AhmedSSoliman commited on
Commit
9d9026f
1 Parent(s): a211b63

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +1681 -0
trainer_state.json ADDED
@@ -0,0 +1,1681 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.6102247834205627,
3
+ "best_model_checkpoint": "./luke-marian-training-1/checkpoint-123440",
4
+ "epoch": 10.0,
5
+ "global_step": 123440,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.04,
12
+ "learning_rate": 8.052495139338951e-07,
13
+ "loss": 0.7856,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.08,
18
+ "learning_rate": 1.6153596889176928e-06,
19
+ "loss": 0.748,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.12,
24
+ "learning_rate": 2.4254698639014908e-06,
25
+ "loss": 0.7809,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.16,
30
+ "learning_rate": 3.2355800388852886e-06,
31
+ "loss": 0.7923,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.2,
36
+ "learning_rate": 4.044069993519119e-06,
37
+ "loss": 0.7799,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.24,
42
+ "learning_rate": 4.854180168502917e-06,
43
+ "loss": 0.8085,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.28,
48
+ "learning_rate": 5.664290343486715e-06,
49
+ "loss": 0.7759,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.32,
54
+ "learning_rate": 6.474400518470513e-06,
55
+ "loss": 0.7833,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.36,
60
+ "learning_rate": 7.2828904731043424e-06,
61
+ "loss": 0.7804,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.41,
66
+ "learning_rate": 8.09300064808814e-06,
67
+ "loss": 0.8169,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.45,
72
+ "learning_rate": 8.903110823071939e-06,
73
+ "loss": 0.8058,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.49,
78
+ "learning_rate": 9.713220998055737e-06,
79
+ "loss": 0.8265,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.53,
84
+ "learning_rate": 9.972541528805813e-06,
85
+ "loss": 0.7904,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.57,
90
+ "learning_rate": 9.929904151175087e-06,
91
+ "loss": 0.7829,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.61,
96
+ "learning_rate": 9.88726677354436e-06,
97
+ "loss": 0.8126,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.65,
102
+ "learning_rate": 9.844629395913635e-06,
103
+ "loss": 0.8133,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.69,
108
+ "learning_rate": 9.802077293038171e-06,
109
+ "loss": 0.7928,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.73,
114
+ "learning_rate": 9.759525190162705e-06,
115
+ "loss": 0.8564,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.77,
120
+ "learning_rate": 9.716887812531979e-06,
121
+ "loss": 0.7624,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.81,
126
+ "learning_rate": 9.674250434901252e-06,
127
+ "loss": 0.7662,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.85,
132
+ "learning_rate": 9.631613057270527e-06,
133
+ "loss": 0.8363,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.89,
138
+ "learning_rate": 9.5889756796398e-06,
139
+ "loss": 0.7793,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.93,
144
+ "learning_rate": 9.546338302009073e-06,
145
+ "loss": 0.7561,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.97,
150
+ "learning_rate": 9.50378619913361e-06,
151
+ "loss": 0.774,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 1.0,
156
+ "eval_BLEU": 32.0811169920738,
157
+ "eval_BLEU-Bigram-Precision": 37.1092045950269,
158
+ "eval_BLEU-Trigram-Precision": 26.487573476332887,
159
+ "eval_BLEU-Unigram-Precision": 56.60325589903054,
160
+ "eval_ROUGE-2": 25.923225512351976,
161
+ "eval_ROUGE-L": 40.443599801980355,
162
+ "eval_Sacre-Bigram-Precision": 34.86367988711385,
163
+ "eval_Sacre-Trigram-Precision": 26.20122541115769,
164
+ "eval_Sacre-Unigram-Precision": 54.42125237191651,
165
+ "eval_SacreBLEU": 30.767354925943692,
166
+ "eval_loss": 0.9131098985671997,
167
+ "eval_runtime": 174.9979,
168
+ "eval_samples_per_second": 7.069,
169
+ "eval_steps_per_second": 7.069,
170
+ "step": 12344
171
+ },
172
+ {
173
+ "epoch": 1.01,
174
+ "learning_rate": 9.461148821502884e-06,
175
+ "loss": 0.7846,
176
+ "step": 12500
177
+ },
178
+ {
179
+ "epoch": 1.05,
180
+ "learning_rate": 9.418511443872157e-06,
181
+ "loss": 0.7368,
182
+ "step": 13000
183
+ },
184
+ {
185
+ "epoch": 1.09,
186
+ "learning_rate": 9.37587406624143e-06,
187
+ "loss": 0.7279,
188
+ "step": 13500
189
+ },
190
+ {
191
+ "epoch": 1.13,
192
+ "learning_rate": 9.333236688610704e-06,
193
+ "loss": 0.7253,
194
+ "step": 14000
195
+ },
196
+ {
197
+ "epoch": 1.17,
198
+ "learning_rate": 9.290599310979979e-06,
199
+ "loss": 0.7259,
200
+ "step": 14500
201
+ },
202
+ {
203
+ "epoch": 1.22,
204
+ "learning_rate": 9.247961933349252e-06,
205
+ "loss": 0.7432,
206
+ "step": 15000
207
+ },
208
+ {
209
+ "epoch": 1.26,
210
+ "learning_rate": 9.205324555718525e-06,
211
+ "loss": 0.7437,
212
+ "step": 15500
213
+ },
214
+ {
215
+ "epoch": 1.3,
216
+ "learning_rate": 9.1626871780878e-06,
217
+ "loss": 0.6641,
218
+ "step": 16000
219
+ },
220
+ {
221
+ "epoch": 1.34,
222
+ "learning_rate": 9.120135075212335e-06,
223
+ "loss": 0.7239,
224
+ "step": 16500
225
+ },
226
+ {
227
+ "epoch": 1.38,
228
+ "learning_rate": 9.077497697581608e-06,
229
+ "loss": 0.7007,
230
+ "step": 17000
231
+ },
232
+ {
233
+ "epoch": 1.42,
234
+ "learning_rate": 9.034860319950883e-06,
235
+ "loss": 0.7107,
236
+ "step": 17500
237
+ },
238
+ {
239
+ "epoch": 1.46,
240
+ "learning_rate": 8.992222942320156e-06,
241
+ "loss": 0.7033,
242
+ "step": 18000
243
+ },
244
+ {
245
+ "epoch": 1.5,
246
+ "learning_rate": 8.949670839444692e-06,
247
+ "loss": 0.6932,
248
+ "step": 18500
249
+ },
250
+ {
251
+ "epoch": 1.54,
252
+ "learning_rate": 8.907118736569226e-06,
253
+ "loss": 0.7351,
254
+ "step": 19000
255
+ },
256
+ {
257
+ "epoch": 1.58,
258
+ "learning_rate": 8.8644813589385e-06,
259
+ "loss": 0.7511,
260
+ "step": 19500
261
+ },
262
+ {
263
+ "epoch": 1.62,
264
+ "learning_rate": 8.821843981307775e-06,
265
+ "loss": 0.6994,
266
+ "step": 20000
267
+ },
268
+ {
269
+ "epoch": 1.66,
270
+ "learning_rate": 8.779206603677048e-06,
271
+ "loss": 0.6792,
272
+ "step": 20500
273
+ },
274
+ {
275
+ "epoch": 1.7,
276
+ "learning_rate": 8.736569226046321e-06,
277
+ "loss": 0.6701,
278
+ "step": 21000
279
+ },
280
+ {
281
+ "epoch": 1.74,
282
+ "learning_rate": 8.693931848415596e-06,
283
+ "loss": 0.6885,
284
+ "step": 21500
285
+ },
286
+ {
287
+ "epoch": 1.78,
288
+ "learning_rate": 8.65129447078487e-06,
289
+ "loss": 0.6796,
290
+ "step": 22000
291
+ },
292
+ {
293
+ "epoch": 1.82,
294
+ "learning_rate": 8.608657093154144e-06,
295
+ "loss": 0.6812,
296
+ "step": 22500
297
+ },
298
+ {
299
+ "epoch": 1.86,
300
+ "learning_rate": 8.566104990278679e-06,
301
+ "loss": 0.704,
302
+ "step": 23000
303
+ },
304
+ {
305
+ "epoch": 1.9,
306
+ "learning_rate": 8.523467612647952e-06,
307
+ "loss": 0.6443,
308
+ "step": 23500
309
+ },
310
+ {
311
+ "epoch": 1.94,
312
+ "learning_rate": 8.480830235017227e-06,
313
+ "loss": 0.6513,
314
+ "step": 24000
315
+ },
316
+ {
317
+ "epoch": 1.98,
318
+ "learning_rate": 8.4381928573865e-06,
319
+ "loss": 0.6532,
320
+ "step": 24500
321
+ },
322
+ {
323
+ "epoch": 2.0,
324
+ "eval_BLEU": 34.14564345469897,
325
+ "eval_BLEU-Bigram-Precision": 42.337954289996254,
326
+ "eval_BLEU-Trigram-Precision": 32.169876203576344,
327
+ "eval_BLEU-Unigram-Precision": 61.27510040160643,
328
+ "eval_ROUGE-2": 28.971835230430752,
329
+ "eval_ROUGE-L": 43.230417531342944,
330
+ "eval_Sacre-Bigram-Precision": 40.19043401240035,
331
+ "eval_Sacre-Trigram-Precision": 31.907054139181078,
332
+ "eval_Sacre-Unigram-Precision": 59.04875395057251,
333
+ "eval_SacreBLEU": 33.00279853845508,
334
+ "eval_loss": 0.8338403701782227,
335
+ "eval_runtime": 167.0515,
336
+ "eval_samples_per_second": 7.405,
337
+ "eval_steps_per_second": 7.405,
338
+ "step": 24688
339
+ },
340
+ {
341
+ "epoch": 2.03,
342
+ "learning_rate": 8.395555479755773e-06,
343
+ "loss": 0.6209,
344
+ "step": 25000
345
+ },
346
+ {
347
+ "epoch": 2.07,
348
+ "learning_rate": 8.35300337688031e-06,
349
+ "loss": 0.6246,
350
+ "step": 25500
351
+ },
352
+ {
353
+ "epoch": 2.11,
354
+ "learning_rate": 8.310365999249583e-06,
355
+ "loss": 0.6061,
356
+ "step": 26000
357
+ },
358
+ {
359
+ "epoch": 2.15,
360
+ "learning_rate": 8.267728621618858e-06,
361
+ "loss": 0.6347,
362
+ "step": 26500
363
+ },
364
+ {
365
+ "epoch": 2.19,
366
+ "learning_rate": 8.22509124398813e-06,
367
+ "loss": 0.609,
368
+ "step": 27000
369
+ },
370
+ {
371
+ "epoch": 2.23,
372
+ "learning_rate": 8.182539141112665e-06,
373
+ "loss": 0.6059,
374
+ "step": 27500
375
+ },
376
+ {
377
+ "epoch": 2.27,
378
+ "learning_rate": 8.13990176348194e-06,
379
+ "loss": 0.6301,
380
+ "step": 28000
381
+ },
382
+ {
383
+ "epoch": 2.31,
384
+ "learning_rate": 8.097264385851213e-06,
385
+ "loss": 0.5859,
386
+ "step": 28500
387
+ },
388
+ {
389
+ "epoch": 2.35,
390
+ "learning_rate": 8.054627008220487e-06,
391
+ "loss": 0.6133,
392
+ "step": 29000
393
+ },
394
+ {
395
+ "epoch": 2.39,
396
+ "learning_rate": 8.012074905345023e-06,
397
+ "loss": 0.5813,
398
+ "step": 29500
399
+ },
400
+ {
401
+ "epoch": 2.43,
402
+ "learning_rate": 7.969437527714296e-06,
403
+ "loss": 0.6049,
404
+ "step": 30000
405
+ },
406
+ {
407
+ "epoch": 2.47,
408
+ "learning_rate": 7.92680015008357e-06,
409
+ "loss": 0.5982,
410
+ "step": 30500
411
+ },
412
+ {
413
+ "epoch": 2.51,
414
+ "learning_rate": 7.884162772452844e-06,
415
+ "loss": 0.5807,
416
+ "step": 31000
417
+ },
418
+ {
419
+ "epoch": 2.55,
420
+ "learning_rate": 7.841525394822117e-06,
421
+ "loss": 0.5809,
422
+ "step": 31500
423
+ },
424
+ {
425
+ "epoch": 2.59,
426
+ "learning_rate": 7.798973291946653e-06,
427
+ "loss": 0.5984,
428
+ "step": 32000
429
+ },
430
+ {
431
+ "epoch": 2.63,
432
+ "learning_rate": 7.756335914315927e-06,
433
+ "loss": 0.6308,
434
+ "step": 32500
435
+ },
436
+ {
437
+ "epoch": 2.67,
438
+ "learning_rate": 7.7136985366852e-06,
439
+ "loss": 0.5964,
440
+ "step": 33000
441
+ },
442
+ {
443
+ "epoch": 2.71,
444
+ "learning_rate": 7.671061159054473e-06,
445
+ "loss": 0.5783,
446
+ "step": 33500
447
+ },
448
+ {
449
+ "epoch": 2.75,
450
+ "learning_rate": 7.62850905617901e-06,
451
+ "loss": 0.5914,
452
+ "step": 34000
453
+ },
454
+ {
455
+ "epoch": 2.79,
456
+ "learning_rate": 7.585871678548283e-06,
457
+ "loss": 0.5716,
458
+ "step": 34500
459
+ },
460
+ {
461
+ "epoch": 2.84,
462
+ "learning_rate": 7.543319575672818e-06,
463
+ "loss": 0.613,
464
+ "step": 35000
465
+ },
466
+ {
467
+ "epoch": 2.88,
468
+ "learning_rate": 7.500682198042092e-06,
469
+ "loss": 0.561,
470
+ "step": 35500
471
+ },
472
+ {
473
+ "epoch": 2.92,
474
+ "learning_rate": 7.458044820411367e-06,
475
+ "loss": 0.5731,
476
+ "step": 36000
477
+ },
478
+ {
479
+ "epoch": 2.96,
480
+ "learning_rate": 7.41540744278064e-06,
481
+ "loss": 0.565,
482
+ "step": 36500
483
+ },
484
+ {
485
+ "epoch": 3.0,
486
+ "learning_rate": 7.372855339905175e-06,
487
+ "loss": 0.6118,
488
+ "step": 37000
489
+ },
490
+ {
491
+ "epoch": 3.0,
492
+ "eval_BLEU": 39.5916929052258,
493
+ "eval_BLEU-Bigram-Precision": 44.30874450415242,
494
+ "eval_BLEU-Trigram-Precision": 34.66957832891385,
495
+ "eval_BLEU-Unigram-Precision": 61.68977749113189,
496
+ "eval_ROUGE-2": 33.461963127391705,
497
+ "eval_ROUGE-L": 46.65653806735692,
498
+ "eval_Sacre-Bigram-Precision": 42.369437106279214,
499
+ "eval_Sacre-Trigram-Precision": 34.54220198406245,
500
+ "eval_Sacre-Unigram-Precision": 59.68165957650208,
501
+ "eval_SacreBLEU": 38.27100461755832,
502
+ "eval_loss": 0.7759082317352295,
503
+ "eval_runtime": 173.3948,
504
+ "eval_samples_per_second": 7.134,
505
+ "eval_steps_per_second": 7.134,
506
+ "step": 37032
507
+ },
508
+ {
509
+ "epoch": 3.04,
510
+ "learning_rate": 7.33030323702971e-06,
511
+ "loss": 0.4871,
512
+ "step": 37500
513
+ },
514
+ {
515
+ "epoch": 3.08,
516
+ "learning_rate": 7.287665859398984e-06,
517
+ "loss": 0.4991,
518
+ "step": 38000
519
+ },
520
+ {
521
+ "epoch": 3.12,
522
+ "learning_rate": 7.245028481768258e-06,
523
+ "loss": 0.5436,
524
+ "step": 38500
525
+ },
526
+ {
527
+ "epoch": 3.16,
528
+ "learning_rate": 7.202391104137532e-06,
529
+ "loss": 0.5165,
530
+ "step": 39000
531
+ },
532
+ {
533
+ "epoch": 3.2,
534
+ "learning_rate": 7.159753726506805e-06,
535
+ "loss": 0.516,
536
+ "step": 39500
537
+ },
538
+ {
539
+ "epoch": 3.24,
540
+ "learning_rate": 7.11711634887608e-06,
541
+ "loss": 0.5173,
542
+ "step": 40000
543
+ },
544
+ {
545
+ "epoch": 3.28,
546
+ "learning_rate": 7.074478971245353e-06,
547
+ "loss": 0.5239,
548
+ "step": 40500
549
+ },
550
+ {
551
+ "epoch": 3.32,
552
+ "learning_rate": 7.031841593614626e-06,
553
+ "loss": 0.5112,
554
+ "step": 41000
555
+ },
556
+ {
557
+ "epoch": 3.36,
558
+ "learning_rate": 6.9892042159839e-06,
559
+ "loss": 0.516,
560
+ "step": 41500
561
+ },
562
+ {
563
+ "epoch": 3.4,
564
+ "learning_rate": 6.9465668383531745e-06,
565
+ "loss": 0.5384,
566
+ "step": 42000
567
+ },
568
+ {
569
+ "epoch": 3.44,
570
+ "learning_rate": 6.903929460722449e-06,
571
+ "loss": 0.5194,
572
+ "step": 42500
573
+ },
574
+ {
575
+ "epoch": 3.48,
576
+ "learning_rate": 6.861292083091722e-06,
577
+ "loss": 0.5006,
578
+ "step": 43000
579
+ },
580
+ {
581
+ "epoch": 3.52,
582
+ "learning_rate": 6.818739980216257e-06,
583
+ "loss": 0.5362,
584
+ "step": 43500
585
+ },
586
+ {
587
+ "epoch": 3.56,
588
+ "learning_rate": 6.776102602585531e-06,
589
+ "loss": 0.5505,
590
+ "step": 44000
591
+ },
592
+ {
593
+ "epoch": 3.6,
594
+ "learning_rate": 6.733465224954805e-06,
595
+ "loss": 0.5247,
596
+ "step": 44500
597
+ },
598
+ {
599
+ "epoch": 3.65,
600
+ "learning_rate": 6.6908278473240785e-06,
601
+ "loss": 0.5104,
602
+ "step": 45000
603
+ },
604
+ {
605
+ "epoch": 3.69,
606
+ "learning_rate": 6.648275744448614e-06,
607
+ "loss": 0.516,
608
+ "step": 45500
609
+ },
610
+ {
611
+ "epoch": 3.73,
612
+ "learning_rate": 6.605638366817888e-06,
613
+ "loss": 0.5097,
614
+ "step": 46000
615
+ },
616
+ {
617
+ "epoch": 3.77,
618
+ "learning_rate": 6.563000989187162e-06,
619
+ "loss": 0.5147,
620
+ "step": 46500
621
+ },
622
+ {
623
+ "epoch": 3.81,
624
+ "learning_rate": 6.520363611556435e-06,
625
+ "loss": 0.4988,
626
+ "step": 47000
627
+ },
628
+ {
629
+ "epoch": 3.85,
630
+ "learning_rate": 6.47781150868097e-06,
631
+ "loss": 0.5143,
632
+ "step": 47500
633
+ },
634
+ {
635
+ "epoch": 3.89,
636
+ "learning_rate": 6.435174131050244e-06,
637
+ "loss": 0.5146,
638
+ "step": 48000
639
+ },
640
+ {
641
+ "epoch": 3.93,
642
+ "learning_rate": 6.3925367534195185e-06,
643
+ "loss": 0.4949,
644
+ "step": 48500
645
+ },
646
+ {
647
+ "epoch": 3.97,
648
+ "learning_rate": 6.349899375788792e-06,
649
+ "loss": 0.5327,
650
+ "step": 49000
651
+ },
652
+ {
653
+ "epoch": 4.0,
654
+ "eval_BLEU": 41.583977153720554,
655
+ "eval_BLEU-Bigram-Precision": 47.79210232463709,
656
+ "eval_BLEU-Trigram-Precision": 38.44570809639859,
657
+ "eval_BLEU-Unigram-Precision": 64.70223028797936,
658
+ "eval_ROUGE-2": 35.92233849645701,
659
+ "eval_ROUGE-L": 48.89087193511837,
660
+ "eval_Sacre-Bigram-Precision": 46.017043349388665,
661
+ "eval_Sacre-Trigram-Precision": 38.29859537834164,
662
+ "eval_Sacre-Unigram-Precision": 62.97069051167412,
663
+ "eval_SacreBLEU": 40.202754965999326,
664
+ "eval_loss": 0.734529972076416,
665
+ "eval_runtime": 168.3041,
666
+ "eval_samples_per_second": 7.35,
667
+ "eval_steps_per_second": 7.35,
668
+ "step": 49376
669
+ },
670
+ {
671
+ "epoch": 4.01,
672
+ "learning_rate": 6.307347272913327e-06,
673
+ "loss": 0.515,
674
+ "step": 49500
675
+ },
676
+ {
677
+ "epoch": 4.05,
678
+ "learning_rate": 6.264709895282601e-06,
679
+ "loss": 0.4603,
680
+ "step": 50000
681
+ },
682
+ {
683
+ "epoch": 4.09,
684
+ "learning_rate": 6.222072517651875e-06,
685
+ "loss": 0.4565,
686
+ "step": 50500
687
+ },
688
+ {
689
+ "epoch": 4.13,
690
+ "learning_rate": 6.179435140021148e-06,
691
+ "loss": 0.4557,
692
+ "step": 51000
693
+ },
694
+ {
695
+ "epoch": 4.17,
696
+ "learning_rate": 6.136797762390423e-06,
697
+ "loss": 0.4586,
698
+ "step": 51500
699
+ },
700
+ {
701
+ "epoch": 4.21,
702
+ "learning_rate": 6.094245659514958e-06,
703
+ "loss": 0.4295,
704
+ "step": 52000
705
+ },
706
+ {
707
+ "epoch": 4.25,
708
+ "learning_rate": 6.051608281884232e-06,
709
+ "loss": 0.4774,
710
+ "step": 52500
711
+ },
712
+ {
713
+ "epoch": 4.29,
714
+ "learning_rate": 6.009056179008766e-06,
715
+ "loss": 0.4555,
716
+ "step": 53000
717
+ },
718
+ {
719
+ "epoch": 4.33,
720
+ "learning_rate": 5.966418801378041e-06,
721
+ "loss": 0.4448,
722
+ "step": 53500
723
+ },
724
+ {
725
+ "epoch": 4.37,
726
+ "learning_rate": 5.923781423747314e-06,
727
+ "loss": 0.4481,
728
+ "step": 54000
729
+ },
730
+ {
731
+ "epoch": 4.42,
732
+ "learning_rate": 5.881144046116588e-06,
733
+ "loss": 0.4288,
734
+ "step": 54500
735
+ },
736
+ {
737
+ "epoch": 4.46,
738
+ "learning_rate": 5.838506668485862e-06,
739
+ "loss": 0.4459,
740
+ "step": 55000
741
+ },
742
+ {
743
+ "epoch": 4.5,
744
+ "learning_rate": 5.7958692908551366e-06,
745
+ "loss": 0.4616,
746
+ "step": 55500
747
+ },
748
+ {
749
+ "epoch": 4.54,
750
+ "learning_rate": 5.75323191322441e-06,
751
+ "loss": 0.4775,
752
+ "step": 56000
753
+ },
754
+ {
755
+ "epoch": 4.58,
756
+ "learning_rate": 5.710594535593683e-06,
757
+ "loss": 0.4414,
758
+ "step": 56500
759
+ },
760
+ {
761
+ "epoch": 4.62,
762
+ "learning_rate": 5.668042432718218e-06,
763
+ "loss": 0.483,
764
+ "step": 57000
765
+ },
766
+ {
767
+ "epoch": 4.66,
768
+ "learning_rate": 5.625405055087493e-06,
769
+ "loss": 0.4671,
770
+ "step": 57500
771
+ },
772
+ {
773
+ "epoch": 4.7,
774
+ "learning_rate": 5.5827676774567664e-06,
775
+ "loss": 0.4696,
776
+ "step": 58000
777
+ },
778
+ {
779
+ "epoch": 4.74,
780
+ "learning_rate": 5.54013029982604e-06,
781
+ "loss": 0.4628,
782
+ "step": 58500
783
+ },
784
+ {
785
+ "epoch": 4.78,
786
+ "learning_rate": 5.497578196950575e-06,
787
+ "loss": 0.4591,
788
+ "step": 59000
789
+ },
790
+ {
791
+ "epoch": 4.82,
792
+ "learning_rate": 5.45494081931985e-06,
793
+ "loss": 0.4362,
794
+ "step": 59500
795
+ },
796
+ {
797
+ "epoch": 4.86,
798
+ "learning_rate": 5.412303441689123e-06,
799
+ "loss": 0.4767,
800
+ "step": 60000
801
+ },
802
+ {
803
+ "epoch": 4.9,
804
+ "learning_rate": 5.369666064058396e-06,
805
+ "loss": 0.4642,
806
+ "step": 60500
807
+ },
808
+ {
809
+ "epoch": 4.94,
810
+ "learning_rate": 5.3271139611829315e-06,
811
+ "loss": 0.4348,
812
+ "step": 61000
813
+ },
814
+ {
815
+ "epoch": 4.98,
816
+ "learning_rate": 5.2844765835522064e-06,
817
+ "loss": 0.4365,
818
+ "step": 61500
819
+ },
820
+ {
821
+ "epoch": 5.0,
822
+ "eval_BLEU": 43.53907049710156,
823
+ "eval_BLEU-Bigram-Precision": 49.90573248407643,
824
+ "eval_BLEU-Trigram-Precision": 40.825538394605175,
825
+ "eval_BLEU-Unigram-Precision": 66.37906241012367,
826
+ "eval_ROUGE-2": 38.22782051808487,
827
+ "eval_ROUGE-L": 51.36307879612767,
828
+ "eval_Sacre-Bigram-Precision": 48.25423818887177,
829
+ "eval_Sacre-Trigram-Precision": 40.784982935153586,
830
+ "eval_Sacre-Unigram-Precision": 64.79006681958711,
831
+ "eval_SacreBLEU": 42.18614400441328,
832
+ "eval_loss": 0.7037733197212219,
833
+ "eval_runtime": 168.6409,
834
+ "eval_samples_per_second": 7.335,
835
+ "eval_steps_per_second": 7.335,
836
+ "step": 61720
837
+ },
838
+ {
839
+ "epoch": 5.02,
840
+ "learning_rate": 5.241924480676741e-06,
841
+ "loss": 0.4345,
842
+ "step": 62000
843
+ },
844
+ {
845
+ "epoch": 5.06,
846
+ "learning_rate": 5.199372377801276e-06,
847
+ "loss": 0.4124,
848
+ "step": 62500
849
+ },
850
+ {
851
+ "epoch": 5.1,
852
+ "learning_rate": 5.156735000170549e-06,
853
+ "loss": 0.3992,
854
+ "step": 63000
855
+ },
856
+ {
857
+ "epoch": 5.14,
858
+ "learning_rate": 5.114097622539824e-06,
859
+ "loss": 0.3928,
860
+ "step": 63500
861
+ },
862
+ {
863
+ "epoch": 5.18,
864
+ "learning_rate": 5.0714602449090974e-06,
865
+ "loss": 0.4428,
866
+ "step": 64000
867
+ },
868
+ {
869
+ "epoch": 5.23,
870
+ "learning_rate": 5.0288228672783715e-06,
871
+ "loss": 0.4003,
872
+ "step": 64500
873
+ },
874
+ {
875
+ "epoch": 5.27,
876
+ "learning_rate": 4.986185489647646e-06,
877
+ "loss": 0.4077,
878
+ "step": 65000
879
+ },
880
+ {
881
+ "epoch": 5.31,
882
+ "learning_rate": 4.943548112016919e-06,
883
+ "loss": 0.4076,
884
+ "step": 65500
885
+ },
886
+ {
887
+ "epoch": 5.35,
888
+ "learning_rate": 4.900910734386193e-06,
889
+ "loss": 0.4138,
890
+ "step": 66000
891
+ },
892
+ {
893
+ "epoch": 5.39,
894
+ "learning_rate": 4.858273356755466e-06,
895
+ "loss": 0.3901,
896
+ "step": 66500
897
+ },
898
+ {
899
+ "epoch": 5.43,
900
+ "learning_rate": 4.81563597912474e-06,
901
+ "loss": 0.4412,
902
+ "step": 67000
903
+ },
904
+ {
905
+ "epoch": 5.47,
906
+ "learning_rate": 4.7729986014940135e-06,
907
+ "loss": 0.4128,
908
+ "step": 67500
909
+ },
910
+ {
911
+ "epoch": 5.51,
912
+ "learning_rate": 4.730361223863288e-06,
913
+ "loss": 0.4382,
914
+ "step": 68000
915
+ },
916
+ {
917
+ "epoch": 5.55,
918
+ "learning_rate": 4.687809120987823e-06,
919
+ "loss": 0.3855,
920
+ "step": 68500
921
+ },
922
+ {
923
+ "epoch": 5.59,
924
+ "learning_rate": 4.645171743357097e-06,
925
+ "loss": 0.4116,
926
+ "step": 69000
927
+ },
928
+ {
929
+ "epoch": 5.63,
930
+ "learning_rate": 4.60253436572637e-06,
931
+ "loss": 0.3977,
932
+ "step": 69500
933
+ },
934
+ {
935
+ "epoch": 5.67,
936
+ "learning_rate": 4.559896988095644e-06,
937
+ "loss": 0.4232,
938
+ "step": 70000
939
+ },
940
+ {
941
+ "epoch": 5.71,
942
+ "learning_rate": 4.517259610464918e-06,
943
+ "loss": 0.4194,
944
+ "step": 70500
945
+ },
946
+ {
947
+ "epoch": 5.75,
948
+ "learning_rate": 4.4746222328341924e-06,
949
+ "loss": 0.389,
950
+ "step": 71000
951
+ },
952
+ {
953
+ "epoch": 5.79,
954
+ "learning_rate": 4.4319848552034665e-06,
955
+ "loss": 0.4051,
956
+ "step": 71500
957
+ },
958
+ {
959
+ "epoch": 5.83,
960
+ "learning_rate": 4.38934747757274e-06,
961
+ "loss": 0.4337,
962
+ "step": 72000
963
+ },
964
+ {
965
+ "epoch": 5.87,
966
+ "learning_rate": 4.346795374697275e-06,
967
+ "loss": 0.4143,
968
+ "step": 72500
969
+ },
970
+ {
971
+ "epoch": 5.91,
972
+ "learning_rate": 4.304157997066549e-06,
973
+ "loss": 0.4146,
974
+ "step": 73000
975
+ },
976
+ {
977
+ "epoch": 5.95,
978
+ "learning_rate": 4.261520619435823e-06,
979
+ "loss": 0.3878,
980
+ "step": 73500
981
+ },
982
+ {
983
+ "epoch": 5.99,
984
+ "learning_rate": 4.218883241805096e-06,
985
+ "loss": 0.3955,
986
+ "step": 74000
987
+ },
988
+ {
989
+ "epoch": 6.0,
990
+ "eval_BLEU": 46.18598569567297,
991
+ "eval_BLEU-Bigram-Precision": 51.835734293717486,
992
+ "eval_BLEU-Trigram-Precision": 42.84724073580379,
993
+ "eval_BLEU-Unigram-Precision": 67.89297658862876,
994
+ "eval_ROUGE-2": 40.77288879803008,
995
+ "eval_ROUGE-L": 53.47443573476642,
996
+ "eval_Sacre-Bigram-Precision": 50.16359387172163,
997
+ "eval_Sacre-Trigram-Precision": 42.76279276279276,
998
+ "eval_Sacre-Unigram-Precision": 66.28928362287722,
999
+ "eval_SacreBLEU": 44.9656942267261,
1000
+ "eval_loss": 0.6645232439041138,
1001
+ "eval_runtime": 169.8377,
1002
+ "eval_samples_per_second": 7.283,
1003
+ "eval_steps_per_second": 7.283,
1004
+ "step": 74064
1005
+ },
1006
+ {
1007
+ "epoch": 6.04,
1008
+ "learning_rate": 4.176331138929632e-06,
1009
+ "loss": 0.3766,
1010
+ "step": 74500
1011
+ },
1012
+ {
1013
+ "epoch": 6.08,
1014
+ "learning_rate": 4.133693761298906e-06,
1015
+ "loss": 0.3556,
1016
+ "step": 75000
1017
+ },
1018
+ {
1019
+ "epoch": 6.12,
1020
+ "learning_rate": 4.09105638366818e-06,
1021
+ "loss": 0.3511,
1022
+ "step": 75500
1023
+ },
1024
+ {
1025
+ "epoch": 6.16,
1026
+ "learning_rate": 4.048504280792714e-06,
1027
+ "loss": 0.4021,
1028
+ "step": 76000
1029
+ },
1030
+ {
1031
+ "epoch": 6.2,
1032
+ "learning_rate": 4.005952177917249e-06,
1033
+ "loss": 0.3931,
1034
+ "step": 76500
1035
+ },
1036
+ {
1037
+ "epoch": 6.24,
1038
+ "learning_rate": 3.9633148002865234e-06,
1039
+ "loss": 0.3704,
1040
+ "step": 77000
1041
+ },
1042
+ {
1043
+ "epoch": 6.28,
1044
+ "learning_rate": 3.9206774226557975e-06,
1045
+ "loss": 0.3515,
1046
+ "step": 77500
1047
+ },
1048
+ {
1049
+ "epoch": 6.32,
1050
+ "learning_rate": 3.878040045025071e-06,
1051
+ "loss": 0.3907,
1052
+ "step": 78000
1053
+ },
1054
+ {
1055
+ "epoch": 6.36,
1056
+ "learning_rate": 3.835402667394345e-06,
1057
+ "loss": 0.3643,
1058
+ "step": 78500
1059
+ },
1060
+ {
1061
+ "epoch": 6.4,
1062
+ "learning_rate": 3.7927652897636185e-06,
1063
+ "loss": 0.3708,
1064
+ "step": 79000
1065
+ },
1066
+ {
1067
+ "epoch": 6.44,
1068
+ "learning_rate": 3.7501279121328926e-06,
1069
+ "loss": 0.3697,
1070
+ "step": 79500
1071
+ },
1072
+ {
1073
+ "epoch": 6.48,
1074
+ "learning_rate": 3.7074905345021663e-06,
1075
+ "loss": 0.3486,
1076
+ "step": 80000
1077
+ },
1078
+ {
1079
+ "epoch": 6.52,
1080
+ "learning_rate": 3.6648531568714404e-06,
1081
+ "loss": 0.3552,
1082
+ "step": 80500
1083
+ },
1084
+ {
1085
+ "epoch": 6.56,
1086
+ "learning_rate": 3.6222157792407136e-06,
1087
+ "loss": 0.3812,
1088
+ "step": 81000
1089
+ },
1090
+ {
1091
+ "epoch": 6.6,
1092
+ "learning_rate": 3.5795784016099877e-06,
1093
+ "loss": 0.3622,
1094
+ "step": 81500
1095
+ },
1096
+ {
1097
+ "epoch": 6.64,
1098
+ "learning_rate": 3.5369410239792614e-06,
1099
+ "loss": 0.391,
1100
+ "step": 82000
1101
+ },
1102
+ {
1103
+ "epoch": 6.68,
1104
+ "learning_rate": 3.494388921103797e-06,
1105
+ "loss": 0.4055,
1106
+ "step": 82500
1107
+ },
1108
+ {
1109
+ "epoch": 6.72,
1110
+ "learning_rate": 3.4517515434730702e-06,
1111
+ "loss": 0.3967,
1112
+ "step": 83000
1113
+ },
1114
+ {
1115
+ "epoch": 6.76,
1116
+ "learning_rate": 3.4091141658423443e-06,
1117
+ "loss": 0.3949,
1118
+ "step": 83500
1119
+ },
1120
+ {
1121
+ "epoch": 6.8,
1122
+ "learning_rate": 3.3664767882116184e-06,
1123
+ "loss": 0.3898,
1124
+ "step": 84000
1125
+ },
1126
+ {
1127
+ "epoch": 6.85,
1128
+ "learning_rate": 3.3239246853361536e-06,
1129
+ "loss": 0.3997,
1130
+ "step": 84500
1131
+ },
1132
+ {
1133
+ "epoch": 6.89,
1134
+ "learning_rate": 3.281287307705427e-06,
1135
+ "loss": 0.4075,
1136
+ "step": 85000
1137
+ },
1138
+ {
1139
+ "epoch": 6.93,
1140
+ "learning_rate": 3.238649930074701e-06,
1141
+ "loss": 0.3704,
1142
+ "step": 85500
1143
+ },
1144
+ {
1145
+ "epoch": 6.97,
1146
+ "learning_rate": 3.196012552443975e-06,
1147
+ "loss": 0.3809,
1148
+ "step": 86000
1149
+ },
1150
+ {
1151
+ "epoch": 7.0,
1152
+ "eval_BLEU": 48.53041153339425,
1153
+ "eval_BLEU-Bigram-Precision": 52.751822317890515,
1154
+ "eval_BLEU-Trigram-Precision": 44.54280358258696,
1155
+ "eval_BLEU-Unigram-Precision": 67.82913552910786,
1156
+ "eval_ROUGE-2": 42.93136082972705,
1157
+ "eval_ROUGE-L": 54.935588774300534,
1158
+ "eval_Sacre-Bigram-Precision": 51.321117635148816,
1159
+ "eval_Sacre-Trigram-Precision": 44.494843134078515,
1160
+ "eval_Sacre-Unigram-Precision": 66.2363645024532,
1161
+ "eval_SacreBLEU": 47.488510994186036,
1162
+ "eval_loss": 0.6420064568519592,
1163
+ "eval_runtime": 175.4955,
1164
+ "eval_samples_per_second": 7.049,
1165
+ "eval_steps_per_second": 7.049,
1166
+ "step": 86408
1167
+ },
1168
+ {
1169
+ "epoch": 7.01,
1170
+ "learning_rate": 3.1534604495685103e-06,
1171
+ "loss": 0.3682,
1172
+ "step": 86500
1173
+ },
1174
+ {
1175
+ "epoch": 7.05,
1176
+ "learning_rate": 3.110908346693045e-06,
1177
+ "loss": 0.3759,
1178
+ "step": 87000
1179
+ },
1180
+ {
1181
+ "epoch": 7.09,
1182
+ "learning_rate": 3.068270969062319e-06,
1183
+ "loss": 0.3607,
1184
+ "step": 87500
1185
+ },
1186
+ {
1187
+ "epoch": 7.13,
1188
+ "learning_rate": 3.025633591431593e-06,
1189
+ "loss": 0.3478,
1190
+ "step": 88000
1191
+ },
1192
+ {
1193
+ "epoch": 7.17,
1194
+ "learning_rate": 2.982996213800867e-06,
1195
+ "loss": 0.3474,
1196
+ "step": 88500
1197
+ },
1198
+ {
1199
+ "epoch": 7.21,
1200
+ "learning_rate": 2.94035883617014e-06,
1201
+ "loss": 0.3824,
1202
+ "step": 89000
1203
+ },
1204
+ {
1205
+ "epoch": 7.25,
1206
+ "learning_rate": 2.8977214585394142e-06,
1207
+ "loss": 0.3469,
1208
+ "step": 89500
1209
+ },
1210
+ {
1211
+ "epoch": 7.29,
1212
+ "learning_rate": 2.8550840809086883e-06,
1213
+ "loss": 0.3599,
1214
+ "step": 90000
1215
+ },
1216
+ {
1217
+ "epoch": 7.33,
1218
+ "learning_rate": 2.8124467032779616e-06,
1219
+ "loss": 0.3332,
1220
+ "step": 90500
1221
+ },
1222
+ {
1223
+ "epoch": 7.37,
1224
+ "learning_rate": 2.7698093256472356e-06,
1225
+ "loss": 0.3508,
1226
+ "step": 91000
1227
+ },
1228
+ {
1229
+ "epoch": 7.41,
1230
+ "learning_rate": 2.727257222771771e-06,
1231
+ "loss": 0.3382,
1232
+ "step": 91500
1233
+ },
1234
+ {
1235
+ "epoch": 7.45,
1236
+ "learning_rate": 2.684619845141045e-06,
1237
+ "loss": 0.3583,
1238
+ "step": 92000
1239
+ },
1240
+ {
1241
+ "epoch": 7.49,
1242
+ "learning_rate": 2.641982467510318e-06,
1243
+ "loss": 0.3392,
1244
+ "step": 92500
1245
+ },
1246
+ {
1247
+ "epoch": 7.53,
1248
+ "learning_rate": 2.5993450898795923e-06,
1249
+ "loss": 0.3548,
1250
+ "step": 93000
1251
+ },
1252
+ {
1253
+ "epoch": 7.57,
1254
+ "learning_rate": 2.556707712248866e-06,
1255
+ "loss": 0.3372,
1256
+ "step": 93500
1257
+ },
1258
+ {
1259
+ "epoch": 7.62,
1260
+ "learning_rate": 2.5141556093734016e-06,
1261
+ "loss": 0.3386,
1262
+ "step": 94000
1263
+ },
1264
+ {
1265
+ "epoch": 7.66,
1266
+ "learning_rate": 2.471518231742675e-06,
1267
+ "loss": 0.362,
1268
+ "step": 94500
1269
+ },
1270
+ {
1271
+ "epoch": 7.7,
1272
+ "learning_rate": 2.428880854111949e-06,
1273
+ "loss": 0.3454,
1274
+ "step": 95000
1275
+ },
1276
+ {
1277
+ "epoch": 7.74,
1278
+ "learning_rate": 2.3862434764812226e-06,
1279
+ "loss": 0.3757,
1280
+ "step": 95500
1281
+ },
1282
+ {
1283
+ "epoch": 7.78,
1284
+ "learning_rate": 2.3436913736057578e-06,
1285
+ "loss": 0.3785,
1286
+ "step": 96000
1287
+ },
1288
+ {
1289
+ "epoch": 7.82,
1290
+ "learning_rate": 2.3010539959750314e-06,
1291
+ "loss": 0.3656,
1292
+ "step": 96500
1293
+ },
1294
+ {
1295
+ "epoch": 7.86,
1296
+ "learning_rate": 2.2584166183443055e-06,
1297
+ "loss": 0.36,
1298
+ "step": 97000
1299
+ },
1300
+ {
1301
+ "epoch": 7.9,
1302
+ "learning_rate": 2.2157792407135796e-06,
1303
+ "loss": 0.3522,
1304
+ "step": 97500
1305
+ },
1306
+ {
1307
+ "epoch": 7.94,
1308
+ "learning_rate": 2.1732271378381144e-06,
1309
+ "loss": 0.3511,
1310
+ "step": 98000
1311
+ },
1312
+ {
1313
+ "epoch": 7.98,
1314
+ "learning_rate": 2.130589760207388e-06,
1315
+ "loss": 0.3349,
1316
+ "step": 98500
1317
+ },
1318
+ {
1319
+ "epoch": 8.0,
1320
+ "eval_BLEU": 49.57199736076417,
1321
+ "eval_BLEU-Bigram-Precision": 55.58697298661954,
1322
+ "eval_BLEU-Trigram-Precision": 47.143087942269375,
1323
+ "eval_BLEU-Unigram-Precision": 70.68244463454045,
1324
+ "eval_ROUGE-2": 44.4865922111988,
1325
+ "eval_ROUGE-L": 56.71133192128113,
1326
+ "eval_Sacre-Bigram-Precision": 53.98119122257053,
1327
+ "eval_Sacre-Trigram-Precision": 47.064737753449144,
1328
+ "eval_Sacre-Unigram-Precision": 69.11714187564411,
1329
+ "eval_SacreBLEU": 48.47735653922784,
1330
+ "eval_loss": 0.6274014711380005,
1331
+ "eval_runtime": 169.1165,
1332
+ "eval_samples_per_second": 7.314,
1333
+ "eval_steps_per_second": 7.314,
1334
+ "step": 98752
1335
+ },
1336
+ {
1337
+ "epoch": 8.02,
1338
+ "learning_rate": 2.087952382576662e-06,
1339
+ "loss": 0.3635,
1340
+ "step": 99000
1341
+ },
1342
+ {
1343
+ "epoch": 8.06,
1344
+ "learning_rate": 2.0453150049459363e-06,
1345
+ "loss": 0.3149,
1346
+ "step": 99500
1347
+ },
1348
+ {
1349
+ "epoch": 8.1,
1350
+ "learning_rate": 2.00267762731521e-06,
1351
+ "loss": 0.3348,
1352
+ "step": 100000
1353
+ },
1354
+ {
1355
+ "epoch": 8.14,
1356
+ "learning_rate": 1.9600402496844836e-06,
1357
+ "loss": 0.3358,
1358
+ "step": 100500
1359
+ },
1360
+ {
1361
+ "epoch": 8.18,
1362
+ "learning_rate": 1.9174028720537573e-06,
1363
+ "loss": 0.3405,
1364
+ "step": 101000
1365
+ },
1366
+ {
1367
+ "epoch": 8.22,
1368
+ "learning_rate": 1.8747654944230311e-06,
1369
+ "loss": 0.328,
1370
+ "step": 101500
1371
+ },
1372
+ {
1373
+ "epoch": 8.26,
1374
+ "learning_rate": 1.8322133915475666e-06,
1375
+ "loss": 0.3487,
1376
+ "step": 102000
1377
+ },
1378
+ {
1379
+ "epoch": 8.3,
1380
+ "learning_rate": 1.7895760139168402e-06,
1381
+ "loss": 0.3012,
1382
+ "step": 102500
1383
+ },
1384
+ {
1385
+ "epoch": 8.34,
1386
+ "learning_rate": 1.746938636286114e-06,
1387
+ "loss": 0.3456,
1388
+ "step": 103000
1389
+ },
1390
+ {
1391
+ "epoch": 8.38,
1392
+ "learning_rate": 1.7043012586553878e-06,
1393
+ "loss": 0.3472,
1394
+ "step": 103500
1395
+ },
1396
+ {
1397
+ "epoch": 8.43,
1398
+ "learning_rate": 1.6617491557799232e-06,
1399
+ "loss": 0.3476,
1400
+ "step": 104000
1401
+ },
1402
+ {
1403
+ "epoch": 8.47,
1404
+ "learning_rate": 1.6191117781491968e-06,
1405
+ "loss": 0.3262,
1406
+ "step": 104500
1407
+ },
1408
+ {
1409
+ "epoch": 8.51,
1410
+ "learning_rate": 1.5765596752737323e-06,
1411
+ "loss": 0.3607,
1412
+ "step": 105000
1413
+ },
1414
+ {
1415
+ "epoch": 8.55,
1416
+ "learning_rate": 1.533922297643006e-06,
1417
+ "loss": 0.341,
1418
+ "step": 105500
1419
+ },
1420
+ {
1421
+ "epoch": 8.59,
1422
+ "learning_rate": 1.4912849200122798e-06,
1423
+ "loss": 0.3164,
1424
+ "step": 106000
1425
+ },
1426
+ {
1427
+ "epoch": 8.63,
1428
+ "learning_rate": 1.4486475423815535e-06,
1429
+ "loss": 0.354,
1430
+ "step": 106500
1431
+ },
1432
+ {
1433
+ "epoch": 8.67,
1434
+ "learning_rate": 1.4060101647508274e-06,
1435
+ "loss": 0.3282,
1436
+ "step": 107000
1437
+ },
1438
+ {
1439
+ "epoch": 8.71,
1440
+ "learning_rate": 1.363372787120101e-06,
1441
+ "loss": 0.3419,
1442
+ "step": 107500
1443
+ },
1444
+ {
1445
+ "epoch": 8.75,
1446
+ "learning_rate": 1.320735409489375e-06,
1447
+ "loss": 0.3524,
1448
+ "step": 108000
1449
+ },
1450
+ {
1451
+ "epoch": 8.79,
1452
+ "learning_rate": 1.2780980318586486e-06,
1453
+ "loss": 0.3372,
1454
+ "step": 108500
1455
+ },
1456
+ {
1457
+ "epoch": 8.83,
1458
+ "learning_rate": 1.2354606542279224e-06,
1459
+ "loss": 0.358,
1460
+ "step": 109000
1461
+ },
1462
+ {
1463
+ "epoch": 8.87,
1464
+ "learning_rate": 1.1929085513524577e-06,
1465
+ "loss": 0.3426,
1466
+ "step": 109500
1467
+ },
1468
+ {
1469
+ "epoch": 8.91,
1470
+ "learning_rate": 1.150356448476993e-06,
1471
+ "loss": 0.3449,
1472
+ "step": 110000
1473
+ },
1474
+ {
1475
+ "epoch": 8.95,
1476
+ "learning_rate": 1.1077190708462667e-06,
1477
+ "loss": 0.3547,
1478
+ "step": 110500
1479
+ },
1480
+ {
1481
+ "epoch": 8.99,
1482
+ "learning_rate": 1.0650816932155406e-06,
1483
+ "loss": 0.347,
1484
+ "step": 111000
1485
+ },
1486
+ {
1487
+ "epoch": 9.0,
1488
+ "eval_BLEU": 50.42765483523289,
1489
+ "eval_BLEU-Bigram-Precision": 55.31231468669697,
1490
+ "eval_BLEU-Trigram-Precision": 47.11826938259908,
1491
+ "eval_BLEU-Unigram-Precision": 69.82722488706749,
1492
+ "eval_ROUGE-2": 44.880113409125535,
1493
+ "eval_ROUGE-L": 56.82064108206701,
1494
+ "eval_Sacre-Bigram-Precision": 53.78361795684631,
1495
+ "eval_Sacre-Trigram-Precision": 47.04437530702472,
1496
+ "eval_Sacre-Unigram-Precision": 68.36258716037509,
1497
+ "eval_SacreBLEU": 49.36901330124516,
1498
+ "eval_loss": 0.6147586703300476,
1499
+ "eval_runtime": 169.9129,
1500
+ "eval_samples_per_second": 7.28,
1501
+ "eval_steps_per_second": 7.28,
1502
+ "step": 111096
1503
+ },
1504
+ {
1505
+ "epoch": 9.03,
1506
+ "learning_rate": 1.0224443155848143e-06,
1507
+ "loss": 0.3398,
1508
+ "step": 111500
1509
+ },
1510
+ {
1511
+ "epoch": 9.07,
1512
+ "learning_rate": 9.798069379540882e-07,
1513
+ "loss": 0.3255,
1514
+ "step": 112000
1515
+ },
1516
+ {
1517
+ "epoch": 9.11,
1518
+ "learning_rate": 9.37169560323362e-07,
1519
+ "loss": 0.3178,
1520
+ "step": 112500
1521
+ },
1522
+ {
1523
+ "epoch": 9.15,
1524
+ "learning_rate": 8.946174574478971e-07,
1525
+ "loss": 0.3281,
1526
+ "step": 113000
1527
+ },
1528
+ {
1529
+ "epoch": 9.19,
1530
+ "learning_rate": 8.520653545724324e-07,
1531
+ "loss": 0.3261,
1532
+ "step": 113500
1533
+ },
1534
+ {
1535
+ "epoch": 9.24,
1536
+ "learning_rate": 8.094279769417062e-07,
1537
+ "loss": 0.3275,
1538
+ "step": 114000
1539
+ },
1540
+ {
1541
+ "epoch": 9.28,
1542
+ "learning_rate": 7.6679059931098e-07,
1543
+ "loss": 0.3359,
1544
+ "step": 114500
1545
+ },
1546
+ {
1547
+ "epoch": 9.32,
1548
+ "learning_rate": 7.241532216802538e-07,
1549
+ "loss": 0.3165,
1550
+ "step": 115000
1551
+ },
1552
+ {
1553
+ "epoch": 9.36,
1554
+ "learning_rate": 6.815158440495275e-07,
1555
+ "loss": 0.3199,
1556
+ "step": 115500
1557
+ },
1558
+ {
1559
+ "epoch": 9.4,
1560
+ "learning_rate": 6.388784664188015e-07,
1561
+ "loss": 0.332,
1562
+ "step": 116000
1563
+ },
1564
+ {
1565
+ "epoch": 9.44,
1566
+ "learning_rate": 5.962410887880752e-07,
1567
+ "loss": 0.3227,
1568
+ "step": 116500
1569
+ },
1570
+ {
1571
+ "epoch": 9.48,
1572
+ "learning_rate": 5.536037111573491e-07,
1573
+ "loss": 0.3203,
1574
+ "step": 117000
1575
+ },
1576
+ {
1577
+ "epoch": 9.52,
1578
+ "learning_rate": 5.109663335266228e-07,
1579
+ "loss": 0.347,
1580
+ "step": 117500
1581
+ },
1582
+ {
1583
+ "epoch": 9.56,
1584
+ "learning_rate": 4.6832895589589667e-07,
1585
+ "loss": 0.3485,
1586
+ "step": 118000
1587
+ },
1588
+ {
1589
+ "epoch": 9.6,
1590
+ "learning_rate": 4.2569157826517044e-07,
1591
+ "loss": 0.3244,
1592
+ "step": 118500
1593
+ },
1594
+ {
1595
+ "epoch": 9.64,
1596
+ "learning_rate": 3.830542006344442e-07,
1597
+ "loss": 0.3351,
1598
+ "step": 119000
1599
+ },
1600
+ {
1601
+ "epoch": 9.68,
1602
+ "learning_rate": 3.4041682300371804e-07,
1603
+ "loss": 0.3341,
1604
+ "step": 119500
1605
+ },
1606
+ {
1607
+ "epoch": 9.72,
1608
+ "learning_rate": 2.9786472012825325e-07,
1609
+ "loss": 0.328,
1610
+ "step": 120000
1611
+ },
1612
+ {
1613
+ "epoch": 9.76,
1614
+ "learning_rate": 2.5522734249752707e-07,
1615
+ "loss": 0.3447,
1616
+ "step": 120500
1617
+ },
1618
+ {
1619
+ "epoch": 9.8,
1620
+ "learning_rate": 2.1258996486680084e-07,
1621
+ "loss": 0.3326,
1622
+ "step": 121000
1623
+ },
1624
+ {
1625
+ "epoch": 9.84,
1626
+ "learning_rate": 1.6995258723607467e-07,
1627
+ "loss": 0.3259,
1628
+ "step": 121500
1629
+ },
1630
+ {
1631
+ "epoch": 9.88,
1632
+ "learning_rate": 1.274004843606099e-07,
1633
+ "loss": 0.3355,
1634
+ "step": 122000
1635
+ },
1636
+ {
1637
+ "epoch": 9.92,
1638
+ "learning_rate": 8.47631067298837e-08,
1639
+ "loss": 0.3653,
1640
+ "step": 122500
1641
+ },
1642
+ {
1643
+ "epoch": 9.96,
1644
+ "learning_rate": 4.212572909915749e-08,
1645
+ "loss": 0.3359,
1646
+ "step": 123000
1647
+ },
1648
+ {
1649
+ "epoch": 10.0,
1650
+ "eval_BLEU": 50.8713809636162,
1651
+ "eval_BLEU-Bigram-Precision": 55.25529935869192,
1652
+ "eval_BLEU-Trigram-Precision": 47.15997915581032,
1653
+ "eval_BLEU-Unigram-Precision": 69.56702363367799,
1654
+ "eval_ROUGE-2": 45.59251780348448,
1655
+ "eval_ROUGE-L": 57.14390377702154,
1656
+ "eval_Sacre-Bigram-Precision": 53.736780853109344,
1657
+ "eval_Sacre-Trigram-Precision": 47.052790672568285,
1658
+ "eval_Sacre-Unigram-Precision": 68.06666666666666,
1659
+ "eval_SacreBLEU": 49.82287256259556,
1660
+ "eval_loss": 0.6102247834205627,
1661
+ "eval_runtime": 172.6687,
1662
+ "eval_samples_per_second": 7.164,
1663
+ "eval_steps_per_second": 7.164,
1664
+ "step": 123440
1665
+ },
1666
+ {
1667
+ "epoch": 10.0,
1668
+ "step": 123440,
1669
+ "total_flos": 3771355808563200.0,
1670
+ "train_loss": 0.4878676912032082,
1671
+ "train_runtime": 11854.8374,
1672
+ "train_samples_per_second": 20.824,
1673
+ "train_steps_per_second": 10.413
1674
+ }
1675
+ ],
1676
+ "max_steps": 123440,
1677
+ "num_train_epochs": 10,
1678
+ "total_flos": 3771355808563200.0,
1679
+ "trial_name": null,
1680
+ "trial_params": null
1681
+ }