2.375,
280
+ "learning_rate": 4.98907795730263e-05,
281
+ "loss": 0.3234,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.040638016864777,
286
+ "grad_norm": 4.46875,
287
+ "learning_rate": 4.988312323579114e-05,
288
+ "loss": 0.267,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.04165396728639642,
293
+ "grad_norm": 3.75,
294
+ "learning_rate": 4.98752081851971e-05,
295
+ "loss": 0.3081,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.04266991770801585,
300
+ "grad_norm": 2.203125,
301
+ "learning_rate": 4.986703450353809e-05,
302
+ "loss": 0.2917,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.04368586812963527,
307
+ "grad_norm": 1.6015625,
308
+ "learning_rate": 4.985860227579703e-05,
309
+ "loss": 0.2805,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0447018185512547,
314
+ "grad_norm": 3.140625,
315
+ "learning_rate": 4.984991158964499e-05,
316
+ "loss": 0.3534,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.04571776897287412,
321
+ "grad_norm": 3.296875,
322
+ "learning_rate": 4.9840962535440265e-05,
323
+ "loss": 0.335,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.04673371939449355,
328
+ "grad_norm": 3.25,
329
+ "learning_rate": 4.983175520622744e-05,
330
+ "loss": 0.2544,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.04774966981611297,
335
+ "grad_norm": 2.25,
336
+ "learning_rate": 4.982228969773642e-05,
337
+ "loss": 0.3449,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.0487656202377324,
342
+ "grad_norm": 4.9375,
343
+ "learning_rate": 4.9812566108381435e-05,
344
+ "loss": 0.2964,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.04978157065935182,
349
+ "grad_norm": 1.5703125,
350
+ "learning_rate": 4.9802584539260035e-05,
351
+ "loss": 0.2799,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.05079752108097125,
356
+ "grad_norm": 2.828125,
357
+ "learning_rate": 4.979234509415199e-05,
358
+ "loss": 0.3231,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.05181347150259067,
363
+ "grad_norm": 2.9375,
364
+ "learning_rate": 4.978184787951828e-05,
365
+ "loss": 0.2943,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.0528294219242101,
370
+ "grad_norm": 2.34375,
371
+ "learning_rate": 4.977109300449992e-05,
372
+ "loss": 0.2705,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.05384537234582952,
377
+ "grad_norm": 3.140625,
378
+ "learning_rate": 4.9760080580916876e-05,
379
+ "loss": 0.2998,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.05486132276744895,
384
+ "grad_norm": 3.5625,
385
+ "learning_rate": 4.974881072326688e-05,
386
+ "loss": 0.2595,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.05587727318906837,
391
+ "grad_norm": 4.25,
392
+ "learning_rate": 4.9737283548724236e-05,
393
+ "loss": 0.2803,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.0568932236106878,
398
+ "grad_norm": 4.0625,
399
+ "learning_rate": 4.97254991771386e-05,
400
+ "loss": 0.3511,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.05790917403230722,
405
+ "grad_norm": 2.515625,
406
+ "learning_rate": 4.971345773103377e-05,
407
+ "loss": 0.312,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.05892512445392665,
412
+ "grad_norm": 3.21875,
413
+ "learning_rate": 4.9701159335606365e-05,
414
+ "loss": 0.2482,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.05994107487554607,
419
+ "grad_norm": 5.5,
420
+ "learning_rate": 4.968860411872454e-05,
421
+ "loss": 0.2537,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.0609570252971655,
426
+ "grad_norm": 3.546875,
427
+ "learning_rate": 4.967579221092666e-05,
428
+ "loss": 0.3125,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.06197297571878492,
433
+ "grad_norm": 2.984375,
434
+ "learning_rate": 4.966272374541996e-05,
435
+ "loss": 0.2354,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.06298892614040434,
440
+ "grad_norm": 3.6875,
441
+ "learning_rate": 4.964939885807912e-05,
442
+ "loss": 0.3213,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.06400487656202378,
447
+ "grad_norm": 2.140625,
448
+ "learning_rate": 4.9635817687444876e-05,
449
+ "loss": 0.3003,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.0650208269836432,
454
+ "grad_norm": 3.484375,
455
+ "learning_rate": 4.962198037472259e-05,
456
+ "loss": 0.2996,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.06603677740526262,
461
+ "grad_norm": 3.21875,
462
+ "learning_rate": 4.9607887063780776e-05,
463
+ "loss": 0.2257,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.06705272782688204,
468
+ "grad_norm": 5.375,
469
+ "learning_rate": 4.9593537901149564e-05,
470
+ "loss": 0.223,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.06806867824850148,
475
+ "grad_norm": 4.1875,
476
+ "learning_rate": 4.957893303601924e-05,
477
+ "loss": 0.3407,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.0690846286701209,
482
+ "grad_norm": 3.328125,
483
+ "learning_rate": 4.956407262023866e-05,
484
+ "loss": 0.2589,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.07010057909174032,
489
+ "grad_norm": 2.953125,
490
+ "learning_rate": 4.954895680831367e-05,
491
+ "loss": 0.2949,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.07111652951335974,
496
+ "grad_norm": 4.0625,
497
+ "learning_rate": 4.9533585757405506e-05,
498
+ "loss": 0.2995,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.07213247993497918,
503
+ "grad_norm": 4.625,
504
+ "learning_rate": 4.951795962732917e-05,
505
+ "loss": 0.2894,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.0731484303565986,
510
+ "grad_norm": 3.0,
511
+ "learning_rate": 4.9502078580551755e-05,
512
+ "loss": 0.3082,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.07416438077821802,
517
+ "grad_norm": 3.65625,
518
+ "learning_rate": 4.9485942782190734e-05,
519
+ "loss": 0.2308,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.07518033119983744,
524
+ "grad_norm": 4.78125,
525
+ "learning_rate": 4.9469552400012306e-05,
526
+ "loss": 0.2272,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.07619628162145688,
531
+ "grad_norm": 4.25,
532
+ "learning_rate": 4.94529076044296e-05,
533
+ "loss": 0.2701,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.0772122320430763,
538
+ "grad_norm": 3.140625,
539
+ "learning_rate": 4.94360085685009e-05,
540
+ "loss": 0.2686,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.07822818246469572,
545
+ "grad_norm": 0.765625,
546
+ "learning_rate": 4.9418855467927894e-05,
547
+ "loss": 0.2051,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.07924413288631514,
552
+ "grad_norm": 1.796875,
553
+ "learning_rate": 4.940144848105379e-05,
554
+ "loss": 0.2267,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.08026008330793458,
559
+ "grad_norm": 4.5625,
560
+ "learning_rate": 4.93837877888615e-05,
561
+ "loss": 0.2597,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.081276033729554,
566
+ "grad_norm": 3.03125,
567
+ "learning_rate": 4.9365873574971745e-05,
568
+ "loss": 0.3701,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.08229198415117342,
573
+ "grad_norm": 4.5625,
574
+ "learning_rate": 4.9347706025641136e-05,
575
+ "loss": 0.2559,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.08330793457279284,
580
+ "grad_norm": 3.90625,
581
+ "learning_rate": 4.9329285329760275e-05,
582
+ "loss": 0.2799,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.08432388499441228,
587
+ "grad_norm": 3.140625,
588
+ "learning_rate": 4.9310611678851735e-05,
589
+ "loss": 0.2866,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.0853398354160317,
594
+ "grad_norm": 2.46875,
595
+ "learning_rate": 4.929168526706811e-05,
596
+ "loss": 0.3105,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.08635578583765112,
601
+ "grad_norm": 13.625,
602
+ "learning_rate": 4.927250629119e-05,
603
+ "loss": 0.2454,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.08737173625927054,
608
+ "grad_norm": 3.921875,
609
+ "learning_rate": 4.9253074950623925e-05,
610
+ "loss": 0.2424,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.08838768668088998,
615
+ "grad_norm": 2.90625,
616
+ "learning_rate": 4.9233391447400286e-05,
617
+ "loss": 0.2481,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.0894036371025094,
622
+ "grad_norm": 2.96875,
623
+ "learning_rate": 4.921345598617125e-05,
624
+ "loss": 0.2231,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.09041958752412882,
629
+ "grad_norm": 5.375,
630
+ "learning_rate": 4.9193268774208654e-05,
631
+ "loss": 0.3447,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.09143553794574824,
636
+ "grad_norm": 2.0,
637
+ "learning_rate": 4.9172830021401785e-05,
638
+ "loss": 0.229,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.09245148836736768,
643
+ "grad_norm": 3.1875,
644
+ "learning_rate": 4.9152139940255245e-05,
645
+ "loss": 0.2122,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.0934674387889871,
650
+ "grad_norm": 3.40625,
651
+ "learning_rate": 4.913119874588677e-05,
652
+ "loss": 0.2386,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.09448338921060652,
657
+ "grad_norm": 1.4609375,
658
+ "learning_rate": 4.911000665602489e-05,
659
+ "loss": 0.1944,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.09549933963222594,
664
+ "grad_norm": 5.0625,
665
+ "learning_rate": 4.9088563891006786e-05,
666
+ "loss": 0.2038,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.09651529005384538,
671
+ "grad_norm": 4.53125,
672
+ "learning_rate": 4.906687067377592e-05,
673
+ "loss": 0.3122,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.0975312404754648,
678
+ "grad_norm": 2.84375,
679
+ "learning_rate": 4.904492722987976e-05,
680
+ "loss": 0.3157,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.09854719089708422,
685
+ "grad_norm": 2.171875,
686
+ "learning_rate": 4.902273378746738e-05,
687
+ "loss": 0.3077,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.09956314131870364,
692
+ "grad_norm": 2.84375,
693
+ "learning_rate": 4.9000290577287165e-05,
694
+ "loss": 0.2756,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.10057909174032308,
699
+ "grad_norm": 0.99609375,
700
+ "learning_rate": 4.897759783268434e-05,
701
+ "loss": 0.2915,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.1015950421619425,
706
+ "grad_norm": 3.53125,
707
+ "learning_rate": 4.895465578959859e-05,
708
+ "loss": 0.2052,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.10261099258356192,
713
+ "grad_norm": 4.0,
714
+ "learning_rate": 4.893146468656159e-05,
715
+ "loss": 0.2499,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.10362694300518134,
720
+ "grad_norm": 1.65625,
721
+ "learning_rate": 4.890802476469452e-05,
722
+ "loss": 0.278,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.10464289342680078,
727
+ "grad_norm": 3.625,
728
+ "learning_rate": 4.888433626770558e-05,
729
+ "loss": 0.2143,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.1056588438484202,
734
+ "grad_norm": 5.0625,
735
+ "learning_rate": 4.886039944188741e-05,
736
+ "loss": 0.2878,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.10667479427003962,
741
+ "grad_norm": 4.5,
742
+ "learning_rate": 4.883621453611461e-05,
743
+ "loss": 0.2744,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.10769074469165904,
748
+ "grad_norm": 4.5625,
749
+ "learning_rate": 4.881178180184106e-05,
750
+ "loss": 0.2734,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.10870669511327848,
755
+ "grad_norm": 3.125,
756
+ "learning_rate": 4.878710149309735e-05,
757
+ "loss": 0.3574,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.1097226455348979,
762
+ "grad_norm": 3.0625,
763
+ "learning_rate": 4.876217386648816e-05,
764
+ "loss": 0.2625,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.11073859595651732,
769
+ "grad_norm": 4.0625,
770
+ "learning_rate": 4.873699918118955e-05,
771
+ "loss": 0.2437,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.11175454637813674,
776
+ "grad_norm": 1.59375,
777
+ "learning_rate": 4.87115776989463e-05,
778
+ "loss": 0.2051,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.11277049679975618,
783
+ "grad_norm": 4.375,
784
+ "learning_rate": 4.8685909684069153e-05,
785
+ "loss": 0.1727,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.1137864472213756,
790
+ "grad_norm": 2.28125,
791
+ "learning_rate": 4.865999540343211e-05,
792
+ "loss": 0.2256,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.11480239764299502,
797
+ "grad_norm": 2.265625,
798
+ "learning_rate": 4.86338351264696e-05,
799
+ "loss": 0.3529,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.11581834806461444,
804
+ "grad_norm": 2.34375,
805
+ "learning_rate": 4.8607429125173754e-05,
806
+ "loss": 0.2113,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.11683429848623388,
811
+ "grad_norm": 0.7578125,
812
+ "learning_rate": 4.858077767409149e-05,
813
+ "loss": 0.2759,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.1178502489078533,
818
+ "grad_norm": 3.640625,
819
+ "learning_rate": 4.855388105032174e-05,
820
+ "loss": 0.2482,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.11886619932947272,
825
+ "grad_norm": 3.5,
826
+ "learning_rate": 4.852673953351249e-05,
827
+ "loss": 0.1865,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.11988214975109214,
832
+ "grad_norm": 3.75,
833
+ "learning_rate": 4.849935340585796e-05,
834
+ "loss": 0.2659,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.12089810017271158,
839
+ "grad_norm": 3.375,
840
+ "learning_rate": 4.8471722952095586e-05,
841
+ "loss": 0.1506,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.121914050594331,
846
+ "grad_norm": 3.34375,
847
+ "learning_rate": 4.844384845950312e-05,
848
+ "loss": 0.307,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.12293000101595042,
853
+ "grad_norm": 1.578125,
854
+ "learning_rate": 4.841573021789561e-05,
855
+ "loss": 0.1952,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.12394595143756984,
860
+ "grad_norm": 1.2890625,
861
+ "learning_rate": 4.838736851962239e-05,
862
+ "loss": 0.1779,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.12496190185918928,
867
+ "grad_norm": 1.265625,
868
+ "learning_rate": 4.835876365956408e-05,
869
+ "loss": 0.1235,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.12597785228080868,
874
+ "grad_norm": 1.9609375,
875
+ "learning_rate": 4.8329915935129436e-05,
876
+ "loss": 0.1876,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.12699380270242813,
881
+ "grad_norm": 1.6328125,
882
+ "learning_rate": 4.830082564625235e-05,
883
+ "loss": 0.2188,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.12800975312404755,
888
+ "grad_norm": 3.96875,
889
+ "learning_rate": 4.8271493095388684e-05,
890
+ "loss": 0.2622,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.12902570354566698,
895
+ "grad_norm": 3.765625,
896
+ "learning_rate": 4.824191858751312e-05,
897
+ "loss": 0.2724,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.1300416539672864,
902
+ "grad_norm": 5.59375,
903
+ "learning_rate": 4.821210243011601e-05,
904
+ "loss": 0.2413,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.13105760438890582,
909
+ "grad_norm": 3.34375,
910
+ "learning_rate": 4.818204493320016e-05,
911
+ "loss": 0.2618,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.13207355481052524,
916
+ "grad_norm": 2.78125,
917
+ "learning_rate": 4.8151746409277634e-05,
918
+ "loss": 0.2295,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.13308950523214466,
923
+ "grad_norm": 3.1875,
924
+ "learning_rate": 4.8121207173366484e-05,
925
+ "loss": 0.2733,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.13410545565376408,
930
+ "grad_norm": 2.28125,
931
+ "learning_rate": 4.809042754298746e-05,
932
+ "loss": 0.2311,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.13512140607538353,
937
+ "grad_norm": 2.171875,
938
+ "learning_rate": 4.805940783816075e-05,
939
+ "loss": 0.2059,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.13613735649700295,
944
+ "grad_norm": 2.796875,
945
+ "learning_rate": 4.8028148381402625e-05,
946
+ "loss": 0.2102,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.13715330691862238,
951
+ "grad_norm": 2.96875,
952
+ "learning_rate": 4.7996649497722084e-05,
953
+ "loss": 0.2708,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.1381692573402418,
958
+ "grad_norm": 2.4375,
959
+ "learning_rate": 4.7964911514617485e-05,
960
+ "loss": 0.2429,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.13918520776186122,
965
+ "grad_norm": 5.8125,
966
+ "learning_rate": 4.793293476207312e-05,
967
+ "loss": 0.2725,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.14020115818348064,
972
+ "grad_norm": 2.40625,
973
+ "learning_rate": 4.790071957255585e-05,
974
+ "loss": 0.2098,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.14121710860510006,
979
+ "grad_norm": 4.25,
980
+ "learning_rate": 4.786826628101154e-05,
981
+ "loss": 0.2101,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.14223305902671948,
986
+ "grad_norm": 2.578125,
987
+ "learning_rate": 4.783557522486167e-05,
988
+ "loss": 0.2624,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.14324900944833893,
993
+ "grad_norm": 3.125,
994
+ "learning_rate": 4.780264674399978e-05,
995
+ "loss": 0.2518,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.14426495986995835,
1000
+ "grad_norm": 3.671875,
1001
+ "learning_rate": 4.7769481180787966e-05,
1002
+ "loss": 0.3112,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.14528091029157778,
1007
+ "grad_norm": 3.984375,
1008
+ "learning_rate": 4.773607888005327e-05,
1009
+ "loss": 0.2747,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1462968607131972,
1014
+ "grad_norm": 3.234375,
1015
+ "learning_rate": 4.770244018908416e-05,
1016
+ "loss": 0.1572,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.14731281113481662,
1021
+ "grad_norm": 4.09375,
1022
+ "learning_rate": 4.766856545762687e-05,
1023
+ "loss": 0.2148,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.14832876155643604,
1028
+ "grad_norm": 1.6875,
1029
+ "learning_rate": 4.763445503788178e-05,
1030
+ "loss": 0.2531,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.14934471197805546,
1035
+ "grad_norm": 2.375,
1036
+ "learning_rate": 4.760010928449976e-05,
1037
+ "loss": 0.199,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.15036066239967488,
1042
+ "grad_norm": 4.6875,
1043
+ "learning_rate": 4.7565528554578485e-05,
1044
+ "loss": 0.2366,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.15137661282129433,
1049
+ "grad_norm": 5.4375,
1050
+ "learning_rate": 4.75307132076587e-05,
1051
+ "loss": 0.1862,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.15239256324291375,
1056
+ "grad_norm": 2.484375,
1057
+ "learning_rate": 4.749566360572049e-05,
1058
+ "loss": 0.2143,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.15340851366453317,
1063
+ "grad_norm": 2.1875,
1064
+ "learning_rate": 4.746038011317955e-05,
1065
+ "loss": 0.1877,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.1544244640861526,
1070
+ "grad_norm": 2.84375,
1071
+ "learning_rate": 4.742486309688333e-05,
1072
+ "loss": 0.2831,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.15544041450777202,
1077
+ "grad_norm": 2.015625,
1078
+ "learning_rate": 4.738911292610732e-05,
1079
+ "loss": 0.1708,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.15645636492939144,
1084
+ "grad_norm": 3.953125,
1085
+ "learning_rate": 4.735312997255107e-05,
1086
+ "loss": 0.192,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.15747231535101086,
1091
+ "grad_norm": 2.09375,
1092
+ "learning_rate": 4.7316914610334475e-05,
1093
+ "loss": 0.2586,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.15848826577263028,
1098
+ "grad_norm": 3.6875,
1099
+ "learning_rate": 4.728046721599378e-05,
1100
+ "loss": 0.2141,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.15950421619424973,
1105
+ "grad_norm": 2.9375,
1106
+ "learning_rate": 4.724378816847771e-05,
1107
+ "loss": 0.193,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.16052016661586915,
1112
+ "grad_norm": 1.5625,
1113
+ "learning_rate": 4.720687784914352e-05,
1114
+ "loss": 0.191,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.16153611703748857,
1119
+ "grad_norm": 3.75,
1120
+ "learning_rate": 4.716973664175304e-05,
1121
+ "loss": 0.2172,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.162552067459108,
1126
+ "grad_norm": 3.125,
1127
+ "learning_rate": 4.7132364932468645e-05,
1128
+ "loss": 0.2134,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.16356801788072742,
1133
+ "grad_norm": 4.09375,
1134
+ "learning_rate": 4.709476310984932e-05,
1135
+ "loss": 0.2055,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.16458396830234684,
1140
+ "grad_norm": 3.875,
1141
+ "learning_rate": 4.705693156484652e-05,
1142
+ "loss": 0.2136,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.16559991872396626,
1147
+ "grad_norm": 1.1796875,
1148
+ "learning_rate": 4.7018870690800196e-05,
1149
+ "loss": 0.1471,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.16661586914558568,
1154
+ "grad_norm": 2.5,
1155
+ "learning_rate": 4.698058088343465e-05,
1156
+ "loss": 0.2308,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.16763181956720513,
1161
+ "grad_norm": 1.390625,
1162
+ "learning_rate": 4.6942062540854425e-05,
1163
+ "loss": 0.2456,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.16864776998882455,
1168
+ "grad_norm": 3.125,
1169
+ "learning_rate": 4.69033160635402e-05,
1170
+ "loss": 0.2654,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.16966372041044397,
1175
+ "grad_norm": 3.984375,
1176
+ "learning_rate": 4.6864341854344587e-05,
1177
+ "loss": 0.2226,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.1706796708320634,
1182
+ "grad_norm": 2.328125,
1183
+ "learning_rate": 4.682514031848795e-05,
1184
+ "loss": 0.2438,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.17169562125368282,
1189
+ "grad_norm": 3.078125,
1190
+ "learning_rate": 4.678571186355423e-05,
1191
+ "loss": 0.1889,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.17271157167530224,
1196
+ "grad_norm": 3.328125,
1197
+ "learning_rate": 4.6746056899486644e-05,
1198
+ "loss": 0.2117,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.17372752209692166,
1203
+ "grad_norm": 2.78125,
1204
+ "learning_rate": 4.67061758385835e-05,
1205
+ "loss": 0.1953,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.17474347251854108,
1210
+ "grad_norm": 3.09375,
1211
+ "learning_rate": 4.6666069095493816e-05,
1212
+ "loss": 0.1844,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.17575942294016053,
1217
+ "grad_norm": 3.234375,
1218
+ "learning_rate": 4.662573708721309e-05,
1219
+ "loss": 0.2774,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.17677537336177995,
1224
+ "grad_norm": 4.03125,
1225
+ "learning_rate": 4.658518023307894e-05,
1226
+ "loss": 0.2527,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.17779132378339937,
1231
+ "grad_norm": 3.21875,
1232
+ "learning_rate": 4.654439895476671e-05,
1233
+ "loss": 0.2164,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.1788072742050188,
1238
+ "grad_norm": 2.390625,
1239
+ "learning_rate": 4.6503393676285146e-05,
1240
+ "loss": 0.2424,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.17982322462663822,
1245
+ "grad_norm": 1.8359375,
1246
+ "learning_rate": 4.646216482397192e-05,
1247
+ "loss": 0.2428,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.18083917504825764,
1252
+ "grad_norm": 2.796875,
1253
+ "learning_rate": 4.6420712826489275e-05,
1254
+ "loss": 0.2155,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.18185512546987706,
1259
+ "grad_norm": 0.69921875,
1260
+ "learning_rate": 4.6379038114819485e-05,
1261
+ "loss": 0.1544,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.18287107589149648,
1266
+ "grad_norm": 3.40625,
1267
+ "learning_rate": 4.6337141122260444e-05,
1268
+ "loss": 0.2029,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.18388702631311593,
1273
+ "grad_norm": 2.359375,
1274
+ "learning_rate": 4.629502228442112e-05,
1275
+ "loss": 0.1489,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.18490297673473535,
1280
+ "grad_norm": 1.4453125,
1281
+ "learning_rate": 4.6252682039217045e-05,
1282
+ "loss": 0.2101,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.18591892715635477,
1287
+ "grad_norm": 2.71875,
1288
+ "learning_rate": 4.621012082686573e-05,
1289
+ "loss": 0.2076,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.1869348775779742,
1294
+ "grad_norm": 3.0625,
1295
+ "learning_rate": 4.616733908988216e-05,
1296
+ "loss": 0.2719,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.18795082799959362,
1301
+ "grad_norm": 1.953125,
1302
+ "learning_rate": 4.612433727307409e-05,
1303
+ "loss": 0.2105,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.18896677842121304,
1308
+ "grad_norm": 3.46875,
1309
+ "learning_rate": 4.608111582353751e-05,
1310
+ "loss": 0.1877,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.18998272884283246,
1315
+ "grad_norm": 2.546875,
1316
+ "learning_rate": 4.603767519065197e-05,
1317
+ "loss": 0.2238,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.19099867926445188,
1322
+ "grad_norm": 1.5703125,
1323
+ "learning_rate": 4.599401582607589e-05,
1324
+ "loss": 0.243,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.19201462968607133,
1329
+ "grad_norm": 2.4375,
1330
+ "learning_rate": 4.595013818374185e-05,
1331
+ "loss": 0.1867,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.19303058010769075,
1336
+ "grad_norm": 2.203125,
1337
+ "learning_rate": 4.5906042719851925e-05,
1338
+ "loss": 0.1994,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.19404653052931017,
1343
+ "grad_norm": 3.984375,
1344
+ "learning_rate": 4.586172989287291e-05,
1345
+ "loss": 0.1899,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.1950624809509296,
1350
+ "grad_norm": 2.6875,
1351
+ "learning_rate": 4.5817200163531534e-05,
1352
+ "loss": 0.2528,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.19607843137254902,
1357
+ "grad_norm": 2.71875,
1358
+ "learning_rate": 4.577245399480972e-05,
1359
+ "loss": 0.2336,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.19709438179416844,
1364
+ "grad_norm": 2.640625,
1365
+ "learning_rate": 4.5727491851939715e-05,
1366
+ "loss": 0.2204,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.19811033221578786,
1371
+ "grad_norm": 1.78125,
1372
+ "learning_rate": 4.568231420239929e-05,
1373
+ "loss": 0.1656,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.19912628263740728,
1378
+ "grad_norm": 3.15625,
1379
+ "learning_rate": 4.563692151590687e-05,
1380
+ "loss": 0.2105,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.20014223305902673,
1385
+ "grad_norm": 1.3671875,
1386
+ "learning_rate": 4.5591314264416666e-05,
1387
+ "loss": 0.1464,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.20115818348064615,
1392
+ "grad_norm": 4.25,
1393
+ "learning_rate": 4.554549292211371e-05,
1394
+ "loss": 0.2103,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.20217413390226557,
1399
+ "grad_norm": 2.984375,
1400
+ "learning_rate": 4.549945796540901e-05,
1401
+ "loss": 0.144,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.203190084323885,
1406
+ "grad_norm": 1.859375,
1407
+ "learning_rate": 4.545320987293453e-05,
1408
+ "loss": 0.1963,
1409
+ "step": 2000
1410
+ }
1411
+ ],
1412
+ "logging_steps": 10,
1413
+ "max_steps": 9843,
1414
+ "num_input_tokens_seen": 0,
1415
+ "num_train_epochs": 1,
1416
+ "save_steps": 500,
1417
+ "total_flos": 0.0,
1418
+ "train_batch_size": 7,
1419
+ "trial_name": null,
1420
+ "trial_params": null
1421
+ }
checkpoints/checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
3
+ size 4984
checkpoints/checkpoint-2500/.DS_Store ADDED
Binary file (6.15 kB). View file
 
checkpoints/checkpoint-2500/adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }
checkpoints/checkpoint-2500/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56179446117ddaff95696487189a032d48a215d8456339c2ae5eda1870df93b9
3
+ size 5919456
checkpoints/checkpoint-2500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd04baeeb8fe514953f04bca66abb944e11003693bffa63c977e38849273cb49
3
+ size 11930938
checkpoints/checkpoint-2500/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:460a81dc2425be030afb2c9930b7e9fc9de54ad9cf988c330851fefe47a118c2
3
+ size 15024
checkpoints/checkpoint-2500/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f39c713bcb800d5d9d55e44b2d2e744bbee1e449a935b9a681868e507ac58f86
3
+ size 15024
checkpoints/checkpoint-2500/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6b1c81cf4c5daaf0fe1a5aa4e87259ede9f631f0867a6d085b8eba9a03f7275
3
+ size 15024
checkpoints/checkpoint-2500/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7ae3d71695a016b734120d0e84d8510429872cccd4cdfc5ad051249bdceb709
3
+ size 15024
checkpoints/checkpoint-2500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0992ea36b796a9efd875f53514a1a7b72426fb94846549bcad84ea3eae0acee
3
+ size 1064
checkpoints/checkpoint-2500/trainer_state.json ADDED
@@ -0,0 +1,1771 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.25398760540485626,
5
+ "eval_steps": 500,
6
+ "global_step": 2500,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001015950421619425,
13
+ "grad_norm": 17.625,
14
+ "learning_rate": 5e-06,
15
+ "loss": 3.4264,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.00203190084323885,
20
+ "grad_norm": 12.5625,
21
+ "learning_rate": 1e-05,
22
+ "loss": 3.432,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.003047851264858275,
27
+ "grad_norm": 14.0625,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 3.23,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0040638016864777,
34
+ "grad_norm": 12.4375,
35
+ "learning_rate": 2e-05,
36
+ "loss": 2.9762,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.005079752108097125,
41
+ "grad_norm": 10.0625,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 2.6173,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.00609570252971655,
48
+ "grad_norm": 10.1875,
49
+ "learning_rate": 3e-05,
50
+ "loss": 2.2004,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.007111652951335975,
55
+ "grad_norm": 7.03125,
56
+ "learning_rate": 3.5e-05,
57
+ "loss": 1.4176,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.0081276033729554,
62
+ "grad_norm": 4.375,
63
+ "learning_rate": 4e-05,
64
+ "loss": 1.0122,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.009143553794574825,
69
+ "grad_norm": 6.5625,
70
+ "learning_rate": 4.5e-05,
71
+ "loss": 0.9116,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.01015950421619425,
76
+ "grad_norm": 5.28125,
77
+ "learning_rate": 5e-05,
78
+ "loss": 0.6832,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.011175454637813675,
83
+ "grad_norm": 5.5,
84
+ "learning_rate": 4.9999870035728426e-05,
85
+ "loss": 0.7355,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.0121914050594331,
90
+ "grad_norm": 5.1875,
91
+ "learning_rate": 4.9999480144264944e-05,
92
+ "loss": 0.6673,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.013207355481052525,
97
+ "grad_norm": 4.5,
98
+ "learning_rate": 4.9998830329663314e-05,
99
+ "loss": 0.6792,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.01422330590267195,
104
+ "grad_norm": 3.9375,
105
+ "learning_rate": 4.9997920598679756e-05,
106
+ "loss": 0.6207,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.015239256324291375,
111
+ "grad_norm": 3.15625,
112
+ "learning_rate": 4.999675096077286e-05,
113
+ "loss": 0.483,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.0162552067459108,
118
+ "grad_norm": 5.28125,
119
+ "learning_rate": 4.999532142810354e-05,
120
+ "loss": 0.5319,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.017271157167530225,
125
+ "grad_norm": 4.59375,
126
+ "learning_rate": 4.999363201553483e-05,
127
+ "loss": 0.6052,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.01828710758914965,
132
+ "grad_norm": 5.03125,
133
+ "learning_rate": 4.9991682740631794e-05,
134
+ "loss": 0.4258,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.019303058010769075,
139
+ "grad_norm": 3.859375,
140
+ "learning_rate": 4.998947362366133e-05,
141
+ "loss": 0.4309,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.0203190084323885,
146
+ "grad_norm": 3.328125,
147
+ "learning_rate": 4.998700468759193e-05,
148
+ "loss": 0.3957,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.021334958854007924,
153
+ "grad_norm": 4.9375,
154
+ "learning_rate": 4.9984275958093475e-05,
155
+ "loss": 0.4777,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.02235090927562735,
160
+ "grad_norm": 4.78125,
161
+ "learning_rate": 4.998128746353695e-05,
162
+ "loss": 0.3549,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.023366859697246774,
167
+ "grad_norm": 4.0625,
168
+ "learning_rate": 4.997803923499417e-05,
169
+ "loss": 0.4447,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.0243828101188662,
174
+ "grad_norm": 6.375,
175
+ "learning_rate": 4.99745313062374e-05,
176
+ "loss": 0.3808,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.025398760540485624,
181
+ "grad_norm": 3.59375,
182
+ "learning_rate": 4.99707637137391e-05,
183
+ "loss": 0.3827,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.02641471096210505,
188
+ "grad_norm": 3.015625,
189
+ "learning_rate": 4.996673649667145e-05,
190
+ "loss": 0.3694,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.027430661383724474,
195
+ "grad_norm": 2.296875,
196
+ "learning_rate": 4.9962449696906e-05,
197
+ "loss": 0.3586,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.0284466118053439,
202
+ "grad_norm": 4.125,
203
+ "learning_rate": 4.9957903359013214e-05,
204
+ "loss": 0.3832,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.029462562226963324,
209
+ "grad_norm": 3.296875,
210
+ "learning_rate": 4.995309753026201e-05,
211
+ "loss": 0.328,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.03047851264858275,
216
+ "grad_norm": 4.5,
217
+ "learning_rate": 4.994803226061927e-05,
218
+ "loss": 0.3667,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.03149446307020217,
223
+ "grad_norm": 4.3125,
224
+ "learning_rate": 4.994270760274933e-05,
225
+ "loss": 0.3811,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.0325104134918216,
230
+ "grad_norm": 3.421875,
231
+ "learning_rate": 4.99371236120134e-05,
232
+ "loss": 0.3065,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.03352636391344102,
237
+ "grad_norm": 4.6875,
238
+ "learning_rate": 4.993128034646902e-05,
239
+ "loss": 0.4177,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.03454231433506045,
244
+ "grad_norm": 3.046875,
245
+ "learning_rate": 4.992517786686947e-05,
246
+ "loss": 0.33,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.03555826475667987,
251
+ "grad_norm": 1.8828125,
252
+ "learning_rate": 4.9918816236663077e-05,
253
+ "loss": 0.3287,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.0365742151782993,
258
+ "grad_norm": 3.8125,
259
+ "learning_rate": 4.991219552199262e-05,
260
+ "loss": 0.2934,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.03759016559991872,
265
+ "grad_norm": 4.28125,
266
+ "learning_rate": 4.99053157916946e-05,
267
+ "loss": 0.3176,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.03860611602153815,
272
+ "grad_norm": 2.609375,
273
+ "learning_rate": 4.989817711729856e-05,
274
+ "loss": 0.3318,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.03962206644315757,
279
+ "grad_norm": 2.375,
280
+ "learning_rate": 4.98907795730263e-05,
281
+ "loss": 0.3234,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.040638016864777,
286
+ "grad_norm": 4.46875,
287
+ "learning_rate": 4.988312323579114e-05,
288
+ "loss": 0.267,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.04165396728639642,
293
+ "grad_norm": 3.75,
294
+ "learning_rate": 4.98752081851971e-05,
295
+ "loss": 0.3081,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.04266991770801585,
300
+ "grad_norm": 2.203125,
301
+ "learning_rate": 4.986703450353809e-05,
302
+ "loss": 0.2917,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.04368586812963527,
307
+ "grad_norm": 1.6015625,
308
+ "learning_rate": 4.985860227579703e-05,
309
+ "loss": 0.2805,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.0447018185512547,
314
+ "grad_norm": 3.140625,
315
+ "learning_rate": 4.984991158964499e-05,
316
+ "loss": 0.3534,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.04571776897287412,
321
+ "grad_norm": 3.296875,
322
+ "learning_rate": 4.9840962535440265e-05,
323
+ "loss": 0.335,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.04673371939449355,
328
+ "grad_norm": 3.25,
329
+ "learning_rate": 4.983175520622744e-05,
330
+ "loss": 0.2544,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.04774966981611297,
335
+ "grad_norm": 2.25,
336
+ "learning_rate": 4.982228969773642e-05,
337
+ "loss": 0.3449,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.0487656202377324,
342
+ "grad_norm": 4.9375,
343
+ "learning_rate": 4.9812566108381435e-05,
344
+ "loss": 0.2964,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.04978157065935182,
349
+ "grad_norm": 1.5703125,
350
+ "learning_rate": 4.9802584539260035e-05,
351
+ "loss": 0.2799,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.05079752108097125,
356
+ "grad_norm": 2.828125,
357
+ "learning_rate": 4.979234509415199e-05,
358
+ "loss": 0.3231,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.05181347150259067,
363
+ "grad_norm": 2.9375,
364
+ "learning_rate": 4.978184787951828e-05,
365
+ "loss": 0.2943,
366
+ "step": 510
367
+ },
368
+ {
369
+ "epoch": 0.0528294219242101,
370
+ "grad_norm": 2.34375,
371
+ "learning_rate": 4.977109300449992e-05,
372
+ "loss": 0.2705,
373
+ "step": 520
374
+ },
375
+ {
376
+ "epoch": 0.05384537234582952,
377
+ "grad_norm": 3.140625,
378
+ "learning_rate": 4.9760080580916876e-05,
379
+ "loss": 0.2998,
380
+ "step": 530
381
+ },
382
+ {
383
+ "epoch": 0.05486132276744895,
384
+ "grad_norm": 3.5625,
385
+ "learning_rate": 4.974881072326688e-05,
386
+ "loss": 0.2595,
387
+ "step": 540
388
+ },
389
+ {
390
+ "epoch": 0.05587727318906837,
391
+ "grad_norm": 4.25,
392
+ "learning_rate": 4.9737283548724236e-05,
393
+ "loss": 0.2803,
394
+ "step": 550
395
+ },
396
+ {
397
+ "epoch": 0.0568932236106878,
398
+ "grad_norm": 4.0625,
399
+ "learning_rate": 4.97254991771386e-05,
400
+ "loss": 0.3511,
401
+ "step": 560
402
+ },
403
+ {
404
+ "epoch": 0.05790917403230722,
405
+ "grad_norm": 2.515625,
406
+ "learning_rate": 4.971345773103377e-05,
407
+ "loss": 0.312,
408
+ "step": 570
409
+ },
410
+ {
411
+ "epoch": 0.05892512445392665,
412
+ "grad_norm": 3.21875,
413
+ "learning_rate": 4.9701159335606365e-05,
414
+ "loss": 0.2482,
415
+ "step": 580
416
+ },
417
+ {
418
+ "epoch": 0.05994107487554607,
419
+ "grad_norm": 5.5,
420
+ "learning_rate": 4.968860411872454e-05,
421
+ "loss": 0.2537,
422
+ "step": 590
423
+ },
424
+ {
425
+ "epoch": 0.0609570252971655,
426
+ "grad_norm": 3.546875,
427
+ "learning_rate": 4.967579221092666e-05,
428
+ "loss": 0.3125,
429
+ "step": 600
430
+ },
431
+ {
432
+ "epoch": 0.06197297571878492,
433
+ "grad_norm": 2.984375,
434
+ "learning_rate": 4.966272374541996e-05,
435
+ "loss": 0.2354,
436
+ "step": 610
437
+ },
438
+ {
439
+ "epoch": 0.06298892614040434,
440
+ "grad_norm": 3.6875,
441
+ "learning_rate": 4.964939885807912e-05,
442
+ "loss": 0.3213,
443
+ "step": 620
444
+ },
445
+ {
446
+ "epoch": 0.06400487656202378,
447
+ "grad_norm": 2.140625,
448
+ "learning_rate": 4.9635817687444876e-05,
449
+ "loss": 0.3003,
450
+ "step": 630
451
+ },
452
+ {
453
+ "epoch": 0.0650208269836432,
454
+ "grad_norm": 3.484375,
455
+ "learning_rate": 4.962198037472259e-05,
456
+ "loss": 0.2996,
457
+ "step": 640
458
+ },
459
+ {
460
+ "epoch": 0.06603677740526262,
461
+ "grad_norm": 3.21875,
462
+ "learning_rate": 4.9607887063780776e-05,
463
+ "loss": 0.2257,
464
+ "step": 650
465
+ },
466
+ {
467
+ "epoch": 0.06705272782688204,
468
+ "grad_norm": 5.375,
469
+ "learning_rate": 4.9593537901149564e-05,
470
+ "loss": 0.223,
471
+ "step": 660
472
+ },
473
+ {
474
+ "epoch": 0.06806867824850148,
475
+ "grad_norm": 4.1875,
476
+ "learning_rate": 4.957893303601924e-05,
477
+ "loss": 0.3407,
478
+ "step": 670
479
+ },
480
+ {
481
+ "epoch": 0.0690846286701209,
482
+ "grad_norm": 3.328125,
483
+ "learning_rate": 4.956407262023866e-05,
484
+ "loss": 0.2589,
485
+ "step": 680
486
+ },
487
+ {
488
+ "epoch": 0.07010057909174032,
489
+ "grad_norm": 2.953125,
490
+ "learning_rate": 4.954895680831367e-05,
491
+ "loss": 0.2949,
492
+ "step": 690
493
+ },
494
+ {
495
+ "epoch": 0.07111652951335974,
496
+ "grad_norm": 4.0625,
497
+ "learning_rate": 4.9533585757405506e-05,
498
+ "loss": 0.2995,
499
+ "step": 700
500
+ },
501
+ {
502
+ "epoch": 0.07213247993497918,
503
+ "grad_norm": 4.625,
504
+ "learning_rate": 4.951795962732917e-05,
505
+ "loss": 0.2894,
506
+ "step": 710
507
+ },
508
+ {
509
+ "epoch": 0.0731484303565986,
510
+ "grad_norm": 3.0,
511
+ "learning_rate": 4.9502078580551755e-05,
512
+ "loss": 0.3082,
513
+ "step": 720
514
+ },
515
+ {
516
+ "epoch": 0.07416438077821802,
517
+ "grad_norm": 3.65625,
518
+ "learning_rate": 4.9485942782190734e-05,
519
+ "loss": 0.2308,
520
+ "step": 730
521
+ },
522
+ {
523
+ "epoch": 0.07518033119983744,
524
+ "grad_norm": 4.78125,
525
+ "learning_rate": 4.9469552400012306e-05,
526
+ "loss": 0.2272,
527
+ "step": 740
528
+ },
529
+ {
530
+ "epoch": 0.07619628162145688,
531
+ "grad_norm": 4.25,
532
+ "learning_rate": 4.94529076044296e-05,
533
+ "loss": 0.2701,
534
+ "step": 750
535
+ },
536
+ {
537
+ "epoch": 0.0772122320430763,
538
+ "grad_norm": 3.140625,
539
+ "learning_rate": 4.94360085685009e-05,
540
+ "loss": 0.2686,
541
+ "step": 760
542
+ },
543
+ {
544
+ "epoch": 0.07822818246469572,
545
+ "grad_norm": 0.765625,
546
+ "learning_rate": 4.9418855467927894e-05,
547
+ "loss": 0.2051,
548
+ "step": 770
549
+ },
550
+ {
551
+ "epoch": 0.07924413288631514,
552
+ "grad_norm": 1.796875,
553
+ "learning_rate": 4.940144848105379e-05,
554
+ "loss": 0.2267,
555
+ "step": 780
556
+ },
557
+ {
558
+ "epoch": 0.08026008330793458,
559
+ "grad_norm": 4.5625,
560
+ "learning_rate": 4.93837877888615e-05,
561
+ "loss": 0.2597,
562
+ "step": 790
563
+ },
564
+ {
565
+ "epoch": 0.081276033729554,
566
+ "grad_norm": 3.03125,
567
+ "learning_rate": 4.9365873574971745e-05,
568
+ "loss": 0.3701,
569
+ "step": 800
570
+ },
571
+ {
572
+ "epoch": 0.08229198415117342,
573
+ "grad_norm": 4.5625,
574
+ "learning_rate": 4.9347706025641136e-05,
575
+ "loss": 0.2559,
576
+ "step": 810
577
+ },
578
+ {
579
+ "epoch": 0.08330793457279284,
580
+ "grad_norm": 3.90625,
581
+ "learning_rate": 4.9329285329760275e-05,
582
+ "loss": 0.2799,
583
+ "step": 820
584
+ },
585
+ {
586
+ "epoch": 0.08432388499441228,
587
+ "grad_norm": 3.140625,
588
+ "learning_rate": 4.9310611678851735e-05,
589
+ "loss": 0.2866,
590
+ "step": 830
591
+ },
592
+ {
593
+ "epoch": 0.0853398354160317,
594
+ "grad_norm": 2.46875,
595
+ "learning_rate": 4.929168526706811e-05,
596
+ "loss": 0.3105,
597
+ "step": 840
598
+ },
599
+ {
600
+ "epoch": 0.08635578583765112,
601
+ "grad_norm": 13.625,
602
+ "learning_rate": 4.927250629119e-05,
603
+ "loss": 0.2454,
604
+ "step": 850
605
+ },
606
+ {
607
+ "epoch": 0.08737173625927054,
608
+ "grad_norm": 3.921875,
609
+ "learning_rate": 4.9253074950623925e-05,
610
+ "loss": 0.2424,
611
+ "step": 860
612
+ },
613
+ {
614
+ "epoch": 0.08838768668088998,
615
+ "grad_norm": 2.90625,
616
+ "learning_rate": 4.9233391447400286e-05,
617
+ "loss": 0.2481,
618
+ "step": 870
619
+ },
620
+ {
621
+ "epoch": 0.0894036371025094,
622
+ "grad_norm": 2.96875,
623
+ "learning_rate": 4.921345598617125e-05,
624
+ "loss": 0.2231,
625
+ "step": 880
626
+ },
627
+ {
628
+ "epoch": 0.09041958752412882,
629
+ "grad_norm": 5.375,
630
+ "learning_rate": 4.9193268774208654e-05,
631
+ "loss": 0.3447,
632
+ "step": 890
633
+ },
634
+ {
635
+ "epoch": 0.09143553794574824,
636
+ "grad_norm": 2.0,
637
+ "learning_rate": 4.9172830021401785e-05,
638
+ "loss": 0.229,
639
+ "step": 900
640
+ },
641
+ {
642
+ "epoch": 0.09245148836736768,
643
+ "grad_norm": 3.1875,
644
+ "learning_rate": 4.9152139940255245e-05,
645
+ "loss": 0.2122,
646
+ "step": 910
647
+ },
648
+ {
649
+ "epoch": 0.0934674387889871,
650
+ "grad_norm": 3.40625,
651
+ "learning_rate": 4.913119874588677e-05,
652
+ "loss": 0.2386,
653
+ "step": 920
654
+ },
655
+ {
656
+ "epoch": 0.09448338921060652,
657
+ "grad_norm": 1.4609375,
658
+ "learning_rate": 4.911000665602489e-05,
659
+ "loss": 0.1944,
660
+ "step": 930
661
+ },
662
+ {
663
+ "epoch": 0.09549933963222594,
664
+ "grad_norm": 5.0625,
665
+ "learning_rate": 4.9088563891006786e-05,
666
+ "loss": 0.2038,
667
+ "step": 940
668
+ },
669
+ {
670
+ "epoch": 0.09651529005384538,
671
+ "grad_norm": 4.53125,
672
+ "learning_rate": 4.906687067377592e-05,
673
+ "loss": 0.3122,
674
+ "step": 950
675
+ },
676
+ {
677
+ "epoch": 0.0975312404754648,
678
+ "grad_norm": 2.84375,
679
+ "learning_rate": 4.904492722987976e-05,
680
+ "loss": 0.3157,
681
+ "step": 960
682
+ },
683
+ {
684
+ "epoch": 0.09854719089708422,
685
+ "grad_norm": 2.171875,
686
+ "learning_rate": 4.902273378746738e-05,
687
+ "loss": 0.3077,
688
+ "step": 970
689
+ },
690
+ {
691
+ "epoch": 0.09956314131870364,
692
+ "grad_norm": 2.84375,
693
+ "learning_rate": 4.9000290577287165e-05,
694
+ "loss": 0.2756,
695
+ "step": 980
696
+ },
697
+ {
698
+ "epoch": 0.10057909174032308,
699
+ "grad_norm": 0.99609375,
700
+ "learning_rate": 4.897759783268434e-05,
701
+ "loss": 0.2915,
702
+ "step": 990
703
+ },
704
+ {
705
+ "epoch": 0.1015950421619425,
706
+ "grad_norm": 3.53125,
707
+ "learning_rate": 4.895465578959859e-05,
708
+ "loss": 0.2052,
709
+ "step": 1000
710
+ },
711
+ {
712
+ "epoch": 0.10261099258356192,
713
+ "grad_norm": 4.0,
714
+ "learning_rate": 4.893146468656159e-05,
715
+ "loss": 0.2499,
716
+ "step": 1010
717
+ },
718
+ {
719
+ "epoch": 0.10362694300518134,
720
+ "grad_norm": 1.65625,
721
+ "learning_rate": 4.890802476469452e-05,
722
+ "loss": 0.278,
723
+ "step": 1020
724
+ },
725
+ {
726
+ "epoch": 0.10464289342680078,
727
+ "grad_norm": 3.625,
728
+ "learning_rate": 4.888433626770558e-05,
729
+ "loss": 0.2143,
730
+ "step": 1030
731
+ },
732
+ {
733
+ "epoch": 0.1056588438484202,
734
+ "grad_norm": 5.0625,
735
+ "learning_rate": 4.886039944188741e-05,
736
+ "loss": 0.2878,
737
+ "step": 1040
738
+ },
739
+ {
740
+ "epoch": 0.10667479427003962,
741
+ "grad_norm": 4.5,
742
+ "learning_rate": 4.883621453611461e-05,
743
+ "loss": 0.2744,
744
+ "step": 1050
745
+ },
746
+ {
747
+ "epoch": 0.10769074469165904,
748
+ "grad_norm": 4.5625,
749
+ "learning_rate": 4.881178180184106e-05,
750
+ "loss": 0.2734,
751
+ "step": 1060
752
+ },
753
+ {
754
+ "epoch": 0.10870669511327848,
755
+ "grad_norm": 3.125,
756
+ "learning_rate": 4.878710149309735e-05,
757
+ "loss": 0.3574,
758
+ "step": 1070
759
+ },
760
+ {
761
+ "epoch": 0.1097226455348979,
762
+ "grad_norm": 3.0625,
763
+ "learning_rate": 4.876217386648816e-05,
764
+ "loss": 0.2625,
765
+ "step": 1080
766
+ },
767
+ {
768
+ "epoch": 0.11073859595651732,
769
+ "grad_norm": 4.0625,
770
+ "learning_rate": 4.873699918118955e-05,
771
+ "loss": 0.2437,
772
+ "step": 1090
773
+ },
774
+ {
775
+ "epoch": 0.11175454637813674,
776
+ "grad_norm": 1.59375,
777
+ "learning_rate": 4.87115776989463e-05,
778
+ "loss": 0.2051,
779
+ "step": 1100
780
+ },
781
+ {
782
+ "epoch": 0.11277049679975618,
783
+ "grad_norm": 4.375,
784
+ "learning_rate": 4.8685909684069153e-05,
785
+ "loss": 0.1727,
786
+ "step": 1110
787
+ },
788
+ {
789
+ "epoch": 0.1137864472213756,
790
+ "grad_norm": 2.28125,
791
+ "learning_rate": 4.865999540343211e-05,
792
+ "loss": 0.2256,
793
+ "step": 1120
794
+ },
795
+ {
796
+ "epoch": 0.11480239764299502,
797
+ "grad_norm": 2.265625,
798
+ "learning_rate": 4.86338351264696e-05,
799
+ "loss": 0.3529,
800
+ "step": 1130
801
+ },
802
+ {
803
+ "epoch": 0.11581834806461444,
804
+ "grad_norm": 2.34375,
805
+ "learning_rate": 4.8607429125173754e-05,
806
+ "loss": 0.2113,
807
+ "step": 1140
808
+ },
809
+ {
810
+ "epoch": 0.11683429848623388,
811
+ "grad_norm": 0.7578125,
812
+ "learning_rate": 4.858077767409149e-05,
813
+ "loss": 0.2759,
814
+ "step": 1150
815
+ },
816
+ {
817
+ "epoch": 0.1178502489078533,
818
+ "grad_norm": 3.640625,
819
+ "learning_rate": 4.855388105032174e-05,
820
+ "loss": 0.2482,
821
+ "step": 1160
822
+ },
823
+ {
824
+ "epoch": 0.11886619932947272,
825
+ "grad_norm": 3.5,
826
+ "learning_rate": 4.852673953351249e-05,
827
+ "loss": 0.1865,
828
+ "step": 1170
829
+ },
830
+ {
831
+ "epoch": 0.11988214975109214,
832
+ "grad_norm": 3.75,
833
+ "learning_rate": 4.849935340585796e-05,
834
+ "loss": 0.2659,
835
+ "step": 1180
836
+ },
837
+ {
838
+ "epoch": 0.12089810017271158,
839
+ "grad_norm": 3.375,
840
+ "learning_rate": 4.8471722952095586e-05,
841
+ "loss": 0.1506,
842
+ "step": 1190
843
+ },
844
+ {
845
+ "epoch": 0.121914050594331,
846
+ "grad_norm": 3.34375,
847
+ "learning_rate": 4.844384845950312e-05,
848
+ "loss": 0.307,
849
+ "step": 1200
850
+ },
851
+ {
852
+ "epoch": 0.12293000101595042,
853
+ "grad_norm": 1.578125,
854
+ "learning_rate": 4.841573021789561e-05,
855
+ "loss": 0.1952,
856
+ "step": 1210
857
+ },
858
+ {
859
+ "epoch": 0.12394595143756984,
860
+ "grad_norm": 1.2890625,
861
+ "learning_rate": 4.838736851962239e-05,
862
+ "loss": 0.1779,
863
+ "step": 1220
864
+ },
865
+ {
866
+ "epoch": 0.12496190185918928,
867
+ "grad_norm": 1.265625,
868
+ "learning_rate": 4.835876365956408e-05,
869
+ "loss": 0.1235,
870
+ "step": 1230
871
+ },
872
+ {
873
+ "epoch": 0.12597785228080868,
874
+ "grad_norm": 1.9609375,
875
+ "learning_rate": 4.8329915935129436e-05,
876
+ "loss": 0.1876,
877
+ "step": 1240
878
+ },
879
+ {
880
+ "epoch": 0.12699380270242813,
881
+ "grad_norm": 1.6328125,
882
+ "learning_rate": 4.830082564625235e-05,
883
+ "loss": 0.2188,
884
+ "step": 1250
885
+ },
886
+ {
887
+ "epoch": 0.12800975312404755,
888
+ "grad_norm": 3.96875,
889
+ "learning_rate": 4.8271493095388684e-05,
890
+ "loss": 0.2622,
891
+ "step": 1260
892
+ },
893
+ {
894
+ "epoch": 0.12902570354566698,
895
+ "grad_norm": 3.765625,
896
+ "learning_rate": 4.824191858751312e-05,
897
+ "loss": 0.2724,
898
+ "step": 1270
899
+ },
900
+ {
901
+ "epoch": 0.1300416539672864,
902
+ "grad_norm": 5.59375,
903
+ "learning_rate": 4.821210243011601e-05,
904
+ "loss": 0.2413,
905
+ "step": 1280
906
+ },
907
+ {
908
+ "epoch": 0.13105760438890582,
909
+ "grad_norm": 3.34375,
910
+ "learning_rate": 4.818204493320016e-05,
911
+ "loss": 0.2618,
912
+ "step": 1290
913
+ },
914
+ {
915
+ "epoch": 0.13207355481052524,
916
+ "grad_norm": 2.78125,
917
+ "learning_rate": 4.8151746409277634e-05,
918
+ "loss": 0.2295,
919
+ "step": 1300
920
+ },
921
+ {
922
+ "epoch": 0.13308950523214466,
923
+ "grad_norm": 3.1875,
924
+ "learning_rate": 4.8121207173366484e-05,
925
+ "loss": 0.2733,
926
+ "step": 1310
927
+ },
928
+ {
929
+ "epoch": 0.13410545565376408,
930
+ "grad_norm": 2.28125,
931
+ "learning_rate": 4.809042754298746e-05,
932
+ "loss": 0.2311,
933
+ "step": 1320
934
+ },
935
+ {
936
+ "epoch": 0.13512140607538353,
937
+ "grad_norm": 2.171875,
938
+ "learning_rate": 4.805940783816075e-05,
939
+ "loss": 0.2059,
940
+ "step": 1330
941
+ },
942
+ {
943
+ "epoch": 0.13613735649700295,
944
+ "grad_norm": 2.796875,
945
+ "learning_rate": 4.8028148381402625e-05,
946
+ "loss": 0.2102,
947
+ "step": 1340
948
+ },
949
+ {
950
+ "epoch": 0.13715330691862238,
951
+ "grad_norm": 2.96875,
952
+ "learning_rate": 4.7996649497722084e-05,
953
+ "loss": 0.2708,
954
+ "step": 1350
955
+ },
956
+ {
957
+ "epoch": 0.1381692573402418,
958
+ "grad_norm": 2.4375,
959
+ "learning_rate": 4.7964911514617485e-05,
960
+ "loss": 0.2429,
961
+ "step": 1360
962
+ },
963
+ {
964
+ "epoch": 0.13918520776186122,
965
+ "grad_norm": 5.8125,
966
+ "learning_rate": 4.793293476207312e-05,
967
+ "loss": 0.2725,
968
+ "step": 1370
969
+ },
970
+ {
971
+ "epoch": 0.14020115818348064,
972
+ "grad_norm": 2.40625,
973
+ "learning_rate": 4.790071957255585e-05,
974
+ "loss": 0.2098,
975
+ "step": 1380
976
+ },
977
+ {
978
+ "epoch": 0.14121710860510006,
979
+ "grad_norm": 4.25,
980
+ "learning_rate": 4.786826628101154e-05,
981
+ "loss": 0.2101,
982
+ "step": 1390
983
+ },
984
+ {
985
+ "epoch": 0.14223305902671948,
986
+ "grad_norm": 2.578125,
987
+ "learning_rate": 4.783557522486167e-05,
988
+ "loss": 0.2624,
989
+ "step": 1400
990
+ },
991
+ {
992
+ "epoch": 0.14324900944833893,
993
+ "grad_norm": 3.125,
994
+ "learning_rate": 4.780264674399978e-05,
995
+ "loss": 0.2518,
996
+ "step": 1410
997
+ },
998
+ {
999
+ "epoch": 0.14426495986995835,
1000
+ "grad_norm": 3.671875,
1001
+ "learning_rate": 4.7769481180787966e-05,
1002
+ "loss": 0.3112,
1003
+ "step": 1420
1004
+ },
1005
+ {
1006
+ "epoch": 0.14528091029157778,
1007
+ "grad_norm": 3.984375,
1008
+ "learning_rate": 4.773607888005327e-05,
1009
+ "loss": 0.2747,
1010
+ "step": 1430
1011
+ },
1012
+ {
1013
+ "epoch": 0.1462968607131972,
1014
+ "grad_norm": 3.234375,
1015
+ "learning_rate": 4.770244018908416e-05,
1016
+ "loss": 0.1572,
1017
+ "step": 1440
1018
+ },
1019
+ {
1020
+ "epoch": 0.14731281113481662,
1021
+ "grad_norm": 4.09375,
1022
+ "learning_rate": 4.766856545762687e-05,
1023
+ "loss": 0.2148,
1024
+ "step": 1450
1025
+ },
1026
+ {
1027
+ "epoch": 0.14832876155643604,
1028
+ "grad_norm": 1.6875,
1029
+ "learning_rate": 4.763445503788178e-05,
1030
+ "loss": 0.2531,
1031
+ "step": 1460
1032
+ },
1033
+ {
1034
+ "epoch": 0.14934471197805546,
1035
+ "grad_norm": 2.375,
1036
+ "learning_rate": 4.760010928449976e-05,
1037
+ "loss": 0.199,
1038
+ "step": 1470
1039
+ },
1040
+ {
1041
+ "epoch": 0.15036066239967488,
1042
+ "grad_norm": 4.6875,
1043
+ "learning_rate": 4.7565528554578485e-05,
1044
+ "loss": 0.2366,
1045
+ "step": 1480
1046
+ },
1047
+ {
1048
+ "epoch": 0.15137661282129433,
1049
+ "grad_norm": 5.4375,
1050
+ "learning_rate": 4.75307132076587e-05,
1051
+ "loss": 0.1862,
1052
+ "step": 1490
1053
+ },
1054
+ {
1055
+ "epoch": 0.15239256324291375,
1056
+ "grad_norm": 2.484375,
1057
+ "learning_rate": 4.749566360572049e-05,
1058
+ "loss": 0.2143,
1059
+ "step": 1500
1060
+ },
1061
+ {
1062
+ "epoch": 0.15340851366453317,
1063
+ "grad_norm": 2.1875,
1064
+ "learning_rate": 4.746038011317955e-05,
1065
+ "loss": 0.1877,
1066
+ "step": 1510
1067
+ },
1068
+ {
1069
+ "epoch": 0.1544244640861526,
1070
+ "grad_norm": 2.84375,
1071
+ "learning_rate": 4.742486309688333e-05,
1072
+ "loss": 0.2831,
1073
+ "step": 1520
1074
+ },
1075
+ {
1076
+ "epoch": 0.15544041450777202,
1077
+ "grad_norm": 2.015625,
1078
+ "learning_rate": 4.738911292610732e-05,
1079
+ "loss": 0.1708,
1080
+ "step": 1530
1081
+ },
1082
+ {
1083
+ "epoch": 0.15645636492939144,
1084
+ "grad_norm": 3.953125,
1085
+ "learning_rate": 4.735312997255107e-05,
1086
+ "loss": 0.192,
1087
+ "step": 1540
1088
+ },
1089
+ {
1090
+ "epoch": 0.15747231535101086,
1091
+ "grad_norm": 2.09375,
1092
+ "learning_rate": 4.7316914610334475e-05,
1093
+ "loss": 0.2586,
1094
+ "step": 1550
1095
+ },
1096
+ {
1097
+ "epoch": 0.15848826577263028,
1098
+ "grad_norm": 3.6875,
1099
+ "learning_rate": 4.728046721599378e-05,
1100
+ "loss": 0.2141,
1101
+ "step": 1560
1102
+ },
1103
+ {
1104
+ "epoch": 0.15950421619424973,
1105
+ "grad_norm": 2.9375,
1106
+ "learning_rate": 4.724378816847771e-05,
1107
+ "loss": 0.193,
1108
+ "step": 1570
1109
+ },
1110
+ {
1111
+ "epoch": 0.16052016661586915,
1112
+ "grad_norm": 1.5625,
1113
+ "learning_rate": 4.720687784914352e-05,
1114
+ "loss": 0.191,
1115
+ "step": 1580
1116
+ },
1117
+ {
1118
+ "epoch": 0.16153611703748857,
1119
+ "grad_norm": 3.75,
1120
+ "learning_rate": 4.716973664175304e-05,
1121
+ "loss": 0.2172,
1122
+ "step": 1590
1123
+ },
1124
+ {
1125
+ "epoch": 0.162552067459108,
1126
+ "grad_norm": 3.125,
1127
+ "learning_rate": 4.7132364932468645e-05,
1128
+ "loss": 0.2134,
1129
+ "step": 1600
1130
+ },
1131
+ {
1132
+ "epoch": 0.16356801788072742,
1133
+ "grad_norm": 4.09375,
1134
+ "learning_rate": 4.709476310984932e-05,
1135
+ "loss": 0.2055,
1136
+ "step": 1610
1137
+ },
1138
+ {
1139
+ "epoch": 0.16458396830234684,
1140
+ "grad_norm": 3.875,
1141
+ "learning_rate": 4.705693156484652e-05,
1142
+ "loss": 0.2136,
1143
+ "step": 1620
1144
+ },
1145
+ {
1146
+ "epoch": 0.16559991872396626,
1147
+ "grad_norm": 1.1796875,
1148
+ "learning_rate": 4.7018870690800196e-05,
1149
+ "loss": 0.1471,
1150
+ "step": 1630
1151
+ },
1152
+ {
1153
+ "epoch": 0.16661586914558568,
1154
+ "grad_norm": 2.5,
1155
+ "learning_rate": 4.698058088343465e-05,
1156
+ "loss": 0.2308,
1157
+ "step": 1640
1158
+ },
1159
+ {
1160
+ "epoch": 0.16763181956720513,
1161
+ "grad_norm": 1.390625,
1162
+ "learning_rate": 4.6942062540854425e-05,
1163
+ "loss": 0.2456,
1164
+ "step": 1650
1165
+ },
1166
+ {
1167
+ "epoch": 0.16864776998882455,
1168
+ "grad_norm": 3.125,
1169
+ "learning_rate": 4.69033160635402e-05,
1170
+ "loss": 0.2654,
1171
+ "step": 1660
1172
+ },
1173
+ {
1174
+ "epoch": 0.16966372041044397,
1175
+ "grad_norm": 3.984375,
1176
+ "learning_rate": 4.6864341854344587e-05,
1177
+ "loss": 0.2226,
1178
+ "step": 1670
1179
+ },
1180
+ {
1181
+ "epoch": 0.1706796708320634,
1182
+ "grad_norm": 2.328125,
1183
+ "learning_rate": 4.682514031848795e-05,
1184
+ "loss": 0.2438,
1185
+ "step": 1680
1186
+ },
1187
+ {
1188
+ "epoch": 0.17169562125368282,
1189
+ "grad_norm": 3.078125,
1190
+ "learning_rate": 4.678571186355423e-05,
1191
+ "loss": 0.1889,
1192
+ "step": 1690
1193
+ },
1194
+ {
1195
+ "epoch": 0.17271157167530224,
1196
+ "grad_norm": 3.328125,
1197
+ "learning_rate": 4.6746056899486644e-05,
1198
+ "loss": 0.2117,
1199
+ "step": 1700
1200
+ },
1201
+ {
1202
+ "epoch": 0.17372752209692166,
1203
+ "grad_norm": 2.78125,
1204
+ "learning_rate": 4.67061758385835e-05,
1205
+ "loss": 0.1953,
1206
+ "step": 1710
1207
+ },
1208
+ {
1209
+ "epoch": 0.17474347251854108,
1210
+ "grad_norm": 3.09375,
1211
+ "learning_rate": 4.6666069095493816e-05,
1212
+ "loss": 0.1844,
1213
+ "step": 1720
1214
+ },
1215
+ {
1216
+ "epoch": 0.17575942294016053,
1217
+ "grad_norm": 3.234375,
1218
+ "learning_rate": 4.662573708721309e-05,
1219
+ "loss": 0.2774,
1220
+ "step": 1730
1221
+ },
1222
+ {
1223
+ "epoch": 0.17677537336177995,
1224
+ "grad_norm": 4.03125,
1225
+ "learning_rate": 4.658518023307894e-05,
1226
+ "loss": 0.2527,
1227
+ "step": 1740
1228
+ },
1229
+ {
1230
+ "epoch": 0.17779132378339937,
1231
+ "grad_norm": 3.21875,
1232
+ "learning_rate": 4.654439895476671e-05,
1233
+ "loss": 0.2164,
1234
+ "step": 1750
1235
+ },
1236
+ {
1237
+ "epoch": 0.1788072742050188,
1238
+ "grad_norm": 2.390625,
1239
+ "learning_rate": 4.6503393676285146e-05,
1240
+ "loss": 0.2424,
1241
+ "step": 1760
1242
+ },
1243
+ {
1244
+ "epoch": 0.17982322462663822,
1245
+ "grad_norm": 1.8359375,
1246
+ "learning_rate": 4.646216482397192e-05,
1247
+ "loss": 0.2428,
1248
+ "step": 1770
1249
+ },
1250
+ {
1251
+ "epoch": 0.18083917504825764,
1252
+ "grad_norm": 2.796875,
1253
+ "learning_rate": 4.6420712826489275e-05,
1254
+ "loss": 0.2155,
1255
+ "step": 1780
1256
+ },
1257
+ {
1258
+ "epoch": 0.18185512546987706,
1259
+ "grad_norm": 0.69921875,
1260
+ "learning_rate": 4.6379038114819485e-05,
1261
+ "loss": 0.1544,
1262
+ "step": 1790
1263
+ },
1264
+ {
1265
+ "epoch": 0.18287107589149648,
1266
+ "grad_norm": 3.40625,
1267
+ "learning_rate": 4.6337141122260444e-05,
1268
+ "loss": 0.2029,
1269
+ "step": 1800
1270
+ },
1271
+ {
1272
+ "epoch": 0.18388702631311593,
1273
+ "grad_norm": 2.359375,
1274
+ "learning_rate": 4.629502228442112e-05,
1275
+ "loss": 0.1489,
1276
+ "step": 1810
1277
+ },
1278
+ {
1279
+ "epoch": 0.18490297673473535,
1280
+ "grad_norm": 1.4453125,
1281
+ "learning_rate": 4.6252682039217045e-05,
1282
+ "loss": 0.2101,
1283
+ "step": 1820
1284
+ },
1285
+ {
1286
+ "epoch": 0.18591892715635477,
1287
+ "grad_norm": 2.71875,
1288
+ "learning_rate": 4.621012082686573e-05,
1289
+ "loss": 0.2076,
1290
+ "step": 1830
1291
+ },
1292
+ {
1293
+ "epoch": 0.1869348775779742,
1294
+ "grad_norm": 3.0625,
1295
+ "learning_rate": 4.616733908988216e-05,
1296
+ "loss": 0.2719,
1297
+ "step": 1840
1298
+ },
1299
+ {
1300
+ "epoch": 0.18795082799959362,
1301
+ "grad_norm": 1.953125,
1302
+ "learning_rate": 4.612433727307409e-05,
1303
+ "loss": 0.2105,
1304
+ "step": 1850
1305
+ },
1306
+ {
1307
+ "epoch": 0.18896677842121304,
1308
+ "grad_norm": 3.46875,
1309
+ "learning_rate": 4.608111582353751e-05,
1310
+ "loss": 0.1877,
1311
+ "step": 1860
1312
+ },
1313
+ {
1314
+ "epoch": 0.18998272884283246,
1315
+ "grad_norm": 2.546875,
1316
+ "learning_rate": 4.603767519065197e-05,
1317
+ "loss": 0.2238,
1318
+ "step": 1870
1319
+ },
1320
+ {
1321
+ "epoch": 0.19099867926445188,
1322
+ "grad_norm": 1.5703125,
1323
+ "learning_rate": 4.599401582607589e-05,
1324
+ "loss": 0.243,
1325
+ "step": 1880
1326
+ },
1327
+ {
1328
+ "epoch": 0.19201462968607133,
1329
+ "grad_norm": 2.4375,
1330
+ "learning_rate": 4.595013818374185e-05,
1331
+ "loss": 0.1867,
1332
+ "step": 1890
1333
+ },
1334
+ {
1335
+ "epoch": 0.19303058010769075,
1336
+ "grad_norm": 2.203125,
1337
+ "learning_rate": 4.5906042719851925e-05,
1338
+ "loss": 0.1994,
1339
+ "step": 1900
1340
+ },
1341
+ {
1342
+ "epoch": 0.19404653052931017,
1343
+ "grad_norm": 3.984375,
1344
+ "learning_rate": 4.586172989287291e-05,
1345
+ "loss": 0.1899,
1346
+ "step": 1910
1347
+ },
1348
+ {
1349
+ "epoch": 0.1950624809509296,
1350
+ "grad_norm": 2.6875,
1351
+ "learning_rate": 4.5817200163531534e-05,
1352
+ "loss": 0.2528,
1353
+ "step": 1920
1354
+ },
1355
+ {
1356
+ "epoch": 0.19607843137254902,
1357
+ "grad_norm": 2.71875,
1358
+ "learning_rate": 4.577245399480972e-05,
1359
+ "loss": 0.2336,
1360
+ "step": 1930
1361
+ },
1362
+ {
1363
+ "epoch": 0.19709438179416844,
1364
+ "grad_norm": 2.640625,
1365
+ "learning_rate": 4.5727491851939715e-05,
1366
+ "loss": 0.2204,
1367
+ "step": 1940
1368
+ },
1369
+ {
1370
+ "epoch": 0.19811033221578786,
1371
+ "grad_norm": 1.78125,
1372
+ "learning_rate": 4.568231420239929e-05,
1373
+ "loss": 0.1656,
1374
+ "step": 1950
1375
+ },
1376
+ {
1377
+ "epoch": 0.19912628263740728,
1378
+ "grad_norm": 3.15625,
1379
+ "learning_rate": 4.563692151590687e-05,
1380
+ "loss": 0.2105,
1381
+ "step": 1960
1382
+ },
1383
+ {
1384
+ "epoch": 0.20014223305902673,
1385
+ "grad_norm": 1.3671875,
1386
+ "learning_rate": 4.5591314264416666e-05,
1387
+ "loss": 0.1464,
1388
+ "step": 1970
1389
+ },
1390
+ {
1391
+ "epoch": 0.20115818348064615,
1392
+ "grad_norm": 4.25,
1393
+ "learning_rate": 4.554549292211371e-05,
1394
+ "loss": 0.2103,
1395
+ "step": 1980
1396
+ },
1397
+ {
1398
+ "epoch": 0.20217413390226557,
1399
+ "grad_norm": 2.984375,
1400
+ "learning_rate": 4.549945796540901e-05,
1401
+ "loss": 0.144,
1402
+ "step": 1990
1403
+ },
1404
+ {
1405
+ "epoch": 0.203190084323885,
1406
+ "grad_norm": 1.859375,
1407
+ "learning_rate": 4.545320987293453e-05,
1408
+ "loss": 0.1963,
1409
+ "step": 2000
1410
+ },
1411
+ {
1412
+ "epoch": 0.20420603474550442,
1413
+ "grad_norm": 1.078125,
1414
+ "learning_rate": 4.540674912553824e-05,
1415
+ "loss": 0.2115,
1416
+ "step": 2010
1417
+ },
1418
+ {
1419
+ "epoch": 0.20522198516712384,
1420
+ "grad_norm": 4.25,
1421
+ "learning_rate": 4.536007620627911e-05,
1422
+ "loss": 0.1682,
1423
+ "step": 2020
1424
+ },
1425
+ {
1426
+ "epoch": 0.20623793558874326,
1427
+ "grad_norm": 2.71875,
1428
+ "learning_rate": 4.531319160042212e-05,
1429
+ "loss": 0.1992,
1430
+ "step": 2030
1431
+ },
1432
+ {
1433
+ "epoch": 0.20725388601036268,
1434
+ "grad_norm": 1.2890625,
1435
+ "learning_rate": 4.5266095795433126e-05,
1436
+ "loss": 0.1134,
1437
+ "step": 2040
1438
+ },
1439
+ {
1440
+ "epoch": 0.20826983643198213,
1441
+ "grad_norm": 3.296875,
1442
+ "learning_rate": 4.5218789280973925e-05,
1443
+ "loss": 0.1474,
1444
+ "step": 2050
1445
+ },
1446
+ {
1447
+ "epoch": 0.20928578685360155,
1448
+ "grad_norm": 1.9375,
1449
+ "learning_rate": 4.5171272548897024e-05,
1450
+ "loss": 0.1955,
1451
+ "step": 2060
1452
+ },
1453
+ {
1454
+ "epoch": 0.21030173727522097,
1455
+ "grad_norm": 2.734375,
1456
+ "learning_rate": 4.512354609324063e-05,
1457
+ "loss": 0.2042,
1458
+ "step": 2070
1459
+ },
1460
+ {
1461
+ "epoch": 0.2113176876968404,
1462
+ "grad_norm": 2.921875,
1463
+ "learning_rate": 4.507561041022347e-05,
1464
+ "loss": 0.2174,
1465
+ "step": 2080
1466
+ },
1467
+ {
1468
+ "epoch": 0.21233363811845982,
1469
+ "grad_norm": 2.40625,
1470
+ "learning_rate": 4.502746599823963e-05,
1471
+ "loss": 0.2634,
1472
+ "step": 2090
1473
+ },
1474
+ {
1475
+ "epoch": 0.21334958854007924,
1476
+ "grad_norm": 1.71875,
1477
+ "learning_rate": 4.497911335785339e-05,
1478
+ "loss": 0.1884,
1479
+ "step": 2100
1480
+ },
1481
+ {
1482
+ "epoch": 0.21436553896169866,
1483
+ "grad_norm": 0.79296875,
1484
+ "learning_rate": 4.4930552991794e-05,
1485
+ "loss": 0.1872,
1486
+ "step": 2110
1487
+ },
1488
+ {
1489
+ "epoch": 0.21538148938331808,
1490
+ "grad_norm": 3.171875,
1491
+ "learning_rate": 4.4881785404950474e-05,
1492
+ "loss": 0.2233,
1493
+ "step": 2120
1494
+ },
1495
+ {
1496
+ "epoch": 0.21639743980493753,
1497
+ "grad_norm": 2.59375,
1498
+ "learning_rate": 4.483281110436631e-05,
1499
+ "loss": 0.2374,
1500
+ "step": 2130
1501
+ },
1502
+ {
1503
+ "epoch": 0.21741339022655695,
1504
+ "grad_norm": 3.328125,
1505
+ "learning_rate": 4.478363059923426e-05,
1506
+ "loss": 0.2545,
1507
+ "step": 2140
1508
+ },
1509
+ {
1510
+ "epoch": 0.21842934064817637,
1511
+ "grad_norm": 2.3125,
1512
+ "learning_rate": 4.4734244400891014e-05,
1513
+ "loss": 0.2063,
1514
+ "step": 2150
1515
+ },
1516
+ {
1517
+ "epoch": 0.2194452910697958,
1518
+ "grad_norm": 3.40625,
1519
+ "learning_rate": 4.4684653022811865e-05,
1520
+ "loss": 0.1219,
1521
+ "step": 2160
1522
+ },
1523
+ {
1524
+ "epoch": 0.22046124149141522,
1525
+ "grad_norm": 4.1875,
1526
+ "learning_rate": 4.463485698060541e-05,
1527
+ "loss": 0.2805,
1528
+ "step": 2170
1529
+ },
1530
+ {
1531
+ "epoch": 0.22147719191303464,
1532
+ "grad_norm": 2.3125,
1533
+ "learning_rate": 4.458485679200814e-05,
1534
+ "loss": 0.1998,
1535
+ "step": 2180
1536
+ },
1537
+ {
1538
+ "epoch": 0.22249314233465406,
1539
+ "grad_norm": 3.578125,
1540
+ "learning_rate": 4.453465297687912e-05,
1541
+ "loss": 0.2489,
1542
+ "step": 2190
1543
+ },
1544
+ {
1545
+ "epoch": 0.22350909275627348,
1546
+ "grad_norm": 2.59375,
1547
+ "learning_rate": 4.448424605719452e-05,
1548
+ "loss": 0.2731,
1549
+ "step": 2200
1550
+ },
1551
+ {
1552
+ "epoch": 0.22452504317789293,
1553
+ "grad_norm": 3.28125,
1554
+ "learning_rate": 4.443363655704224e-05,
1555
+ "loss": 0.2425,
1556
+ "step": 2210
1557
+ },
1558
+ {
1559
+ "epoch": 0.22554099359951235,
1560
+ "grad_norm": 2.78125,
1561
+ "learning_rate": 4.438282500261641e-05,
1562
+ "loss": 0.2938,
1563
+ "step": 2220
1564
+ },
1565
+ {
1566
+ "epoch": 0.22655694402113177,
1567
+ "grad_norm": 1.1953125,
1568
+ "learning_rate": 4.433181192221197e-05,
1569
+ "loss": 0.1728,
1570
+ "step": 2230
1571
+ },
1572
+ {
1573
+ "epoch": 0.2275728944427512,
1574
+ "grad_norm": 1.34375,
1575
+ "learning_rate": 4.4280597846219155e-05,
1576
+ "loss": 0.216,
1577
+ "step": 2240
1578
+ },
1579
+ {
1580
+ "epoch": 0.22858884486437062,
1581
+ "grad_norm": 1.8515625,
1582
+ "learning_rate": 4.422918330711796e-05,
1583
+ "loss": 0.1612,
1584
+ "step": 2250
1585
+ },
1586
+ {
1587
+ "epoch": 0.22960479528599004,
1588
+ "grad_norm": 1.90625,
1589
+ "learning_rate": 4.417756883947263e-05,
1590
+ "loss": 0.107,
1591
+ "step": 2260
1592
+ },
1593
+ {
1594
+ "epoch": 0.23062074570760946,
1595
+ "grad_norm": 3.375,
1596
+ "learning_rate": 4.412575497992611e-05,
1597
+ "loss": 0.1756,
1598
+ "step": 2270
1599
+ },
1600
+ {
1601
+ "epoch": 0.23163669612922888,
1602
+ "grad_norm": 4.375,
1603
+ "learning_rate": 4.407374226719445e-05,
1604
+ "loss": 0.234,
1605
+ "step": 2280
1606
+ },
1607
+ {
1608
+ "epoch": 0.23265264655084833,
1609
+ "grad_norm": 3.25,
1610
+ "learning_rate": 4.402153124206119e-05,
1611
+ "loss": 0.2144,
1612
+ "step": 2290
1613
+ },
1614
+ {
1615
+ "epoch": 0.23366859697246775,
1616
+ "grad_norm": 1.703125,
1617
+ "learning_rate": 4.396912244737173e-05,
1618
+ "loss": 0.1696,
1619
+ "step": 2300
1620
+ },
1621
+ {
1622
+ "epoch": 0.23468454739408717,
1623
+ "grad_norm": 2.84375,
1624
+ "learning_rate": 4.391651642802778e-05,
1625
+ "loss": 0.2506,
1626
+ "step": 2310
1627
+ },
1628
+ {
1629
+ "epoch": 0.2357004978157066,
1630
+ "grad_norm": 4.5,
1631
+ "learning_rate": 4.386371373098155e-05,
1632
+ "loss": 0.1686,
1633
+ "step": 2320
1634
+ },
1635
+ {
1636
+ "epoch": 0.23671644823732602,
1637
+ "grad_norm": 2.515625,
1638
+ "learning_rate": 4.381071490523018e-05,
1639
+ "loss": 0.2403,
1640
+ "step": 2330
1641
+ },
1642
+ {
1643
+ "epoch": 0.23773239865894544,
1644
+ "grad_norm": 4.4375,
1645
+ "learning_rate": 4.3757520501809955e-05,
1646
+ "loss": 0.1611,
1647
+ "step": 2340
1648
+ },
1649
+ {
1650
+ "epoch": 0.23874834908056486,
1651
+ "grad_norm": 1.609375,
1652
+ "learning_rate": 4.370413107379065e-05,
1653
+ "loss": 0.1698,
1654
+ "step": 2350
1655
+ },
1656
+ {
1657
+ "epoch": 0.23976429950218428,
1658
+ "grad_norm": 4.96875,
1659
+ "learning_rate": 4.36505471762697e-05,
1660
+ "loss": 0.1928,
1661
+ "step": 2360
1662
+ },
1663
+ {
1664
+ "epoch": 0.24078024992380373,
1665
+ "grad_norm": 0.8984375,
1666
+ "learning_rate": 4.3596769366366474e-05,
1667
+ "loss": 0.2035,
1668
+ "step": 2370
1669
+ },
1670
+ {
1671
+ "epoch": 0.24179620034542315,
1672
+ "grad_norm": 5.75,
1673
+ "learning_rate": 4.354279820321649e-05,
1674
+ "loss": 0.16,
1675
+ "step": 2380
1676
+ },
1677
+ {
1678
+ "epoch": 0.24281215076704257,
1679
+ "grad_norm": 1.9453125,
1680
+ "learning_rate": 4.34886342479656e-05,
1681
+ "loss": 0.1851,
1682
+ "step": 2390
1683
+ },
1684
+ {
1685
+ "epoch": 0.243828101188662,
1686
+ "grad_norm": 1.015625,
1687
+ "learning_rate": 4.34342780637641e-05,
1688
+ "loss": 0.1726,
1689
+ "step": 2400
1690
+ },
1691
+ {
1692
+ "epoch": 0.24484405161028142,
1693
+ "grad_norm": 4.59375,
1694
+ "learning_rate": 4.337973021576095e-05,
1695
+ "loss": 0.2847,
1696
+ "step": 2410
1697
+ },
1698
+ {
1699
+ "epoch": 0.24586000203190084,
1700
+ "grad_norm": 1.03125,
1701
+ "learning_rate": 4.3324991271097846e-05,
1702
+ "loss": 0.2528,
1703
+ "step": 2420
1704
+ },
1705
+ {
1706
+ "epoch": 0.24687595245352026,
1707
+ "grad_norm": 2.1875,
1708
+ "learning_rate": 4.3270061798903374e-05,
1709
+ "loss": 0.1573,
1710
+ "step": 2430
1711
+ },
1712
+ {
1713
+ "epoch": 0.24789190287513968,
1714
+ "grad_norm": 0.98046875,
1715
+ "learning_rate": 4.321494237028701e-05,
1716
+ "loss": 0.1703,
1717
+ "step": 2440
1718
+ },
1719
+ {
1720
+ "epoch": 0.24890785329675913,
1721
+ "grad_norm": 3.8125,
1722
+ "learning_rate": 4.31596335583333e-05,
1723
+ "loss": 0.2613,
1724
+ "step": 2450
1725
+ },
1726
+ {
1727
+ "epoch": 0.24992380371837855,
1728
+ "grad_norm": 4.0625,
1729
+ "learning_rate": 4.310413593809579e-05,
1730
+ "loss": 0.22,
1731
+ "step": 2460
1732
+ },
1733
+ {
1734
+ "epoch": 0.250939754139998,
1735
+ "grad_norm": 3.15625,
1736
+ "learning_rate": 4.304845008659108e-05,
1737
+ "loss": 0.1263,
1738
+ "step": 2470
1739
+ },
1740
+ {
1741
+ "epoch": 0.25195570456161737,
1742
+ "grad_norm": 3.046875,
1743
+ "learning_rate": 4.2992576582792895e-05,
1744
+ "loss": 0.1639,
1745
+ "step": 2480
1746
+ },
1747
+ {
1748
+ "epoch": 0.2529716549832368,
1749
+ "grad_norm": 9.8125,
1750
+ "learning_rate": 4.293651600762595e-05,
1751
+ "loss": 0.2681,
1752
+ "step": 2490
1753
+ },
1754
+ {
1755
+ "epoch": 0.25398760540485626,
1756
+ "grad_norm": 3.734375,
1757
+ "learning_rate": 4.288026894395999e-05,
1758
+ "loss": 0.2292,
1759
+ "step": 2500
1760
+ }
1761
+ ],
1762
+ "logging_steps": 10,
1763
+ "max_steps": 9843,
1764
+ "num_input_tokens_seen": 0,
1765
+ "num_train_epochs": 1,
1766
+ "save_steps": 500,
1767
+ "total_flos": 0.0,
1768
+ "train_batch_size": 7,
1769
+ "trial_name": null,
1770
+ "trial_params": null
1771
+ }
checkpoints/checkpoint-2500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:678fdb0261b74cfbf22c48f7675247cc333daf413df3ae34c37afbda83a36025
3
+ size 4984
checkpoints/checkpoint-3000/.DS_Store ADDED
Binary file (6.15 kB). View file
 
checkpoints/checkpoint-3000/adapter_config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "openbmb/MiniCPM-2B-dpo-bf16",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "v_proj",
24
+ "q_proj"
25
+ ],
26
+ "task_type": "CAUSAL_LM",
27
+ "use_dora": false,
28
+ "use_rslora": false
29
+ }