Training in progress, epoch 3

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:52fc98c43906faab05c1813e058c87fd219fea801456c4e76953508aacba20bb
+oid sha256:fd1093982678959f2ce3f5820d09f6d1b6fa6918e8bf38d9c67203a789600c48
 size 4949453792

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c68a024292cde277d36cc83965b268b68fa1244aae0d59b667d9afa9aa2bd3ea
+oid sha256:e489b4e230d7f572cae58d69df77043a6449f776aaadba0e54f9029b6a454b51
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:035a6b88018731b0495043d2dcbbf3cb07256ba56b758c65ac172bb95207e490
+oid sha256:d845e6fb19a0ee1488dbd5f5a18bc0dfd997f8c15d7e4716e7ecfc9c5fc67abb
 size 4546807800

trainer_log.jsonl

@@ -97,3 +97,49 @@
 {"current_steps": 950, "total_steps": 1419, "loss": 0.2124, "lr": 1.7154609981874945e-06, "epoch": 2.0084566596194504, "percentage": 66.95, "elapsed_time": "1:36:23", "remaining_time": "0:47:35"}
 {"current_steps": 960, "total_steps": 1419, "loss": 0.1655, "lr": 1.6691802874310402e-06, "epoch": 2.029598308668076, "percentage": 67.65, "elapsed_time": "1:37:24", "remaining_time": "0:46:34"}
 {"current_steps": 970, "total_steps": 1419, "loss": 0.1695, "lr": 1.6234865975701169e-06, "epoch": 2.050739957716702, "percentage": 68.36, "elapsed_time": "1:38:23", "remaining_time": "0:45:32"}
+{"current_steps": 980, "total_steps": 1419, "loss": 0.1628, "lr": 1.5784047460191638e-06, "epoch": 2.0718816067653276, "percentage": 69.06, "elapsed_time": "1:39:22", "remaining_time": "0:44:31"}
+{"current_steps": 990, "total_steps": 1419, "loss": 0.1629, "lr": 1.5339592178875502e-06, "epoch": 2.0930232558139537, "percentage": 69.77, "elapsed_time": "1:40:21", "remaining_time": "0:43:29"}
+{"current_steps": 1000, "total_steps": 1419, "loss": 0.1632, "lr": 1.4901741526810793e-06, "epoch": 2.1141649048625792, "percentage": 70.47, "elapsed_time": "1:41:21", "remaining_time": "0:42:28"}
+{"current_steps": 1010, "total_steps": 1419, "loss": 0.1648, "lr": 1.4470733311912094e-06, "epoch": 2.1353065539112053, "percentage": 71.18, "elapsed_time": "1:42:20", "remaining_time": "0:41:26"}
+{"current_steps": 1020, "total_steps": 1419, "loss": 0.1635, "lr": 1.4046801625790954e-06, "epoch": 2.156448202959831, "percentage": 71.88, "elapsed_time": "1:43:19", "remaining_time": "0:40:25"}
+{"current_steps": 1030, "total_steps": 1419, "loss": 0.1645, "lr": 1.3630176716614696e-06, "epoch": 2.177589852008457, "percentage": 72.59, "elapsed_time": "1:44:19", "remaining_time": "0:39:23"}
+{"current_steps": 1040, "total_steps": 1419, "loss": 0.167, "lr": 1.3221084864052755e-06, "epoch": 2.1987315010570825, "percentage": 73.29, "elapsed_time": "1:45:18", "remaining_time": "0:38:22"}
+{"current_steps": 1050, "total_steps": 1419, "loss": 0.1621, "lr": 1.2819748256378395e-06, "epoch": 2.219873150105708, "percentage": 74.0, "elapsed_time": "1:46:17", "remaining_time": "0:37:21"}
+{"current_steps": 1060, "total_steps": 1419, "loss": 0.1626, "lr": 1.2426384869792552e-06, "epoch": 2.241014799154334, "percentage": 74.7, "elapsed_time": "1:47:17", "remaining_time": "0:36:20"}
+{"current_steps": 1070, "total_steps": 1419, "loss": 0.1642, "lr": 1.2041208350035386e-06, "epoch": 2.2621564482029597, "percentage": 75.41, "elapsed_time": "1:48:16", "remaining_time": "0:35:18"}
+{"current_steps": 1080, "total_steps": 1419, "loss": 0.1642, "lr": 1.1664427896349775e-06, "epoch": 2.2832980972515857, "percentage": 76.11, "elapsed_time": "1:49:15", "remaining_time": "0:34:17"}
+{"current_steps": 1090, "total_steps": 1419, "loss": 0.1647, "lr": 1.129624814785987e-06, "epoch": 2.3044397463002113, "percentage": 76.81, "elapsed_time": "1:50:15", "remaining_time": "0:33:16"}
+{"current_steps": 1100, "total_steps": 1419, "loss": 0.163, "lr": 1.0936869072426324e-06, "epoch": 2.3255813953488373, "percentage": 77.52, "elapsed_time": "1:51:14", "remaining_time": "0:32:15"}
+{"current_steps": 1110, "total_steps": 1419, "loss": 0.1626, "lr": 1.0586485858038623e-06, "epoch": 2.346723044397463, "percentage": 78.22, "elapsed_time": "1:52:14", "remaining_time": "0:31:14"}
+{"current_steps": 1120, "total_steps": 1419, "loss": 0.1669, "lr": 1.0245288806803492e-06, "epoch": 2.367864693446089, "percentage": 78.93, "elapsed_time": "1:53:13", "remaining_time": "0:30:13"}
+{"current_steps": 1130, "total_steps": 1419, "loss": 0.1629, "lr": 9.913463231586928e-07, "epoch": 2.3890063424947146, "percentage": 79.63, "elapsed_time": "1:54:13", "remaining_time": "0:29:12"}
+{"current_steps": 1140, "total_steps": 1419, "loss": 0.1644, "lr": 9.591189355366054e-07, "epoch": 2.41014799154334, "percentage": 80.34, "elapsed_time": "1:55:12", "remaining_time": "0:28:11"}
+{"current_steps": 1150, "total_steps": 1419, "loss": 0.167, "lr": 9.278642213345369e-07, "epoch": 2.431289640591966, "percentage": 81.04, "elapsed_time": "1:56:13", "remaining_time": "0:27:11"}
+{"current_steps": 1160, "total_steps": 1419, "loss": 0.1638, "lr": 8.975991557890668e-07, "epoch": 2.452431289640592, "percentage": 81.75, "elapsed_time": "1:57:14", "remaining_time": "0:26:10"}
+{"current_steps": 1170, "total_steps": 1419, "loss": 0.1642, "lr": 8.683401766332171e-07, "epoch": 2.473572938689218, "percentage": 82.45, "elapsed_time": "1:58:13", "remaining_time": "0:25:09"}
+{"current_steps": 1180, "total_steps": 1419, "loss": 0.1676, "lr": 8.401031751686978e-07, "epoch": 2.4947145877378434, "percentage": 83.16, "elapsed_time": "1:59:13", "remaining_time": "0:24:08"}
+{"current_steps": 1190, "total_steps": 1419, "loss": 0.1644, "lr": 8.129034876349334e-07, "epoch": 2.5158562367864694, "percentage": 83.86, "elapsed_time": "2:00:13", "remaining_time": "0:23:08"}
+{"current_steps": 1200, "total_steps": 1419, "loss": 0.165, "lr": 7.867558868795578e-07, "epoch": 2.536997885835095, "percentage": 84.57, "elapsed_time": "2:01:12", "remaining_time": "0:22:07"}
+{"current_steps": 1210, "total_steps": 1419, "loss": 0.1638, "lr": 7.616745743349038e-07, "epoch": 2.558139534883721, "percentage": 85.27, "elapsed_time": "2:02:11", "remaining_time": "0:21:06"}
+{"current_steps": 1220, "total_steps": 1419, "loss": 0.1652, "lr": 7.376731723048383e-07, "epoch": 2.5792811839323466, "percentage": 85.98, "elapsed_time": "2:03:11", "remaining_time": "0:20:05"}
+{"current_steps": 1230, "total_steps": 1419, "loss": 0.1636, "lr": 7.147647165661439e-07, "epoch": 2.6004228329809727, "percentage": 86.68, "elapsed_time": "2:04:10", "remaining_time": "0:19:04"}
+{"current_steps": 1240, "total_steps": 1419, "loss": 0.1625, "lr": 6.929616492884497e-07, "epoch": 2.6215644820295982, "percentage": 87.39, "elapsed_time": "2:05:10", "remaining_time": "0:18:04"}
+{"current_steps": 1250, "total_steps": 1419, "loss": 0.1638, "lr": 6.722758122765728e-07, "epoch": 2.6427061310782243, "percentage": 88.09, "elapsed_time": "2:06:10", "remaining_time": "0:17:03"}
+{"current_steps": 1260, "total_steps": 1419, "loss": 0.1611, "lr": 6.527184405389298e-07, "epoch": 2.66384778012685, "percentage": 88.79, "elapsed_time": "2:07:09", "remaining_time": "0:16:02"}
+{"current_steps": 1270, "total_steps": 1419, "loss": 0.1605, "lr": 6.343001561855149e-07, "epoch": 2.6849894291754755, "percentage": 89.5, "elapsed_time": "2:08:10", "remaining_time": "0:15:02"}
+{"current_steps": 1280, "total_steps": 1419, "loss": 0.1655, "lr": 6.17030962658762e-07, "epoch": 2.7061310782241015, "percentage": 90.2, "elapsed_time": "2:09:10", "remaining_time": "0:14:01"}
+{"current_steps": 1290, "total_steps": 1419, "loss": 0.1636, "lr": 6.009202393004164e-07, "epoch": 2.7272727272727275, "percentage": 90.91, "elapsed_time": "2:10:11", "remaining_time": "0:13:01"}
+{"current_steps": 1300, "total_steps": 1419, "loss": 0.1644, "lr": 5.859767362573741e-07, "epoch": 2.748414376321353, "percentage": 91.61, "elapsed_time": "2:11:11", "remaining_time": "0:12:00"}
+{"current_steps": 1310, "total_steps": 1419, "loss": 0.1659, "lr": 5.722085697292536e-07, "epoch": 2.7695560253699787, "percentage": 92.32, "elapsed_time": "2:12:10", "remaining_time": "0:10:59"}
+{"current_steps": 1320, "total_steps": 1419, "loss": 0.1667, "lr": 5.596232175602791e-07, "epoch": 2.7906976744186047, "percentage": 93.02, "elapsed_time": "2:13:10", "remaining_time": "0:09:59"}
+{"current_steps": 1330, "total_steps": 1419, "loss": 0.164, "lr": 5.482275151778719e-07, "epoch": 2.8118393234672303, "percentage": 93.73, "elapsed_time": "2:14:09", "remaining_time": "0:08:58"}
+{"current_steps": 1340, "total_steps": 1419, "loss": 0.1632, "lr": 5.380276518801548e-07, "epoch": 2.8329809725158563, "percentage": 94.43, "elapsed_time": "2:15:09", "remaining_time": "0:07:58"}
+{"current_steps": 1350, "total_steps": 1419, "loss": 0.1642, "lr": 5.290291674743844e-07, "epoch": 2.854122621564482, "percentage": 95.14, "elapsed_time": "2:16:08", "remaining_time": "0:06:57"}
+{"current_steps": 1360, "total_steps": 1419, "loss": 0.1641, "lr": 5.212369492681416e-07, "epoch": 2.875264270613108, "percentage": 95.84, "elapsed_time": "2:17:08", "remaining_time": "0:05:56"}
+{"current_steps": 1370, "total_steps": 1419, "loss": 0.1629, "lr": 5.146552294149075e-07, "epoch": 2.8964059196617336, "percentage": 96.55, "elapsed_time": "2:18:08", "remaining_time": "0:04:56"}
+{"current_steps": 1380, "total_steps": 1419, "loss": 0.1648, "lr": 5.092875826154744e-07, "epoch": 2.9175475687103596, "percentage": 97.25, "elapsed_time": "2:19:08", "remaining_time": "0:03:55"}
+{"current_steps": 1390, "total_steps": 1419, "loss": 0.1621, "lr": 5.051369241764327e-07, "epoch": 2.938689217758985, "percentage": 97.96, "elapsed_time": "2:20:10", "remaining_time": "0:02:55"}
+{"current_steps": 1400, "total_steps": 1419, "loss": 0.1623, "lr": 5.022055084267933e-07, "epoch": 2.9598308668076108, "percentage": 98.66, "elapsed_time": "2:21:10", "remaining_time": "0:01:54"}
+{"current_steps": 1410, "total_steps": 1419, "loss": 0.1634, "lr": 5.004949274936044e-07, "epoch": 2.980972515856237, "percentage": 99.37, "elapsed_time": "2:22:10", "remaining_time": "0:00:54"}
+{"current_steps": 1419, "total_steps": 1419, "eval_loss": 0.3471969962120056, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2:24:57", "remaining_time": "0:00:00"}
+{"current_steps": 1419, "total_steps": 1419, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2:25:55", "remaining_time": "0:00:00"}