Training in progress, step 17884

model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2d6122486fd6f9d1f569a5ca7bbaedc91343f5402896e87d48de5f15621c7b91
+oid sha256:d233b027c7b45d8e9215855924e8c694a77173b91e4ffd4632fd6c5d90c3b19b
 size 4996670464

model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:07aa787d09e10b0fac649bf33baee6d42094543511cf8b57ef06f4a177e196fc
+oid sha256:4334aa90bb8a9679d00153b73ec1dba55eca27deddf7eafa7ff57392c53ace30
 size 1178224960

trainer_log.jsonl

@@ -1700,3 +1700,90 @@
 {"current_steps": 17000, "total_steps": 17884, "loss": 0.0167, "lr": 1.4849592300382964e-07, "epoch": 1.9010875338980682, "percentage": 95.06, "elapsed_time": "1 day, 18:13:25", "remaining_time": "2:11:44"}
 {"current_steps": 17010, "total_steps": 17884, "loss": 0.0182, "lr": 1.451633953345799e-07, "epoch": 1.9022058206827142, "percentage": 95.11, "elapsed_time": "1 day, 18:15:41", "remaining_time": "2:10:17"}
 {"current_steps": 17020, "total_steps": 17884, "loss": 0.016, "lr": 1.4186841395906536e-07, "epoch": 1.90332410746736, "percentage": 95.17, "elapsed_time": "1 day, 18:17:00", "remaining_time": "2:08:47"}
+{"current_steps": 17030, "total_steps": 17884, "loss": 0.0147, "lr": 1.3861099143095302e-07, "epoch": 1.904442394252006, "percentage": 95.22, "elapsed_time": "1 day, 18:18:23", "remaining_time": "2:07:17"}
+{"current_steps": 17040, "total_steps": 17884, "loss": 0.0154, "lr": 1.3539114016081434e-07, "epoch": 1.9055606810366519, "percentage": 95.28, "elapsed_time": "1 day, 18:19:50", "remaining_time": "2:05:47"}
+{"current_steps": 17050, "total_steps": 17884, "loss": 0.0185, "lr": 1.3220887241607304e-07, "epoch": 1.9066789678212976, "percentage": 95.34, "elapsed_time": "1 day, 18:21:26", "remaining_time": "2:04:18"}
+{"current_steps": 17060, "total_steps": 17884, "loss": 0.013, "lr": 1.2906420032096745e-07, "epoch": 1.9077972546059438, "percentage": 95.39, "elapsed_time": "1 day, 18:22:50", "remaining_time": "2:02:49"}
+{"current_steps": 17070, "total_steps": 17884, "loss": 0.0195, "lr": 1.2595713585649482e-07, "epoch": 1.9089155413905896, "percentage": 95.45, "elapsed_time": "1 day, 18:24:16", "remaining_time": "2:01:19"}
+{"current_steps": 17080, "total_steps": 17884, "loss": 0.0158, "lr": 1.2288769086037046e-07, "epoch": 1.9100338281752356, "percentage": 95.5, "elapsed_time": "1 day, 18:25:49", "remaining_time": "1:59:50"}
+{"current_steps": 17090, "total_steps": 17884, "loss": 0.0156, "lr": 1.198558770269842e-07, "epoch": 1.9111521149598816, "percentage": 95.56, "elapsed_time": "1 day, 18:27:01", "remaining_time": "1:58:20"}
+{"current_steps": 17100, "total_steps": 17884, "loss": 0.0186, "lr": 1.1686170590734957e-07, "epoch": 1.9122704017445273, "percentage": 95.62, "elapsed_time": "1 day, 18:28:24", "remaining_time": "1:56:50"}
+{"current_steps": 17110, "total_steps": 17884, "loss": 0.0169, "lr": 1.1390518890906588e-07, "epoch": 1.9133886885291733, "percentage": 95.67, "elapsed_time": "1 day, 18:29:47", "remaining_time": "1:55:20"}
+{"current_steps": 17120, "total_steps": 17884, "loss": 0.0235, "lr": 1.1098633729627606e-07, "epoch": 1.9145069753138193, "percentage": 95.73, "elapsed_time": "1 day, 18:31:22", "remaining_time": "1:53:51"}
+{"current_steps": 17130, "total_steps": 17884, "loss": 0.0198, "lr": 1.0810516218961343e-07, "epoch": 1.915625262098465, "percentage": 95.78, "elapsed_time": "1 day, 18:33:04", "remaining_time": "1:52:22"}
+{"current_steps": 17140, "total_steps": 17884, "loss": 0.0159, "lr": 1.0526167456617165e-07, "epoch": 1.916743548883111, "percentage": 95.84, "elapsed_time": "1 day, 18:34:45", "remaining_time": "1:50:53"}
+{"current_steps": 17150, "total_steps": 17884, "loss": 0.0173, "lr": 1.0245588525945483e-07, "epoch": 1.917861835667757, "percentage": 95.9, "elapsed_time": "1 day, 18:36:23", "remaining_time": "1:49:24"}
+{"current_steps": 17160, "total_steps": 17884, "loss": 0.0184, "lr": 9.968780495934082e-08, "epoch": 1.9189801224524028, "percentage": 95.95, "elapsed_time": "1 day, 18:37:34", "remaining_time": "1:47:54"}
+{"current_steps": 17170, "total_steps": 17884, "loss": 0.0204, "lr": 9.69574442120369e-08, "epoch": 1.920098409237049, "percentage": 96.01, "elapsed_time": "1 day, 18:39:03", "remaining_time": "1:46:24"}
+{"current_steps": 17180, "total_steps": 17884, "loss": 0.0205, "lr": 9.42648134200419e-08, "epoch": 1.9212166960216948, "percentage": 96.06, "elapsed_time": "1 day, 18:40:28", "remaining_time": "1:44:55"}
+{"current_steps": 17190, "total_steps": 17884, "loss": 0.0204, "lr": 9.160992284210634e-08, "epoch": 1.9223349828063405, "percentage": 96.12, "elapsed_time": "1 day, 18:42:09", "remaining_time": "1:43:26"}
+{"current_steps": 17200, "total_steps": 17884, "loss": 0.0155, "lr": 8.899278259319244e-08, "epoch": 1.9234532695909867, "percentage": 96.18, "elapsed_time": "1 day, 18:43:31", "remaining_time": "1:41:56"}
+{"current_steps": 17210, "total_steps": 17884, "loss": 0.0135, "lr": 8.641340264443632e-08, "epoch": 1.9245715563756325, "percentage": 96.23, "elapsed_time": "1 day, 18:44:56", "remaining_time": "1:40:27"}
+{"current_steps": 17220, "total_steps": 17884, "loss": 0.0174, "lr": 8.38717928231092e-08, "epoch": 1.9256898431602785, "percentage": 96.29, "elapsed_time": "1 day, 18:46:26", "remaining_time": "1:38:57"}
+{"current_steps": 17230, "total_steps": 17884, "loss": 0.0222, "lr": 8.136796281258185e-08, "epoch": 1.9268081299449245, "percentage": 96.34, "elapsed_time": "1 day, 18:48:05", "remaining_time": "1:37:28"}
+{"current_steps": 17240, "total_steps": 17884, "loss": 0.026, "lr": 7.890192215228576e-08, "epoch": 1.9279264167295702, "percentage": 96.4, "elapsed_time": "1 day, 18:49:39", "remaining_time": "1:35:59"}
+{"current_steps": 17250, "total_steps": 17884, "loss": 0.0191, "lr": 7.647368023767532e-08, "epoch": 1.9290447035142162, "percentage": 96.45, "elapsed_time": "1 day, 18:51:07", "remaining_time": "1:34:29"}
+{"current_steps": 17260, "total_steps": 17884, "loss": 0.0181, "lr": 7.408324632019903e-08, "epoch": 1.9301629902988622, "percentage": 96.51, "elapsed_time": "1 day, 18:52:42", "remaining_time": "1:33:00"}
+{"current_steps": 17270, "total_steps": 17884, "loss": 0.0193, "lr": 7.173062950725284e-08, "epoch": 1.931281277083508, "percentage": 96.57, "elapsed_time": "1 day, 18:54:22", "remaining_time": "1:31:31"}
+{"current_steps": 17280, "total_steps": 17884, "loss": 0.0139, "lr": 6.941583876215907e-08, "epoch": 1.9323995638681541, "percentage": 96.62, "elapsed_time": "1 day, 18:55:43", "remaining_time": "1:30:01"}
+{"current_steps": 17290, "total_steps": 17884, "loss": 0.0159, "lr": 6.713888290411974e-08, "epoch": 1.9335178506528, "percentage": 96.68, "elapsed_time": "1 day, 18:57:18", "remaining_time": "1:28:32"}
+{"current_steps": 17300, "total_steps": 17884, "loss": 0.0157, "lr": 6.489977060818998e-08, "epoch": 1.9346361374374457, "percentage": 96.73, "elapsed_time": "1 day, 18:58:48", "remaining_time": "1:27:03"}
+{"current_steps": 17310, "total_steps": 17884, "loss": 0.0217, "lr": 6.269851040524466e-08, "epoch": 1.9357544242220919, "percentage": 96.79, "elapsed_time": "1 day, 19:00:16", "remaining_time": "1:25:33"}
+{"current_steps": 17320, "total_steps": 17884, "loss": 0.0157, "lr": 6.053511068194406e-08, "epoch": 1.9368727110067376, "percentage": 96.85, "elapsed_time": "1 day, 19:01:46", "remaining_time": "1:24:04"}
+{"current_steps": 17330, "total_steps": 17884, "loss": 0.0163, "lr": 5.840957968070049e-08, "epoch": 1.9379909977913836, "percentage": 96.9, "elapsed_time": "1 day, 19:03:19", "remaining_time": "1:22:34"}
+{"current_steps": 17340, "total_steps": 17884, "loss": 0.0167, "lr": 5.632192549965054e-08, "epoch": 1.9391092845760296, "percentage": 96.96, "elapsed_time": "1 day, 19:04:53", "remaining_time": "1:21:05"}
+{"current_steps": 17350, "total_steps": 17884, "loss": 0.0204, "lr": 5.427215609262293e-08, "epoch": 1.9402275713606754, "percentage": 97.01, "elapsed_time": "1 day, 19:06:11", "remaining_time": "1:19:35"}
+{"current_steps": 17360, "total_steps": 17884, "loss": 0.0151, "lr": 5.226027926910737e-08, "epoch": 1.9413458581453213, "percentage": 97.07, "elapsed_time": "1 day, 19:07:36", "remaining_time": "1:18:06"}
+{"current_steps": 17370, "total_steps": 17884, "loss": 0.0188, "lr": 5.028630269422574e-08, "epoch": 1.9424641449299673, "percentage": 97.13, "elapsed_time": "1 day, 19:09:09", "remaining_time": "1:16:36"}
+{"current_steps": 17380, "total_steps": 17884, "loss": 0.0215, "lr": 4.835023388869986e-08, "epoch": 1.943582431714613, "percentage": 97.18, "elapsed_time": "1 day, 19:10:39", "remaining_time": "1:15:07"}
+{"current_steps": 17390, "total_steps": 17884, "loss": 0.0179, "lr": 4.6452080228829296e-08, "epoch": 1.9447007184992593, "percentage": 97.24, "elapsed_time": "1 day, 19:12:04", "remaining_time": "1:13:38"}
+{"current_steps": 17400, "total_steps": 17884, "loss": 0.0195, "lr": 4.4591848946456964e-08, "epoch": 1.945819005283905, "percentage": 97.29, "elapsed_time": "1 day, 19:13:24", "remaining_time": "1:12:08"}
+{"current_steps": 17410, "total_steps": 17884, "loss": 0.015, "lr": 4.2769547128943544e-08, "epoch": 1.9469372920685508, "percentage": 97.35, "elapsed_time": "1 day, 19:14:40", "remaining_time": "1:10:38"}
+{"current_steps": 17420, "total_steps": 17884, "loss": 0.0211, "lr": 4.0985181719143116e-08, "epoch": 1.948055578853197, "percentage": 97.41, "elapsed_time": "1 day, 19:16:06", "remaining_time": "1:09:09"}
+{"current_steps": 17430, "total_steps": 17884, "loss": 0.0195, "lr": 3.923875951537204e-08, "epoch": 1.9491738656378428, "percentage": 97.46, "elapsed_time": "1 day, 19:17:36", "remaining_time": "1:07:39"}
+{"current_steps": 17440, "total_steps": 17884, "loss": 0.0149, "lr": 3.753028717138785e-08, "epoch": 1.9502921524224888, "percentage": 97.52, "elapsed_time": "1 day, 19:18:51", "remaining_time": "1:06:09"}
+{"current_steps": 17450, "total_steps": 17884, "loss": 0.0161, "lr": 3.585977119635819e-08, "epoch": 1.9514104392071348, "percentage": 97.57, "elapsed_time": "1 day, 19:20:28", "remaining_time": "1:04:40"}
+{"current_steps": 17460, "total_steps": 17884, "loss": 0.0189, "lr": 3.422721795484418e-08, "epoch": 1.9525287259917805, "percentage": 97.63, "elapsed_time": "1 day, 19:22:10", "remaining_time": "1:03:11"}
+{"current_steps": 17470, "total_steps": 17884, "loss": 0.0186, "lr": 3.2632633666767055e-08, "epoch": 1.9536470127764265, "percentage": 97.69, "elapsed_time": "1 day, 19:23:54", "remaining_time": "1:01:42"}
+{"current_steps": 17480, "total_steps": 17884, "loss": 0.0164, "lr": 3.1076024407390436e-08, "epoch": 1.9547652995610725, "percentage": 97.74, "elapsed_time": "1 day, 19:25:20", "remaining_time": "1:00:12"}
+{"current_steps": 17490, "total_steps": 17884, "loss": 0.0204, "lr": 2.9557396107295912e-08, "epoch": 1.9558835863457182, "percentage": 97.8, "elapsed_time": "1 day, 19:26:46", "remaining_time": "0:58:43"}
+{"current_steps": 17500, "total_steps": 17884, "loss": 0.0198, "lr": 2.8076754552358587e-08, "epoch": 1.9570018731303644, "percentage": 97.85, "elapsed_time": "1 day, 19:28:17", "remaining_time": "0:57:13"}
+{"current_steps": 17510, "total_steps": 17884, "loss": 0.0175, "lr": 2.6634105383729347e-08, "epoch": 1.9581201599150102, "percentage": 97.91, "elapsed_time": "1 day, 19:29:35", "remaining_time": "0:55:44"}
+{"current_steps": 17520, "total_steps": 17884, "loss": 0.0165, "lr": 2.522945409780819e-08, "epoch": 1.959238446699656, "percentage": 97.96, "elapsed_time": "1 day, 19:31:07", "remaining_time": "0:54:14"}
+{"current_steps": 17530, "total_steps": 17884, "loss": 0.0134, "lr": 2.386280604622537e-08, "epoch": 1.9603567334843022, "percentage": 98.02, "elapsed_time": "1 day, 19:32:25", "remaining_time": "0:52:45"}
+{"current_steps": 17540, "total_steps": 17884, "loss": 0.0295, "lr": 2.2534166435823623e-08, "epoch": 1.961475020268948, "percentage": 98.08, "elapsed_time": "1 day, 19:34:00", "remaining_time": "0:51:15"}
+{"current_steps": 17550, "total_steps": 17884, "loss": 0.0228, "lr": 2.1243540328632627e-08, "epoch": 1.962593307053594, "percentage": 98.13, "elapsed_time": "1 day, 19:35:28", "remaining_time": "0:49:46"}
+{"current_steps": 17560, "total_steps": 17884, "loss": 0.0187, "lr": 1.9990932641854586e-08, "epoch": 1.96371159383824, "percentage": 98.19, "elapsed_time": "1 day, 19:37:26", "remaining_time": "0:48:17"}
+{"current_steps": 17570, "total_steps": 17884, "loss": 0.0135, "lr": 1.8776348147846458e-08, "epoch": 1.9648298806228857, "percentage": 98.24, "elapsed_time": "1 day, 19:39:06", "remaining_time": "0:46:48"}
+{"current_steps": 17580, "total_steps": 17884, "loss": 0.0167, "lr": 1.759979147409552e-08, "epoch": 1.9659481674075316, "percentage": 98.3, "elapsed_time": "1 day, 19:40:40", "remaining_time": "0:45:19"}
+{"current_steps": 17590, "total_steps": 17884, "loss": 0.0145, "lr": 1.6461267103208278e-08, "epoch": 1.9670664541921776, "percentage": 98.36, "elapsed_time": "1 day, 19:42:06", "remaining_time": "0:43:49"}
+{"current_steps": 17600, "total_steps": 17884, "loss": 0.0154, "lr": 1.5360779372887157e-08, "epoch": 1.9681847409768234, "percentage": 98.41, "elapsed_time": "1 day, 19:43:21", "remaining_time": "0:42:19"}
+{"current_steps": 17610, "total_steps": 17884, "loss": 0.0153, "lr": 1.4298332475921606e-08, "epoch": 1.9693030277614696, "percentage": 98.47, "elapsed_time": "1 day, 19:45:01", "remaining_time": "0:40:50"}
+{"current_steps": 17620, "total_steps": 17884, "loss": 0.0148, "lr": 1.3273930460165896e-08, "epoch": 1.9704213145461154, "percentage": 98.52, "elapsed_time": "1 day, 19:46:49", "remaining_time": "0:39:21"}
+{"current_steps": 17630, "total_steps": 17884, "loss": 0.0185, "lr": 1.2287577228524695e-08, "epoch": 1.9715396013307611, "percentage": 98.58, "elapsed_time": "1 day, 19:48:12", "remaining_time": "0:37:51"}
+{"current_steps": 17640, "total_steps": 17884, "loss": 0.0151, "lr": 1.1339276538940846e-08, "epoch": 1.9726578881154073, "percentage": 98.64, "elapsed_time": "1 day, 19:49:38", "remaining_time": "0:36:22"}
+{"current_steps": 17650, "total_steps": 17884, "loss": 0.0171, "lr": 1.0429032004378726e-08, "epoch": 1.973776174900053, "percentage": 98.69, "elapsed_time": "1 day, 19:51:04", "remaining_time": "0:34:52"}
+{"current_steps": 17660, "total_steps": 17884, "loss": 0.0185, "lr": 9.556847092813126e-09, "epoch": 1.974894461684699, "percentage": 98.75, "elapsed_time": "1 day, 19:52:41", "remaining_time": "0:33:23"}
+{"current_steps": 17670, "total_steps": 17884, "loss": 0.0119, "lr": 8.722725127210396e-09, "epoch": 1.976012748469345, "percentage": 98.8, "elapsed_time": "1 day, 19:54:19", "remaining_time": "0:31:54"}
+{"current_steps": 17680, "total_steps": 17884, "loss": 0.0194, "lr": 7.926669285522881e-09, "epoch": 1.9771310352539908, "percentage": 98.86, "elapsed_time": "1 day, 19:55:58", "remaining_time": "0:30:24"}
+{"current_steps": 17690, "total_steps": 17884, "loss": 0.0244, "lr": 7.168682600671162e-09, "epoch": 1.9782493220386368, "percentage": 98.92, "elapsed_time": "1 day, 19:57:19", "remaining_time": "0:28:55"}
+{"current_steps": 17700, "total_steps": 17884, "loss": 0.0175, "lr": 6.448767960536284e-09, "epoch": 1.9793676088232828, "percentage": 98.97, "elapsed_time": "1 day, 19:58:48", "remaining_time": "0:27:25"}
+{"current_steps": 17710, "total_steps": 17884, "loss": 0.0148, "lr": 5.7669281079475446e-09, "epoch": 1.9804858956079285, "percentage": 99.03, "elapsed_time": "1 day, 20:00:16", "remaining_time": "0:25:56"}
+{"current_steps": 17720, "total_steps": 17884, "loss": 0.0209, "lr": 5.12316564066917e-09, "epoch": 1.9816041823925747, "percentage": 99.08, "elapsed_time": "1 day, 20:01:41", "remaining_time": "0:24:26"}
+{"current_steps": 17730, "total_steps": 17884, "loss": 0.0193, "lr": 4.517483011395873e-09, "epoch": 1.9827224691772205, "percentage": 99.14, "elapsed_time": "1 day, 20:03:07", "remaining_time": "0:22:57"}
+{"current_steps": 17740, "total_steps": 17884, "loss": 0.0262, "lr": 3.9498825277395345e-09, "epoch": 1.9838407559618663, "percentage": 99.19, "elapsed_time": "1 day, 20:04:34", "remaining_time": "0:21:27"}
+{"current_steps": 17750, "total_steps": 17884, "loss": 0.0164, "lr": 3.420366352221427e-09, "epoch": 1.9849590427465125, "percentage": 99.25, "elapsed_time": "1 day, 20:06:09", "remaining_time": "0:19:58"}
+{"current_steps": 17760, "total_steps": 17884, "loss": 0.0204, "lr": 2.9289365022644455e-09, "epoch": 1.9860773295311582, "percentage": 99.31, "elapsed_time": "1 day, 20:07:37", "remaining_time": "0:18:29"}
+{"current_steps": 17770, "total_steps": 17884, "loss": 0.0162, "lr": 2.4755948501831163e-09, "epoch": 1.9871956163158042, "percentage": 99.36, "elapsed_time": "1 day, 20:09:27", "remaining_time": "0:16:59"}
+{"current_steps": 17780, "total_steps": 17884, "loss": 0.0194, "lr": 2.0603431231824845e-09, "epoch": 1.9883139031004502, "percentage": 99.42, "elapsed_time": "1 day, 20:10:56", "remaining_time": "0:15:30"}
+{"current_steps": 17790, "total_steps": 17884, "loss": 0.0184, "lr": 1.6831829033425728e-09, "epoch": 1.989432189885096, "percentage": 99.47, "elapsed_time": "1 day, 20:12:34", "remaining_time": "0:14:00"}
+{"current_steps": 17800, "total_steps": 17884, "loss": 0.0191, "lr": 1.3441156276217115e-09, "epoch": 1.990550476669742, "percentage": 99.53, "elapsed_time": "1 day, 20:14:05", "remaining_time": "0:12:31"}
+{"current_steps": 17810, "total_steps": 17884, "loss": 0.0148, "lr": 1.0431425878421053e-09, "epoch": 1.991668763454388, "percentage": 99.59, "elapsed_time": "1 day, 20:15:27", "remaining_time": "0:11:02"}
+{"current_steps": 17820, "total_steps": 17884, "loss": 0.0194, "lr": 7.802649306931642e-10, "epoch": 1.9927870502390337, "percentage": 99.64, "elapsed_time": "1 day, 20:17:01", "remaining_time": "0:09:32"}
+{"current_steps": 17830, "total_steps": 17884, "loss": 0.0161, "lr": 5.554836577215117e-10, "epoch": 1.9939053370236797, "percentage": 99.7, "elapsed_time": "1 day, 20:18:26", "remaining_time": "0:08:03"}
+{"current_steps": 17840, "total_steps": 17884, "loss": 0.0167, "lr": 3.6879962532987424e-10, "epoch": 1.9950236238083257, "percentage": 99.75, "elapsed_time": "1 day, 20:19:54", "remaining_time": "0:06:33"}
+{"current_steps": 17850, "total_steps": 17884, "loss": 0.0185, "lr": 2.2021354477153034e-10, "epoch": 1.9961419105929714, "percentage": 99.81, "elapsed_time": "1 day, 20:21:21", "remaining_time": "0:05:04"}
+{"current_steps": 17860, "total_steps": 17884, "loss": 0.0159, "lr": 1.0972598215142072e-10, "epoch": 1.9972601973776176, "percentage": 99.87, "elapsed_time": "1 day, 20:22:44", "remaining_time": "0:03:34"}
+{"current_steps": 17870, "total_steps": 17884, "loss": 0.0199, "lr": 3.733735841726649e-11, "epoch": 1.9983784841622634, "percentage": 99.92, "elapsed_time": "1 day, 20:24:20", "remaining_time": "0:02:05"}
+{"current_steps": 17880, "total_steps": 17884, "loss": 0.0176, "lr": 3.0479493673407633e-12, "epoch": 1.9994967709469094, "percentage": 99.98, "elapsed_time": "1 day, 20:25:56", "remaining_time": "0:00:35"}
+{"current_steps": 17884, "total_steps": 17884, "epoch": 1.9999440856607678, "percentage": 100.0, "elapsed_time": "1 day, 20:27:12", "remaining_time": "0:00:00"}