Training in progress, epoch 4

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6af0b9cd4f03a09dfa8986eb8d48f90df1aaf920e6763f13a2e2ab492be6f941
+oid sha256:1f2be67e18476b13b782c835c24bd6bf40fa1e868867a0f5498ffd7461bfa6f9
 size 4877660776

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2dfb4b0eb69df6ffc1021329377d51432d8c9cb63b427398e6be14110599dbaa
+oid sha256:fd9d591d9ab44eac1a99d1a49cd2fb9f20980a1b46dbfdf27c5fb26be47c6e5b
 size 4932751008

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7855db0f5da04906cdf7744c5d2e22be151071f8072ee6288d914ed1b88d5fc8
+oid sha256:63aecdd832d583905127b9a34c4aa3fbc691f17690868b613e9671eafd4e29a8
 size 4330865200

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae63275163a3555b3bcd2e37f6624f5801ae5311973fe0b1a239bd8efb348acf
+oid sha256:117f06a60fce7f93c58e2c565816b1269a9fdf32260d333707e3e9c7047606dd
 size 1089994880

trainer_log.jsonl

@@ -767,3 +767,190 @@
 {"current_steps": 767, "total_steps": 955, "loss": 0.9412, "lr": 9.088284229726572e-06, "epoch": 4.0, "percentage": 80.31, "elapsed_time": "11:06:02", "remaining_time": "2:43:15"}
 {"current_steps": 768, "total_steps": 955, "loss": 0.7375, "lr": 8.995646566503838e-06, "epoch": 4.005215123859192, "percentage": 80.42, "elapsed_time": "11:08:13", "remaining_time": "2:42:42"}
 {"current_steps": 769, "total_steps": 955, "loss": 0.7373, "lr": 8.90342360478723e-06, "epoch": 4.010430247718383, "percentage": 80.52, "elapsed_time": "11:09:04", "remaining_time": "2:41:49"}
+{"current_steps": 770, "total_steps": 955, "loss": 0.7453, "lr": 8.81161657811318e-06, "epoch": 4.015645371577575, "percentage": 80.63, "elapsed_time": "11:09:54", "remaining_time": "2:40:57"}
+{"current_steps": 771, "total_steps": 955, "loss": 0.7421, "lr": 8.720226714454831e-06, "epoch": 4.020860495436767, "percentage": 80.73, "elapsed_time": "11:10:44", "remaining_time": "2:40:04"}
+{"current_steps": 772, "total_steps": 955, "loss": 0.7393, "lr": 8.629255236205475e-06, "epoch": 4.026075619295958, "percentage": 80.84, "elapsed_time": "11:11:35", "remaining_time": "2:39:11"}
+{"current_steps": 773, "total_steps": 955, "loss": 0.7262, "lr": 8.538703360162267e-06, "epoch": 4.03129074315515, "percentage": 80.94, "elapsed_time": "11:12:25", "remaining_time": "2:38:19"}
+{"current_steps": 774, "total_steps": 955, "loss": 0.7229, "lr": 8.44857229750994e-06, "epoch": 4.036505867014341, "percentage": 81.05, "elapsed_time": "11:13:17", "remaining_time": "2:37:27"}
+{"current_steps": 775, "total_steps": 955, "loss": 0.7406, "lr": 8.3588632538046e-06, "epoch": 4.041720990873533, "percentage": 81.15, "elapsed_time": "11:14:19", "remaining_time": "2:36:37"}
+{"current_steps": 776, "total_steps": 955, "loss": 0.7375, "lr": 8.269577428957571e-06, "epoch": 4.046936114732725, "percentage": 81.26, "elapsed_time": "11:15:20", "remaining_time": "2:35:46"}
+{"current_steps": 777, "total_steps": 955, "loss": 0.7315, "lr": 8.180716017219433e-06, "epoch": 4.052151238591916, "percentage": 81.36, "elapsed_time": "11:16:10", "remaining_time": "2:34:54"}
+{"current_steps": 778, "total_steps": 955, "loss": 0.7439, "lr": 8.092280207163941e-06, "epoch": 4.057366362451108, "percentage": 81.47, "elapsed_time": "11:17:01", "remaining_time": "2:34:01"}
+{"current_steps": 779, "total_steps": 955, "loss": 0.7219, "lr": 8.004271181672178e-06, "epoch": 4.0625814863103, "percentage": 81.57, "elapsed_time": "11:17:51", "remaining_time": "2:33:08"}
+{"current_steps": 780, "total_steps": 955, "loss": 0.7409, "lr": 7.916690117916772e-06, "epoch": 4.067796610169491, "percentage": 81.68, "elapsed_time": "11:18:41", "remaining_time": "2:32:16"}
+{"current_steps": 781, "total_steps": 955, "loss": 0.7377, "lr": 7.829538187346077e-06, "epoch": 4.073011734028683, "percentage": 81.78, "elapsed_time": "11:19:32", "remaining_time": "2:31:23"}
+{"current_steps": 782, "total_steps": 955, "loss": 0.7399, "lr": 7.742816555668532e-06, "epoch": 4.078226857887874, "percentage": 81.88, "elapsed_time": "11:20:22", "remaining_time": "2:30:31"}
+{"current_steps": 783, "total_steps": 955, "loss": 0.7327, "lr": 7.656526382837106e-06, "epoch": 4.083441981747066, "percentage": 81.99, "elapsed_time": "11:21:13", "remaining_time": "2:29:38"}
+{"current_steps": 784, "total_steps": 955, "loss": 0.7309, "lr": 7.570668823033727e-06, "epoch": 4.088657105606258, "percentage": 82.09, "elapsed_time": "11:22:03", "remaining_time": "2:28:45"}
+{"current_steps": 785, "total_steps": 955, "loss": 0.7343, "lr": 7.4852450246538685e-06, "epoch": 4.0938722294654495, "percentage": 82.2, "elapsed_time": "11:22:54", "remaining_time": "2:27:53"}
+{"current_steps": 786, "total_steps": 955, "loss": 0.736, "lr": 7.40025613029121e-06, "epoch": 4.0990873533246415, "percentage": 82.3, "elapsed_time": "11:23:44", "remaining_time": "2:27:00"}
+{"current_steps": 787, "total_steps": 955, "loss": 0.7342, "lr": 7.315703276722317e-06, "epoch": 4.1043024771838335, "percentage": 82.41, "elapsed_time": "11:24:35", "remaining_time": "2:26:08"}
+{"current_steps": 788, "total_steps": 955, "loss": 0.7458, "lr": 7.231587594891438e-06, "epoch": 4.1095176010430245, "percentage": 82.51, "elapsed_time": "11:25:25", "remaining_time": "2:25:15"}
+{"current_steps": 789, "total_steps": 955, "loss": 0.7318, "lr": 7.147910209895435e-06, "epoch": 4.1147327249022165, "percentage": 82.62, "elapsed_time": "11:26:18", "remaining_time": "2:24:23"}
+{"current_steps": 790, "total_steps": 955, "loss": 0.7442, "lr": 7.064672240968638e-06, "epoch": 4.1199478487614085, "percentage": 82.72, "elapsed_time": "11:27:22", "remaining_time": "2:23:33"}
+{"current_steps": 791, "total_steps": 955, "loss": 0.7403, "lr": 6.9818748014679785e-06, "epoch": 4.1251629726206, "percentage": 82.83, "elapsed_time": "11:28:22", "remaining_time": "2:22:43"}
+{"current_steps": 792, "total_steps": 955, "loss": 0.7405, "lr": 6.899518998858017e-06, "epoch": 4.130378096479792, "percentage": 82.93, "elapsed_time": "11:29:13", "remaining_time": "2:21:50"}
+{"current_steps": 793, "total_steps": 955, "loss": 0.7346, "lr": 6.81760593469615e-06, "epoch": 4.135593220338983, "percentage": 83.04, "elapsed_time": "11:30:03", "remaining_time": "2:20:58"}
+{"current_steps": 794, "total_steps": 955, "loss": 0.7303, "lr": 6.7361367046179325e-06, "epoch": 4.140808344198175, "percentage": 83.14, "elapsed_time": "11:30:53", "remaining_time": "2:20:05"}
+{"current_steps": 795, "total_steps": 955, "loss": 0.731, "lr": 6.655112398322332e-06, "epoch": 4.146023468057367, "percentage": 83.25, "elapsed_time": "11:31:44", "remaining_time": "2:19:13"}
+{"current_steps": 796, "total_steps": 955, "loss": 0.7307, "lr": 6.574534099557204e-06, "epoch": 4.151238591916558, "percentage": 83.35, "elapsed_time": "11:32:34", "remaining_time": "2:18:20"}
+{"current_steps": 797, "total_steps": 955, "loss": 0.7275, "lr": 6.49440288610482e-06, "epoch": 4.15645371577575, "percentage": 83.46, "elapsed_time": "11:33:24", "remaining_time": "2:17:27"}
+{"current_steps": 798, "total_steps": 955, "loss": 0.7339, "lr": 6.414719829767406e-06, "epoch": 4.161668839634942, "percentage": 83.56, "elapsed_time": "11:34:15", "remaining_time": "2:16:35"}
+{"current_steps": 799, "total_steps": 955, "loss": 0.7418, "lr": 6.335485996352782e-06, "epoch": 4.166883963494133, "percentage": 83.66, "elapsed_time": "11:35:05", "remaining_time": "2:15:42"}
+{"current_steps": 800, "total_steps": 955, "loss": 0.7304, "lr": 6.256702445660221e-06, "epoch": 4.172099087353325, "percentage": 83.77, "elapsed_time": "11:35:56", "remaining_time": "2:14:50"}
+{"current_steps": 801, "total_steps": 955, "loss": 0.743, "lr": 6.178370231466142e-06, "epoch": 4.177314211212516, "percentage": 83.87, "elapsed_time": "11:36:46", "remaining_time": "2:13:57"}
+{"current_steps": 802, "total_steps": 955, "loss": 0.7437, "lr": 6.100490401510089e-06, "epoch": 4.182529335071708, "percentage": 83.98, "elapsed_time": "11:37:37", "remaining_time": "2:13:05"}
+{"current_steps": 803, "total_steps": 955, "loss": 0.7383, "lr": 6.0230639974807206e-06, "epoch": 4.1877444589309, "percentage": 84.08, "elapsed_time": "11:38:27", "remaining_time": "2:12:12"}
+{"current_steps": 804, "total_steps": 955, "loss": 0.7329, "lr": 5.946092055001824e-06, "epoch": 4.192959582790091, "percentage": 84.19, "elapsed_time": "11:39:20", "remaining_time": "2:11:20"}
+{"current_steps": 805, "total_steps": 955, "loss": 0.7417, "lr": 5.869575603618507e-06, "epoch": 4.198174706649283, "percentage": 84.29, "elapsed_time": "11:40:23", "remaining_time": "2:10:30"}
+{"current_steps": 806, "total_steps": 955, "loss": 0.7337, "lr": 5.793515666783429e-06, "epoch": 4.203389830508475, "percentage": 84.4, "elapsed_time": "11:41:23", "remaining_time": "2:09:39"}
+{"current_steps": 807, "total_steps": 955, "loss": 0.733, "lr": 5.717913261843078e-06, "epoch": 4.208604954367666, "percentage": 84.5, "elapsed_time": "11:42:14", "remaining_time": "2:08:47"}
+{"current_steps": 808, "total_steps": 955, "loss": 0.7422, "lr": 5.642769400024199e-06, "epoch": 4.213820078226858, "percentage": 84.61, "elapsed_time": "11:43:04", "remaining_time": "2:07:54"}
+{"current_steps": 809, "total_steps": 955, "loss": 0.7404, "lr": 5.568085086420247e-06, "epoch": 4.219035202086049, "percentage": 84.71, "elapsed_time": "11:43:55", "remaining_time": "2:07:02"}
+{"current_steps": 810, "total_steps": 955, "loss": 0.7334, "lr": 5.4938613199779334e-06, "epoch": 4.224250325945241, "percentage": 84.82, "elapsed_time": "11:44:45", "remaining_time": "2:06:09"}
+{"current_steps": 811, "total_steps": 955, "loss": 0.7222, "lr": 5.420099093483911e-06, "epoch": 4.229465449804433, "percentage": 84.92, "elapsed_time": "11:45:35", "remaining_time": "2:05:17"}
+{"current_steps": 812, "total_steps": 955, "loss": 0.7307, "lr": 5.346799393551436e-06, "epoch": 4.234680573663624, "percentage": 85.03, "elapsed_time": "11:46:26", "remaining_time": "2:04:24"}
+{"current_steps": 813, "total_steps": 955, "loss": 0.7387, "lr": 5.273963200607206e-06, "epoch": 4.239895697522816, "percentage": 85.13, "elapsed_time": "11:47:16", "remaining_time": "2:03:32"}
+{"current_steps": 814, "total_steps": 955, "loss": 0.7251, "lr": 5.201591488878257e-06, "epoch": 4.245110821382008, "percentage": 85.24, "elapsed_time": "11:48:07", "remaining_time": "2:02:39"}
+{"current_steps": 815, "total_steps": 955, "loss": 0.7234, "lr": 5.129685226378893e-06, "epoch": 4.250325945241199, "percentage": 85.34, "elapsed_time": "11:48:57", "remaining_time": "2:01:47"}
+{"current_steps": 816, "total_steps": 955, "loss": 0.7271, "lr": 5.058245374897763e-06, "epoch": 4.255541069100391, "percentage": 85.45, "elapsed_time": "11:49:48", "remaining_time": "2:00:54"}
+{"current_steps": 817, "total_steps": 955, "loss": 0.7393, "lr": 4.987272889985009e-06, "epoch": 4.260756192959583, "percentage": 85.55, "elapsed_time": "11:50:38", "remaining_time": "2:00:02"}
+{"current_steps": 818, "total_steps": 955, "loss": 0.7332, "lr": 4.916768720939451e-06, "epoch": 4.265971316818774, "percentage": 85.65, "elapsed_time": "11:51:29", "remaining_time": "1:59:09"}
+{"current_steps": 819, "total_steps": 955, "loss": 0.7297, "lr": 4.846733810795914e-06, "epoch": 4.271186440677966, "percentage": 85.76, "elapsed_time": "11:52:22", "remaining_time": "1:58:17"}
+{"current_steps": 820, "total_steps": 955, "loss": 0.7276, "lr": 4.777169096312624e-06, "epoch": 4.276401564537157, "percentage": 85.86, "elapsed_time": "11:53:27", "remaining_time": "1:57:27"}
+{"current_steps": 821, "total_steps": 955, "loss": 0.7281, "lr": 4.708075507958642e-06, "epoch": 4.281616688396349, "percentage": 85.97, "elapsed_time": "11:54:26", "remaining_time": "1:56:36"}
+{"current_steps": 822, "total_steps": 955, "loss": 0.7276, "lr": 4.639453969901442e-06, "epoch": 4.286831812255541, "percentage": 86.07, "elapsed_time": "11:55:17", "remaining_time": "1:55:44"}
+{"current_steps": 823, "total_steps": 955, "loss": 0.7409, "lr": 4.571305399994561e-06, "epoch": 4.292046936114732, "percentage": 86.18, "elapsed_time": "11:56:07", "remaining_time": "1:54:51"}
+{"current_steps": 824, "total_steps": 955, "loss": 0.741, "lr": 4.503630709765294e-06, "epoch": 4.297262059973924, "percentage": 86.28, "elapsed_time": "11:56:58", "remaining_time": "1:53:59"}
+{"current_steps": 825, "total_steps": 955, "loss": 0.7331, "lr": 4.436430804402521e-06, "epoch": 4.302477183833116, "percentage": 86.39, "elapsed_time": "11:57:48", "remaining_time": "1:53:06"}
+{"current_steps": 826, "total_steps": 955, "loss": 0.7376, "lr": 4.3697065827446e-06, "epoch": 4.3076923076923075, "percentage": 86.49, "elapsed_time": "11:58:39", "remaining_time": "1:52:14"}
+{"current_steps": 827, "total_steps": 955, "loss": 0.7267, "lr": 4.303458937267326e-06, "epoch": 4.3129074315514995, "percentage": 86.6, "elapsed_time": "11:59:29", "remaining_time": "1:51:21"}
+{"current_steps": 828, "total_steps": 955, "loss": 0.7309, "lr": 4.237688754072022e-06, "epoch": 4.318122555410691, "percentage": 86.7, "elapsed_time": "12:00:19", "remaining_time": "1:50:29"}
+{"current_steps": 829, "total_steps": 955, "loss": 0.7342, "lr": 4.172396912873659e-06, "epoch": 4.3233376792698825, "percentage": 86.81, "elapsed_time": "12:01:10", "remaining_time": "1:49:36"}
+{"current_steps": 830, "total_steps": 955, "loss": 0.7392, "lr": 4.107584286989097e-06, "epoch": 4.3285528031290745, "percentage": 86.91, "elapsed_time": "12:02:00", "remaining_time": "1:48:44"}
+{"current_steps": 831, "total_steps": 955, "loss": 0.7303, "lr": 4.043251743325436e-06, "epoch": 4.333767926988266, "percentage": 87.02, "elapsed_time": "12:02:51", "remaining_time": "1:47:51"}
+{"current_steps": 832, "total_steps": 955, "loss": 0.7301, "lr": 3.979400142368368e-06, "epoch": 4.338983050847458, "percentage": 87.12, "elapsed_time": "12:03:41", "remaining_time": "1:46:59"}
+{"current_steps": 833, "total_steps": 955, "loss": 0.7374, "lr": 3.916030338170696e-06, "epoch": 4.34419817470665, "percentage": 87.23, "elapsed_time": "12:04:32", "remaining_time": "1:46:06"}
+{"current_steps": 834, "total_steps": 955, "loss": 0.7387, "lr": 3.8531431783409165e-06, "epoch": 4.349413298565841, "percentage": 87.33, "elapsed_time": "12:05:25", "remaining_time": "1:45:14"}
+{"current_steps": 835, "total_steps": 955, "loss": 0.7385, "lr": 3.7907395040318685e-06, "epoch": 4.354628422425033, "percentage": 87.43, "elapsed_time": "12:06:29", "remaining_time": "1:44:24"}
+{"current_steps": 836, "total_steps": 955, "loss": 0.7268, "lr": 3.7288201499294797e-06, "epoch": 4.359843546284225, "percentage": 87.54, "elapsed_time": "12:07:27", "remaining_time": "1:43:32"}
+{"current_steps": 837, "total_steps": 955, "loss": 0.7453, "lr": 3.667385944241626e-06, "epoch": 4.365058670143416, "percentage": 87.64, "elapsed_time": "12:08:17", "remaining_time": "1:42:40"}
+{"current_steps": 838, "total_steps": 955, "loss": 0.7395, "lr": 3.606437708687023e-06, "epoch": 4.370273794002608, "percentage": 87.75, "elapsed_time": "12:09:07", "remaining_time": "1:41:47"}
+{"current_steps": 839, "total_steps": 955, "loss": 0.7354, "lr": 3.545976258484247e-06, "epoch": 4.375488917861799, "percentage": 87.85, "elapsed_time": "12:09:58", "remaining_time": "1:40:55"}
+{"current_steps": 840, "total_steps": 955, "loss": 0.7459, "lr": 3.4860024023408577e-06, "epoch": 4.380704041720991, "percentage": 87.96, "elapsed_time": "12:10:48", "remaining_time": "1:40:03"}
+{"current_steps": 841, "total_steps": 955, "loss": 0.7509, "lr": 3.426516942442524e-06, "epoch": 4.385919165580183, "percentage": 88.06, "elapsed_time": "12:11:38", "remaining_time": "1:39:10"}
+{"current_steps": 842, "total_steps": 955, "loss": 0.7375, "lr": 3.3675206744423394e-06, "epoch": 4.391134289439374, "percentage": 88.17, "elapsed_time": "12:12:29", "remaining_time": "1:38:18"}
+{"current_steps": 843, "total_steps": 955, "loss": 0.7302, "lr": 3.309014387450189e-06, "epoch": 4.396349413298566, "percentage": 88.27, "elapsed_time": "12:13:19", "remaining_time": "1:37:25"}
+{"current_steps": 844, "total_steps": 955, "loss": 0.7407, "lr": 3.2509988640221414e-06, "epoch": 4.401564537157758, "percentage": 88.38, "elapsed_time": "12:14:10", "remaining_time": "1:36:33"}
+{"current_steps": 845, "total_steps": 955, "loss": 0.7255, "lr": 3.1934748801500184e-06, "epoch": 4.406779661016949, "percentage": 88.48, "elapsed_time": "12:15:00", "remaining_time": "1:35:40"}
+{"current_steps": 846, "total_steps": 955, "loss": 0.7369, "lr": 3.136443205251034e-06, "epoch": 4.411994784876141, "percentage": 88.59, "elapsed_time": "12:15:51", "remaining_time": "1:34:48"}
+{"current_steps": 847, "total_steps": 955, "loss": 0.7337, "lr": 3.0799046021574574e-06, "epoch": 4.417209908735332, "percentage": 88.69, "elapsed_time": "12:16:41", "remaining_time": "1:33:56"}
+{"current_steps": 848, "total_steps": 955, "loss": 0.7319, "lr": 3.0238598271064323e-06, "epoch": 4.422425032594524, "percentage": 88.8, "elapsed_time": "12:17:32", "remaining_time": "1:33:03"}
+{"current_steps": 849, "total_steps": 955, "loss": 0.7226, "lr": 2.9683096297298753e-06, "epoch": 4.427640156453716, "percentage": 88.9, "elapsed_time": "12:18:26", "remaining_time": "1:32:11"}
+{"current_steps": 850, "total_steps": 955, "loss": 0.74, "lr": 2.9132547530444254e-06, "epoch": 4.432855280312907, "percentage": 89.01, "elapsed_time": "12:19:30", "remaining_time": "1:31:21"}
+{"current_steps": 851, "total_steps": 955, "loss": 0.7325, "lr": 2.8586959334415064e-06, "epoch": 4.438070404172099, "percentage": 89.11, "elapsed_time": "12:20:28", "remaining_time": "1:30:29"}
+{"current_steps": 852, "total_steps": 955, "loss": 0.7197, "lr": 2.8046339006775115e-06, "epoch": 4.443285528031291, "percentage": 89.21, "elapsed_time": "12:21:18", "remaining_time": "1:29:37"}
+{"current_steps": 853, "total_steps": 955, "loss": 0.7302, "lr": 2.751069377863993e-06, "epoch": 4.448500651890482, "percentage": 89.32, "elapsed_time": "12:22:08", "remaining_time": "1:28:44"}
+{"current_steps": 854, "total_steps": 955, "loss": 0.7289, "lr": 2.698003081458018e-06, "epoch": 4.453715775749674, "percentage": 89.42, "elapsed_time": "12:22:59", "remaining_time": "1:27:52"}
+{"current_steps": 855, "total_steps": 955, "loss": 0.7279, "lr": 2.6454357212526026e-06, "epoch": 4.458930899608866, "percentage": 89.53, "elapsed_time": "12:23:49", "remaining_time": "1:26:59"}
+{"current_steps": 856, "total_steps": 955, "loss": 0.7335, "lr": 2.5933680003671625e-06, "epoch": 4.464146023468057, "percentage": 89.63, "elapsed_time": "12:24:40", "remaining_time": "1:26:07"}
+{"current_steps": 857, "total_steps": 955, "loss": 0.7266, "lr": 2.541800615238175e-06, "epoch": 4.469361147327249, "percentage": 89.74, "elapsed_time": "12:25:30", "remaining_time": "1:25:15"}
+{"current_steps": 858, "total_steps": 955, "loss": 0.7269, "lr": 2.4907342556098125e-06, "epoch": 4.47457627118644, "percentage": 89.84, "elapsed_time": "12:26:21", "remaining_time": "1:24:22"}
+{"current_steps": 859, "total_steps": 955, "loss": 0.7453, "lr": 2.440169604524738e-06, "epoch": 4.479791395045632, "percentage": 89.95, "elapsed_time": "12:27:11", "remaining_time": "1:23:30"}
+{"current_steps": 860, "total_steps": 955, "loss": 0.7319, "lr": 2.390107338314982e-06, "epoch": 4.485006518904824, "percentage": 90.05, "elapsed_time": "12:28:02", "remaining_time": "1:22:37"}
+{"current_steps": 861, "total_steps": 955, "loss": 0.737, "lr": 2.340548126592861e-06, "epoch": 4.490221642764015, "percentage": 90.16, "elapsed_time": "12:28:52", "remaining_time": "1:21:45"}
+{"current_steps": 862, "total_steps": 955, "loss": 0.7343, "lr": 2.2914926322420515e-06, "epoch": 4.495436766623207, "percentage": 90.26, "elapsed_time": "12:29:42", "remaining_time": "1:20:53"}
+{"current_steps": 863, "total_steps": 955, "loss": 0.737, "lr": 2.242941511408714e-06, "epoch": 4.500651890482399, "percentage": 90.37, "elapsed_time": "12:30:33", "remaining_time": "1:20:00"}
+{"current_steps": 864, "total_steps": 955, "loss": 0.7377, "lr": 2.1948954134927103e-06, "epoch": 4.50586701434159, "percentage": 90.47, "elapsed_time": "12:31:27", "remaining_time": "1:19:08"}
+{"current_steps": 865, "total_steps": 955, "loss": 0.7257, "lr": 2.147354981138925e-06, "epoch": 4.511082138200782, "percentage": 90.58, "elapsed_time": "12:32:31", "remaining_time": "1:18:17"}
+{"current_steps": 866, "total_steps": 955, "loss": 0.729, "lr": 2.1003208502286744e-06, "epoch": 4.5162972620599735, "percentage": 90.68, "elapsed_time": "12:33:28", "remaining_time": "1:17:26"}
+{"current_steps": 867, "total_steps": 955, "loss": 0.7316, "lr": 2.0537936498711898e-06, "epoch": 4.5215123859191655, "percentage": 90.79, "elapsed_time": "12:34:18", "remaining_time": "1:16:33"}
+{"current_steps": 868, "total_steps": 955, "loss": 0.7337, "lr": 2.007774002395193e-06, "epoch": 4.5267275097783575, "percentage": 90.89, "elapsed_time": "12:35:09", "remaining_time": "1:15:41"}
+{"current_steps": 869, "total_steps": 955, "loss": 0.7268, "lr": 1.9622625233406143e-06, "epoch": 4.531942633637549, "percentage": 90.99, "elapsed_time": "12:35:59", "remaining_time": "1:14:49"}
+{"current_steps": 870, "total_steps": 955, "loss": 0.736, "lr": 1.9172598214503146e-06, "epoch": 4.5371577574967406, "percentage": 91.1, "elapsed_time": "12:36:50", "remaining_time": "1:13:56"}
+{"current_steps": 871, "total_steps": 955, "loss": 0.7359, "lr": 1.8727664986619576e-06, "epoch": 4.5423728813559325, "percentage": 91.2, "elapsed_time": "12:37:40", "remaining_time": "1:13:04"}
+{"current_steps": 872, "total_steps": 955, "loss": 0.733, "lr": 1.8287831500999775e-06, "epoch": 4.547588005215124, "percentage": 91.31, "elapsed_time": "12:38:31", "remaining_time": "1:12:11"}
+{"current_steps": 873, "total_steps": 955, "loss": 0.7274, "lr": 1.7853103640675852e-06, "epoch": 4.552803129074316, "percentage": 91.41, "elapsed_time": "12:39:21", "remaining_time": "1:11:19"}
+{"current_steps": 874, "total_steps": 955, "loss": 0.7346, "lr": 1.74234872203892e-06, "epoch": 4.558018252933508, "percentage": 91.52, "elapsed_time": "12:40:11", "remaining_time": "1:10:27"}
+{"current_steps": 875, "total_steps": 955, "loss": 0.7355, "lr": 1.699898798651285e-06, "epoch": 4.563233376792699, "percentage": 91.62, "elapsed_time": "12:41:02", "remaining_time": "1:09:34"}
+{"current_steps": 876, "total_steps": 955, "loss": 0.7323, "lr": 1.6579611616974212e-06, "epoch": 4.568448500651891, "percentage": 91.73, "elapsed_time": "12:41:52", "remaining_time": "1:08:42"}
+{"current_steps": 877, "total_steps": 955, "loss": 0.7333, "lr": 1.6165363721179471e-06, "epoch": 4.573663624511082, "percentage": 91.83, "elapsed_time": "12:42:43", "remaining_time": "1:07:50"}
+{"current_steps": 878, "total_steps": 955, "loss": 0.7198, "lr": 1.575624983993853e-06, "epoch": 4.578878748370274, "percentage": 91.94, "elapsed_time": "12:43:33", "remaining_time": "1:06:57"}
+{"current_steps": 879, "total_steps": 955, "loss": 0.7265, "lr": 1.5352275445390752e-06, "epoch": 4.584093872229466, "percentage": 92.04, "elapsed_time": "12:44:28", "remaining_time": "1:06:05"}
+{"current_steps": 880, "total_steps": 955, "loss": 0.7328, "lr": 1.4953445940931688e-06, "epoch": 4.589308996088657, "percentage": 92.15, "elapsed_time": "12:45:33", "remaining_time": "1:05:14"}
+{"current_steps": 881, "total_steps": 955, "loss": 0.7305, "lr": 1.4559766661141317e-06, "epoch": 4.594524119947849, "percentage": 92.25, "elapsed_time": "12:46:30", "remaining_time": "1:04:23"}
+{"current_steps": 882, "total_steps": 955, "loss": 0.7432, "lr": 1.4171242871711922e-06, "epoch": 4.599739243807041, "percentage": 92.36, "elapsed_time": "12:47:21", "remaining_time": "1:03:30"}
+{"current_steps": 883, "total_steps": 955, "loss": 0.7317, "lr": 1.3787879769378277e-06, "epoch": 4.604954367666232, "percentage": 92.46, "elapsed_time": "12:48:11", "remaining_time": "1:02:38"}
+{"current_steps": 884, "total_steps": 955, "loss": 0.7379, "lr": 1.3409682481848063e-06, "epoch": 4.610169491525424, "percentage": 92.57, "elapsed_time": "12:49:01", "remaining_time": "1:01:45"}
+{"current_steps": 885, "total_steps": 955, "loss": 0.739, "lr": 1.3036656067732679e-06, "epoch": 4.615384615384615, "percentage": 92.67, "elapsed_time": "12:49:52", "remaining_time": "1:00:53"}
+{"current_steps": 886, "total_steps": 955, "loss": 0.743, "lr": 1.2668805516480577e-06, "epoch": 4.620599739243807, "percentage": 92.77, "elapsed_time": "12:50:42", "remaining_time": "1:00:01"}
+{"current_steps": 887, "total_steps": 955, "loss": 0.746, "lr": 1.2306135748309633e-06, "epoch": 4.625814863102999, "percentage": 92.88, "elapsed_time": "12:51:33", "remaining_time": "0:59:08"}
+{"current_steps": 888, "total_steps": 955, "loss": 0.7317, "lr": 1.1948651614141783e-06, "epoch": 4.63102998696219, "percentage": 92.98, "elapsed_time": "12:52:23", "remaining_time": "0:58:16"}
+{"current_steps": 889, "total_steps": 955, "loss": 0.7213, "lr": 1.1596357895538213e-06, "epoch": 4.636245110821382, "percentage": 93.09, "elapsed_time": "12:53:14", "remaining_time": "0:57:24"}
+{"current_steps": 890, "total_steps": 955, "loss": 0.7398, "lr": 1.1249259304634986e-06, "epoch": 4.641460234680574, "percentage": 93.19, "elapsed_time": "12:54:04", "remaining_time": "0:56:32"}
+{"current_steps": 891, "total_steps": 955, "loss": 0.7368, "lr": 1.0907360484080409e-06, "epoch": 4.646675358539765, "percentage": 93.3, "elapsed_time": "12:54:55", "remaining_time": "0:55:39"}
+{"current_steps": 892, "total_steps": 955, "loss": 0.7543, "lr": 1.0570666006972875e-06, "epoch": 4.651890482398957, "percentage": 93.4, "elapsed_time": "12:55:45", "remaining_time": "0:54:47"}
+{"current_steps": 893, "total_steps": 955, "loss": 0.7243, "lr": 1.023918037679943e-06, "epoch": 4.657105606258149, "percentage": 93.51, "elapsed_time": "12:56:36", "remaining_time": "0:53:55"}
+{"current_steps": 894, "total_steps": 955, "loss": 0.7373, "lr": 9.912908027375745e-07, "epoch": 4.66232073011734, "percentage": 93.61, "elapsed_time": "12:57:32", "remaining_time": "0:53:03"}
+{"current_steps": 895, "total_steps": 955, "loss": 0.7412, "lr": 9.591853322787003e-07, "epoch": 4.667535853976532, "percentage": 93.72, "elapsed_time": "12:58:36", "remaining_time": "0:52:11"}
+{"current_steps": 896, "total_steps": 955, "loss": 0.729, "lr": 9.276020557329101e-07, "epoch": 4.672750977835723, "percentage": 93.82, "elapsed_time": "12:59:34", "remaining_time": "0:51:20"}
+{"current_steps": 897, "total_steps": 955, "loss": 0.7329, "lr": 8.965413955451363e-07, "epoch": 4.677966101694915, "percentage": 93.93, "elapsed_time": "13:00:24", "remaining_time": "0:50:27"}
+{"current_steps": 898, "total_steps": 955, "loss": 0.7387, "lr": 8.660037671700405e-07, "epoch": 4.683181225554107, "percentage": 94.03, "elapsed_time": "13:01:15", "remaining_time": "0:49:35"}
+{"current_steps": 899, "total_steps": 955, "loss": 0.7377, "lr": 8.359895790663963e-07, "epoch": 4.688396349413298, "percentage": 94.14, "elapsed_time": "13:02:05", "remaining_time": "0:48:43"}
+{"current_steps": 900, "total_steps": 955, "loss": 0.7353, "lr": 8.064992326916577e-07, "epoch": 4.69361147327249, "percentage": 94.24, "elapsed_time": "13:02:55", "remaining_time": "0:47:50"}
+{"current_steps": 901, "total_steps": 955, "loss": 0.7272, "lr": 7.775331224966076e-07, "epoch": 4.698826597131681, "percentage": 94.35, "elapsed_time": "13:03:46", "remaining_time": "0:46:58"}
+{"current_steps": 902, "total_steps": 955, "loss": 0.7292, "lr": 7.49091635920034e-07, "epoch": 4.704041720990873, "percentage": 94.45, "elapsed_time": "13:04:36", "remaining_time": "0:46:06"}
+{"current_steps": 903, "total_steps": 955, "loss": 0.7416, "lr": 7.211751533835998e-07, "epoch": 4.709256844850065, "percentage": 94.55, "elapsed_time": "13:05:26", "remaining_time": "0:45:13"}
+{"current_steps": 904, "total_steps": 955, "loss": 0.7337, "lr": 6.937840482867142e-07, "epoch": 4.7144719687092564, "percentage": 94.66, "elapsed_time": "13:06:17", "remaining_time": "0:44:21"}
+{"current_steps": 905, "total_steps": 955, "loss": 0.7324, "lr": 6.669186870015454e-07, "epoch": 4.719687092568448, "percentage": 94.76, "elapsed_time": "13:07:07", "remaining_time": "0:43:29"}
+{"current_steps": 906, "total_steps": 955, "loss": 0.7374, "lr": 6.405794288681577e-07, "epoch": 4.72490221642764, "percentage": 94.87, "elapsed_time": "13:07:58", "remaining_time": "0:42:36"}
+{"current_steps": 907, "total_steps": 955, "loss": 0.7326, "lr": 6.147666261896445e-07, "epoch": 4.7301173402868315, "percentage": 94.97, "elapsed_time": "13:08:48", "remaining_time": "0:41:44"}
+{"current_steps": 908, "total_steps": 955, "loss": 0.7375, "lr": 5.894806242274565e-07, "epoch": 4.7353324641460235, "percentage": 95.08, "elapsed_time": "13:09:38", "remaining_time": "0:40:52"}
+{"current_steps": 909, "total_steps": 955, "loss": 0.7366, "lr": 5.647217611967914e-07, "epoch": 4.7405475880052155, "percentage": 95.18, "elapsed_time": "13:10:34", "remaining_time": "0:40:00"}
+{"current_steps": 910, "total_steps": 955, "loss": 0.7282, "lr": 5.404903682620299e-07, "epoch": 4.745762711864407, "percentage": 95.29, "elapsed_time": "13:11:40", "remaining_time": "0:39:08"}
+{"current_steps": 911, "total_steps": 955, "loss": 0.7234, "lr": 5.167867695323426e-07, "epoch": 4.7509778357235986, "percentage": 95.39, "elapsed_time": "13:12:35", "remaining_time": "0:38:16"}
+{"current_steps": 912, "total_steps": 955, "loss": 0.734, "lr": 4.936112820573335e-07, "epoch": 4.7561929595827905, "percentage": 95.5, "elapsed_time": "13:13:25", "remaining_time": "0:37:24"}
+{"current_steps": 913, "total_steps": 955, "loss": 0.7319, "lr": 4.709642158228134e-07, "epoch": 4.761408083441982, "percentage": 95.6, "elapsed_time": "13:14:15", "remaining_time": "0:36:32"}
+{"current_steps": 914, "total_steps": 955, "loss": 0.7345, "lr": 4.4884587374663727e-07, "epoch": 4.766623207301174, "percentage": 95.71, "elapsed_time": "13:15:06", "remaining_time": "0:35:40"}
+{"current_steps": 915, "total_steps": 955, "loss": 0.7328, "lr": 4.2725655167468626e-07, "epoch": 4.771838331160365, "percentage": 95.81, "elapsed_time": "13:15:56", "remaining_time": "0:34:47"}
+{"current_steps": 916, "total_steps": 955, "loss": 0.7395, "lr": 4.061965383768529e-07, "epoch": 4.777053455019557, "percentage": 95.92, "elapsed_time": "13:16:47", "remaining_time": "0:33:55"}
+{"current_steps": 917, "total_steps": 955, "loss": 0.7282, "lr": 3.856661155432484e-07, "epoch": 4.782268578878749, "percentage": 96.02, "elapsed_time": "13:17:37", "remaining_time": "0:33:03"}
+{"current_steps": 918, "total_steps": 955, "loss": 0.7321, "lr": 3.656655577803703e-07, "epoch": 4.78748370273794, "percentage": 96.13, "elapsed_time": "13:18:28", "remaining_time": "0:32:10"}
+{"current_steps": 919, "total_steps": 955, "loss": 0.721, "lr": 3.4619513260746973e-07, "epoch": 4.792698826597132, "percentage": 96.23, "elapsed_time": "13:19:18", "remaining_time": "0:31:18"}
+{"current_steps": 920, "total_steps": 955, "loss": 0.7383, "lr": 3.2725510045297225e-07, "epoch": 4.797913950456323, "percentage": 96.34, "elapsed_time": "13:20:08", "remaining_time": "0:30:26"}
+{"current_steps": 921, "total_steps": 955, "loss": 0.7356, "lr": 3.0884571465096936e-07, "epoch": 4.803129074315515, "percentage": 96.44, "elapsed_time": "13:20:59", "remaining_time": "0:29:34"}
+{"current_steps": 922, "total_steps": 955, "loss": 0.7375, "lr": 2.909672214378434e-07, "epoch": 4.808344198174707, "percentage": 96.54, "elapsed_time": "13:21:49", "remaining_time": "0:28:41"}
+{"current_steps": 923, "total_steps": 955, "loss": 0.7399, "lr": 2.736198599489903e-07, "epoch": 4.813559322033898, "percentage": 96.65, "elapsed_time": "13:22:40", "remaining_time": "0:27:49"}
+{"current_steps": 924, "total_steps": 955, "loss": 0.7509, "lr": 2.568038622155955e-07, "epoch": 4.81877444589309, "percentage": 96.75, "elapsed_time": "13:23:36", "remaining_time": "0:26:57"}
+{"current_steps": 925, "total_steps": 955, "loss": 0.7284, "lr": 2.4051945316153846e-07, "epoch": 4.823989569752282, "percentage": 96.86, "elapsed_time": "13:24:41", "remaining_time": "0:26:05"}
+{"current_steps": 926, "total_steps": 955, "loss": 0.7299, "lr": 2.2476685060039527e-07, "epoch": 4.829204693611473, "percentage": 96.96, "elapsed_time": "13:25:36", "remaining_time": "0:25:13"}
+{"current_steps": 927, "total_steps": 955, "loss": 0.7276, "lr": 2.0954626523251197e-07, "epoch": 4.834419817470665, "percentage": 97.07, "elapsed_time": "13:26:26", "remaining_time": "0:24:21"}
+{"current_steps": 928, "total_steps": 955, "loss": 0.7428, "lr": 1.948579006421847e-07, "epoch": 4.839634941329857, "percentage": 97.17, "elapsed_time": "13:27:16", "remaining_time": "0:23:29"}
+{"current_steps": 929, "total_steps": 955, "loss": 0.7401, "lr": 1.8070195329495944e-07, "epoch": 4.844850065189048, "percentage": 97.28, "elapsed_time": "13:28:07", "remaining_time": "0:22:37"}
+{"current_steps": 930, "total_steps": 955, "loss": 0.7409, "lr": 1.6707861253497214e-07, "epoch": 4.85006518904824, "percentage": 97.38, "elapsed_time": "13:28:57", "remaining_time": "0:21:44"}
+{"current_steps": 931, "total_steps": 955, "loss": 0.7362, "lr": 1.539880605824351e-07, "epoch": 4.855280312907432, "percentage": 97.49, "elapsed_time": "13:29:48", "remaining_time": "0:20:52"}
+{"current_steps": 932, "total_steps": 955, "loss": 0.7332, "lr": 1.4143047253120322e-07, "epoch": 4.860495436766623, "percentage": 97.59, "elapsed_time": "13:30:38", "remaining_time": "0:20:00"}
+{"current_steps": 933, "total_steps": 955, "loss": 0.7233, "lr": 1.2940601634642059e-07, "epoch": 4.865710560625815, "percentage": 97.7, "elapsed_time": "13:31:28", "remaining_time": "0:19:08"}
+{"current_steps": 934, "total_steps": 955, "loss": 0.7395, "lr": 1.1791485286227311e-07, "epoch": 4.870925684485006, "percentage": 97.8, "elapsed_time": "13:32:19", "remaining_time": "0:18:15"}
+{"current_steps": 935, "total_steps": 955, "loss": 0.7231, "lr": 1.0695713577984824e-07, "epoch": 4.876140808344198, "percentage": 97.91, "elapsed_time": "13:33:09", "remaining_time": "0:17:23"}
+{"current_steps": 936, "total_steps": 955, "loss": 0.7351, "lr": 9.653301166507422e-08, "epoch": 4.88135593220339, "percentage": 98.01, "elapsed_time": "13:34:00", "remaining_time": "0:16:31"}
+{"current_steps": 937, "total_steps": 955, "loss": 0.7333, "lr": 8.664261994675738e-08, "epoch": 4.886571056062581, "percentage": 98.12, "elapsed_time": "13:34:50", "remaining_time": "0:15:39"}
+{"current_steps": 938, "total_steps": 955, "loss": 0.7353, "lr": 7.728609291471678e-08, "epoch": 4.891786179921773, "percentage": 98.22, "elapsed_time": "13:35:41", "remaining_time": "0:14:46"}
+{"current_steps": 939, "total_steps": 955, "loss": 0.7387, "lr": 6.846355571801688e-08, "epoch": 4.897001303780964, "percentage": 98.32, "elapsed_time": "13:36:37", "remaining_time": "0:13:54"}
+{"current_steps": 940, "total_steps": 955, "loss": 0.7327, "lr": 6.017512636329325e-08, "epoch": 4.902216427640156, "percentage": 98.43, "elapsed_time": "13:37:42", "remaining_time": "0:13:02"}
+{"current_steps": 941, "total_steps": 955, "loss": 0.7323, "lr": 5.242091571318053e-08, "epoch": 4.907431551499348, "percentage": 98.53, "elapsed_time": "13:38:37", "remaining_time": "0:12:10"}
+{"current_steps": 942, "total_steps": 955, "loss": 0.7369, "lr": 4.5201027484811365e-08, "epoch": 4.912646675358539, "percentage": 98.64, "elapsed_time": "13:39:27", "remaining_time": "0:11:18"}
+{"current_steps": 943, "total_steps": 955, "loss": 0.7269, "lr": 3.851555824844866e-08, "epoch": 4.917861799217731, "percentage": 98.74, "elapsed_time": "13:40:17", "remaining_time": "0:10:26"}
+{"current_steps": 944, "total_steps": 955, "loss": 0.72, "lr": 3.23645974261888e-08, "epoch": 4.923076923076923, "percentage": 98.85, "elapsed_time": "13:41:08", "remaining_time": "0:09:34"}
+{"current_steps": 945, "total_steps": 955, "loss": 0.7302, "lr": 2.674822729075377e-08, "epoch": 4.9282920469361144, "percentage": 98.95, "elapsed_time": "13:41:58", "remaining_time": "0:08:41"}
+{"current_steps": 946, "total_steps": 955, "loss": 0.7389, "lr": 2.166652296440752e-08, "epoch": 4.933507170795306, "percentage": 99.06, "elapsed_time": "13:42:49", "remaining_time": "0:07:49"}
+{"current_steps": 947, "total_steps": 955, "loss": 0.7384, "lr": 1.7119552417943496e-08, "epoch": 4.938722294654498, "percentage": 99.16, "elapsed_time": "13:43:39", "remaining_time": "0:06:57"}
+{"current_steps": 948, "total_steps": 955, "loss": 0.7352, "lr": 1.3107376469769783e-08, "epoch": 4.9439374185136895, "percentage": 99.27, "elapsed_time": "13:44:29", "remaining_time": "0:06:05"}
+{"current_steps": 949, "total_steps": 955, "loss": 0.7202, "lr": 9.630048785105318e-09, "epoch": 4.9491525423728815, "percentage": 99.37, "elapsed_time": "13:45:20", "remaining_time": "0:05:13"}
+{"current_steps": 950, "total_steps": 955, "loss": 0.7352, "lr": 6.687615875264897e-09, "epoch": 4.9543676662320735, "percentage": 99.48, "elapsed_time": "13:46:10", "remaining_time": "0:04:20"}
+{"current_steps": 951, "total_steps": 955, "loss": 0.7342, "lr": 4.280117097015257e-09, "epoch": 4.959582790091265, "percentage": 99.58, "elapsed_time": "13:47:01", "remaining_time": "0:03:28"}
+{"current_steps": 952, "total_steps": 955, "loss": 0.7357, "lr": 2.407584652073247e-09, "epoch": 4.9647979139504566, "percentage": 99.69, "elapsed_time": "13:47:51", "remaining_time": "0:02:36"}
+{"current_steps": 953, "total_steps": 955, "loss": 0.7212, "lr": 1.0700435866706216e-09, "epoch": 4.970013037809648, "percentage": 99.79, "elapsed_time": "13:48:42", "remaining_time": "0:01:44"}
+{"current_steps": 954, "total_steps": 955, "loss": 0.7353, "lr": 2.6751179119877124e-10, "epoch": 4.97522816166884, "percentage": 99.9, "elapsed_time": "13:49:38", "remaining_time": "0:00:52"}
+{"current_steps": 955, "total_steps": 955, "loss": 0.7386, "lr": 0.0, "epoch": 4.980443285528032, "percentage": 100.0, "elapsed_time": "13:50:43", "remaining_time": "0:00:00"}
+{"current_steps": 955, "total_steps": 955, "epoch": 4.980443285528032, "percentage": 100.0, "elapsed_time": "13:52:14", "remaining_time": "0:00:00"}