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
61489bb
·
verified ·
1 Parent(s): 6fd93e9

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 +176 -0
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b3c29d3537bf106b4606fb06a15dcc338a026c5e2b6465c07bb0ea1386578ef6
3
  size 4877660776
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae86f77d1632e6e91291bf2752dd442fcd5e451f3971a7c79a2ea16c6d1eb8cc
3
  size 4877660776
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:989381031726a828e90c6c0d5fbe86c54d6694499c72e677e2ccae58ec39015d
3
  size 4932751008
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b53acd1e459671113652ffb74b84952f8f182ce9325238e2fabd3f451dd28bdc
3
  size 4932751008
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:20431c183f2cb0bc9cebed70cf843cca533f9ca666e2d9f5b6ccf739ffdb5536
3
  size 4330865200
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:133f1fce880b7c0edb4b13ea9e6d9674f3a3e7b5b23d91d9f2beeaed58fef76a
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:f5c855c01b949a2397111b1059756af9c285d61b2823bba1555fd8f020884c28
3
  size 1089994880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c621d5b583e97bd5e4e79f142e2d7d316898455919a9e4e529cc0a4aeea193f5
3
  size 1089994880
trainer_log.jsonl CHANGED
@@ -355,3 +355,179 @@
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"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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"}
358
+ {"current_steps": 3560, "total_steps": 5316, "loss": 0.3547, "lr": 5e-06, "epoch": 2.008462623413258, "percentage": 66.97, "elapsed_time": "13:14:53", "remaining_time": "6:32:04"}
359
+ {"current_steps": 3570, "total_steps": 5316, "loss": 0.356, "lr": 5e-06, "epoch": 2.01410437235543, "percentage": 67.16, "elapsed_time": "13:17:04", "remaining_time": "6:29:49"}
360
+ {"current_steps": 3580, "total_steps": 5316, "loss": 0.3528, "lr": 5e-06, "epoch": 2.0197461212976022, "percentage": 67.34, "elapsed_time": "13:19:15", "remaining_time": "6:27:34"}
361
+ {"current_steps": 3590, "total_steps": 5316, "loss": 0.3442, "lr": 5e-06, "epoch": 2.0253878702397743, "percentage": 67.53, "elapsed_time": "13:21:27", "remaining_time": "6:25:19"}
362
+ {"current_steps": 3600, "total_steps": 5316, "loss": 0.3432, "lr": 5e-06, "epoch": 2.0310296191819464, "percentage": 67.72, "elapsed_time": "13:23:39", "remaining_time": "6:23:04"}
363
+ {"current_steps": 3610, "total_steps": 5316, "loss": 0.3466, "lr": 5e-06, "epoch": 2.0366713681241184, "percentage": 67.91, "elapsed_time": "13:25:50", "remaining_time": "6:20:49"}
364
+ {"current_steps": 3620, "total_steps": 5316, "loss": 0.3504, "lr": 5e-06, "epoch": 2.0423131170662905, "percentage": 68.1, "elapsed_time": "13:28:02", "remaining_time": "6:18:34"}
365
+ {"current_steps": 3630, "total_steps": 5316, "loss": 0.3406, "lr": 5e-06, "epoch": 2.0479548660084625, "percentage": 68.28, "elapsed_time": "13:30:13", "remaining_time": "6:16:19"}
366
+ {"current_steps": 3640, "total_steps": 5316, "loss": 0.3431, "lr": 5e-06, "epoch": 2.0535966149506346, "percentage": 68.47, "elapsed_time": "13:32:25", "remaining_time": "6:14:04"}
367
+ {"current_steps": 3650, "total_steps": 5316, "loss": 0.3492, "lr": 5e-06, "epoch": 2.0592383638928067, "percentage": 68.66, "elapsed_time": "13:34:36", "remaining_time": "6:11:49"}
368
+ {"current_steps": 3660, "total_steps": 5316, "loss": 0.3411, "lr": 5e-06, "epoch": 2.0648801128349787, "percentage": 68.85, "elapsed_time": "13:36:48", "remaining_time": "6:09:34"}
369
+ {"current_steps": 3670, "total_steps": 5316, "loss": 0.3476, "lr": 5e-06, "epoch": 2.070521861777151, "percentage": 69.04, "elapsed_time": "13:38:59", "remaining_time": "6:07:19"}
370
+ {"current_steps": 3680, "total_steps": 5316, "loss": 0.3446, "lr": 5e-06, "epoch": 2.076163610719323, "percentage": 69.22, "elapsed_time": "13:41:10", "remaining_time": "6:05:04"}
371
+ {"current_steps": 3690, "total_steps": 5316, "loss": 0.3511, "lr": 5e-06, "epoch": 2.081805359661495, "percentage": 69.41, "elapsed_time": "13:43:22", "remaining_time": "6:02:49"}
372
+ {"current_steps": 3700, "total_steps": 5316, "loss": 0.3701, "lr": 5e-06, "epoch": 2.087447108603667, "percentage": 69.6, "elapsed_time": "13:45:33", "remaining_time": "6:00:34"}
373
+ {"current_steps": 3710, "total_steps": 5316, "loss": 0.3563, "lr": 5e-06, "epoch": 2.093088857545839, "percentage": 69.79, "elapsed_time": "13:47:45", "remaining_time": "5:58:19"}
374
+ {"current_steps": 3720, "total_steps": 5316, "loss": 0.3471, "lr": 5e-06, "epoch": 2.098730606488011, "percentage": 69.98, "elapsed_time": "13:49:56", "remaining_time": "5:56:04"}
375
+ {"current_steps": 3730, "total_steps": 5316, "loss": 0.3567, "lr": 5e-06, "epoch": 2.104372355430183, "percentage": 70.17, "elapsed_time": "13:52:08", "remaining_time": "5:53:49"}
376
+ {"current_steps": 3740, "total_steps": 5316, "loss": 0.3529, "lr": 5e-06, "epoch": 2.1100141043723553, "percentage": 70.35, "elapsed_time": "13:54:19", "remaining_time": "5:51:34"}
377
+ {"current_steps": 3750, "total_steps": 5316, "loss": 0.3501, "lr": 5e-06, "epoch": 2.1156558533145273, "percentage": 70.54, "elapsed_time": "13:56:31", "remaining_time": "5:49:19"}
378
+ {"current_steps": 3760, "total_steps": 5316, "loss": 0.3498, "lr": 5e-06, "epoch": 2.1212976022566994, "percentage": 70.73, "elapsed_time": "13:58:42", "remaining_time": "5:47:05"}
379
+ {"current_steps": 3770, "total_steps": 5316, "loss": 0.3519, "lr": 5e-06, "epoch": 2.1269393511988715, "percentage": 70.92, "elapsed_time": "14:00:54", "remaining_time": "5:44:50"}
380
+ {"current_steps": 3780, "total_steps": 5316, "loss": 0.3476, "lr": 5e-06, "epoch": 2.1325811001410435, "percentage": 71.11, "elapsed_time": "14:03:05", "remaining_time": "5:42:35"}
381
+ {"current_steps": 3790, "total_steps": 5316, "loss": 0.3519, "lr": 5e-06, "epoch": 2.138222849083216, "percentage": 71.29, "elapsed_time": "14:05:16", "remaining_time": "5:40:20"}
382
+ {"current_steps": 3800, "total_steps": 5316, "loss": 0.3564, "lr": 5e-06, "epoch": 2.143864598025388, "percentage": 71.48, "elapsed_time": "14:07:27", "remaining_time": "5:38:05"}
383
+ {"current_steps": 3810, "total_steps": 5316, "loss": 0.3518, "lr": 5e-06, "epoch": 2.14950634696756, "percentage": 71.67, "elapsed_time": "14:09:39", "remaining_time": "5:35:50"}
384
+ {"current_steps": 3820, "total_steps": 5316, "loss": 0.3445, "lr": 5e-06, "epoch": 2.155148095909732, "percentage": 71.86, "elapsed_time": "14:11:50", "remaining_time": "5:33:36"}
385
+ {"current_steps": 3830, "total_steps": 5316, "loss": 0.3445, "lr": 5e-06, "epoch": 2.1607898448519043, "percentage": 72.05, "elapsed_time": "14:14:01", "remaining_time": "5:31:21"}
386
+ {"current_steps": 3840, "total_steps": 5316, "loss": 0.356, "lr": 5e-06, "epoch": 2.1664315937940763, "percentage": 72.23, "elapsed_time": "14:16:13", "remaining_time": "5:29:06"}
387
+ {"current_steps": 3850, "total_steps": 5316, "loss": 0.3662, "lr": 5e-06, "epoch": 2.1720733427362484, "percentage": 72.42, "elapsed_time": "14:18:24", "remaining_time": "5:26:51"}
388
+ {"current_steps": 3860, "total_steps": 5316, "loss": 0.3467, "lr": 5e-06, "epoch": 2.1777150916784205, "percentage": 72.61, "elapsed_time": "14:20:35", "remaining_time": "5:24:37"}
389
+ {"current_steps": 3870, "total_steps": 5316, "loss": 0.3448, "lr": 5e-06, "epoch": 2.1833568406205925, "percentage": 72.8, "elapsed_time": "14:22:47", "remaining_time": "5:22:22"}
390
+ {"current_steps": 3880, "total_steps": 5316, "loss": 0.3477, "lr": 5e-06, "epoch": 2.1889985895627646, "percentage": 72.99, "elapsed_time": "14:24:58", "remaining_time": "5:20:07"}
391
+ {"current_steps": 3890, "total_steps": 5316, "loss": 0.3495, "lr": 5e-06, "epoch": 2.1946403385049367, "percentage": 73.18, "elapsed_time": "14:27:09", "remaining_time": "5:17:53"}
392
+ {"current_steps": 3900, "total_steps": 5316, "loss": 0.3487, "lr": 5e-06, "epoch": 2.2002820874471087, "percentage": 73.36, "elapsed_time": "14:29:21", "remaining_time": "5:15:38"}
393
+ {"current_steps": 3910, "total_steps": 5316, "loss": 0.3524, "lr": 5e-06, "epoch": 2.205923836389281, "percentage": 73.55, "elapsed_time": "14:31:33", "remaining_time": "5:13:24"}
394
+ {"current_steps": 3920, "total_steps": 5316, "loss": 0.3471, "lr": 5e-06, "epoch": 2.211565585331453, "percentage": 73.74, "elapsed_time": "14:33:44", "remaining_time": "5:11:09"}
395
+ {"current_steps": 3930, "total_steps": 5316, "loss": 0.3592, "lr": 5e-06, "epoch": 2.217207334273625, "percentage": 73.93, "elapsed_time": "14:35:55", "remaining_time": "5:08:54"}
396
+ {"current_steps": 3940, "total_steps": 5316, "loss": 0.3557, "lr": 5e-06, "epoch": 2.222849083215797, "percentage": 74.12, "elapsed_time": "14:38:07", "remaining_time": "5:06:40"}
397
+ {"current_steps": 3950, "total_steps": 5316, "loss": 0.3423, "lr": 5e-06, "epoch": 2.228490832157969, "percentage": 74.3, "elapsed_time": "14:40:18", "remaining_time": "5:04:25"}
398
+ {"current_steps": 3960, "total_steps": 5316, "loss": 0.3546, "lr": 5e-06, "epoch": 2.234132581100141, "percentage": 74.49, "elapsed_time": "14:42:29", "remaining_time": "5:02:11"}
399
+ {"current_steps": 3970, "total_steps": 5316, "loss": 0.3567, "lr": 5e-06, "epoch": 2.239774330042313, "percentage": 74.68, "elapsed_time": "14:44:41", "remaining_time": "4:59:56"}
400
+ {"current_steps": 3980, "total_steps": 5316, "loss": 0.3584, "lr": 5e-06, "epoch": 2.2454160789844853, "percentage": 74.87, "elapsed_time": "14:46:52", "remaining_time": "4:57:42"}
401
+ {"current_steps": 3990, "total_steps": 5316, "loss": 0.3486, "lr": 5e-06, "epoch": 2.2510578279266573, "percentage": 75.06, "elapsed_time": "14:49:04", "remaining_time": "4:55:27"}
402
+ {"current_steps": 4000, "total_steps": 5316, "loss": 0.3636, "lr": 5e-06, "epoch": 2.2566995768688294, "percentage": 75.24, "elapsed_time": "14:51:15", "remaining_time": "4:53:13"}
403
+ {"current_steps": 4010, "total_steps": 5316, "loss": 0.3537, "lr": 5e-06, "epoch": 2.2623413258110014, "percentage": 75.43, "elapsed_time": "14:53:26", "remaining_time": "4:50:58"}
404
+ {"current_steps": 4020, "total_steps": 5316, "loss": 0.3501, "lr": 5e-06, "epoch": 2.2679830747531735, "percentage": 75.62, "elapsed_time": "14:55:38", "remaining_time": "4:48:44"}
405
+ {"current_steps": 4030, "total_steps": 5316, "loss": 0.3541, "lr": 5e-06, "epoch": 2.2736248236953456, "percentage": 75.81, "elapsed_time": "14:57:49", "remaining_time": "4:46:30"}
406
+ {"current_steps": 4040, "total_steps": 5316, "loss": 0.351, "lr": 5e-06, "epoch": 2.2792665726375176, "percentage": 76.0, "elapsed_time": "15:00:00", "remaining_time": "4:44:15"}
407
+ {"current_steps": 4050, "total_steps": 5316, "loss": 0.3485, "lr": 5e-06, "epoch": 2.2849083215796897, "percentage": 76.19, "elapsed_time": "15:02:12", "remaining_time": "4:42:01"}
408
+ {"current_steps": 4060, "total_steps": 5316, "loss": 0.3502, "lr": 5e-06, "epoch": 2.2905500705218618, "percentage": 76.37, "elapsed_time": "15:04:23", "remaining_time": "4:39:46"}
409
+ {"current_steps": 4070, "total_steps": 5316, "loss": 0.3651, "lr": 5e-06, "epoch": 2.296191819464034, "percentage": 76.56, "elapsed_time": "15:06:34", "remaining_time": "4:37:32"}
410
+ {"current_steps": 4080, "total_steps": 5316, "loss": 0.3544, "lr": 5e-06, "epoch": 2.301833568406206, "percentage": 76.75, "elapsed_time": "15:08:46", "remaining_time": "4:35:18"}
411
+ {"current_steps": 4090, "total_steps": 5316, "loss": 0.3685, "lr": 5e-06, "epoch": 2.307475317348378, "percentage": 76.94, "elapsed_time": "15:10:57", "remaining_time": "4:33:03"}
412
+ {"current_steps": 4100, "total_steps": 5316, "loss": 0.3619, "lr": 5e-06, "epoch": 2.31311706629055, "percentage": 77.13, "elapsed_time": "15:13:09", "remaining_time": "4:30:49"}
413
+ {"current_steps": 4110, "total_steps": 5316, "loss": 0.352, "lr": 5e-06, "epoch": 2.318758815232722, "percentage": 77.31, "elapsed_time": "15:15:20", "remaining_time": "4:28:35"}
414
+ {"current_steps": 4120, "total_steps": 5316, "loss": 0.3383, "lr": 5e-06, "epoch": 2.324400564174894, "percentage": 77.5, "elapsed_time": "15:17:31", "remaining_time": "4:26:21"}
415
+ {"current_steps": 4130, "total_steps": 5316, "loss": 0.3536, "lr": 5e-06, "epoch": 2.330042313117066, "percentage": 77.69, "elapsed_time": "15:19:42", "remaining_time": "4:24:06"}
416
+ {"current_steps": 4140, "total_steps": 5316, "loss": 0.3563, "lr": 5e-06, "epoch": 2.3356840620592383, "percentage": 77.88, "elapsed_time": "15:21:54", "remaining_time": "4:21:52"}
417
+ {"current_steps": 4150, "total_steps": 5316, "loss": 0.3467, "lr": 5e-06, "epoch": 2.3413258110014104, "percentage": 78.07, "elapsed_time": "15:24:05", "remaining_time": "4:19:38"}
418
+ {"current_steps": 4160, "total_steps": 5316, "loss": 0.3503, "lr": 5e-06, "epoch": 2.3469675599435824, "percentage": 78.25, "elapsed_time": "15:26:16", "remaining_time": "4:17:23"}
419
+ {"current_steps": 4170, "total_steps": 5316, "loss": 0.352, "lr": 5e-06, "epoch": 2.3526093088857545, "percentage": 78.44, "elapsed_time": "15:28:28", "remaining_time": "4:15:09"}
420
+ {"current_steps": 4180, "total_steps": 5316, "loss": 0.3774, "lr": 5e-06, "epoch": 2.3582510578279265, "percentage": 78.63, "elapsed_time": "15:30:39", "remaining_time": "4:12:55"}
421
+ {"current_steps": 4190, "total_steps": 5316, "loss": 0.3598, "lr": 5e-06, "epoch": 2.3638928067700986, "percentage": 78.82, "elapsed_time": "15:32:50", "remaining_time": "4:10:41"}
422
+ {"current_steps": 4200, "total_steps": 5316, "loss": 0.362, "lr": 5e-06, "epoch": 2.3695345557122707, "percentage": 79.01, "elapsed_time": "15:35:02", "remaining_time": "4:08:27"}
423
+ {"current_steps": 4210, "total_steps": 5316, "loss": 0.354, "lr": 5e-06, "epoch": 2.3751763046544427, "percentage": 79.19, "elapsed_time": "15:37:13", "remaining_time": "4:06:12"}
424
+ {"current_steps": 4220, "total_steps": 5316, "loss": 0.3602, "lr": 5e-06, "epoch": 2.380818053596615, "percentage": 79.38, "elapsed_time": "15:39:24", "remaining_time": "4:03:58"}
425
+ {"current_steps": 4230, "total_steps": 5316, "loss": 0.3516, "lr": 5e-06, "epoch": 2.386459802538787, "percentage": 79.57, "elapsed_time": "15:41:35", "remaining_time": "4:01:44"}
426
+ {"current_steps": 4240, "total_steps": 5316, "loss": 0.3489, "lr": 5e-06, "epoch": 2.392101551480959, "percentage": 79.76, "elapsed_time": "15:43:47", "remaining_time": "3:59:30"}
427
+ {"current_steps": 4250, "total_steps": 5316, "loss": 0.3566, "lr": 5e-06, "epoch": 2.397743300423131, "percentage": 79.95, "elapsed_time": "15:45:58", "remaining_time": "3:57:16"}
428
+ {"current_steps": 4260, "total_steps": 5316, "loss": 0.355, "lr": 5e-06, "epoch": 2.403385049365303, "percentage": 80.14, "elapsed_time": "15:48:09", "remaining_time": "3:55:02"}
429
+ {"current_steps": 4270, "total_steps": 5316, "loss": 0.3606, "lr": 5e-06, "epoch": 2.409026798307475, "percentage": 80.32, "elapsed_time": "15:50:20", "remaining_time": "3:52:48"}
430
+ {"current_steps": 4280, "total_steps": 5316, "loss": 0.3551, "lr": 5e-06, "epoch": 2.414668547249647, "percentage": 80.51, "elapsed_time": "15:52:31", "remaining_time": "3:50:33"}
431
+ {"current_steps": 4290, "total_steps": 5316, "loss": 0.3521, "lr": 5e-06, "epoch": 2.4203102961918193, "percentage": 80.7, "elapsed_time": "15:54:42", "remaining_time": "3:48:19"}
432
+ {"current_steps": 4300, "total_steps": 5316, "loss": 0.3736, "lr": 5e-06, "epoch": 2.4259520451339913, "percentage": 80.89, "elapsed_time": "15:56:53", "remaining_time": "3:46:05"}
433
+ {"current_steps": 4310, "total_steps": 5316, "loss": 0.3467, "lr": 5e-06, "epoch": 2.4315937940761634, "percentage": 81.08, "elapsed_time": "15:59:04", "remaining_time": "3:43:51"}
434
+ {"current_steps": 4320, "total_steps": 5316, "loss": 0.348, "lr": 5e-06, "epoch": 2.4372355430183354, "percentage": 81.26, "elapsed_time": "16:01:15", "remaining_time": "3:41:37"}
435
+ {"current_steps": 4330, "total_steps": 5316, "loss": 0.3521, "lr": 5e-06, "epoch": 2.4428772919605075, "percentage": 81.45, "elapsed_time": "16:03:26", "remaining_time": "3:39:23"}
436
+ {"current_steps": 4340, "total_steps": 5316, "loss": 0.3456, "lr": 5e-06, "epoch": 2.44851904090268, "percentage": 81.64, "elapsed_time": "16:05:37", "remaining_time": "3:37:09"}
437
+ {"current_steps": 4350, "total_steps": 5316, "loss": 0.3433, "lr": 5e-06, "epoch": 2.454160789844852, "percentage": 81.83, "elapsed_time": "16:07:49", "remaining_time": "3:34:55"}
438
+ {"current_steps": 4360, "total_steps": 5316, "loss": 0.349, "lr": 5e-06, "epoch": 2.459802538787024, "percentage": 82.02, "elapsed_time": "16:10:00", "remaining_time": "3:32:41"}
439
+ {"current_steps": 4370, "total_steps": 5316, "loss": 0.3578, "lr": 5e-06, "epoch": 2.465444287729196, "percentage": 82.2, "elapsed_time": "16:12:12", "remaining_time": "3:30:27"}
440
+ {"current_steps": 4380, "total_steps": 5316, "loss": 0.3485, "lr": 5e-06, "epoch": 2.4710860366713683, "percentage": 82.39, "elapsed_time": "16:14:23", "remaining_time": "3:28:13"}
441
+ {"current_steps": 4390, "total_steps": 5316, "loss": 0.3471, "lr": 5e-06, "epoch": 2.4767277856135403, "percentage": 82.58, "elapsed_time": "16:16:34", "remaining_time": "3:25:59"}
442
+ {"current_steps": 4400, "total_steps": 5316, "loss": 0.3562, "lr": 5e-06, "epoch": 2.4823695345557124, "percentage": 82.77, "elapsed_time": "16:18:45", "remaining_time": "3:23:45"}
443
+ {"current_steps": 4410, "total_steps": 5316, "loss": 0.3482, "lr": 5e-06, "epoch": 2.4880112834978845, "percentage": 82.96, "elapsed_time": "16:20:56", "remaining_time": "3:21:31"}
444
+ {"current_steps": 4420, "total_steps": 5316, "loss": 0.355, "lr": 5e-06, "epoch": 2.4936530324400565, "percentage": 83.15, "elapsed_time": "16:23:07", "remaining_time": "3:19:17"}
445
+ {"current_steps": 4430, "total_steps": 5316, "loss": 0.3538, "lr": 5e-06, "epoch": 2.4992947813822286, "percentage": 83.33, "elapsed_time": "16:25:18", "remaining_time": "3:17:03"}
446
+ {"current_steps": 4440, "total_steps": 5316, "loss": 0.3747, "lr": 5e-06, "epoch": 2.5049365303244007, "percentage": 83.52, "elapsed_time": "16:27:29", "remaining_time": "3:14:49"}
447
+ {"current_steps": 4450, "total_steps": 5316, "loss": 0.3519, "lr": 5e-06, "epoch": 2.5105782792665727, "percentage": 83.71, "elapsed_time": "16:29:40", "remaining_time": "3:12:35"}
448
+ {"current_steps": 4460, "total_steps": 5316, "loss": 0.342, "lr": 5e-06, "epoch": 2.516220028208745, "percentage": 83.9, "elapsed_time": "16:31:51", "remaining_time": "3:10:22"}
449
+ {"current_steps": 4470, "total_steps": 5316, "loss": 0.3607, "lr": 5e-06, "epoch": 2.521861777150917, "percentage": 84.09, "elapsed_time": "16:34:02", "remaining_time": "3:08:08"}
450
+ {"current_steps": 4480, "total_steps": 5316, "loss": 0.3468, "lr": 5e-06, "epoch": 2.527503526093089, "percentage": 84.27, "elapsed_time": "16:36:13", "remaining_time": "3:05:54"}
451
+ {"current_steps": 4490, "total_steps": 5316, "loss": 0.3443, "lr": 5e-06, "epoch": 2.533145275035261, "percentage": 84.46, "elapsed_time": "16:38:24", "remaining_time": "3:03:40"}
452
+ {"current_steps": 4500, "total_steps": 5316, "loss": 0.3524, "lr": 5e-06, "epoch": 2.538787023977433, "percentage": 84.65, "elapsed_time": "16:40:35", "remaining_time": "3:01:26"}
453
+ {"current_steps": 4510, "total_steps": 5316, "loss": 0.3479, "lr": 5e-06, "epoch": 2.544428772919605, "percentage": 84.84, "elapsed_time": "16:42:46", "remaining_time": "2:59:12"}
454
+ {"current_steps": 4520, "total_steps": 5316, "loss": 0.3695, "lr": 5e-06, "epoch": 2.550070521861777, "percentage": 85.03, "elapsed_time": "16:44:57", "remaining_time": "2:56:58"}
455
+ {"current_steps": 4530, "total_steps": 5316, "loss": 0.353, "lr": 5e-06, "epoch": 2.5557122708039492, "percentage": 85.21, "elapsed_time": "16:47:08", "remaining_time": "2:54:44"}
456
+ {"current_steps": 4540, "total_steps": 5316, "loss": 0.3539, "lr": 5e-06, "epoch": 2.5613540197461213, "percentage": 85.4, "elapsed_time": "16:49:19", "remaining_time": "2:52:31"}
457
+ {"current_steps": 4550, "total_steps": 5316, "loss": 0.3598, "lr": 5e-06, "epoch": 2.5669957686882934, "percentage": 85.59, "elapsed_time": "16:51:30", "remaining_time": "2:50:17"}
458
+ {"current_steps": 4560, "total_steps": 5316, "loss": 0.3485, "lr": 5e-06, "epoch": 2.5726375176304654, "percentage": 85.78, "elapsed_time": "16:53:41", "remaining_time": "2:48:03"}
459
+ {"current_steps": 4570, "total_steps": 5316, "loss": 0.3479, "lr": 5e-06, "epoch": 2.5782792665726375, "percentage": 85.97, "elapsed_time": "16:55:52", "remaining_time": "2:45:49"}
460
+ {"current_steps": 4580, "total_steps": 5316, "loss": 0.3543, "lr": 5e-06, "epoch": 2.5839210155148096, "percentage": 86.16, "elapsed_time": "16:58:03", "remaining_time": "2:43:36"}
461
+ {"current_steps": 4590, "total_steps": 5316, "loss": 0.3475, "lr": 5e-06, "epoch": 2.5895627644569816, "percentage": 86.34, "elapsed_time": "17:00:14", "remaining_time": "2:41:22"}
462
+ {"current_steps": 4600, "total_steps": 5316, "loss": 0.3484, "lr": 5e-06, "epoch": 2.5952045133991537, "percentage": 86.53, "elapsed_time": "17:02:25", "remaining_time": "2:39:08"}
463
+ {"current_steps": 4610, "total_steps": 5316, "loss": 0.3478, "lr": 5e-06, "epoch": 2.6008462623413258, "percentage": 86.72, "elapsed_time": "17:04:36", "remaining_time": "2:36:54"}
464
+ {"current_steps": 4620, "total_steps": 5316, "loss": 0.3532, "lr": 5e-06, "epoch": 2.606488011283498, "percentage": 86.91, "elapsed_time": "17:06:47", "remaining_time": "2:34:41"}
465
+ {"current_steps": 4630, "total_steps": 5316, "loss": 0.3667, "lr": 5e-06, "epoch": 2.61212976022567, "percentage": 87.1, "elapsed_time": "17:08:58", "remaining_time": "2:32:27"}
466
+ {"current_steps": 4640, "total_steps": 5316, "loss": 0.3566, "lr": 5e-06, "epoch": 2.617771509167842, "percentage": 87.28, "elapsed_time": "17:11:10", "remaining_time": "2:30:13"}
467
+ {"current_steps": 4650, "total_steps": 5316, "loss": 0.3466, "lr": 5e-06, "epoch": 2.623413258110014, "percentage": 87.47, "elapsed_time": "17:13:21", "remaining_time": "2:28:00"}
468
+ {"current_steps": 4660, "total_steps": 5316, "loss": 0.3532, "lr": 5e-06, "epoch": 2.629055007052186, "percentage": 87.66, "elapsed_time": "17:15:32", "remaining_time": "2:25:46"}
469
+ {"current_steps": 4670, "total_steps": 5316, "loss": 0.352, "lr": 5e-06, "epoch": 2.634696755994358, "percentage": 87.85, "elapsed_time": "17:17:43", "remaining_time": "2:23:32"}
470
+ {"current_steps": 4680, "total_steps": 5316, "loss": 0.3479, "lr": 5e-06, "epoch": 2.64033850493653, "percentage": 88.04, "elapsed_time": "17:19:54", "remaining_time": "2:21:19"}
471
+ {"current_steps": 4690, "total_steps": 5316, "loss": 0.3399, "lr": 5e-06, "epoch": 2.6459802538787023, "percentage": 88.22, "elapsed_time": "17:22:05", "remaining_time": "2:19:05"}
472
+ {"current_steps": 4700, "total_steps": 5316, "loss": 0.3496, "lr": 5e-06, "epoch": 2.6516220028208743, "percentage": 88.41, "elapsed_time": "17:24:16", "remaining_time": "2:16:51"}
473
+ {"current_steps": 4710, "total_steps": 5316, "loss": 0.3698, "lr": 5e-06, "epoch": 2.6572637517630464, "percentage": 88.6, "elapsed_time": "17:26:27", "remaining_time": "2:14:38"}
474
+ {"current_steps": 4720, "total_steps": 5316, "loss": 0.3557, "lr": 5e-06, "epoch": 2.6629055007052185, "percentage": 88.79, "elapsed_time": "17:28:38", "remaining_time": "2:12:24"}
475
+ {"current_steps": 4730, "total_steps": 5316, "loss": 0.3555, "lr": 5e-06, "epoch": 2.6685472496473905, "percentage": 88.98, "elapsed_time": "17:30:49", "remaining_time": "2:10:11"}
476
+ {"current_steps": 4740, "total_steps": 5316, "loss": 0.3533, "lr": 5e-06, "epoch": 2.6741889985895626, "percentage": 89.16, "elapsed_time": "17:32:59", "remaining_time": "2:07:57"}
477
+ {"current_steps": 4750, "total_steps": 5316, "loss": 0.3593, "lr": 5e-06, "epoch": 2.679830747531735, "percentage": 89.35, "elapsed_time": "17:35:10", "remaining_time": "2:05:43"}
478
+ {"current_steps": 4760, "total_steps": 5316, "loss": 0.35, "lr": 5e-06, "epoch": 2.685472496473907, "percentage": 89.54, "elapsed_time": "17:37:21", "remaining_time": "2:03:30"}
479
+ {"current_steps": 4770, "total_steps": 5316, "loss": 0.3508, "lr": 5e-06, "epoch": 2.6911142454160792, "percentage": 89.73, "elapsed_time": "17:39:32", "remaining_time": "2:01:16"}
480
+ {"current_steps": 4780, "total_steps": 5316, "loss": 0.3659, "lr": 5e-06, "epoch": 2.6967559943582513, "percentage": 89.92, "elapsed_time": "17:41:43", "remaining_time": "1:59:03"}
481
+ {"current_steps": 4790, "total_steps": 5316, "loss": 0.3488, "lr": 5e-06, "epoch": 2.7023977433004234, "percentage": 90.11, "elapsed_time": "17:43:54", "remaining_time": "1:56:49"}
482
+ {"current_steps": 4800, "total_steps": 5316, "loss": 0.3622, "lr": 5e-06, "epoch": 2.7080394922425954, "percentage": 90.29, "elapsed_time": "17:46:05", "remaining_time": "1:54:36"}
483
+ {"current_steps": 4810, "total_steps": 5316, "loss": 0.3659, "lr": 5e-06, "epoch": 2.7136812411847675, "percentage": 90.48, "elapsed_time": "17:48:16", "remaining_time": "1:52:22"}
484
+ {"current_steps": 4820, "total_steps": 5316, "loss": 0.3521, "lr": 5e-06, "epoch": 2.7193229901269396, "percentage": 90.67, "elapsed_time": "17:50:27", "remaining_time": "1:50:09"}
485
+ {"current_steps": 4830, "total_steps": 5316, "loss": 0.3609, "lr": 5e-06, "epoch": 2.7249647390691116, "percentage": 90.86, "elapsed_time": "17:52:38", "remaining_time": "1:47:55"}
486
+ {"current_steps": 4840, "total_steps": 5316, "loss": 0.3608, "lr": 5e-06, "epoch": 2.7306064880112837, "percentage": 91.05, "elapsed_time": "17:54:49", "remaining_time": "1:45:42"}
487
+ {"current_steps": 4850, "total_steps": 5316, "loss": 0.3516, "lr": 5e-06, "epoch": 2.7362482369534558, "percentage": 91.23, "elapsed_time": "17:57:00", "remaining_time": "1:43:28"}
488
+ {"current_steps": 4860, "total_steps": 5316, "loss": 0.3575, "lr": 5e-06, "epoch": 2.741889985895628, "percentage": 91.42, "elapsed_time": "17:59:11", "remaining_time": "1:41:15"}
489
+ {"current_steps": 4870, "total_steps": 5316, "loss": 0.3508, "lr": 5e-06, "epoch": 2.7475317348378, "percentage": 91.61, "elapsed_time": "18:01:22", "remaining_time": "1:39:02"}
490
+ {"current_steps": 4880, "total_steps": 5316, "loss": 0.3548, "lr": 5e-06, "epoch": 2.753173483779972, "percentage": 91.8, "elapsed_time": "18:03:33", "remaining_time": "1:36:48"}
491
+ {"current_steps": 4890, "total_steps": 5316, "loss": 0.3518, "lr": 5e-06, "epoch": 2.758815232722144, "percentage": 91.99, "elapsed_time": "18:05:44", "remaining_time": "1:34:35"}
492
+ {"current_steps": 4900, "total_steps": 5316, "loss": 0.3501, "lr": 5e-06, "epoch": 2.764456981664316, "percentage": 92.17, "elapsed_time": "18:07:55", "remaining_time": "1:32:21"}
493
+ {"current_steps": 4910, "total_steps": 5316, "loss": 0.3452, "lr": 5e-06, "epoch": 2.770098730606488, "percentage": 92.36, "elapsed_time": "18:10:06", "remaining_time": "1:30:08"}
494
+ {"current_steps": 4920, "total_steps": 5316, "loss": 0.3576, "lr": 5e-06, "epoch": 2.77574047954866, "percentage": 92.55, "elapsed_time": "18:12:17", "remaining_time": "1:27:54"}
495
+ {"current_steps": 4930, "total_steps": 5316, "loss": 0.3669, "lr": 5e-06, "epoch": 2.7813822284908323, "percentage": 92.74, "elapsed_time": "18:14:28", "remaining_time": "1:25:41"}
496
+ {"current_steps": 4940, "total_steps": 5316, "loss": 0.3558, "lr": 5e-06, "epoch": 2.7870239774330043, "percentage": 92.93, "elapsed_time": "18:16:40", "remaining_time": "1:23:28"}
497
+ {"current_steps": 4950, "total_steps": 5316, "loss": 0.353, "lr": 5e-06, "epoch": 2.7926657263751764, "percentage": 93.12, "elapsed_time": "18:18:51", "remaining_time": "1:21:14"}
498
+ {"current_steps": 4960, "total_steps": 5316, "loss": 0.3575, "lr": 5e-06, "epoch": 2.7983074753173485, "percentage": 93.3, "elapsed_time": "18:21:02", "remaining_time": "1:19:01"}
499
+ {"current_steps": 4970, "total_steps": 5316, "loss": 0.376, "lr": 5e-06, "epoch": 2.8039492242595205, "percentage": 93.49, "elapsed_time": "18:23:13", "remaining_time": "1:16:48"}
500
+ {"current_steps": 4980, "total_steps": 5316, "loss": 0.3552, "lr": 5e-06, "epoch": 2.8095909732016926, "percentage": 93.68, "elapsed_time": "18:25:25", "remaining_time": "1:14:34"}
501
+ {"current_steps": 4990, "total_steps": 5316, "loss": 0.3538, "lr": 5e-06, "epoch": 2.8152327221438647, "percentage": 93.87, "elapsed_time": "18:27:36", "remaining_time": "1:12:21"}
502
+ {"current_steps": 5000, "total_steps": 5316, "loss": 0.3596, "lr": 5e-06, "epoch": 2.8208744710860367, "percentage": 94.06, "elapsed_time": "18:29:47", "remaining_time": "1:10:08"}
503
+ {"current_steps": 5010, "total_steps": 5316, "loss": 0.3589, "lr": 5e-06, "epoch": 2.826516220028209, "percentage": 94.24, "elapsed_time": "18:31:59", "remaining_time": "1:07:55"}
504
+ {"current_steps": 5020, "total_steps": 5316, "loss": 0.3717, "lr": 5e-06, "epoch": 2.832157968970381, "percentage": 94.43, "elapsed_time": "18:34:10", "remaining_time": "1:05:41"}
505
+ {"current_steps": 5030, "total_steps": 5316, "loss": 0.3573, "lr": 5e-06, "epoch": 2.837799717912553, "percentage": 94.62, "elapsed_time": "18:36:21", "remaining_time": "1:03:28"}
506
+ {"current_steps": 5040, "total_steps": 5316, "loss": 0.3477, "lr": 5e-06, "epoch": 2.843441466854725, "percentage": 94.81, "elapsed_time": "18:38:32", "remaining_time": "1:01:15"}
507
+ {"current_steps": 5050, "total_steps": 5316, "loss": 0.3514, "lr": 5e-06, "epoch": 2.849083215796897, "percentage": 95.0, "elapsed_time": "18:40:44", "remaining_time": "0:59:01"}
508
+ {"current_steps": 5060, "total_steps": 5316, "loss": 0.3457, "lr": 5e-06, "epoch": 2.854724964739069, "percentage": 95.18, "elapsed_time": "18:42:55", "remaining_time": "0:56:48"}
509
+ {"current_steps": 5070, "total_steps": 5316, "loss": 0.3616, "lr": 5e-06, "epoch": 2.860366713681241, "percentage": 95.37, "elapsed_time": "18:45:06", "remaining_time": "0:54:35"}
510
+ {"current_steps": 5080, "total_steps": 5316, "loss": 0.3527, "lr": 5e-06, "epoch": 2.8660084626234132, "percentage": 95.56, "elapsed_time": "18:47:17", "remaining_time": "0:52:22"}
511
+ {"current_steps": 5090, "total_steps": 5316, "loss": 0.367, "lr": 5e-06, "epoch": 2.8716502115655853, "percentage": 95.75, "elapsed_time": "18:49:29", "remaining_time": "0:50:08"}
512
+ {"current_steps": 5100, "total_steps": 5316, "loss": 0.3511, "lr": 5e-06, "epoch": 2.8772919605077574, "percentage": 95.94, "elapsed_time": "18:51:40", "remaining_time": "0:47:55"}
513
+ {"current_steps": 5110, "total_steps": 5316, "loss": 0.3623, "lr": 5e-06, "epoch": 2.8829337094499294, "percentage": 96.12, "elapsed_time": "18:53:51", "remaining_time": "0:45:42"}
514
+ {"current_steps": 5120, "total_steps": 5316, "loss": 0.3495, "lr": 5e-06, "epoch": 2.8885754583921015, "percentage": 96.31, "elapsed_time": "18:56:02", "remaining_time": "0:43:29"}
515
+ {"current_steps": 5130, "total_steps": 5316, "loss": 0.3566, "lr": 5e-06, "epoch": 2.8942172073342736, "percentage": 96.5, "elapsed_time": "18:58:13", "remaining_time": "0:41:16"}
516
+ {"current_steps": 5140, "total_steps": 5316, "loss": 0.357, "lr": 5e-06, "epoch": 2.8998589562764456, "percentage": 96.69, "elapsed_time": "19:00:24", "remaining_time": "0:39:02"}
517
+ {"current_steps": 5150, "total_steps": 5316, "loss": 0.3559, "lr": 5e-06, "epoch": 2.9055007052186177, "percentage": 96.88, "elapsed_time": "19:02:35", "remaining_time": "0:36:49"}
518
+ {"current_steps": 5160, "total_steps": 5316, "loss": 0.3599, "lr": 5e-06, "epoch": 2.9111424541607898, "percentage": 97.07, "elapsed_time": "19:04:46", "remaining_time": "0:34:36"}
519
+ {"current_steps": 5170, "total_steps": 5316, "loss": 0.3554, "lr": 5e-06, "epoch": 2.916784203102962, "percentage": 97.25, "elapsed_time": "19:06:57", "remaining_time": "0:32:23"}
520
+ {"current_steps": 5180, "total_steps": 5316, "loss": 0.3553, "lr": 5e-06, "epoch": 2.922425952045134, "percentage": 97.44, "elapsed_time": "19:09:08", "remaining_time": "0:30:10"}
521
+ {"current_steps": 5190, "total_steps": 5316, "loss": 0.3537, "lr": 5e-06, "epoch": 2.928067700987306, "percentage": 97.63, "elapsed_time": "19:11:19", "remaining_time": "0:27:57"}
522
+ {"current_steps": 5200, "total_steps": 5316, "loss": 0.3591, "lr": 5e-06, "epoch": 2.933709449929478, "percentage": 97.82, "elapsed_time": "19:13:30", "remaining_time": "0:25:43"}
523
+ {"current_steps": 5210, "total_steps": 5316, "loss": 0.3561, "lr": 5e-06, "epoch": 2.93935119887165, "percentage": 98.01, "elapsed_time": "19:15:42", "remaining_time": "0:23:30"}
524
+ {"current_steps": 5220, "total_steps": 5316, "loss": 0.3633, "lr": 5e-06, "epoch": 2.944992947813822, "percentage": 98.19, "elapsed_time": "19:17:53", "remaining_time": "0:21:17"}
525
+ {"current_steps": 5230, "total_steps": 5316, "loss": 0.3496, "lr": 5e-06, "epoch": 2.950634696755994, "percentage": 98.38, "elapsed_time": "19:20:04", "remaining_time": "0:19:04"}
526
+ {"current_steps": 5240, "total_steps": 5316, "loss": 0.3484, "lr": 5e-06, "epoch": 2.9562764456981663, "percentage": 98.57, "elapsed_time": "19:22:15", "remaining_time": "0:16:51"}
527
+ {"current_steps": 5250, "total_steps": 5316, "loss": 0.3735, "lr": 5e-06, "epoch": 2.9619181946403383, "percentage": 98.76, "elapsed_time": "19:24:27", "remaining_time": "0:14:38"}
528
+ {"current_steps": 5260, "total_steps": 5316, "loss": 0.3517, "lr": 5e-06, "epoch": 2.9675599435825104, "percentage": 98.95, "elapsed_time": "19:26:38", "remaining_time": "0:12:25"}
529
+ {"current_steps": 5270, "total_steps": 5316, "loss": 0.358, "lr": 5e-06, "epoch": 2.9732016925246825, "percentage": 99.13, "elapsed_time": "19:28:49", "remaining_time": "0:10:12"}
530
+ {"current_steps": 5280, "total_steps": 5316, "loss": 0.3743, "lr": 5e-06, "epoch": 2.9788434414668545, "percentage": 99.32, "elapsed_time": "19:31:00", "remaining_time": "0:07:59"}
531
+ {"current_steps": 5290, "total_steps": 5316, "loss": 0.3519, "lr": 5e-06, "epoch": 2.9844851904090266, "percentage": 99.51, "elapsed_time": "19:33:11", "remaining_time": "0:05:45"}
532
+ {"current_steps": 5300, "total_steps": 5316, "loss": 0.3473, "lr": 5e-06, "epoch": 2.9901269393511987, "percentage": 99.7, "elapsed_time": "19:35:22", "remaining_time": "0:03:32"}
533
+ {"current_steps": 5310, "total_steps": 5316, "loss": 0.3623, "lr": 5e-06, "epoch": 2.9957686882933707, "percentage": 99.89, "elapsed_time": "19:37:33", "remaining_time": "0:01:19"}