Hugging Face's logo

mlfoundations-dev
/
oh-dcft-v3.1-llama-3.1-8b-qwen

Text Generation
Transformers
Safetensors
qwen2
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
sedrickkeh commited on
Commit
6fd93e9
·
verified ·
1 Parent(s): 20c59aa

Training in progress, epoch 2

Browse files
Files changed (5) hide show
  1. model-00001-of-00004.safetensors +1 -1
  2. model-00002-of-00004.safetensors +1 -1
  3. model-00003-of-00004.safetensors +1 -1
  4. model-00004-of-00004.safetensors +1 -1
  5. trainer_log.jsonl +178 -0
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1ee5139f41a182dba4a775ed56c035b7117381ccea51f265b1ac7e53e18fba82
3
  size 4877660776
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3c29d3537bf106b4606fb06a15dcc338a026c5e2b6465c07bb0ea1386578ef6
3
  size 4877660776
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e43525a1ef1114550b7f0676ff1ebb388651f8c53a1e7639870d074c89a29461
3
  size 4932751008
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:989381031726a828e90c6c0d5fbe86c54d6694499c72e677e2ccae58ec39015d
3
  size 4932751008
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:63f4cf96383069b35f750194a0cd2dbb9b2ca0b81ff466b9762284bf7a41bf98
3
  size 4330865200
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20431c183f2cb0bc9cebed70cf843cca533f9ca666e2d9f5b6ccf739ffdb5536
3
  size 4330865200
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0b62c8f9fc4e2aad92644ed13e59abfbc27d774b96c5225b2bb837ca08bdf3ab
3
  size 1089994880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5c855c01b949a2397111b1059756af9c285d61b2823bba1555fd8f020884c28
3
  size 1089994880
trainer_log.jsonl CHANGED
@@ -177,3 +177,181 @@
177
  {"current_steps": 1770, "total_steps": 5316, "loss": 0.4318, "lr": 5e-06, "epoch": 0.998589562764457, "percentage": 33.3, "elapsed_time": "6:26:05", "remaining_time": "12:53:29"}
178
  {"current_steps": 1772, "total_steps": 5316, "eval_loss": 0.4379998743534088, "epoch": 0.9997179125528914, "percentage": 33.33, "elapsed_time": "6:34:03", "remaining_time": "13:08:06"}
179
  {"current_steps": 1780, "total_steps": 5316, "loss": 0.4228, "lr": 5e-06, "epoch": 1.004231311706629, "percentage": 33.48, "elapsed_time": "6:36:46", "remaining_time": "13:08:12"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
177
  {"current_steps": 1770, "total_steps": 5316, "loss": 0.4318, "lr": 5e-06, "epoch": 0.998589562764457, "percentage": 33.3, "elapsed_time": "6:26:05", "remaining_time": "12:53:29"}
178
  {"current_steps": 1772, "total_steps": 5316, "eval_loss": 0.4379998743534088, "epoch": 0.9997179125528914, "percentage": 33.33, "elapsed_time": "6:34:03", "remaining_time": "13:08:06"}
179
  {"current_steps": 1780, "total_steps": 5316, "loss": 0.4228, "lr": 5e-06, "epoch": 1.004231311706629, "percentage": 33.48, "elapsed_time": "6:36:46", "remaining_time": "13:08:12"}
180
+ {"current_steps": 1790, "total_steps": 5316, "loss": 0.3888, "lr": 5e-06, "epoch": 1.0098730606488011, "percentage": 33.67, "elapsed_time": "6:38:58", "remaining_time": "13:05:54"}
181
+ {"current_steps": 1800, "total_steps": 5316, "loss": 0.4081, "lr": 5e-06, "epoch": 1.0155148095909732, "percentage": 33.86, "elapsed_time": "6:41:09", "remaining_time": "13:03:36"}
182
+ {"current_steps": 1810, "total_steps": 5316, "loss": 0.4083, "lr": 5e-06, "epoch": 1.0211565585331452, "percentage": 34.05, "elapsed_time": "6:43:21", "remaining_time": "13:01:18"}
183
+ {"current_steps": 1820, "total_steps": 5316, "loss": 0.403, "lr": 5e-06, "epoch": 1.0267983074753173, "percentage": 34.24, "elapsed_time": "6:45:33", "remaining_time": "12:59:00"}
184
+ {"current_steps": 1830, "total_steps": 5316, "loss": 0.3984, "lr": 5e-06, "epoch": 1.0324400564174894, "percentage": 34.42, "elapsed_time": "6:47:44", "remaining_time": "12:56:42"}
185
+ {"current_steps": 1840, "total_steps": 5316, "loss": 0.3908, "lr": 5e-06, "epoch": 1.0380818053596614, "percentage": 34.61, "elapsed_time": "6:49:55", "remaining_time": "12:54:24"}
186
+ {"current_steps": 1850, "total_steps": 5316, "loss": 0.3944, "lr": 5e-06, "epoch": 1.0437235543018335, "percentage": 34.8, "elapsed_time": "6:52:07", "remaining_time": "12:52:06"}
187
+ {"current_steps": 1860, "total_steps": 5316, "loss": 0.3888, "lr": 5e-06, "epoch": 1.0493653032440056, "percentage": 34.99, "elapsed_time": "6:54:18", "remaining_time": "12:49:49"}
188
+ {"current_steps": 1870, "total_steps": 5316, "loss": 0.3936, "lr": 5e-06, "epoch": 1.0550070521861776, "percentage": 35.18, "elapsed_time": "6:56:30", "remaining_time": "12:47:31"}
189
+ {"current_steps": 1880, "total_steps": 5316, "loss": 0.3881, "lr": 5e-06, "epoch": 1.0606488011283497, "percentage": 35.36, "elapsed_time": "6:58:41", "remaining_time": "12:45:13"}
190
+ {"current_steps": 1890, "total_steps": 5316, "loss": 0.3997, "lr": 5e-06, "epoch": 1.0662905500705218, "percentage": 35.55, "elapsed_time": "7:00:52", "remaining_time": "12:42:55"}
191
+ {"current_steps": 1900, "total_steps": 5316, "loss": 0.3945, "lr": 5e-06, "epoch": 1.071932299012694, "percentage": 35.74, "elapsed_time": "7:03:04", "remaining_time": "12:40:38"}
192
+ {"current_steps": 1910, "total_steps": 5316, "loss": 0.393, "lr": 5e-06, "epoch": 1.077574047954866, "percentage": 35.93, "elapsed_time": "7:05:15", "remaining_time": "12:38:20"}
193
+ {"current_steps": 1920, "total_steps": 5316, "loss": 0.3905, "lr": 5e-06, "epoch": 1.0832157968970382, "percentage": 36.12, "elapsed_time": "7:07:27", "remaining_time": "12:36:03"}
194
+ {"current_steps": 1930, "total_steps": 5316, "loss": 0.3945, "lr": 5e-06, "epoch": 1.0888575458392102, "percentage": 36.31, "elapsed_time": "7:09:38", "remaining_time": "12:33:45"}
195
+ {"current_steps": 1940, "total_steps": 5316, "loss": 0.3888, "lr": 5e-06, "epoch": 1.0944992947813823, "percentage": 36.49, "elapsed_time": "7:11:49", "remaining_time": "12:31:28"}
196
+ {"current_steps": 1950, "total_steps": 5316, "loss": 0.3889, "lr": 5e-06, "epoch": 1.1001410437235544, "percentage": 36.68, "elapsed_time": "7:14:01", "remaining_time": "12:29:11"}
197
+ {"current_steps": 1960, "total_steps": 5316, "loss": 0.3883, "lr": 5e-06, "epoch": 1.1057827926657264, "percentage": 36.87, "elapsed_time": "7:16:12", "remaining_time": "12:26:53"}
198
+ {"current_steps": 1970, "total_steps": 5316, "loss": 0.4015, "lr": 5e-06, "epoch": 1.1114245416078985, "percentage": 37.06, "elapsed_time": "7:18:23", "remaining_time": "12:24:36"}
199
+ {"current_steps": 1980, "total_steps": 5316, "loss": 0.391, "lr": 5e-06, "epoch": 1.1170662905500706, "percentage": 37.25, "elapsed_time": "7:20:35", "remaining_time": "12:22:19"}
200
+ {"current_steps": 1990, "total_steps": 5316, "loss": 0.3933, "lr": 5e-06, "epoch": 1.1227080394922426, "percentage": 37.43, "elapsed_time": "7:22:47", "remaining_time": "12:20:03"}
201
+ {"current_steps": 2000, "total_steps": 5316, "loss": 0.4033, "lr": 5e-06, "epoch": 1.1283497884344147, "percentage": 37.62, "elapsed_time": "7:24:58", "remaining_time": "12:17:46"}
202
+ {"current_steps": 2010, "total_steps": 5316, "loss": 0.3962, "lr": 5e-06, "epoch": 1.1339915373765868, "percentage": 37.81, "elapsed_time": "7:27:09", "remaining_time": "12:15:29"}
203
+ {"current_steps": 2020, "total_steps": 5316, "loss": 0.3843, "lr": 5e-06, "epoch": 1.1396332863187588, "percentage": 38.0, "elapsed_time": "7:29:21", "remaining_time": "12:13:12"}
204
+ {"current_steps": 2030, "total_steps": 5316, "loss": 0.3896, "lr": 5e-06, "epoch": 1.1452750352609309, "percentage": 38.19, "elapsed_time": "7:31:32", "remaining_time": "12:10:55"}
205
+ {"current_steps": 2040, "total_steps": 5316, "loss": 0.3903, "lr": 5e-06, "epoch": 1.150916784203103, "percentage": 38.37, "elapsed_time": "7:33:44", "remaining_time": "12:08:39"}
206
+ {"current_steps": 2050, "total_steps": 5316, "loss": 0.3923, "lr": 5e-06, "epoch": 1.156558533145275, "percentage": 38.56, "elapsed_time": "7:35:55", "remaining_time": "12:06:22"}
207
+ {"current_steps": 2060, "total_steps": 5316, "loss": 0.4137, "lr": 5e-06, "epoch": 1.162200282087447, "percentage": 38.75, "elapsed_time": "7:38:07", "remaining_time": "12:04:05"}
208
+ {"current_steps": 2070, "total_steps": 5316, "loss": 0.3952, "lr": 5e-06, "epoch": 1.1678420310296191, "percentage": 38.94, "elapsed_time": "7:40:18", "remaining_time": "12:01:49"}
209
+ {"current_steps": 2080, "total_steps": 5316, "loss": 0.3981, "lr": 5e-06, "epoch": 1.1734837799717912, "percentage": 39.13, "elapsed_time": "7:42:30", "remaining_time": "11:59:33"}
210
+ {"current_steps": 2090, "total_steps": 5316, "loss": 0.397, "lr": 5e-06, "epoch": 1.1791255289139633, "percentage": 39.32, "elapsed_time": "7:44:41", "remaining_time": "11:57:16"}
211
+ {"current_steps": 2100, "total_steps": 5316, "loss": 0.3987, "lr": 5e-06, "epoch": 1.1847672778561353, "percentage": 39.5, "elapsed_time": "7:46:53", "remaining_time": "11:55:00"}
212
+ {"current_steps": 2110, "total_steps": 5316, "loss": 0.3913, "lr": 5e-06, "epoch": 1.1904090267983074, "percentage": 39.69, "elapsed_time": "7:49:04", "remaining_time": "11:52:44"}
213
+ {"current_steps": 2120, "total_steps": 5316, "loss": 0.4001, "lr": 5e-06, "epoch": 1.1960507757404795, "percentage": 39.88, "elapsed_time": "7:51:16", "remaining_time": "11:50:28"}
214
+ {"current_steps": 2130, "total_steps": 5316, "loss": 0.3968, "lr": 5e-06, "epoch": 1.2016925246826515, "percentage": 40.07, "elapsed_time": "7:53:27", "remaining_time": "11:48:11"}
215
+ {"current_steps": 2140, "total_steps": 5316, "loss": 0.4022, "lr": 5e-06, "epoch": 1.2073342736248236, "percentage": 40.26, "elapsed_time": "7:55:39", "remaining_time": "11:45:55"}
216
+ {"current_steps": 2150, "total_steps": 5316, "loss": 0.4083, "lr": 5e-06, "epoch": 1.2129760225669957, "percentage": 40.44, "elapsed_time": "7:57:51", "remaining_time": "11:43:39"}
217
+ {"current_steps": 2160, "total_steps": 5316, "loss": 0.379, "lr": 5e-06, "epoch": 1.2186177715091677, "percentage": 40.63, "elapsed_time": "8:00:02", "remaining_time": "11:41:23"}
218
+ {"current_steps": 2170, "total_steps": 5316, "loss": 0.397, "lr": 5e-06, "epoch": 1.22425952045134, "percentage": 40.82, "elapsed_time": "8:02:14", "remaining_time": "11:39:07"}
219
+ {"current_steps": 2180, "total_steps": 5316, "loss": 0.3988, "lr": 5e-06, "epoch": 1.229901269393512, "percentage": 41.01, "elapsed_time": "8:04:25", "remaining_time": "11:36:51"}
220
+ {"current_steps": 2190, "total_steps": 5316, "loss": 0.3802, "lr": 5e-06, "epoch": 1.2355430183356841, "percentage": 41.2, "elapsed_time": "8:06:36", "remaining_time": "11:34:35"}
221
+ {"current_steps": 2200, "total_steps": 5316, "loss": 0.3966, "lr": 5e-06, "epoch": 1.2411847672778562, "percentage": 41.38, "elapsed_time": "8:08:48", "remaining_time": "11:32:19"}
222
+ {"current_steps": 2210, "total_steps": 5316, "loss": 0.3955, "lr": 5e-06, "epoch": 1.2468265162200283, "percentage": 41.57, "elapsed_time": "8:10:59", "remaining_time": "11:30:03"}
223
+ {"current_steps": 2220, "total_steps": 5316, "loss": 0.3888, "lr": 5e-06, "epoch": 1.2524682651622003, "percentage": 41.76, "elapsed_time": "8:13:11", "remaining_time": "11:27:47"}
224
+ {"current_steps": 2230, "total_steps": 5316, "loss": 0.3969, "lr": 5e-06, "epoch": 1.2581100141043724, "percentage": 41.95, "elapsed_time": "8:15:22", "remaining_time": "11:25:31"}
225
+ {"current_steps": 2240, "total_steps": 5316, "loss": 0.4088, "lr": 5e-06, "epoch": 1.2637517630465445, "percentage": 42.14, "elapsed_time": "8:17:33", "remaining_time": "11:23:15"}
226
+ {"current_steps": 2250, "total_steps": 5316, "loss": 0.3935, "lr": 5e-06, "epoch": 1.2693935119887165, "percentage": 42.33, "elapsed_time": "8:19:45", "remaining_time": "11:20:59"}
227
+ {"current_steps": 2260, "total_steps": 5316, "loss": 0.3944, "lr": 5e-06, "epoch": 1.2750352609308886, "percentage": 42.51, "elapsed_time": "8:21:56", "remaining_time": "11:18:43"}
228
+ {"current_steps": 2270, "total_steps": 5316, "loss": 0.3905, "lr": 5e-06, "epoch": 1.2806770098730607, "percentage": 42.7, "elapsed_time": "8:24:07", "remaining_time": "11:16:27"}
229
+ {"current_steps": 2280, "total_steps": 5316, "loss": 0.3955, "lr": 5e-06, "epoch": 1.2863187588152327, "percentage": 42.89, "elapsed_time": "8:26:18", "remaining_time": "11:14:11"}
230
+ {"current_steps": 2290, "total_steps": 5316, "loss": 0.4013, "lr": 5e-06, "epoch": 1.2919605077574048, "percentage": 43.08, "elapsed_time": "8:28:30", "remaining_time": "11:11:55"}
231
+ {"current_steps": 2300, "total_steps": 5316, "loss": 0.4005, "lr": 5e-06, "epoch": 1.2976022566995769, "percentage": 43.27, "elapsed_time": "8:30:41", "remaining_time": "11:09:40"}
232
+ {"current_steps": 2310, "total_steps": 5316, "loss": 0.3933, "lr": 5e-06, "epoch": 1.303244005641749, "percentage": 43.45, "elapsed_time": "8:32:52", "remaining_time": "11:07:24"}
233
+ {"current_steps": 2320, "total_steps": 5316, "loss": 0.3835, "lr": 5e-06, "epoch": 1.308885754583921, "percentage": 43.64, "elapsed_time": "8:35:03", "remaining_time": "11:05:08"}
234
+ {"current_steps": 2330, "total_steps": 5316, "loss": 0.4045, "lr": 5e-06, "epoch": 1.314527503526093, "percentage": 43.83, "elapsed_time": "8:37:15", "remaining_time": "11:02:52"}
235
+ {"current_steps": 2340, "total_steps": 5316, "loss": 0.4009, "lr": 5e-06, "epoch": 1.320169252468265, "percentage": 44.02, "elapsed_time": "8:39:26", "remaining_time": "11:00:37"}
236
+ {"current_steps": 2350, "total_steps": 5316, "loss": 0.3938, "lr": 5e-06, "epoch": 1.3258110014104372, "percentage": 44.21, "elapsed_time": "8:41:37", "remaining_time": "10:58:21"}
237
+ {"current_steps": 2360, "total_steps": 5316, "loss": 0.3928, "lr": 5e-06, "epoch": 1.3314527503526092, "percentage": 44.39, "elapsed_time": "8:43:48", "remaining_time": "10:56:06"}
238
+ {"current_steps": 2370, "total_steps": 5316, "loss": 0.4033, "lr": 5e-06, "epoch": 1.3370944992947813, "percentage": 44.58, "elapsed_time": "8:46:00", "remaining_time": "10:53:50"}
239
+ {"current_steps": 2380, "total_steps": 5316, "loss": 0.3972, "lr": 5e-06, "epoch": 1.3427362482369536, "percentage": 44.77, "elapsed_time": "8:48:11", "remaining_time": "10:51:34"}
240
+ {"current_steps": 2390, "total_steps": 5316, "loss": 0.3968, "lr": 5e-06, "epoch": 1.3483779971791257, "percentage": 44.96, "elapsed_time": "8:50:22", "remaining_time": "10:49:19"}
241
+ {"current_steps": 2400, "total_steps": 5316, "loss": 0.3924, "lr": 5e-06, "epoch": 1.3540197461212977, "percentage": 45.15, "elapsed_time": "8:52:33", "remaining_time": "10:47:03"}
242
+ {"current_steps": 2410, "total_steps": 5316, "loss": 0.4117, "lr": 5e-06, "epoch": 1.3596614950634698, "percentage": 45.33, "elapsed_time": "8:54:45", "remaining_time": "10:44:48"}
243
+ {"current_steps": 2420, "total_steps": 5316, "loss": 0.39, "lr": 5e-06, "epoch": 1.3653032440056418, "percentage": 45.52, "elapsed_time": "8:56:56", "remaining_time": "10:42:32"}
244
+ {"current_steps": 2430, "total_steps": 5316, "loss": 0.4, "lr": 5e-06, "epoch": 1.370944992947814, "percentage": 45.71, "elapsed_time": "8:59:07", "remaining_time": "10:40:17"}
245
+ {"current_steps": 2440, "total_steps": 5316, "loss": 0.4077, "lr": 5e-06, "epoch": 1.376586741889986, "percentage": 45.9, "elapsed_time": "9:01:18", "remaining_time": "10:38:02"}
246
+ {"current_steps": 2450, "total_steps": 5316, "loss": 0.3959, "lr": 5e-06, "epoch": 1.382228490832158, "percentage": 46.09, "elapsed_time": "9:03:29", "remaining_time": "10:35:46"}
247
+ {"current_steps": 2460, "total_steps": 5316, "loss": 0.4016, "lr": 5e-06, "epoch": 1.38787023977433, "percentage": 46.28, "elapsed_time": "9:05:41", "remaining_time": "10:33:31"}
248
+ {"current_steps": 2470, "total_steps": 5316, "loss": 0.3973, "lr": 5e-06, "epoch": 1.3935119887165022, "percentage": 46.46, "elapsed_time": "9:07:52", "remaining_time": "10:31:16"}
249
+ {"current_steps": 2480, "total_steps": 5316, "loss": 0.3942, "lr": 5e-06, "epoch": 1.3991537376586742, "percentage": 46.65, "elapsed_time": "9:10:03", "remaining_time": "10:29:01"}
250
+ {"current_steps": 2490, "total_steps": 5316, "loss": 0.3998, "lr": 5e-06, "epoch": 1.4047954866008463, "percentage": 46.84, "elapsed_time": "9:12:14", "remaining_time": "10:26:45"}
251
+ {"current_steps": 2500, "total_steps": 5316, "loss": 0.3963, "lr": 5e-06, "epoch": 1.4104372355430184, "percentage": 47.03, "elapsed_time": "9:14:25", "remaining_time": "10:24:30"}
252
+ {"current_steps": 2510, "total_steps": 5316, "loss": 0.3885, "lr": 5e-06, "epoch": 1.4160789844851904, "percentage": 47.22, "elapsed_time": "9:16:37", "remaining_time": "10:22:15"}
253
+ {"current_steps": 2520, "total_steps": 5316, "loss": 0.3917, "lr": 5e-06, "epoch": 1.4217207334273625, "percentage": 47.4, "elapsed_time": "9:18:48", "remaining_time": "10:20:00"}
254
+ {"current_steps": 2530, "total_steps": 5316, "loss": 0.3967, "lr": 5e-06, "epoch": 1.4273624823695346, "percentage": 47.59, "elapsed_time": "9:20:59", "remaining_time": "10:17:45"}
255
+ {"current_steps": 2540, "total_steps": 5316, "loss": 0.395, "lr": 5e-06, "epoch": 1.4330042313117066, "percentage": 47.78, "elapsed_time": "9:23:10", "remaining_time": "10:15:30"}
256
+ {"current_steps": 2550, "total_steps": 5316, "loss": 0.3981, "lr": 5e-06, "epoch": 1.4386459802538787, "percentage": 47.97, "elapsed_time": "9:25:21", "remaining_time": "10:13:15"}
257
+ {"current_steps": 2560, "total_steps": 5316, "loss": 0.3925, "lr": 5e-06, "epoch": 1.4442877291960508, "percentage": 48.16, "elapsed_time": "9:27:32", "remaining_time": "10:11:00"}
258
+ {"current_steps": 2570, "total_steps": 5316, "loss": 0.4025, "lr": 5e-06, "epoch": 1.4499294781382228, "percentage": 48.34, "elapsed_time": "9:29:44", "remaining_time": "10:08:45"}
259
+ {"current_steps": 2580, "total_steps": 5316, "loss": 0.3943, "lr": 5e-06, "epoch": 1.4555712270803949, "percentage": 48.53, "elapsed_time": "9:31:55", "remaining_time": "10:06:30"}
260
+ {"current_steps": 2590, "total_steps": 5316, "loss": 0.4082, "lr": 5e-06, "epoch": 1.461212976022567, "percentage": 48.72, "elapsed_time": "9:34:06", "remaining_time": "10:04:15"}
261
+ {"current_steps": 2600, "total_steps": 5316, "loss": 0.4006, "lr": 5e-06, "epoch": 1.466854724964739, "percentage": 48.91, "elapsed_time": "9:36:17", "remaining_time": "10:02:00"}
262
+ {"current_steps": 2610, "total_steps": 5316, "loss": 0.4043, "lr": 5e-06, "epoch": 1.472496473906911, "percentage": 49.1, "elapsed_time": "9:38:29", "remaining_time": "9:59:45"}
263
+ {"current_steps": 2620, "total_steps": 5316, "loss": 0.4056, "lr": 5e-06, "epoch": 1.4781382228490831, "percentage": 49.29, "elapsed_time": "9:40:40", "remaining_time": "9:57:30"}
264
+ {"current_steps": 2630, "total_steps": 5316, "loss": 0.3907, "lr": 5e-06, "epoch": 1.4837799717912552, "percentage": 49.47, "elapsed_time": "9:42:51", "remaining_time": "9:55:15"}
265
+ {"current_steps": 2640, "total_steps": 5316, "loss": 0.3959, "lr": 5e-06, "epoch": 1.4894217207334273, "percentage": 49.66, "elapsed_time": "9:45:02", "remaining_time": "9:53:00"}
266
+ {"current_steps": 2650, "total_steps": 5316, "loss": 0.3934, "lr": 5e-06, "epoch": 1.4950634696755993, "percentage": 49.85, "elapsed_time": "9:47:13", "remaining_time": "9:50:46"}
267
+ {"current_steps": 2660, "total_steps": 5316, "loss": 0.4035, "lr": 5e-06, "epoch": 1.5007052186177714, "percentage": 50.04, "elapsed_time": "9:49:24", "remaining_time": "9:48:31"}
268
+ {"current_steps": 2670, "total_steps": 5316, "loss": 0.3993, "lr": 5e-06, "epoch": 1.5063469675599435, "percentage": 50.23, "elapsed_time": "9:51:35", "remaining_time": "9:46:16"}
269
+ {"current_steps": 2680, "total_steps": 5316, "loss": 0.3918, "lr": 5e-06, "epoch": 1.5119887165021155, "percentage": 50.41, "elapsed_time": "9:53:46", "remaining_time": "9:44:01"}
270
+ {"current_steps": 2690, "total_steps": 5316, "loss": 0.3909, "lr": 5e-06, "epoch": 1.5176304654442876, "percentage": 50.6, "elapsed_time": "9:55:57", "remaining_time": "9:41:47"}
271
+ {"current_steps": 2700, "total_steps": 5316, "loss": 0.3891, "lr": 5e-06, "epoch": 1.5232722143864597, "percentage": 50.79, "elapsed_time": "9:58:08", "remaining_time": "9:39:32"}
272
+ {"current_steps": 2710, "total_steps": 5316, "loss": 0.4063, "lr": 5e-06, "epoch": 1.5289139633286317, "percentage": 50.98, "elapsed_time": "10:00:20", "remaining_time": "9:37:17"}
273
+ {"current_steps": 2720, "total_steps": 5316, "loss": 0.4004, "lr": 5e-06, "epoch": 1.5345557122708038, "percentage": 51.17, "elapsed_time": "10:02:31", "remaining_time": "9:35:03"}
274
+ {"current_steps": 2730, "total_steps": 5316, "loss": 0.3926, "lr": 5e-06, "epoch": 1.540197461212976, "percentage": 51.35, "elapsed_time": "10:04:42", "remaining_time": "9:32:48"}
275
+ {"current_steps": 2740, "total_steps": 5316, "loss": 0.3896, "lr": 5e-06, "epoch": 1.5458392101551481, "percentage": 51.54, "elapsed_time": "10:06:53", "remaining_time": "9:30:34"}
276
+ {"current_steps": 2750, "total_steps": 5316, "loss": 0.3999, "lr": 5e-06, "epoch": 1.5514809590973202, "percentage": 51.73, "elapsed_time": "10:09:04", "remaining_time": "9:28:19"}
277
+ {"current_steps": 2760, "total_steps": 5316, "loss": 0.4077, "lr": 5e-06, "epoch": 1.5571227080394923, "percentage": 51.92, "elapsed_time": "10:11:16", "remaining_time": "9:26:05"}
278
+ {"current_steps": 2770, "total_steps": 5316, "loss": 0.4009, "lr": 5e-06, "epoch": 1.5627644569816643, "percentage": 52.11, "elapsed_time": "10:13:27", "remaining_time": "9:23:51"}
279
+ {"current_steps": 2780, "total_steps": 5316, "loss": 0.3924, "lr": 5e-06, "epoch": 1.5684062059238364, "percentage": 52.29, "elapsed_time": "10:15:38", "remaining_time": "9:21:36"}
280
+ {"current_steps": 2790, "total_steps": 5316, "loss": 0.3877, "lr": 5e-06, "epoch": 1.5740479548660085, "percentage": 52.48, "elapsed_time": "10:17:50", "remaining_time": "9:19:22"}
281
+ {"current_steps": 2800, "total_steps": 5316, "loss": 0.4036, "lr": 5e-06, "epoch": 1.5796897038081805, "percentage": 52.67, "elapsed_time": "10:20:01", "remaining_time": "9:17:08"}
282
+ {"current_steps": 2810, "total_steps": 5316, "loss": 0.3975, "lr": 5e-06, "epoch": 1.5853314527503526, "percentage": 52.86, "elapsed_time": "10:22:13", "remaining_time": "9:14:54"}
283
+ {"current_steps": 2820, "total_steps": 5316, "loss": 0.3918, "lr": 5e-06, "epoch": 1.5909732016925247, "percentage": 53.05, "elapsed_time": "10:24:24", "remaining_time": "9:12:40"}
284
+ {"current_steps": 2830, "total_steps": 5316, "loss": 0.3882, "lr": 5e-06, "epoch": 1.5966149506346967, "percentage": 53.24, "elapsed_time": "10:26:36", "remaining_time": "9:10:26"}
285
+ {"current_steps": 2840, "total_steps": 5316, "loss": 0.3986, "lr": 5e-06, "epoch": 1.6022566995768688, "percentage": 53.42, "elapsed_time": "10:28:47", "remaining_time": "9:08:11"}
286
+ {"current_steps": 2850, "total_steps": 5316, "loss": 0.4039, "lr": 5e-06, "epoch": 1.607898448519041, "percentage": 53.61, "elapsed_time": "10:30:58", "remaining_time": "9:05:57"}
287
+ {"current_steps": 2860, "total_steps": 5316, "loss": 0.4025, "lr": 5e-06, "epoch": 1.6135401974612131, "percentage": 53.8, "elapsed_time": "10:33:09", "remaining_time": "9:03:43"}
288
+ {"current_steps": 2870, "total_steps": 5316, "loss": 0.3905, "lr": 5e-06, "epoch": 1.6191819464033852, "percentage": 53.99, "elapsed_time": "10:35:20", "remaining_time": "9:01:29"}
289
+ {"current_steps": 2880, "total_steps": 5316, "loss": 0.4042, "lr": 5e-06, "epoch": 1.6248236953455573, "percentage": 54.18, "elapsed_time": "10:37:32", "remaining_time": "8:59:14"}
290
+ {"current_steps": 2890, "total_steps": 5316, "loss": 0.3962, "lr": 5e-06, "epoch": 1.6304654442877293, "percentage": 54.36, "elapsed_time": "10:39:43", "remaining_time": "8:57:00"}
291
+ {"current_steps": 2900, "total_steps": 5316, "loss": 0.3892, "lr": 5e-06, "epoch": 1.6361071932299014, "percentage": 54.55, "elapsed_time": "10:41:54", "remaining_time": "8:54:46"}
292
+ {"current_steps": 2910, "total_steps": 5316, "loss": 0.3881, "lr": 5e-06, "epoch": 1.6417489421720735, "percentage": 54.74, "elapsed_time": "10:44:05", "remaining_time": "8:52:32"}
293
+ {"current_steps": 2920, "total_steps": 5316, "loss": 0.394, "lr": 5e-06, "epoch": 1.6473906911142455, "percentage": 54.93, "elapsed_time": "10:46:16", "remaining_time": "8:50:18"}
294
+ {"current_steps": 2930, "total_steps": 5316, "loss": 0.3972, "lr": 5e-06, "epoch": 1.6530324400564176, "percentage": 55.12, "elapsed_time": "10:48:28", "remaining_time": "8:48:04"}
295
+ {"current_steps": 2940, "total_steps": 5316, "loss": 0.394, "lr": 5e-06, "epoch": 1.6586741889985896, "percentage": 55.3, "elapsed_time": "10:50:39", "remaining_time": "8:45:50"}
296
+ {"current_steps": 2950, "total_steps": 5316, "loss": 0.3862, "lr": 5e-06, "epoch": 1.6643159379407617, "percentage": 55.49, "elapsed_time": "10:52:50", "remaining_time": "8:43:36"}
297
+ {"current_steps": 2960, "total_steps": 5316, "loss": 0.4016, "lr": 5e-06, "epoch": 1.6699576868829338, "percentage": 55.68, "elapsed_time": "10:55:01", "remaining_time": "8:41:22"}
298
+ {"current_steps": 2970, "total_steps": 5316, "loss": 0.4095, "lr": 5e-06, "epoch": 1.6755994358251058, "percentage": 55.87, "elapsed_time": "10:57:13", "remaining_time": "8:39:08"}
299
+ {"current_steps": 2980, "total_steps": 5316, "loss": 0.3999, "lr": 5e-06, "epoch": 1.681241184767278, "percentage": 56.06, "elapsed_time": "10:59:24", "remaining_time": "8:36:54"}
300
+ {"current_steps": 2990, "total_steps": 5316, "loss": 0.3939, "lr": 5e-06, "epoch": 1.68688293370945, "percentage": 56.25, "elapsed_time": "11:01:35", "remaining_time": "8:34:40"}
301
+ {"current_steps": 3000, "total_steps": 5316, "loss": 0.4183, "lr": 5e-06, "epoch": 1.692524682651622, "percentage": 56.43, "elapsed_time": "11:03:46", "remaining_time": "8:32:26"}
302
+ {"current_steps": 3010, "total_steps": 5316, "loss": 0.3983, "lr": 5e-06, "epoch": 1.698166431593794, "percentage": 56.62, "elapsed_time": "11:05:58", "remaining_time": "8:30:12"}
303
+ {"current_steps": 3020, "total_steps": 5316, "loss": 0.3951, "lr": 5e-06, "epoch": 1.7038081805359662, "percentage": 56.81, "elapsed_time": "11:08:09", "remaining_time": "8:27:58"}
304
+ {"current_steps": 3030, "total_steps": 5316, "loss": 0.3961, "lr": 5e-06, "epoch": 1.7094499294781382, "percentage": 57.0, "elapsed_time": "11:10:20", "remaining_time": "8:25:44"}
305
+ {"current_steps": 3040, "total_steps": 5316, "loss": 0.3942, "lr": 5e-06, "epoch": 1.7150916784203103, "percentage": 57.19, "elapsed_time": "11:12:32", "remaining_time": "8:23:30"}
306
+ {"current_steps": 3050, "total_steps": 5316, "loss": 0.3946, "lr": 5e-06, "epoch": 1.7207334273624824, "percentage": 57.37, "elapsed_time": "11:14:43", "remaining_time": "8:21:17"}
307
+ {"current_steps": 3060, "total_steps": 5316, "loss": 0.3925, "lr": 5e-06, "epoch": 1.7263751763046544, "percentage": 57.56, "elapsed_time": "11:16:54", "remaining_time": "8:19:03"}
308
+ {"current_steps": 3070, "total_steps": 5316, "loss": 0.3959, "lr": 5e-06, "epoch": 1.7320169252468265, "percentage": 57.75, "elapsed_time": "11:19:05", "remaining_time": "8:16:49"}
309
+ {"current_steps": 3080, "total_steps": 5316, "loss": 0.4012, "lr": 5e-06, "epoch": 1.7376586741889986, "percentage": 57.94, "elapsed_time": "11:21:17", "remaining_time": "8:14:35"}
310
+ {"current_steps": 3090, "total_steps": 5316, "loss": 0.3881, "lr": 5e-06, "epoch": 1.7433004231311706, "percentage": 58.13, "elapsed_time": "11:23:28", "remaining_time": "8:12:21"}
311
+ {"current_steps": 3100, "total_steps": 5316, "loss": 0.4, "lr": 5e-06, "epoch": 1.7489421720733427, "percentage": 58.31, "elapsed_time": "11:25:39", "remaining_time": "8:10:08"}
312
+ {"current_steps": 3110, "total_steps": 5316, "loss": 0.3891, "lr": 5e-06, "epoch": 1.7545839210155147, "percentage": 58.5, "elapsed_time": "11:27:50", "remaining_time": "8:07:54"}
313
+ {"current_steps": 3120, "total_steps": 5316, "loss": 0.3926, "lr": 5e-06, "epoch": 1.7602256699576868, "percentage": 58.69, "elapsed_time": "11:30:02", "remaining_time": "8:05:40"}
314
+ {"current_steps": 3130, "total_steps": 5316, "loss": 0.419, "lr": 5e-06, "epoch": 1.7658674188998589, "percentage": 58.88, "elapsed_time": "11:32:13", "remaining_time": "8:03:26"}
315
+ {"current_steps": 3140, "total_steps": 5316, "loss": 0.3962, "lr": 5e-06, "epoch": 1.771509167842031, "percentage": 59.07, "elapsed_time": "11:34:24", "remaining_time": "8:01:13"}
316
+ {"current_steps": 3150, "total_steps": 5316, "loss": 0.3956, "lr": 5e-06, "epoch": 1.777150916784203, "percentage": 59.26, "elapsed_time": "11:36:35", "remaining_time": "7:58:59"}
317
+ {"current_steps": 3160, "total_steps": 5316, "loss": 0.4006, "lr": 5e-06, "epoch": 1.782792665726375, "percentage": 59.44, "elapsed_time": "11:38:47", "remaining_time": "7:56:45"}
318
+ {"current_steps": 3170, "total_steps": 5316, "loss": 0.3917, "lr": 5e-06, "epoch": 1.7884344146685471, "percentage": 59.63, "elapsed_time": "11:40:58", "remaining_time": "7:54:32"}
319
+ {"current_steps": 3180, "total_steps": 5316, "loss": 0.4085, "lr": 5e-06, "epoch": 1.7940761636107192, "percentage": 59.82, "elapsed_time": "11:43:09", "remaining_time": "7:52:18"}
320
+ {"current_steps": 3190, "total_steps": 5316, "loss": 0.4003, "lr": 5e-06, "epoch": 1.7997179125528913, "percentage": 60.01, "elapsed_time": "11:45:20", "remaining_time": "7:50:04"}
321
+ {"current_steps": 3200, "total_steps": 5316, "loss": 0.3955, "lr": 5e-06, "epoch": 1.8053596614950633, "percentage": 60.2, "elapsed_time": "11:47:31", "remaining_time": "7:47:51"}
322
+ {"current_steps": 3210, "total_steps": 5316, "loss": 0.4016, "lr": 5e-06, "epoch": 1.8110014104372354, "percentage": 60.38, "elapsed_time": "11:49:42", "remaining_time": "7:45:37"}
323
+ {"current_steps": 3220, "total_steps": 5316, "loss": 0.4033, "lr": 5e-06, "epoch": 1.8166431593794075, "percentage": 60.57, "elapsed_time": "11:51:53", "remaining_time": "7:43:23"}
324
+ {"current_steps": 3230, "total_steps": 5316, "loss": 0.3927, "lr": 5e-06, "epoch": 1.8222849083215797, "percentage": 60.76, "elapsed_time": "11:54:05", "remaining_time": "7:41:10"}
325
+ {"current_steps": 3240, "total_steps": 5316, "loss": 0.3963, "lr": 5e-06, "epoch": 1.8279266572637518, "percentage": 60.95, "elapsed_time": "11:56:16", "remaining_time": "7:38:56"}
326
+ {"current_steps": 3250, "total_steps": 5316, "loss": 0.3983, "lr": 5e-06, "epoch": 1.8335684062059239, "percentage": 61.14, "elapsed_time": "11:58:27", "remaining_time": "7:36:43"}
327
+ {"current_steps": 3260, "total_steps": 5316, "loss": 0.3963, "lr": 5e-06, "epoch": 1.839210155148096, "percentage": 61.32, "elapsed_time": "12:00:39", "remaining_time": "7:34:29"}
328
+ {"current_steps": 3270, "total_steps": 5316, "loss": 0.398, "lr": 5e-06, "epoch": 1.844851904090268, "percentage": 61.51, "elapsed_time": "12:02:50", "remaining_time": "7:32:16"}
329
+ {"current_steps": 3280, "total_steps": 5316, "loss": 0.3804, "lr": 5e-06, "epoch": 1.85049365303244, "percentage": 61.7, "elapsed_time": "12:05:01", "remaining_time": "7:30:03"}
330
+ {"current_steps": 3290, "total_steps": 5316, "loss": 0.3866, "lr": 5e-06, "epoch": 1.8561354019746121, "percentage": 61.89, "elapsed_time": "12:07:13", "remaining_time": "7:27:49"}
331
+ {"current_steps": 3300, "total_steps": 5316, "loss": 0.4071, "lr": 5e-06, "epoch": 1.8617771509167842, "percentage": 62.08, "elapsed_time": "12:09:24", "remaining_time": "7:25:36"}
332
+ {"current_steps": 3310, "total_steps": 5316, "loss": 0.4007, "lr": 5e-06, "epoch": 1.8674188998589563, "percentage": 62.26, "elapsed_time": "12:11:36", "remaining_time": "7:23:22"}
333
+ {"current_steps": 3320, "total_steps": 5316, "loss": 0.3958, "lr": 5e-06, "epoch": 1.8730606488011283, "percentage": 62.45, "elapsed_time": "12:13:47", "remaining_time": "7:21:09"}
334
+ {"current_steps": 3330, "total_steps": 5316, "loss": 0.3965, "lr": 5e-06, "epoch": 1.8787023977433004, "percentage": 62.64, "elapsed_time": "12:15:58", "remaining_time": "7:18:56"}
335
+ {"current_steps": 3340, "total_steps": 5316, "loss": 0.3972, "lr": 5e-06, "epoch": 1.8843441466854725, "percentage": 62.83, "elapsed_time": "12:18:10", "remaining_time": "7:16:42"}
336
+ {"current_steps": 3350, "total_steps": 5316, "loss": 0.3943, "lr": 5e-06, "epoch": 1.8899858956276445, "percentage": 63.02, "elapsed_time": "12:20:21", "remaining_time": "7:14:29"}
337
+ {"current_steps": 3360, "total_steps": 5316, "loss": 0.391, "lr": 5e-06, "epoch": 1.8956276445698168, "percentage": 63.21, "elapsed_time": "12:22:32", "remaining_time": "7:12:15"}
338
+ {"current_steps": 3370, "total_steps": 5316, "loss": 0.4039, "lr": 5e-06, "epoch": 1.9012693935119889, "percentage": 63.39, "elapsed_time": "12:24:43", "remaining_time": "7:10:02"}
339
+ {"current_steps": 3380, "total_steps": 5316, "loss": 0.3988, "lr": 5e-06, "epoch": 1.906911142454161, "percentage": 63.58, "elapsed_time": "12:26:54", "remaining_time": "7:07:49"}
340
+ {"current_steps": 3390, "total_steps": 5316, "loss": 0.3884, "lr": 5e-06, "epoch": 1.912552891396333, "percentage": 63.77, "elapsed_time": "12:29:05", "remaining_time": "7:05:35"}
341
+ {"current_steps": 3400, "total_steps": 5316, "loss": 0.4066, "lr": 5e-06, "epoch": 1.918194640338505, "percentage": 63.96, "elapsed_time": "12:31:16", "remaining_time": "7:03:22"}
342
+ {"current_steps": 3410, "total_steps": 5316, "loss": 0.3973, "lr": 5e-06, "epoch": 1.9238363892806771, "percentage": 64.15, "elapsed_time": "12:33:27", "remaining_time": "7:01:08"}
343
+ {"current_steps": 3420, "total_steps": 5316, "loss": 0.3929, "lr": 5e-06, "epoch": 1.9294781382228492, "percentage": 64.33, "elapsed_time": "12:35:39", "remaining_time": "6:58:55"}
344
+ {"current_steps": 3430, "total_steps": 5316, "loss": 0.3857, "lr": 5e-06, "epoch": 1.9351198871650213, "percentage": 64.52, "elapsed_time": "12:37:50", "remaining_time": "6:56:42"}
345
+ {"current_steps": 3440, "total_steps": 5316, "loss": 0.3904, "lr": 5e-06, "epoch": 1.9407616361071933, "percentage": 64.71, "elapsed_time": "12:40:01", "remaining_time": "6:54:28"}
346
+ {"current_steps": 3450, "total_steps": 5316, "loss": 0.3901, "lr": 5e-06, "epoch": 1.9464033850493654, "percentage": 64.9, "elapsed_time": "12:42:12", "remaining_time": "6:52:15"}
347
+ {"current_steps": 3460, "total_steps": 5316, "loss": 0.3883, "lr": 5e-06, "epoch": 1.9520451339915375, "percentage": 65.09, "elapsed_time": "12:44:24", "remaining_time": "6:50:02"}
348
+ {"current_steps": 3470, "total_steps": 5316, "loss": 0.3933, "lr": 5e-06, "epoch": 1.9576868829337095, "percentage": 65.27, "elapsed_time": "12:46:35", "remaining_time": "6:47:49"}
349
+ {"current_steps": 3480, "total_steps": 5316, "loss": 0.4084, "lr": 5e-06, "epoch": 1.9633286318758816, "percentage": 65.46, "elapsed_time": "12:48:46", "remaining_time": "6:45:35"}
350
+ {"current_steps": 3490, "total_steps": 5316, "loss": 0.3829, "lr": 5e-06, "epoch": 1.9689703808180536, "percentage": 65.65, "elapsed_time": "12:50:58", "remaining_time": "6:43:22"}
351
+ {"current_steps": 3500, "total_steps": 5316, "loss": 0.3924, "lr": 5e-06, "epoch": 1.9746121297602257, "percentage": 65.84, "elapsed_time": "12:53:09", "remaining_time": "6:41:09"}
352
+ {"current_steps": 3510, "total_steps": 5316, "loss": 0.3894, "lr": 5e-06, "epoch": 1.9802538787023978, "percentage": 66.03, "elapsed_time": "12:55:21", "remaining_time": "6:38:56"}
353
+ {"current_steps": 3520, "total_steps": 5316, "loss": 0.4027, "lr": 5e-06, "epoch": 1.9858956276445698, "percentage": 66.22, "elapsed_time": "12:57:32", "remaining_time": "6:36:43"}
354
+ {"current_steps": 3530, "total_steps": 5316, "loss": 0.3859, "lr": 5e-06, "epoch": 1.991537376586742, "percentage": 66.4, "elapsed_time": "12:59:44", "remaining_time": "6:34:30"}
355
+ {"current_steps": 3540, "total_steps": 5316, "loss": 0.3944, "lr": 5e-06, "epoch": 1.997179125528914, "percentage": 66.59, "elapsed_time": "13:01:55", "remaining_time": "6:32:17"}
356
+ {"current_steps": 3545, "total_steps": 5316, "eval_loss": 0.43374887108802795, "epoch": 2.0, "percentage": 66.69, "elapsed_time": "13:10:28", "remaining_time": "6:34:53"}
357
+ {"current_steps": 3550, "total_steps": 5316, "loss": 0.3928, "lr": 5e-06, "epoch": 2.002820874471086, "percentage": 66.78, "elapsed_time": "13:12:41", "remaining_time": "6:34:20"}