diff --git "a/trainer_log.jsonl" "b/trainer_log.jsonl"
new file mode 100644--- /dev/null
+++ "b/trainer_log.jsonl"
@@ -0,0 +1,649 @@
+{"current_steps": 1, "total_steps": 540, "loss": 0.1002, "lr": 1.8518518518518519e-06, "epoch": 0.009259259259259259, "percentage": 0.19, "elapsed_time": "0:00:33", "remaining_time": "5:03:38"}
+{"current_steps": 2, "total_steps": 540, "loss": 0.0936, "lr": 3.7037037037037037e-06, "epoch": 0.018518518518518517, "percentage": 0.37, "elapsed_time": "0:00:49", "remaining_time": "3:40:20"}
+{"current_steps": 3, "total_steps": 540, "loss": 0.0898, "lr": 5.555555555555556e-06, "epoch": 0.027777777777777776, "percentage": 0.56, "elapsed_time": "0:01:08", "remaining_time": "3:25:04"}
+{"current_steps": 4, "total_steps": 540, "loss": 0.087, "lr": 7.4074074074074075e-06, "epoch": 0.037037037037037035, "percentage": 0.74, "elapsed_time": "0:01:24", "remaining_time": "3:07:44"}
+{"current_steps": 5, "total_steps": 540, "loss": 0.0904, "lr": 9.259259259259259e-06, "epoch": 0.046296296296296294, "percentage": 0.93, "elapsed_time": "0:01:40", "remaining_time": "2:59:13"}
+{"current_steps": 5, "total_steps": 540, "eval_loss": 0.0950983464717865, "epoch": 0.046296296296296294, "percentage": 0.93, "elapsed_time": "0:01:52", "remaining_time": "3:20:33"}
+{"current_steps": 6, "total_steps": 540, "loss": 0.0805, "lr": 1.1111111111111112e-05, "epoch": 0.05555555555555555, "percentage": 1.11, "elapsed_time": "0:02:08", "remaining_time": "3:10:42"}
+{"current_steps": 7, "total_steps": 540, "loss": 0.0782, "lr": 1.2962962962962962e-05, "epoch": 0.06481481481481481, "percentage": 1.3, "elapsed_time": "0:02:25", "remaining_time": "3:05:01"}
+{"current_steps": 8, "total_steps": 540, "loss": 0.0636, "lr": 1.4814814814814815e-05, "epoch": 0.07407407407407407, "percentage": 1.48, "elapsed_time": "0:02:41", "remaining_time": "2:58:35"}
+{"current_steps": 9, "total_steps": 540, "loss": 0.056, "lr": 1.6666666666666667e-05, "epoch": 0.08333333333333333, "percentage": 1.67, "elapsed_time": "0:02:58", "remaining_time": "2:55:11"}
+{"current_steps": 10, "total_steps": 540, "loss": 0.053, "lr": 1.8518518518518518e-05, "epoch": 0.09259259259259259, "percentage": 1.85, "elapsed_time": "0:03:13", "remaining_time": "2:50:59"}
+{"current_steps": 10, "total_steps": 540, "eval_loss": 0.04887561500072479, "epoch": 0.09259259259259259, "percentage": 1.85, "elapsed_time": "0:03:22", "remaining_time": "2:59:02"}
+{"current_steps": 11, "total_steps": 540, "loss": 0.0538, "lr": 2.037037037037037e-05, "epoch": 0.10185185185185185, "percentage": 2.04, "elapsed_time": "0:03:38", "remaining_time": "2:55:28"}
+{"current_steps": 12, "total_steps": 540, "loss": 0.0398, "lr": 2.2222222222222223e-05, "epoch": 0.1111111111111111, "percentage": 2.22, "elapsed_time": "0:03:54", "remaining_time": "2:51:46"}
+{"current_steps": 13, "total_steps": 540, "loss": 0.0473, "lr": 2.4074074074074074e-05, "epoch": 0.12037037037037036, "percentage": 2.41, "elapsed_time": "0:04:09", "remaining_time": "2:48:52"}
+{"current_steps": 14, "total_steps": 540, "loss": 0.0386, "lr": 2.5925925925925925e-05, "epoch": 0.12962962962962962, "percentage": 2.59, "elapsed_time": "0:04:26", "remaining_time": "2:46:38"}
+{"current_steps": 15, "total_steps": 540, "loss": 0.0417, "lr": 2.777777777777778e-05, "epoch": 0.1388888888888889, "percentage": 2.78, "elapsed_time": "0:04:42", "remaining_time": "2:44:51"}
+{"current_steps": 15, "total_steps": 540, "eval_loss": 0.04156189784407616, "epoch": 0.1388888888888889, "percentage": 2.78, "elapsed_time": "0:04:51", "remaining_time": "2:50:10"}
+{"current_steps": 16, "total_steps": 540, "loss": 0.0384, "lr": 2.962962962962963e-05, "epoch": 0.14814814814814814, "percentage": 2.96, "elapsed_time": "0:05:07", "remaining_time": "2:47:57"}
+{"current_steps": 17, "total_steps": 540, "loss": 0.0378, "lr": 3.148148148148148e-05, "epoch": 0.1574074074074074, "percentage": 3.15, "elapsed_time": "0:05:23", "remaining_time": "2:45:55"}
+{"current_steps": 18, "total_steps": 540, "loss": 0.0399, "lr": 3.3333333333333335e-05, "epoch": 0.16666666666666666, "percentage": 3.33, "elapsed_time": "0:05:40", "remaining_time": "2:44:44"}
+{"current_steps": 19, "total_steps": 540, "loss": 0.0368, "lr": 3.518518518518519e-05, "epoch": 0.17592592592592593, "percentage": 3.52, "elapsed_time": "0:05:57", "remaining_time": "2:43:14"}
+{"current_steps": 20, "total_steps": 540, "loss": 0.0396, "lr": 3.7037037037037037e-05, "epoch": 0.18518518518518517, "percentage": 3.7, "elapsed_time": "0:06:13", "remaining_time": "2:41:46"}
+{"current_steps": 20, "total_steps": 540, "eval_loss": 0.03584723547101021, "epoch": 0.18518518518518517, "percentage": 3.7, "elapsed_time": "0:06:22", "remaining_time": "2:45:46"}
+{"current_steps": 21, "total_steps": 540, "loss": 0.0364, "lr": 3.888888888888889e-05, "epoch": 0.19444444444444445, "percentage": 3.89, "elapsed_time": "0:06:39", "remaining_time": "2:44:29"}
+{"current_steps": 22, "total_steps": 540, "loss": 0.034, "lr": 4.074074074074074e-05, "epoch": 0.2037037037037037, "percentage": 4.07, "elapsed_time": "0:06:55", "remaining_time": "2:42:56"}
+{"current_steps": 23, "total_steps": 540, "loss": 0.0317, "lr": 4.259259259259259e-05, "epoch": 0.21296296296296297, "percentage": 4.26, "elapsed_time": "0:07:11", "remaining_time": "2:41:33"}
+{"current_steps": 24, "total_steps": 540, "loss": 0.0339, "lr": 4.4444444444444447e-05, "epoch": 0.2222222222222222, "percentage": 4.44, "elapsed_time": "0:07:26", "remaining_time": "2:40:05"}
+{"current_steps": 25, "total_steps": 540, "loss": 0.029, "lr": 4.62962962962963e-05, "epoch": 0.23148148148148148, "percentage": 4.63, "elapsed_time": "0:07:42", "remaining_time": "2:38:55"}
+{"current_steps": 25, "total_steps": 540, "eval_loss": 0.03286580368876457, "epoch": 0.23148148148148148, "percentage": 4.63, "elapsed_time": "0:07:52", "remaining_time": "2:42:03"}
+{"current_steps": 26, "total_steps": 540, "loss": 0.0326, "lr": 4.814814814814815e-05, "epoch": 0.24074074074074073, "percentage": 4.81, "elapsed_time": "0:08:08", "remaining_time": "2:40:51"}
+{"current_steps": 27, "total_steps": 540, "loss": 0.03, "lr": 5e-05, "epoch": 0.25, "percentage": 5.0, "elapsed_time": "0:08:24", "remaining_time": "2:39:45"}
+{"current_steps": 28, "total_steps": 540, "loss": 0.0328, "lr": 5.185185185185185e-05, "epoch": 0.25925925925925924, "percentage": 5.19, "elapsed_time": "0:08:41", "remaining_time": "2:38:54"}
+{"current_steps": 29, "total_steps": 540, "loss": 0.0294, "lr": 5.370370370370371e-05, "epoch": 0.26851851851851855, "percentage": 5.37, "elapsed_time": "0:08:56", "remaining_time": "2:37:36"}
+{"current_steps": 30, "total_steps": 540, "loss": 0.0311, "lr": 5.555555555555556e-05, "epoch": 0.2777777777777778, "percentage": 5.56, "elapsed_time": "0:09:14", "remaining_time": "2:37:01"}
+{"current_steps": 30, "total_steps": 540, "eval_loss": 0.030865700915455818, "epoch": 0.2777777777777778, "percentage": 5.56, "elapsed_time": "0:09:23", "remaining_time": "2:39:36"}
+{"current_steps": 31, "total_steps": 540, "loss": 0.0292, "lr": 5.740740740740741e-05, "epoch": 0.28703703703703703, "percentage": 5.74, "elapsed_time": "0:09:39", "remaining_time": "2:38:40"}
+{"current_steps": 32, "total_steps": 540, "loss": 0.0346, "lr": 5.925925925925926e-05, "epoch": 0.2962962962962963, "percentage": 5.93, "elapsed_time": "0:09:55", "remaining_time": "2:37:41"}
+{"current_steps": 33, "total_steps": 540, "loss": 0.0295, "lr": 6.111111111111112e-05, "epoch": 0.3055555555555556, "percentage": 6.11, "elapsed_time": "0:10:10", "remaining_time": "2:36:19"}
+{"current_steps": 34, "total_steps": 540, "loss": 0.0267, "lr": 6.296296296296296e-05, "epoch": 0.3148148148148148, "percentage": 6.3, "elapsed_time": "0:10:26", "remaining_time": "2:35:17"}
+{"current_steps": 35, "total_steps": 540, "loss": 0.0295, "lr": 6.481481481481482e-05, "epoch": 0.32407407407407407, "percentage": 6.48, "elapsed_time": "0:10:42", "remaining_time": "2:34:28"}
+{"current_steps": 35, "total_steps": 540, "eval_loss": 0.028042705729603767, "epoch": 0.32407407407407407, "percentage": 6.48, "elapsed_time": "0:10:51", "remaining_time": "2:36:40"}
+{"current_steps": 36, "total_steps": 540, "loss": 0.0319, "lr": 6.666666666666667e-05, "epoch": 0.3333333333333333, "percentage": 6.67, "elapsed_time": "0:11:08", "remaining_time": "2:35:59"}
+{"current_steps": 37, "total_steps": 540, "loss": 0.0346, "lr": 6.851851851851852e-05, "epoch": 0.3425925925925926, "percentage": 6.85, "elapsed_time": "0:11:24", "remaining_time": "2:35:03"}
+{"current_steps": 38, "total_steps": 540, "loss": 0.031, "lr": 7.037037037037038e-05, "epoch": 0.35185185185185186, "percentage": 7.04, "elapsed_time": "0:11:41", "remaining_time": "2:34:22"}
+{"current_steps": 39, "total_steps": 540, "loss": 0.0335, "lr": 7.222222222222222e-05, "epoch": 0.3611111111111111, "percentage": 7.22, "elapsed_time": "0:11:56", "remaining_time": "2:33:30"}
+{"current_steps": 40, "total_steps": 540, "loss": 0.0262, "lr": 7.407407407407407e-05, "epoch": 0.37037037037037035, "percentage": 7.41, "elapsed_time": "0:12:12", "remaining_time": "2:32:31"}
+{"current_steps": 40, "total_steps": 540, "eval_loss": 0.026602942496538162, "epoch": 0.37037037037037035, "percentage": 7.41, "elapsed_time": "0:12:21", "remaining_time": "2:34:26"}
+{"current_steps": 41, "total_steps": 540, "loss": 0.0252, "lr": 7.592592592592593e-05, "epoch": 0.37962962962962965, "percentage": 7.59, "elapsed_time": "0:12:36", "remaining_time": "2:33:31"}
+{"current_steps": 42, "total_steps": 540, "loss": 0.0242, "lr": 7.777777777777778e-05, "epoch": 0.3888888888888889, "percentage": 7.78, "elapsed_time": "0:12:52", "remaining_time": "2:32:43"}
+{"current_steps": 43, "total_steps": 540, "loss": 0.025, "lr": 7.962962962962964e-05, "epoch": 0.39814814814814814, "percentage": 7.96, "elapsed_time": "0:13:08", "remaining_time": "2:31:51"}
+{"current_steps": 44, "total_steps": 540, "loss": 0.0212, "lr": 8.148148148148148e-05, "epoch": 0.4074074074074074, "percentage": 8.15, "elapsed_time": "0:13:23", "remaining_time": "2:31:02"}
+{"current_steps": 45, "total_steps": 540, "loss": 0.0272, "lr": 8.333333333333334e-05, "epoch": 0.4166666666666667, "percentage": 8.33, "elapsed_time": "0:13:39", "remaining_time": "2:30:18"}
+{"current_steps": 45, "total_steps": 540, "eval_loss": 0.02608887106180191, "epoch": 0.4166666666666667, "percentage": 8.33, "elapsed_time": "0:13:49", "remaining_time": "2:31:59"}
+{"current_steps": 46, "total_steps": 540, "loss": 0.0274, "lr": 8.518518518518518e-05, "epoch": 0.42592592592592593, "percentage": 8.52, "elapsed_time": "0:14:05", "remaining_time": "2:31:18"}
+{"current_steps": 47, "total_steps": 540, "loss": 0.0283, "lr": 8.703703703703704e-05, "epoch": 0.4351851851851852, "percentage": 8.7, "elapsed_time": "0:14:22", "remaining_time": "2:30:45"}
+{"current_steps": 48, "total_steps": 540, "loss": 0.0242, "lr": 8.888888888888889e-05, "epoch": 0.4444444444444444, "percentage": 8.89, "elapsed_time": "0:14:39", "remaining_time": "2:30:14"}
+{"current_steps": 49, "total_steps": 540, "loss": 0.0239, "lr": 9.074074074074075e-05, "epoch": 0.4537037037037037, "percentage": 9.07, "elapsed_time": "0:14:56", "remaining_time": "2:29:44"}
+{"current_steps": 50, "total_steps": 540, "loss": 0.0224, "lr": 9.25925925925926e-05, "epoch": 0.46296296296296297, "percentage": 9.26, "elapsed_time": "0:15:13", "remaining_time": "2:29:15"}
+{"current_steps": 50, "total_steps": 540, "eval_loss": 0.02490057609975338, "epoch": 0.46296296296296297, "percentage": 9.26, "elapsed_time": "0:15:22", "remaining_time": "2:30:44"}
+{"current_steps": 51, "total_steps": 540, "loss": 0.0252, "lr": 9.444444444444444e-05, "epoch": 0.4722222222222222, "percentage": 9.44, "elapsed_time": "0:15:45", "remaining_time": "2:31:07"}
+{"current_steps": 52, "total_steps": 540, "loss": 0.0228, "lr": 9.62962962962963e-05, "epoch": 0.48148148148148145, "percentage": 9.63, "elapsed_time": "0:16:01", "remaining_time": "2:30:27"}
+{"current_steps": 53, "total_steps": 540, "loss": 0.0259, "lr": 9.814814814814815e-05, "epoch": 0.49074074074074076, "percentage": 9.81, "elapsed_time": "0:16:18", "remaining_time": "2:29:49"}
+{"current_steps": 54, "total_steps": 540, "loss": 0.0239, "lr": 0.0001, "epoch": 0.5, "percentage": 10.0, "elapsed_time": "0:16:33", "remaining_time": "2:28:58"}
+{"current_steps": 55, "total_steps": 540, "loss": 0.0229, "lr": 9.99989553622803e-05, "epoch": 0.5092592592592593, "percentage": 10.19, "elapsed_time": "0:16:49", "remaining_time": "2:28:24"}
+{"current_steps": 55, "total_steps": 540, "eval_loss": 0.02450372651219368, "epoch": 0.5092592592592593, "percentage": 10.19, "elapsed_time": "0:16:58", "remaining_time": "2:29:45"}
+{"current_steps": 56, "total_steps": 540, "loss": 0.0219, "lr": 9.999582149277187e-05, "epoch": 0.5185185185185185, "percentage": 10.37, "elapsed_time": "0:17:15", "remaining_time": "2:29:11"}
+{"current_steps": 57, "total_steps": 540, "loss": 0.0248, "lr": 9.999059852242507e-05, "epoch": 0.5277777777777778, "percentage": 10.56, "elapsed_time": "0:17:32", "remaining_time": "2:28:38"}
+{"current_steps": 58, "total_steps": 540, "loss": 0.0194, "lr": 9.998328666948438e-05, "epoch": 0.5370370370370371, "percentage": 10.74, "elapsed_time": "0:17:48", "remaining_time": "2:27:59"}
+{"current_steps": 59, "total_steps": 540, "loss": 0.0251, "lr": 9.997388623947928e-05, "epoch": 0.5462962962962963, "percentage": 10.93, "elapsed_time": "0:18:04", "remaining_time": "2:27:22"}
+{"current_steps": 60, "total_steps": 540, "loss": 0.0233, "lr": 9.996239762521151e-05, "epoch": 0.5555555555555556, "percentage": 11.11, "elapsed_time": "0:18:20", "remaining_time": "2:26:42"}
+{"current_steps": 60, "total_steps": 540, "eval_loss": 0.023316912353038788, "epoch": 0.5555555555555556, "percentage": 11.11, "elapsed_time": "0:18:29", "remaining_time": "2:27:56"}
+{"current_steps": 61, "total_steps": 540, "loss": 0.0222, "lr": 9.994882130673868e-05, "epoch": 0.5648148148148148, "percentage": 11.3, "elapsed_time": "0:18:44", "remaining_time": "2:27:10"}
+{"current_steps": 62, "total_steps": 540, "loss": 0.0272, "lr": 9.993315785135416e-05, "epoch": 0.5740740740740741, "percentage": 11.48, "elapsed_time": "0:19:00", "remaining_time": "2:26:33"}
+{"current_steps": 63, "total_steps": 540, "loss": 0.0241, "lr": 9.991540791356342e-05, "epoch": 0.5833333333333334, "percentage": 11.67, "elapsed_time": "0:19:17", "remaining_time": "2:26:06"}
+{"current_steps": 64, "total_steps": 540, "loss": 0.0216, "lr": 9.989557223505661e-05, "epoch": 0.5925925925925926, "percentage": 11.85, "elapsed_time": "0:19:34", "remaining_time": "2:25:33"}
+{"current_steps": 65, "total_steps": 540, "loss": 0.0217, "lr": 9.987365164467767e-05, "epoch": 0.6018518518518519, "percentage": 12.04, "elapsed_time": "0:19:49", "remaining_time": "2:24:52"}
+{"current_steps": 65, "total_steps": 540, "eval_loss": 0.02255990356206894, "epoch": 0.6018518518518519, "percentage": 12.04, "elapsed_time": "0:19:58", "remaining_time": "2:25:59"}
+{"current_steps": 66, "total_steps": 540, "loss": 0.0213, "lr": 9.98496470583896e-05, "epoch": 0.6111111111111112, "percentage": 12.22, "elapsed_time": "0:20:16", "remaining_time": "2:25:33"}
+{"current_steps": 67, "total_steps": 540, "loss": 0.018, "lr": 9.982355947923629e-05, "epoch": 0.6203703703703703, "percentage": 12.41, "elapsed_time": "0:20:32", "remaining_time": "2:24:59"}
+{"current_steps": 68, "total_steps": 540, "loss": 0.0199, "lr": 9.979538999730047e-05, "epoch": 0.6296296296296297, "percentage": 12.59, "elapsed_time": "0:20:48", "remaining_time": "2:24:25"}
+{"current_steps": 69, "total_steps": 540, "loss": 0.0239, "lr": 9.976513978965829e-05, "epoch": 0.6388888888888888, "percentage": 12.78, "elapsed_time": "0:21:05", "remaining_time": "2:24:00"}
+{"current_steps": 70, "total_steps": 540, "loss": 0.0247, "lr": 9.973281012033007e-05, "epoch": 0.6481481481481481, "percentage": 12.96, "elapsed_time": "0:21:21", "remaining_time": "2:23:25"}
+{"current_steps": 70, "total_steps": 540, "eval_loss": 0.022898558527231216, "epoch": 0.6481481481481481, "percentage": 12.96, "elapsed_time": "0:21:30", "remaining_time": "2:24:26"}
+{"current_steps": 71, "total_steps": 540, "loss": 0.0197, "lr": 9.969840234022749e-05, "epoch": 0.6574074074074074, "percentage": 13.15, "elapsed_time": "0:21:47", "remaining_time": "2:23:57"}
+{"current_steps": 72, "total_steps": 540, "loss": 0.0207, "lr": 9.966191788709716e-05, "epoch": 0.6666666666666666, "percentage": 13.33, "elapsed_time": "0:22:02", "remaining_time": "2:23:15"}
+{"current_steps": 73, "total_steps": 540, "loss": 0.0214, "lr": 9.962335828546048e-05, "epoch": 0.6759259259259259, "percentage": 13.52, "elapsed_time": "0:22:18", "remaining_time": "2:22:41"}
+{"current_steps": 74, "total_steps": 540, "loss": 0.0205, "lr": 9.958272514655006e-05, "epoch": 0.6851851851851852, "percentage": 13.7, "elapsed_time": "0:22:33", "remaining_time": "2:22:05"}
+{"current_steps": 75, "total_steps": 540, "loss": 0.0193, "lr": 9.954002016824227e-05, "epoch": 0.6944444444444444, "percentage": 13.89, "elapsed_time": "0:22:50", "remaining_time": "2:21:38"}
+{"current_steps": 75, "total_steps": 540, "eval_loss": 0.02283317781984806, "epoch": 0.6944444444444444, "percentage": 13.89, "elapsed_time": "0:22:59", "remaining_time": "2:22:35"}
+{"current_steps": 76, "total_steps": 540, "loss": 0.0206, "lr": 9.949524513498636e-05, "epoch": 0.7037037037037037, "percentage": 14.07, "elapsed_time": "0:23:16", "remaining_time": "2:22:03"}
+{"current_steps": 77, "total_steps": 540, "loss": 0.0217, "lr": 9.944840191772987e-05, "epoch": 0.7129629629629629, "percentage": 14.26, "elapsed_time": "0:23:32", "remaining_time": "2:21:32"}
+{"current_steps": 78, "total_steps": 540, "loss": 0.0196, "lr": 9.939949247384046e-05, "epoch": 0.7222222222222222, "percentage": 14.44, "elapsed_time": "0:23:47", "remaining_time": "2:20:57"}
+{"current_steps": 79, "total_steps": 540, "loss": 0.0223, "lr": 9.934851884702414e-05, "epoch": 0.7314814814814815, "percentage": 14.63, "elapsed_time": "0:24:04", "remaining_time": "2:20:31"}
+{"current_steps": 80, "total_steps": 540, "loss": 0.0173, "lr": 9.929548316723982e-05, "epoch": 0.7407407407407407, "percentage": 14.81, "elapsed_time": "0:24:20", "remaining_time": "2:19:58"}
+{"current_steps": 80, "total_steps": 540, "eval_loss": 0.021335698664188385, "epoch": 0.7407407407407407, "percentage": 14.81, "elapsed_time": "0:24:29", "remaining_time": "2:20:51"}
+{"current_steps": 81, "total_steps": 540, "loss": 0.0231, "lr": 9.924038765061042e-05, "epoch": 0.75, "percentage": 15.0, "elapsed_time": "0:24:46", "remaining_time": "2:20:24"}
+{"current_steps": 82, "total_steps": 540, "loss": 0.0224, "lr": 9.918323459933005e-05, "epoch": 0.7592592592592593, "percentage": 15.19, "elapsed_time": "0:25:04", "remaining_time": "2:20:00"}
+{"current_steps": 83, "total_steps": 540, "loss": 0.0223, "lr": 9.912402640156811e-05, "epoch": 0.7685185185185185, "percentage": 15.37, "elapsed_time": "0:25:18", "remaining_time": "2:19:23"}
+{"current_steps": 84, "total_steps": 540, "loss": 0.0198, "lr": 9.906276553136923e-05, "epoch": 0.7777777777777778, "percentage": 15.56, "elapsed_time": "0:25:34", "remaining_time": "2:18:48"}
+{"current_steps": 85, "total_steps": 540, "loss": 0.0207, "lr": 9.899945454855006e-05, "epoch": 0.7870370370370371, "percentage": 15.74, "elapsed_time": "0:25:50", "remaining_time": "2:18:20"}
+{"current_steps": 85, "total_steps": 540, "eval_loss": 0.020375357940793037, "epoch": 0.7870370370370371, "percentage": 15.74, "elapsed_time": "0:25:59", "remaining_time": "2:19:09"}
+{"current_steps": 86, "total_steps": 540, "loss": 0.0197, "lr": 9.893409609859222e-05, "epoch": 0.7962962962962963, "percentage": 15.93, "elapsed_time": "0:26:16", "remaining_time": "2:18:40"}
+{"current_steps": 87, "total_steps": 540, "loss": 0.0199, "lr": 9.88666929125318e-05, "epoch": 0.8055555555555556, "percentage": 16.11, "elapsed_time": "0:26:33", "remaining_time": "2:18:15"}
+{"current_steps": 88, "total_steps": 540, "loss": 0.0169, "lr": 9.879724780684519e-05, "epoch": 0.8148148148148148, "percentage": 16.3, "elapsed_time": "0:26:50", "remaining_time": "2:17:50"}
+{"current_steps": 89, "total_steps": 540, "loss": 0.0209, "lr": 9.872576368333151e-05, "epoch": 0.8240740740740741, "percentage": 16.48, "elapsed_time": "0:27:06", "remaining_time": "2:17:21"}
+{"current_steps": 90, "total_steps": 540, "loss": 0.0213, "lr": 9.865224352899119e-05, "epoch": 0.8333333333333334, "percentage": 16.67, "elapsed_time": "0:27:22", "remaining_time": "2:16:52"}
+{"current_steps": 90, "total_steps": 540, "eval_loss": 0.019939038902521133, "epoch": 0.8333333333333334, "percentage": 16.67, "elapsed_time": "0:27:31", "remaining_time": "2:17:38"}
+{"current_steps": 91, "total_steps": 540, "loss": 0.021, "lr": 9.857669041590134e-05, "epoch": 0.8425925925925926, "percentage": 16.85, "elapsed_time": "0:27:48", "remaining_time": "2:17:12"}
+{"current_steps": 92, "total_steps": 540, "loss": 0.0207, "lr": 9.849910750108717e-05, "epoch": 0.8518518518518519, "percentage": 17.04, "elapsed_time": "0:28:04", "remaining_time": "2:16:42"}
+{"current_steps": 93, "total_steps": 540, "loss": 0.0196, "lr": 9.84194980263903e-05, "epoch": 0.8611111111111112, "percentage": 17.22, "elapsed_time": "0:28:21", "remaining_time": "2:16:16"}
+{"current_steps": 94, "total_steps": 540, "loss": 0.0178, "lr": 9.83378653183331e-05, "epoch": 0.8703703703703703, "percentage": 17.41, "elapsed_time": "0:28:38", "remaining_time": "2:15:51"}
+{"current_steps": 95, "total_steps": 540, "loss": 0.0199, "lr": 9.825421278797983e-05, "epoch": 0.8796296296296297, "percentage": 17.59, "elapsed_time": "0:28:54", "remaining_time": "2:15:24"}
+{"current_steps": 95, "total_steps": 540, "eval_loss": 0.020193172618746758, "epoch": 0.8796296296296297, "percentage": 17.59, "elapsed_time": "0:29:03", "remaining_time": "2:16:07"}
+{"current_steps": 96, "total_steps": 540, "loss": 0.0219, "lr": 9.816854393079403e-05, "epoch": 0.8888888888888888, "percentage": 17.78, "elapsed_time": "0:29:19", "remaining_time": "2:15:36"}
+{"current_steps": 97, "total_steps": 540, "loss": 0.0185, "lr": 9.808086232649246e-05, "epoch": 0.8981481481481481, "percentage": 17.96, "elapsed_time": "0:29:34", "remaining_time": "2:15:01"}
+{"current_steps": 98, "total_steps": 540, "loss": 0.021, "lr": 9.799117163889559e-05, "epoch": 0.9074074074074074, "percentage": 18.15, "elapsed_time": "0:29:50", "remaining_time": "2:14:36"}
+{"current_steps": 99, "total_steps": 540, "loss": 0.02, "lr": 9.789947561577445e-05, "epoch": 0.9166666666666666, "percentage": 18.33, "elapsed_time": "0:30:06", "remaining_time": "2:14:08"}
+{"current_steps": 100, "total_steps": 540, "loss": 0.0188, "lr": 9.780577808869398e-05, "epoch": 0.9259259259259259, "percentage": 18.52, "elapsed_time": "0:30:22", "remaining_time": "2:13:40"}
+{"current_steps": 100, "total_steps": 540, "eval_loss": 0.02070247381925583, "epoch": 0.9259259259259259, "percentage": 18.52, "elapsed_time": "0:30:32", "remaining_time": "2:14:21"}
+{"current_steps": 101, "total_steps": 540, "loss": 0.0218, "lr": 9.771008297285307e-05, "epoch": 0.9351851851851852, "percentage": 18.7, "elapsed_time": "0:30:55", "remaining_time": "2:14:24"}
+{"current_steps": 102, "total_steps": 540, "loss": 0.0202, "lr": 9.761239426692077e-05, "epoch": 0.9444444444444444, "percentage": 18.89, "elapsed_time": "0:31:12", "remaining_time": "2:14:00"}
+{"current_steps": 103, "total_steps": 540, "loss": 0.0197, "lr": 9.751271605286941e-05, "epoch": 0.9537037037037037, "percentage": 19.07, "elapsed_time": "0:31:29", "remaining_time": "2:13:37"}
+{"current_steps": 104, "total_steps": 540, "loss": 0.02, "lr": 9.741105249580383e-05, "epoch": 0.9629629629629629, "percentage": 19.26, "elapsed_time": "0:31:46", "remaining_time": "2:13:12"}
+{"current_steps": 105, "total_steps": 540, "loss": 0.0193, "lr": 9.730740784378753e-05, "epoch": 0.9722222222222222, "percentage": 19.44, "elapsed_time": "0:32:02", "remaining_time": "2:12:45"}
+{"current_steps": 105, "total_steps": 540, "eval_loss": 0.020300446078181267, "epoch": 0.9722222222222222, "percentage": 19.44, "elapsed_time": "0:32:11", "remaining_time": "2:13:23"}
+{"current_steps": 106, "total_steps": 540, "loss": 0.0202, "lr": 9.7201786427665e-05, "epoch": 0.9814814814814815, "percentage": 19.63, "elapsed_time": "0:32:29", "remaining_time": "2:13:01"}
+{"current_steps": 107, "total_steps": 540, "loss": 0.0188, "lr": 9.709419266088086e-05, "epoch": 0.9907407407407407, "percentage": 19.81, "elapsed_time": "0:32:46", "remaining_time": "2:12:35"}
+{"current_steps": 108, "total_steps": 540, "loss": 0.022, "lr": 9.698463103929542e-05, "epoch": 1.0, "percentage": 20.0, "elapsed_time": "0:33:02", "remaining_time": "2:12:10"}
+{"current_steps": 109, "total_steps": 540, "loss": 0.0159, "lr": 9.687310614099675e-05, "epoch": 1.0092592592592593, "percentage": 20.19, "elapsed_time": "0:33:18", "remaining_time": "2:11:43"}
+{"current_steps": 110, "total_steps": 540, "loss": 0.016, "lr": 9.67596226261095e-05, "epoch": 1.0185185185185186, "percentage": 20.37, "elapsed_time": "0:33:34", "remaining_time": "2:11:15"}
+{"current_steps": 110, "total_steps": 540, "eval_loss": 0.01969613879919052, "epoch": 1.0185185185185186, "percentage": 20.37, "elapsed_time": "0:33:43", "remaining_time": "2:11:50"}
+{"current_steps": 111, "total_steps": 540, "loss": 0.014, "lr": 9.664418523660004e-05, "epoch": 1.0277777777777777, "percentage": 20.56, "elapsed_time": "0:33:58", "remaining_time": "2:11:18"}
+{"current_steps": 112, "total_steps": 540, "loss": 0.0172, "lr": 9.652679879607843e-05, "epoch": 1.037037037037037, "percentage": 20.74, "elapsed_time": "0:34:13", "remaining_time": "2:10:48"}
+{"current_steps": 113, "total_steps": 540, "loss": 0.014, "lr": 9.640746820959684e-05, "epoch": 1.0462962962962963, "percentage": 20.93, "elapsed_time": "0:34:29", "remaining_time": "2:10:21"}
+{"current_steps": 114, "total_steps": 540, "loss": 0.0172, "lr": 9.628619846344454e-05, "epoch": 1.0555555555555556, "percentage": 21.11, "elapsed_time": "0:34:46", "remaining_time": "2:09:57"}
+{"current_steps": 115, "total_steps": 540, "loss": 0.0166, "lr": 9.616299462493952e-05, "epoch": 1.0648148148148149, "percentage": 21.3, "elapsed_time": "0:35:03", "remaining_time": "2:09:32"}
+{"current_steps": 115, "total_steps": 540, "eval_loss": 0.019864549860358238, "epoch": 1.0648148148148149, "percentage": 21.3, "elapsed_time": "0:35:12", "remaining_time": "2:10:05"}
+{"current_steps": 116, "total_steps": 540, "loss": 0.0195, "lr": 9.603786184221693e-05, "epoch": 1.074074074074074, "percentage": 21.48, "elapsed_time": "0:35:28", "remaining_time": "2:09:40"}
+{"current_steps": 117, "total_steps": 540, "loss": 0.015, "lr": 9.591080534401371e-05, "epoch": 1.0833333333333333, "percentage": 21.67, "elapsed_time": "0:35:45", "remaining_time": "2:09:15"}
+{"current_steps": 118, "total_steps": 540, "loss": 0.0183, "lr": 9.57818304394503e-05, "epoch": 1.0925925925925926, "percentage": 21.85, "elapsed_time": "0:36:01", "remaining_time": "2:08:50"}
+{"current_steps": 119, "total_steps": 540, "loss": 0.0172, "lr": 9.565094251780871e-05, "epoch": 1.1018518518518519, "percentage": 22.04, "elapsed_time": "0:36:17", "remaining_time": "2:08:22"}
+{"current_steps": 120, "total_steps": 540, "loss": 0.0189, "lr": 9.551814704830734e-05, "epoch": 1.1111111111111112, "percentage": 22.22, "elapsed_time": "0:36:31", "remaining_time": "2:07:51"}
+{"current_steps": 120, "total_steps": 540, "eval_loss": 0.019504941999912262, "epoch": 1.1111111111111112, "percentage": 22.22, "elapsed_time": "0:36:41", "remaining_time": "2:08:24"}
+{"current_steps": 121, "total_steps": 540, "loss": 0.0132, "lr": 9.538344957987244e-05, "epoch": 1.1203703703703705, "percentage": 22.41, "elapsed_time": "0:36:57", "remaining_time": "2:07:59"}
+{"current_steps": 122, "total_steps": 540, "loss": 0.0184, "lr": 9.524685574090627e-05, "epoch": 1.1296296296296295, "percentage": 22.59, "elapsed_time": "0:37:15", "remaining_time": "2:07:41"}
+{"current_steps": 123, "total_steps": 540, "loss": 0.0155, "lr": 9.51083712390519e-05, "epoch": 1.1388888888888888, "percentage": 22.78, "elapsed_time": "0:37:31", "remaining_time": "2:07:13"}
+{"current_steps": 124, "total_steps": 540, "loss": 0.0156, "lr": 9.496800186095466e-05, "epoch": 1.1481481481481481, "percentage": 22.96, "elapsed_time": "0:37:48", "remaining_time": "2:06:50"}
+{"current_steps": 125, "total_steps": 540, "loss": 0.0211, "lr": 9.482575347202047e-05, "epoch": 1.1574074074074074, "percentage": 23.15, "elapsed_time": "0:38:04", "remaining_time": "2:06:25"}
+{"current_steps": 125, "total_steps": 540, "eval_loss": 0.018362991511821747, "epoch": 1.1574074074074074, "percentage": 23.15, "elapsed_time": "0:38:13", "remaining_time": "2:06:55"}
+{"current_steps": 126, "total_steps": 540, "loss": 0.0178, "lr": 9.468163201617062e-05, "epoch": 1.1666666666666667, "percentage": 23.33, "elapsed_time": "0:38:31", "remaining_time": "2:06:35"}
+{"current_steps": 127, "total_steps": 540, "loss": 0.0148, "lr": 9.453564351559348e-05, "epoch": 1.175925925925926, "percentage": 23.52, "elapsed_time": "0:38:45", "remaining_time": "2:06:02"}
+{"current_steps": 128, "total_steps": 540, "loss": 0.0174, "lr": 9.438779407049281e-05, "epoch": 1.1851851851851851, "percentage": 23.7, "elapsed_time": "0:39:01", "remaining_time": "2:05:36"}
+{"current_steps": 129, "total_steps": 540, "loss": 0.0174, "lr": 9.423808985883289e-05, "epoch": 1.1944444444444444, "percentage": 23.89, "elapsed_time": "0:39:16", "remaining_time": "2:05:09"}
+{"current_steps": 130, "total_steps": 540, "loss": 0.0171, "lr": 9.40865371360804e-05, "epoch": 1.2037037037037037, "percentage": 24.07, "elapsed_time": "0:39:32", "remaining_time": "2:04:43"}
+{"current_steps": 130, "total_steps": 540, "eval_loss": 0.018851976841688156, "epoch": 1.2037037037037037, "percentage": 24.07, "elapsed_time": "0:39:42", "remaining_time": "2:05:12"}
+{"current_steps": 131, "total_steps": 540, "loss": 0.0172, "lr": 9.393314223494296e-05, "epoch": 1.212962962962963, "percentage": 24.26, "elapsed_time": "0:39:58", "remaining_time": "2:04:49"}
+{"current_steps": 132, "total_steps": 540, "loss": 0.016, "lr": 9.377791156510455e-05, "epoch": 1.2222222222222223, "percentage": 24.44, "elapsed_time": "0:40:14", "remaining_time": "2:04:24"}
+{"current_steps": 133, "total_steps": 540, "loss": 0.0163, "lr": 9.362085161295769e-05, "epoch": 1.2314814814814814, "percentage": 24.63, "elapsed_time": "0:40:31", "remaining_time": "2:04:00"}
+{"current_steps": 134, "total_steps": 540, "loss": 0.0165, "lr": 9.346196894133239e-05, "epoch": 1.2407407407407407, "percentage": 24.81, "elapsed_time": "0:40:48", "remaining_time": "2:03:37"}
+{"current_steps": 135, "total_steps": 540, "loss": 0.0191, "lr": 9.330127018922194e-05, "epoch": 1.25, "percentage": 25.0, "elapsed_time": "0:41:05", "remaining_time": "2:03:15"}
+{"current_steps": 135, "total_steps": 540, "eval_loss": 0.019194327294826508, "epoch": 1.25, "percentage": 25.0, "elapsed_time": "0:41:14", "remaining_time": "2:03:42"}
+{"current_steps": 136, "total_steps": 540, "loss": 0.0165, "lr": 9.313876207150543e-05, "epoch": 1.2592592592592593, "percentage": 25.19, "elapsed_time": "0:41:31", "remaining_time": "2:03:19"}
+{"current_steps": 137, "total_steps": 540, "loss": 0.0162, "lr": 9.297445137866727e-05, "epoch": 1.2685185185185186, "percentage": 25.37, "elapsed_time": "0:41:46", "remaining_time": "2:02:53"}
+{"current_steps": 138, "total_steps": 540, "loss": 0.0192, "lr": 9.280834497651334e-05, "epoch": 1.2777777777777777, "percentage": 25.56, "elapsed_time": "0:42:03", "remaining_time": "2:02:30"}
+{"current_steps": 139, "total_steps": 540, "loss": 0.015, "lr": 9.264044980588416e-05, "epoch": 1.287037037037037, "percentage": 25.74, "elapsed_time": "0:42:19", "remaining_time": "2:02:06"}
+{"current_steps": 140, "total_steps": 540, "loss": 0.0184, "lr": 9.247077288236488e-05, "epoch": 1.2962962962962963, "percentage": 25.93, "elapsed_time": "0:42:34", "remaining_time": "2:01:39"}
+{"current_steps": 140, "total_steps": 540, "eval_loss": 0.018648317083716393, "epoch": 1.2962962962962963, "percentage": 25.93, "elapsed_time": "0:42:44", "remaining_time": "2:02:06"}
+{"current_steps": 141, "total_steps": 540, "loss": 0.0166, "lr": 9.229932129599205e-05, "epoch": 1.3055555555555556, "percentage": 26.11, "elapsed_time": "0:43:00", "remaining_time": "2:01:42"}
+{"current_steps": 142, "total_steps": 540, "loss": 0.0157, "lr": 9.212610221095748e-05, "epoch": 1.3148148148148149, "percentage": 26.3, "elapsed_time": "0:43:17", "remaining_time": "2:01:19"}
+{"current_steps": 143, "total_steps": 540, "loss": 0.0178, "lr": 9.195112286530873e-05, "epoch": 1.324074074074074, "percentage": 26.48, "elapsed_time": "0:43:34", "remaining_time": "2:00:59"}
+{"current_steps": 144, "total_steps": 540, "loss": 0.0164, "lr": 9.177439057064683e-05, "epoch": 1.3333333333333333, "percentage": 26.67, "elapsed_time": "0:43:50", "remaining_time": "2:00:34"}
+{"current_steps": 145, "total_steps": 540, "loss": 0.0162, "lr": 9.159591271182058e-05, "epoch": 1.3425925925925926, "percentage": 26.85, "elapsed_time": "0:44:07", "remaining_time": "2:00:11"}
+{"current_steps": 145, "total_steps": 540, "eval_loss": 0.018656810745596886, "epoch": 1.3425925925925926, "percentage": 26.85, "elapsed_time": "0:44:16", "remaining_time": "2:00:35"}
+{"current_steps": 146, "total_steps": 540, "loss": 0.021, "lr": 9.141569674661817e-05, "epoch": 1.3518518518518519, "percentage": 27.04, "elapsed_time": "0:44:32", "remaining_time": "2:00:10"}
+{"current_steps": 147, "total_steps": 540, "loss": 0.0162, "lr": 9.123375020545535e-05, "epoch": 1.3611111111111112, "percentage": 27.22, "elapsed_time": "0:44:48", "remaining_time": "1:59:47"}
+{"current_steps": 148, "total_steps": 540, "loss": 0.0169, "lr": 9.105008069106093e-05, "epoch": 1.3703703703703702, "percentage": 27.41, "elapsed_time": "0:45:04", "remaining_time": "1:59:21"}
+{"current_steps": 149, "total_steps": 540, "loss": 0.0162, "lr": 9.086469587815904e-05, "epoch": 1.3796296296296298, "percentage": 27.59, "elapsed_time": "0:45:21", "remaining_time": "1:59:00"}
+{"current_steps": 150, "total_steps": 540, "loss": 0.0165, "lr": 9.067760351314838e-05, "epoch": 1.3888888888888888, "percentage": 27.78, "elapsed_time": "0:45:37", "remaining_time": "1:58:37"}
+{"current_steps": 150, "total_steps": 540, "eval_loss": 0.018213987350463867, "epoch": 1.3888888888888888, "percentage": 27.78, "elapsed_time": "0:45:46", "remaining_time": "1:59:00"}
+{"current_steps": 151, "total_steps": 540, "loss": 0.0204, "lr": 9.048881141377863e-05, "epoch": 1.3981481481481481, "percentage": 27.96, "elapsed_time": "0:46:10", "remaining_time": "1:58:58"}
+{"current_steps": 152, "total_steps": 540, "loss": 0.0164, "lr": 9.029832746882371e-05, "epoch": 1.4074074074074074, "percentage": 28.15, "elapsed_time": "0:46:27", "remaining_time": "1:58:34"}
+{"current_steps": 153, "total_steps": 540, "loss": 0.018, "lr": 9.01061596377522e-05, "epoch": 1.4166666666666667, "percentage": 28.33, "elapsed_time": "0:46:43", "remaining_time": "1:58:10"}
+{"current_steps": 154, "total_steps": 540, "loss": 0.0156, "lr": 8.991231595039465e-05, "epoch": 1.425925925925926, "percentage": 28.52, "elapsed_time": "0:46:58", "remaining_time": "1:57:45"}
+{"current_steps": 155, "total_steps": 540, "loss": 0.0157, "lr": 8.97168045066082e-05, "epoch": 1.4351851851851851, "percentage": 28.7, "elapsed_time": "0:47:13", "remaining_time": "1:57:18"}
+{"current_steps": 155, "total_steps": 540, "eval_loss": 0.01855114847421646, "epoch": 1.4351851851851851, "percentage": 28.7, "elapsed_time": "0:47:22", "remaining_time": "1:57:40"}
+{"current_steps": 156, "total_steps": 540, "loss": 0.0165, "lr": 8.951963347593797e-05, "epoch": 1.4444444444444444, "percentage": 28.89, "elapsed_time": "0:47:40", "remaining_time": "1:57:20"}
+{"current_steps": 157, "total_steps": 540, "loss": 0.0201, "lr": 8.932081109727582e-05, "epoch": 1.4537037037037037, "percentage": 29.07, "elapsed_time": "0:47:57", "remaining_time": "1:56:59"}
+{"current_steps": 158, "total_steps": 540, "loss": 0.0182, "lr": 8.912034567851599e-05, "epoch": 1.462962962962963, "percentage": 29.26, "elapsed_time": "0:48:14", "remaining_time": "1:56:37"}
+{"current_steps": 159, "total_steps": 540, "loss": 0.0153, "lr": 8.891824559620801e-05, "epoch": 1.4722222222222223, "percentage": 29.44, "elapsed_time": "0:48:29", "remaining_time": "1:56:12"}
+{"current_steps": 160, "total_steps": 540, "loss": 0.0159, "lr": 8.871451929520663e-05, "epoch": 1.4814814814814814, "percentage": 29.63, "elapsed_time": "0:48:46", "remaining_time": "1:55:50"}
+{"current_steps": 160, "total_steps": 540, "eval_loss": 0.01888095587491989, "epoch": 1.4814814814814814, "percentage": 29.63, "elapsed_time": "0:48:55", "remaining_time": "1:56:12"}
+{"current_steps": 161, "total_steps": 540, "loss": 0.0163, "lr": 8.850917528831899e-05, "epoch": 1.4907407407407407, "percentage": 29.81, "elapsed_time": "0:49:13", "remaining_time": "1:55:52"}
+{"current_steps": 162, "total_steps": 540, "loss": 0.0125, "lr": 8.83022221559489e-05, "epoch": 1.5, "percentage": 30.0, "elapsed_time": "0:49:27", "remaining_time": "1:55:25"}
+{"current_steps": 163, "total_steps": 540, "loss": 0.0175, "lr": 8.809366854573831e-05, "epoch": 1.5092592592592593, "percentage": 30.19, "elapsed_time": "0:49:43", "remaining_time": "1:55:01"}
+{"current_steps": 164, "total_steps": 540, "loss": 0.0164, "lr": 8.78835231722059e-05, "epoch": 1.5185185185185186, "percentage": 30.37, "elapsed_time": "0:49:58", "remaining_time": "1:54:33"}
+{"current_steps": 165, "total_steps": 540, "loss": 0.0174, "lr": 8.767179481638303e-05, "epoch": 1.5277777777777777, "percentage": 30.56, "elapsed_time": "0:50:14", "remaining_time": "1:54:10"}
+{"current_steps": 165, "total_steps": 540, "eval_loss": 0.018690049648284912, "epoch": 1.5277777777777777, "percentage": 30.56, "elapsed_time": "0:50:23", "remaining_time": "1:54:31"}
+{"current_steps": 166, "total_steps": 540, "loss": 0.0179, "lr": 8.745849232544681e-05, "epoch": 1.5370370370370372, "percentage": 30.74, "elapsed_time": "0:50:39", "remaining_time": "1:54:09"}
+{"current_steps": 167, "total_steps": 540, "loss": 0.0169, "lr": 8.724362461235029e-05, "epoch": 1.5462962962962963, "percentage": 30.93, "elapsed_time": "0:50:55", "remaining_time": "1:53:44"}
+{"current_steps": 168, "total_steps": 540, "loss": 0.0168, "lr": 8.702720065545024e-05, "epoch": 1.5555555555555556, "percentage": 31.11, "elapsed_time": "0:51:12", "remaining_time": "1:53:23"}
+{"current_steps": 169, "total_steps": 540, "loss": 0.0162, "lr": 8.680922949813178e-05, "epoch": 1.5648148148148149, "percentage": 31.3, "elapsed_time": "0:51:29", "remaining_time": "1:53:01"}
+{"current_steps": 170, "total_steps": 540, "loss": 0.0184, "lr": 8.658972024843062e-05, "epoch": 1.574074074074074, "percentage": 31.48, "elapsed_time": "0:51:44", "remaining_time": "1:52:36"}
+{"current_steps": 170, "total_steps": 540, "eval_loss": 0.018272995948791504, "epoch": 1.574074074074074, "percentage": 31.48, "elapsed_time": "0:51:53", "remaining_time": "1:52:56"}
+{"current_steps": 171, "total_steps": 540, "loss": 0.0152, "lr": 8.636868207865244e-05, "epoch": 1.5833333333333335, "percentage": 31.67, "elapsed_time": "0:52:07", "remaining_time": "1:52:29"}
+{"current_steps": 172, "total_steps": 540, "loss": 0.0153, "lr": 8.614612422498964e-05, "epoch": 1.5925925925925926, "percentage": 31.85, "elapsed_time": "0:52:24", "remaining_time": "1:52:08"}
+{"current_steps": 173, "total_steps": 540, "loss": 0.017, "lr": 8.592205598713539e-05, "epoch": 1.6018518518518519, "percentage": 32.04, "elapsed_time": "0:52:41", "remaining_time": "1:51:47"}
+{"current_steps": 174, "total_steps": 540, "loss": 0.0158, "lr": 8.569648672789497e-05, "epoch": 1.6111111111111112, "percentage": 32.22, "elapsed_time": "0:52:55", "remaining_time": "1:51:20"}
+{"current_steps": 175, "total_steps": 540, "loss": 0.0165, "lr": 8.546942587279465e-05, "epoch": 1.6203703703703702, "percentage": 32.41, "elapsed_time": "0:53:12", "remaining_time": "1:50:58"}
+{"current_steps": 175, "total_steps": 540, "eval_loss": 0.018273252993822098, "epoch": 1.6203703703703702, "percentage": 32.41, "elapsed_time": "0:53:21", "remaining_time": "1:51:17"}
+{"current_steps": 176, "total_steps": 540, "loss": 0.0187, "lr": 8.524088290968781e-05, "epoch": 1.6296296296296298, "percentage": 32.59, "elapsed_time": "0:53:38", "remaining_time": "1:50:56"}
+{"current_steps": 177, "total_steps": 540, "loss": 0.0181, "lr": 8.501086738835843e-05, "epoch": 1.6388888888888888, "percentage": 32.78, "elapsed_time": "0:53:53", "remaining_time": "1:50:31"}
+{"current_steps": 178, "total_steps": 540, "loss": 0.0171, "lr": 8.47793889201221e-05, "epoch": 1.6481481481481481, "percentage": 32.96, "elapsed_time": "0:54:10", "remaining_time": "1:50:10"}
+{"current_steps": 179, "total_steps": 540, "loss": 0.021, "lr": 8.45464571774244e-05, "epoch": 1.6574074074074074, "percentage": 33.15, "elapsed_time": "0:54:27", "remaining_time": "1:49:50"}
+{"current_steps": 180, "total_steps": 540, "loss": 0.0173, "lr": 8.43120818934367e-05, "epoch": 1.6666666666666665, "percentage": 33.33, "elapsed_time": "0:54:43", "remaining_time": "1:49:27"}
+{"current_steps": 180, "total_steps": 540, "eval_loss": 0.01778573729097843, "epoch": 1.6666666666666665, "percentage": 33.33, "elapsed_time": "0:54:53", "remaining_time": "1:49:46"}
+{"current_steps": 181, "total_steps": 540, "loss": 0.015, "lr": 8.407627286164948e-05, "epoch": 1.675925925925926, "percentage": 33.52, "elapsed_time": "0:55:09", "remaining_time": "1:49:23"}
+{"current_steps": 182, "total_steps": 540, "loss": 0.0157, "lr": 8.383903993546311e-05, "epoch": 1.6851851851851851, "percentage": 33.7, "elapsed_time": "0:55:25", "remaining_time": "1:49:00"}
+{"current_steps": 183, "total_steps": 540, "loss": 0.0181, "lr": 8.360039302777612e-05, "epoch": 1.6944444444444444, "percentage": 33.89, "elapsed_time": "0:55:42", "remaining_time": "1:48:40"}
+{"current_steps": 184, "total_steps": 540, "loss": 0.0153, "lr": 8.336034211057098e-05, "epoch": 1.7037037037037037, "percentage": 34.07, "elapsed_time": "0:56:00", "remaining_time": "1:48:21"}
+{"current_steps": 185, "total_steps": 540, "loss": 0.0131, "lr": 8.31188972144974e-05, "epoch": 1.7129629629629628, "percentage": 34.26, "elapsed_time": "0:56:16", "remaining_time": "1:47:59"}
+{"current_steps": 185, "total_steps": 540, "eval_loss": 0.017187727615237236, "epoch": 1.7129629629629628, "percentage": 34.26, "elapsed_time": "0:56:25", "remaining_time": "1:48:16"}
+{"current_steps": 186, "total_steps": 540, "loss": 0.0162, "lr": 8.28760684284532e-05, "epoch": 1.7222222222222223, "percentage": 34.44, "elapsed_time": "0:56:41", "remaining_time": "1:47:53"}
+{"current_steps": 187, "total_steps": 540, "loss": 0.0137, "lr": 8.263186589916273e-05, "epoch": 1.7314814814814814, "percentage": 34.63, "elapsed_time": "0:56:58", "remaining_time": "1:47:33"}
+{"current_steps": 188, "total_steps": 540, "loss": 0.0143, "lr": 8.238629983075294e-05, "epoch": 1.7407407407407407, "percentage": 34.81, "elapsed_time": "0:57:14", "remaining_time": "1:47:10"}
+{"current_steps": 189, "total_steps": 540, "loss": 0.0144, "lr": 8.213938048432697e-05, "epoch": 1.75, "percentage": 35.0, "elapsed_time": "0:57:29", "remaining_time": "1:46:46"}
+{"current_steps": 190, "total_steps": 540, "loss": 0.0132, "lr": 8.18911181775353e-05, "epoch": 1.7592592592592593, "percentage": 35.19, "elapsed_time": "0:57:44", "remaining_time": "1:46:22"}
+{"current_steps": 190, "total_steps": 540, "eval_loss": 0.018012873828411102, "epoch": 1.7592592592592593, "percentage": 35.19, "elapsed_time": "0:57:54", "remaining_time": "1:46:39"}
+{"current_steps": 191, "total_steps": 540, "loss": 0.0156, "lr": 8.164152328414476e-05, "epoch": 1.7685185185185186, "percentage": 35.37, "elapsed_time": "0:58:10", "remaining_time": "1:46:18"}
+{"current_steps": 192, "total_steps": 540, "loss": 0.0121, "lr": 8.139060623360493e-05, "epoch": 1.7777777777777777, "percentage": 35.56, "elapsed_time": "0:58:25", "remaining_time": "1:45:54"}
+{"current_steps": 193, "total_steps": 540, "loss": 0.0156, "lr": 8.113837751061246e-05, "epoch": 1.7870370370370372, "percentage": 35.74, "elapsed_time": "0:58:41", "remaining_time": "1:45:31"}
+{"current_steps": 194, "total_steps": 540, "loss": 0.0202, "lr": 8.088484765467286e-05, "epoch": 1.7962962962962963, "percentage": 35.93, "elapsed_time": "0:58:58", "remaining_time": "1:45:11"}
+{"current_steps": 195, "total_steps": 540, "loss": 0.0157, "lr": 8.063002725966015e-05, "epoch": 1.8055555555555556, "percentage": 36.11, "elapsed_time": "0:59:13", "remaining_time": "1:44:47"}
+{"current_steps": 195, "total_steps": 540, "eval_loss": 0.018071575090289116, "epoch": 1.8055555555555556, "percentage": 36.11, "elapsed_time": "0:59:23", "remaining_time": "1:45:03"}
+{"current_steps": 196, "total_steps": 540, "loss": 0.0152, "lr": 8.037392697337418e-05, "epoch": 1.8148148148148149, "percentage": 36.3, "elapsed_time": "0:59:38", "remaining_time": "1:44:40"}
+{"current_steps": 197, "total_steps": 540, "loss": 0.0147, "lr": 8.011655749709575e-05, "epoch": 1.824074074074074, "percentage": 36.48, "elapsed_time": "0:59:55", "remaining_time": "1:44:20"}
+{"current_steps": 198, "total_steps": 540, "loss": 0.0142, "lr": 7.985792958513931e-05, "epoch": 1.8333333333333335, "percentage": 36.67, "elapsed_time": "1:00:11", "remaining_time": "1:43:57"}
+{"current_steps": 199, "total_steps": 540, "loss": 0.0181, "lr": 7.95980540444038e-05, "epoch": 1.8425925925925926, "percentage": 36.85, "elapsed_time": "1:00:26", "remaining_time": "1:43:33"}
+{"current_steps": 200, "total_steps": 540, "loss": 0.0154, "lr": 7.93369417339209e-05, "epoch": 1.8518518518518519, "percentage": 37.04, "elapsed_time": "1:00:42", "remaining_time": "1:43:12"}
+{"current_steps": 200, "total_steps": 540, "eval_loss": 0.01711750030517578, "epoch": 1.8518518518518519, "percentage": 37.04, "elapsed_time": "1:00:52", "remaining_time": "1:43:28"}
+{"current_steps": 201, "total_steps": 540, "loss": 0.0156, "lr": 7.907460356440133e-05, "epoch": 1.8611111111111112, "percentage": 37.22, "elapsed_time": "1:01:16", "remaining_time": "1:43:20"}
+{"current_steps": 202, "total_steps": 540, "loss": 0.0182, "lr": 7.881105049777901e-05, "epoch": 1.8703703703703702, "percentage": 37.41, "elapsed_time": "1:01:32", "remaining_time": "1:42:58"}
+{"current_steps": 203, "total_steps": 540, "loss": 0.0145, "lr": 7.854629354675291e-05, "epoch": 1.8796296296296298, "percentage": 37.59, "elapsed_time": "1:01:47", "remaining_time": "1:42:35"}
+{"current_steps": 204, "total_steps": 540, "loss": 0.0161, "lr": 7.828034377432693e-05, "epoch": 1.8888888888888888, "percentage": 37.78, "elapsed_time": "1:02:03", "remaining_time": "1:42:13"}
+{"current_steps": 205, "total_steps": 540, "loss": 0.0139, "lr": 7.801321229334764e-05, "epoch": 1.8981481481481481, "percentage": 37.96, "elapsed_time": "1:02:19", "remaining_time": "1:41:51"}
+{"current_steps": 205, "total_steps": 540, "eval_loss": 0.01687374897301197, "epoch": 1.8981481481481481, "percentage": 37.96, "elapsed_time": "1:02:28", "remaining_time": "1:42:06"}
+{"current_steps": 206, "total_steps": 540, "loss": 0.0172, "lr": 7.774491026603985e-05, "epoch": 1.9074074074074074, "percentage": 38.15, "elapsed_time": "1:02:46", "remaining_time": "1:41:46"}
+{"current_steps": 207, "total_steps": 540, "loss": 0.0182, "lr": 7.74754489035403e-05, "epoch": 1.9166666666666665, "percentage": 38.33, "elapsed_time": "1:03:02", "remaining_time": "1:41:24"}
+{"current_steps": 208, "total_steps": 540, "loss": 0.0176, "lr": 7.720483946542914e-05, "epoch": 1.925925925925926, "percentage": 38.52, "elapsed_time": "1:03:18", "remaining_time": "1:41:03"}
+{"current_steps": 209, "total_steps": 540, "loss": 0.0149, "lr": 7.69330932592594e-05, "epoch": 1.9351851851851851, "percentage": 38.7, "elapsed_time": "1:03:35", "remaining_time": "1:40:42"}
+{"current_steps": 210, "total_steps": 540, "loss": 0.0169, "lr": 7.666022164008457e-05, "epoch": 1.9444444444444444, "percentage": 38.89, "elapsed_time": "1:03:51", "remaining_time": "1:40:20"}
+{"current_steps": 210, "total_steps": 540, "eval_loss": 0.016974864527583122, "epoch": 1.9444444444444444, "percentage": 38.89, "elapsed_time": "1:04:00", "remaining_time": "1:40:34"}
+{"current_steps": 211, "total_steps": 540, "loss": 0.0148, "lr": 7.63862360099841e-05, "epoch": 1.9537037037037037, "percentage": 39.07, "elapsed_time": "1:04:16", "remaining_time": "1:40:13"}
+{"current_steps": 212, "total_steps": 540, "loss": 0.0158, "lr": 7.611114781758692e-05, "epoch": 1.9629629629629628, "percentage": 39.26, "elapsed_time": "1:04:33", "remaining_time": "1:39:53"}
+{"current_steps": 213, "total_steps": 540, "loss": 0.0172, "lr": 7.583496855759316e-05, "epoch": 1.9722222222222223, "percentage": 39.44, "elapsed_time": "1:04:50", "remaining_time": "1:39:32"}
+{"current_steps": 214, "total_steps": 540, "loss": 0.0149, "lr": 7.555770977029367e-05, "epoch": 1.9814814814814814, "percentage": 39.63, "elapsed_time": "1:05:07", "remaining_time": "1:39:12"}
+{"current_steps": 215, "total_steps": 540, "loss": 0.0158, "lr": 7.527938304108795e-05, "epoch": 1.9907407407407407, "percentage": 39.81, "elapsed_time": "1:05:24", "remaining_time": "1:38:52"}
+{"current_steps": 215, "total_steps": 540, "eval_loss": 0.017042405903339386, "epoch": 1.9907407407407407, "percentage": 39.81, "elapsed_time": "1:05:33", "remaining_time": "1:39:06"}
+{"current_steps": 216, "total_steps": 540, "loss": 0.0126, "lr": 7.500000000000001e-05, "epoch": 2.0, "percentage": 40.0, "elapsed_time": "1:05:48", "remaining_time": "1:38:42"}
+{"current_steps": 217, "total_steps": 540, "loss": 0.0142, "lr": 7.471957232119234e-05, "epoch": 2.009259259259259, "percentage": 40.19, "elapsed_time": "1:06:05", "remaining_time": "1:38:22"}
+{"current_steps": 218, "total_steps": 540, "loss": 0.0151, "lr": 7.443811172247821e-05, "epoch": 2.0185185185185186, "percentage": 40.37, "elapsed_time": "1:06:22", "remaining_time": "1:38:03"}
+{"current_steps": 219, "total_steps": 540, "loss": 0.0132, "lr": 7.415562996483192e-05, "epoch": 2.0277777777777777, "percentage": 40.56, "elapsed_time": "1:06:38", "remaining_time": "1:37:40"}
+{"current_steps": 220, "total_steps": 540, "loss": 0.0139, "lr": 7.387213885189746e-05, "epoch": 2.037037037037037, "percentage": 40.74, "elapsed_time": "1:06:55", "remaining_time": "1:37:21"}
+{"current_steps": 220, "total_steps": 540, "eval_loss": 0.01702064275741577, "epoch": 2.037037037037037, "percentage": 40.74, "elapsed_time": "1:07:05", "remaining_time": "1:37:34"}
+{"current_steps": 221, "total_steps": 540, "loss": 0.0152, "lr": 7.358765022949519e-05, "epoch": 2.0462962962962963, "percentage": 40.93, "elapsed_time": "1:07:20", "remaining_time": "1:37:12"}
+{"current_steps": 222, "total_steps": 540, "loss": 0.0136, "lr": 7.330217598512695e-05, "epoch": 2.0555555555555554, "percentage": 41.11, "elapsed_time": "1:07:38", "remaining_time": "1:36:52"}
+{"current_steps": 223, "total_steps": 540, "loss": 0.0134, "lr": 7.30157280474793e-05, "epoch": 2.064814814814815, "percentage": 41.3, "elapsed_time": "1:07:54", "remaining_time": "1:36:31"}
+{"current_steps": 224, "total_steps": 540, "loss": 0.0158, "lr": 7.272831838592503e-05, "epoch": 2.074074074074074, "percentage": 41.48, "elapsed_time": "1:08:10", "remaining_time": "1:36:10"}
+{"current_steps": 225, "total_steps": 540, "loss": 0.0146, "lr": 7.243995901002312e-05, "epoch": 2.0833333333333335, "percentage": 41.67, "elapsed_time": "1:08:27", "remaining_time": "1:35:51"}
+{"current_steps": 225, "total_steps": 540, "eval_loss": 0.017011733725667, "epoch": 2.0833333333333335, "percentage": 41.67, "elapsed_time": "1:08:37", "remaining_time": "1:36:03"}
+{"current_steps": 226, "total_steps": 540, "loss": 0.0149, "lr": 7.215066196901676e-05, "epoch": 2.0925925925925926, "percentage": 41.85, "elapsed_time": "1:08:52", "remaining_time": "1:35:42"}
+{"current_steps": 227, "total_steps": 540, "loss": 0.0105, "lr": 7.186043935133005e-05, "epoch": 2.1018518518518516, "percentage": 42.04, "elapsed_time": "1:09:09", "remaining_time": "1:35:21"}
+{"current_steps": 228, "total_steps": 540, "loss": 0.0144, "lr": 7.156930328406268e-05, "epoch": 2.111111111111111, "percentage": 42.22, "elapsed_time": "1:09:26", "remaining_time": "1:35:01"}
+{"current_steps": 229, "total_steps": 540, "loss": 0.0134, "lr": 7.127726593248337e-05, "epoch": 2.1203703703703702, "percentage": 42.41, "elapsed_time": "1:09:44", "remaining_time": "1:34:42"}
+{"current_steps": 230, "total_steps": 540, "loss": 0.0115, "lr": 7.098433949952146e-05, "epoch": 2.1296296296296298, "percentage": 42.59, "elapsed_time": "1:10:00", "remaining_time": "1:34:21"}
+{"current_steps": 230, "total_steps": 540, "eval_loss": 0.017404422163963318, "epoch": 2.1296296296296298, "percentage": 42.59, "elapsed_time": "1:10:09", "remaining_time": "1:34:33"}
+{"current_steps": 231, "total_steps": 540, "loss": 0.0135, "lr": 7.069053622525696e-05, "epoch": 2.138888888888889, "percentage": 42.78, "elapsed_time": "1:10:24", "remaining_time": "1:34:10"}
+{"current_steps": 232, "total_steps": 540, "loss": 0.013, "lr": 7.039586838640919e-05, "epoch": 2.148148148148148, "percentage": 42.96, "elapsed_time": "1:10:40", "remaining_time": "1:33:49"}
+{"current_steps": 233, "total_steps": 540, "loss": 0.0112, "lr": 7.01003482958237e-05, "epoch": 2.1574074074074074, "percentage": 43.15, "elapsed_time": "1:10:57", "remaining_time": "1:33:29"}
+{"current_steps": 234, "total_steps": 540, "loss": 0.0114, "lr": 6.980398830195785e-05, "epoch": 2.1666666666666665, "percentage": 43.33, "elapsed_time": "1:11:13", "remaining_time": "1:33:08"}
+{"current_steps": 235, "total_steps": 540, "loss": 0.0138, "lr": 6.950680078836474e-05, "epoch": 2.175925925925926, "percentage": 43.52, "elapsed_time": "1:11:29", "remaining_time": "1:32:47"}
+{"current_steps": 235, "total_steps": 540, "eval_loss": 0.016838619485497475, "epoch": 2.175925925925926, "percentage": 43.52, "elapsed_time": "1:11:38", "remaining_time": "1:32:59"}
+{"current_steps": 236, "total_steps": 540, "loss": 0.0156, "lr": 6.920879817317589e-05, "epoch": 2.185185185185185, "percentage": 43.7, "elapsed_time": "1:11:55", "remaining_time": "1:32:39"}
+{"current_steps": 237, "total_steps": 540, "loss": 0.0111, "lr": 6.890999290858214e-05, "epoch": 2.1944444444444446, "percentage": 43.89, "elapsed_time": "1:12:11", "remaining_time": "1:32:17"}
+{"current_steps": 238, "total_steps": 540, "loss": 0.0155, "lr": 6.861039748031351e-05, "epoch": 2.2037037037037037, "percentage": 44.07, "elapsed_time": "1:12:26", "remaining_time": "1:31:55"}
+{"current_steps": 239, "total_steps": 540, "loss": 0.0127, "lr": 6.83100244071174e-05, "epoch": 2.212962962962963, "percentage": 44.26, "elapsed_time": "1:12:43", "remaining_time": "1:31:35"}
+{"current_steps": 240, "total_steps": 540, "loss": 0.0138, "lr": 6.800888624023553e-05, "epoch": 2.2222222222222223, "percentage": 44.44, "elapsed_time": "1:12:59", "remaining_time": "1:31:14"}
+{"current_steps": 240, "total_steps": 540, "eval_loss": 0.017057882621884346, "epoch": 2.2222222222222223, "percentage": 44.44, "elapsed_time": "1:13:08", "remaining_time": "1:31:25"}
+{"current_steps": 241, "total_steps": 540, "loss": 0.0138, "lr": 6.770699556287939e-05, "epoch": 2.2314814814814814, "percentage": 44.63, "elapsed_time": "1:13:25", "remaining_time": "1:31:05"}
+{"current_steps": 242, "total_steps": 540, "loss": 0.0128, "lr": 6.740436498970452e-05, "epoch": 2.240740740740741, "percentage": 44.81, "elapsed_time": "1:13:41", "remaining_time": "1:30:44"}
+{"current_steps": 243, "total_steps": 540, "loss": 0.0142, "lr": 6.710100716628344e-05, "epoch": 2.25, "percentage": 45.0, "elapsed_time": "1:13:57", "remaining_time": "1:30:23"}
+{"current_steps": 244, "total_steps": 540, "loss": 0.0137, "lr": 6.679693476857711e-05, "epoch": 2.259259259259259, "percentage": 45.19, "elapsed_time": "1:14:13", "remaining_time": "1:30:02"}
+{"current_steps": 245, "total_steps": 540, "loss": 0.0134, "lr": 6.649216050240539e-05, "epoch": 2.2685185185185186, "percentage": 45.37, "elapsed_time": "1:14:29", "remaining_time": "1:29:41"}
+{"current_steps": 245, "total_steps": 540, "eval_loss": 0.016679909080266953, "epoch": 2.2685185185185186, "percentage": 45.37, "elapsed_time": "1:14:38", "remaining_time": "1:29:52"}
+{"current_steps": 246, "total_steps": 540, "loss": 0.0116, "lr": 6.618669710291606e-05, "epoch": 2.2777777777777777, "percentage": 45.56, "elapsed_time": "1:14:53", "remaining_time": "1:29:30"}
+{"current_steps": 247, "total_steps": 540, "loss": 0.014, "lr": 6.588055733405266e-05, "epoch": 2.287037037037037, "percentage": 45.74, "elapsed_time": "1:15:08", "remaining_time": "1:29:08"}
+{"current_steps": 248, "total_steps": 540, "loss": 0.0167, "lr": 6.557375398802123e-05, "epoch": 2.2962962962962963, "percentage": 45.93, "elapsed_time": "1:15:25", "remaining_time": "1:28:48"}
+{"current_steps": 249, "total_steps": 540, "loss": 0.013, "lr": 6.526629988475567e-05, "epoch": 2.3055555555555554, "percentage": 46.11, "elapsed_time": "1:15:41", "remaining_time": "1:28:28"}
+{"current_steps": 250, "total_steps": 540, "loss": 0.0167, "lr": 6.495820787138209e-05, "epoch": 2.314814814814815, "percentage": 46.3, "elapsed_time": "1:15:58", "remaining_time": "1:28:07"}
+{"current_steps": 250, "total_steps": 540, "eval_loss": 0.016377143561840057, "epoch": 2.314814814814815, "percentage": 46.3, "elapsed_time": "1:16:07", "remaining_time": "1:28:17"}
+{"current_steps": 251, "total_steps": 540, "loss": 0.0125, "lr": 6.464949082168204e-05, "epoch": 2.324074074074074, "percentage": 46.48, "elapsed_time": "1:16:30", "remaining_time": "1:28:05"}
+{"current_steps": 252, "total_steps": 540, "loss": 0.0121, "lr": 6.434016163555452e-05, "epoch": 2.3333333333333335, "percentage": 46.67, "elapsed_time": "1:16:45", "remaining_time": "1:27:43"}
+{"current_steps": 253, "total_steps": 540, "loss": 0.0159, "lr": 6.403023323847695e-05, "epoch": 2.3425925925925926, "percentage": 46.85, "elapsed_time": "1:17:01", "remaining_time": "1:27:22"}
+{"current_steps": 254, "total_steps": 540, "loss": 0.0137, "lr": 6.371971858096508e-05, "epoch": 2.351851851851852, "percentage": 47.04, "elapsed_time": "1:17:18", "remaining_time": "1:27:02"}
+{"current_steps": 255, "total_steps": 540, "loss": 0.0123, "lr": 6.340863063803188e-05, "epoch": 2.361111111111111, "percentage": 47.22, "elapsed_time": "1:17:34", "remaining_time": "1:26:42"}
+{"current_steps": 255, "total_steps": 540, "eval_loss": 0.016414109617471695, "epoch": 2.361111111111111, "percentage": 47.22, "elapsed_time": "1:17:43", "remaining_time": "1:26:52"}
+{"current_steps": 256, "total_steps": 540, "loss": 0.012, "lr": 6.30969824086453e-05, "epoch": 2.3703703703703702, "percentage": 47.41, "elapsed_time": "1:17:59", "remaining_time": "1:26:31"}
+{"current_steps": 257, "total_steps": 540, "loss": 0.0127, "lr": 6.27847869151852e-05, "epoch": 2.3796296296296298, "percentage": 47.59, "elapsed_time": "1:18:16", "remaining_time": "1:26:11"}
+{"current_steps": 258, "total_steps": 540, "loss": 0.0141, "lr": 6.247205720289907e-05, "epoch": 2.388888888888889, "percentage": 47.78, "elapsed_time": "1:18:32", "remaining_time": "1:25:51"}
+{"current_steps": 259, "total_steps": 540, "loss": 0.0135, "lr": 6.215880633935708e-05, "epoch": 2.398148148148148, "percentage": 47.96, "elapsed_time": "1:18:47", "remaining_time": "1:25:29"}
+{"current_steps": 260, "total_steps": 540, "loss": 0.0139, "lr": 6.184504741390596e-05, "epoch": 2.4074074074074074, "percentage": 48.15, "elapsed_time": "1:19:04", "remaining_time": "1:25:09"}
+{"current_steps": 260, "total_steps": 540, "eval_loss": 0.016279693692922592, "epoch": 2.4074074074074074, "percentage": 48.15, "elapsed_time": "1:19:13", "remaining_time": "1:25:19"}
+{"current_steps": 261, "total_steps": 540, "loss": 0.0129, "lr": 6.153079353712201e-05, "epoch": 2.4166666666666665, "percentage": 48.33, "elapsed_time": "1:19:30", "remaining_time": "1:24:59"}
+{"current_steps": 262, "total_steps": 540, "loss": 0.0114, "lr": 6.121605784026339e-05, "epoch": 2.425925925925926, "percentage": 48.52, "elapsed_time": "1:19:45", "remaining_time": "1:24:37"}
+{"current_steps": 263, "total_steps": 540, "loss": 0.0138, "lr": 6.09008534747213e-05, "epoch": 2.435185185185185, "percentage": 48.7, "elapsed_time": "1:20:02", "remaining_time": "1:24:18"}
+{"current_steps": 264, "total_steps": 540, "loss": 0.0118, "lr": 6.058519361147055e-05, "epoch": 2.4444444444444446, "percentage": 48.89, "elapsed_time": "1:20:18", "remaining_time": "1:23:57"}
+{"current_steps": 265, "total_steps": 540, "loss": 0.0125, "lr": 6.02690914405191e-05, "epoch": 2.4537037037037037, "percentage": 49.07, "elapsed_time": "1:20:34", "remaining_time": "1:23:37"}
+{"current_steps": 265, "total_steps": 540, "eval_loss": 0.016143780201673508, "epoch": 2.4537037037037037, "percentage": 49.07, "elapsed_time": "1:20:43", "remaining_time": "1:23:46"}
+{"current_steps": 266, "total_steps": 540, "loss": 0.0139, "lr": 5.995256017035703e-05, "epoch": 2.462962962962963, "percentage": 49.26, "elapsed_time": "1:20:59", "remaining_time": "1:23:26"}
+{"current_steps": 267, "total_steps": 540, "loss": 0.0162, "lr": 5.963561302740449e-05, "epoch": 2.4722222222222223, "percentage": 49.44, "elapsed_time": "1:21:16", "remaining_time": "1:23:05"}
+{"current_steps": 268, "total_steps": 540, "loss": 0.0123, "lr": 5.9318263255459116e-05, "epoch": 2.4814814814814814, "percentage": 49.63, "elapsed_time": "1:21:32", "remaining_time": "1:22:45"}
+{"current_steps": 269, "total_steps": 540, "loss": 0.015, "lr": 5.900052411514257e-05, "epoch": 2.490740740740741, "percentage": 49.81, "elapsed_time": "1:21:49", "remaining_time": "1:22:26"}
+{"current_steps": 270, "total_steps": 540, "loss": 0.0126, "lr": 5.868240888334653e-05, "epoch": 2.5, "percentage": 50.0, "elapsed_time": "1:22:05", "remaining_time": "1:22:05"}
+{"current_steps": 270, "total_steps": 540, "eval_loss": 0.016046511009335518, "epoch": 2.5, "percentage": 50.0, "elapsed_time": "1:22:14", "remaining_time": "1:22:14"}
+{"current_steps": 271, "total_steps": 540, "loss": 0.0133, "lr": 5.836393085267776e-05, "epoch": 2.5092592592592595, "percentage": 50.19, "elapsed_time": "1:22:31", "remaining_time": "1:21:55"}
+{"current_steps": 272, "total_steps": 540, "loss": 0.0175, "lr": 5.804510333090287e-05, "epoch": 2.5185185185185186, "percentage": 50.37, "elapsed_time": "1:22:49", "remaining_time": "1:21:35"}
+{"current_steps": 273, "total_steps": 540, "loss": 0.0116, "lr": 5.772593964039203e-05, "epoch": 2.5277777777777777, "percentage": 50.56, "elapsed_time": "1:23:03", "remaining_time": "1:21:13"}
+{"current_steps": 274, "total_steps": 540, "loss": 0.0125, "lr": 5.740645311756245e-05, "epoch": 2.537037037037037, "percentage": 50.74, "elapsed_time": "1:23:19", "remaining_time": "1:20:53"}
+{"current_steps": 275, "total_steps": 540, "loss": 0.0138, "lr": 5.708665711232103e-05, "epoch": 2.5462962962962963, "percentage": 50.93, "elapsed_time": "1:23:35", "remaining_time": "1:20:32"}
+{"current_steps": 275, "total_steps": 540, "eval_loss": 0.016013609245419502, "epoch": 2.5462962962962963, "percentage": 50.93, "elapsed_time": "1:23:44", "remaining_time": "1:20:41"}
+{"current_steps": 276, "total_steps": 540, "loss": 0.0136, "lr": 5.6766564987506566e-05, "epoch": 2.5555555555555554, "percentage": 51.11, "elapsed_time": "1:24:01", "remaining_time": "1:20:22"}
+{"current_steps": 277, "total_steps": 540, "loss": 0.0131, "lr": 5.644619011833133e-05, "epoch": 2.564814814814815, "percentage": 51.3, "elapsed_time": "1:24:17", "remaining_time": "1:20:02"}
+{"current_steps": 278, "total_steps": 540, "loss": 0.0143, "lr": 5.6125545891822274e-05, "epoch": 2.574074074074074, "percentage": 51.48, "elapsed_time": "1:24:34", "remaining_time": "1:19:42"}
+{"current_steps": 279, "total_steps": 540, "loss": 0.0148, "lr": 5.5804645706261514e-05, "epoch": 2.5833333333333335, "percentage": 51.67, "elapsed_time": "1:24:50", "remaining_time": "1:19:21"}
+{"current_steps": 280, "total_steps": 540, "loss": 0.0125, "lr": 5.548350297062659e-05, "epoch": 2.5925925925925926, "percentage": 51.85, "elapsed_time": "1:25:06", "remaining_time": "1:19:01"}
+{"current_steps": 280, "total_steps": 540, "eval_loss": 0.015153205953538418, "epoch": 2.5925925925925926, "percentage": 51.85, "elapsed_time": "1:25:15", "remaining_time": "1:19:10"}
+{"current_steps": 281, "total_steps": 540, "loss": 0.0093, "lr": 5.516213110403009e-05, "epoch": 2.601851851851852, "percentage": 52.04, "elapsed_time": "1:25:30", "remaining_time": "1:18:48"}
+{"current_steps": 282, "total_steps": 540, "loss": 0.0138, "lr": 5.484054353515896e-05, "epoch": 2.611111111111111, "percentage": 52.22, "elapsed_time": "1:25:47", "remaining_time": "1:18:29"}
+{"current_steps": 283, "total_steps": 540, "loss": 0.0121, "lr": 5.451875370171341e-05, "epoch": 2.6203703703703702, "percentage": 52.41, "elapsed_time": "1:26:04", "remaining_time": "1:18:10"}
+{"current_steps": 284, "total_steps": 540, "loss": 0.0126, "lr": 5.419677504984534e-05, "epoch": 2.6296296296296298, "percentage": 52.59, "elapsed_time": "1:26:22", "remaining_time": "1:17:51"}
+{"current_steps": 285, "total_steps": 540, "loss": 0.0133, "lr": 5.387462103359655e-05, "epoch": 2.638888888888889, "percentage": 52.78, "elapsed_time": "1:26:37", "remaining_time": "1:17:30"}
+{"current_steps": 285, "total_steps": 540, "eval_loss": 0.016152961179614067, "epoch": 2.638888888888889, "percentage": 52.78, "elapsed_time": "1:26:46", "remaining_time": "1:17:38"}
+{"current_steps": 286, "total_steps": 540, "loss": 0.0136, "lr": 5.355230511433651e-05, "epoch": 2.648148148148148, "percentage": 52.96, "elapsed_time": "1:27:02", "remaining_time": "1:17:18"}
+{"current_steps": 287, "total_steps": 540, "loss": 0.0133, "lr": 5.32298407601999e-05, "epoch": 2.6574074074074074, "percentage": 53.15, "elapsed_time": "1:27:17", "remaining_time": "1:16:57"}
+{"current_steps": 288, "total_steps": 540, "loss": 0.0143, "lr": 5.290724144552379e-05, "epoch": 2.6666666666666665, "percentage": 53.33, "elapsed_time": "1:27:33", "remaining_time": "1:16:36"}
+{"current_steps": 289, "total_steps": 540, "loss": 0.0137, "lr": 5.258452065028473e-05, "epoch": 2.675925925925926, "percentage": 53.52, "elapsed_time": "1:27:48", "remaining_time": "1:16:15"}
+{"current_steps": 290, "total_steps": 540, "loss": 0.0125, "lr": 5.226169185953532e-05, "epoch": 2.685185185185185, "percentage": 53.7, "elapsed_time": "1:28:04", "remaining_time": "1:15:55"}
+{"current_steps": 290, "total_steps": 540, "eval_loss": 0.016076602041721344, "epoch": 2.685185185185185, "percentage": 53.7, "elapsed_time": "1:28:13", "remaining_time": "1:16:03"}
+{"current_steps": 291, "total_steps": 540, "loss": 0.012, "lr": 5.193876856284085e-05, "epoch": 2.6944444444444446, "percentage": 53.89, "elapsed_time": "1:28:30", "remaining_time": "1:15:44"}
+{"current_steps": 292, "total_steps": 540, "loss": 0.0136, "lr": 5.1615764253715536e-05, "epoch": 2.7037037037037037, "percentage": 54.07, "elapsed_time": "1:28:47", "remaining_time": "1:15:24"}
+{"current_steps": 293, "total_steps": 540, "loss": 0.012, "lr": 5.129269242905882e-05, "epoch": 2.712962962962963, "percentage": 54.26, "elapsed_time": "1:29:02", "remaining_time": "1:15:03"}
+{"current_steps": 294, "total_steps": 540, "loss": 0.0137, "lr": 5.096956658859122e-05, "epoch": 2.7222222222222223, "percentage": 54.44, "elapsed_time": "1:29:20", "remaining_time": "1:14:45"}
+{"current_steps": 295, "total_steps": 540, "loss": 0.0147, "lr": 5.064640023429043e-05, "epoch": 2.7314814814814814, "percentage": 54.63, "elapsed_time": "1:29:37", "remaining_time": "1:14:25"}
+{"current_steps": 295, "total_steps": 540, "eval_loss": 0.01584070920944214, "epoch": 2.7314814814814814, "percentage": 54.63, "elapsed_time": "1:29:46", "remaining_time": "1:14:33"}
+{"current_steps": 296, "total_steps": 540, "loss": 0.0111, "lr": 5.0323206869826966e-05, "epoch": 2.7407407407407405, "percentage": 54.81, "elapsed_time": "1:30:03", "remaining_time": "1:14:14"}
+{"current_steps": 297, "total_steps": 540, "loss": 0.0134, "lr": 5e-05, "epoch": 2.75, "percentage": 55.0, "elapsed_time": "1:30:19", "remaining_time": "1:13:54"}
+{"current_steps": 298, "total_steps": 540, "loss": 0.0124, "lr": 4.967679313017303e-05, "epoch": 2.7592592592592595, "percentage": 55.19, "elapsed_time": "1:30:34", "remaining_time": "1:13:33"}
+{"current_steps": 299, "total_steps": 540, "loss": 0.0144, "lr": 4.9353599765709584e-05, "epoch": 2.7685185185185186, "percentage": 55.37, "elapsed_time": "1:30:51", "remaining_time": "1:13:13"}
+{"current_steps": 300, "total_steps": 540, "loss": 0.0134, "lr": 4.903043341140879e-05, "epoch": 2.7777777777777777, "percentage": 55.56, "elapsed_time": "1:31:07", "remaining_time": "1:12:53"}
+{"current_steps": 300, "total_steps": 540, "eval_loss": 0.016122175380587578, "epoch": 2.7777777777777777, "percentage": 55.56, "elapsed_time": "1:31:16", "remaining_time": "1:13:01"}
+{"current_steps": 301, "total_steps": 540, "loss": 0.0123, "lr": 4.870730757094121e-05, "epoch": 2.787037037037037, "percentage": 55.74, "elapsed_time": "1:31:40", "remaining_time": "1:12:47"}
+{"current_steps": 302, "total_steps": 540, "loss": 0.015, "lr": 4.8384235746284476e-05, "epoch": 2.7962962962962963, "percentage": 55.93, "elapsed_time": "1:31:57", "remaining_time": "1:12:27"}
+{"current_steps": 303, "total_steps": 540, "loss": 0.0142, "lr": 4.806123143715916e-05, "epoch": 2.8055555555555554, "percentage": 56.11, "elapsed_time": "1:32:13", "remaining_time": "1:12:08"}
+{"current_steps": 304, "total_steps": 540, "loss": 0.0131, "lr": 4.7738308140464685e-05, "epoch": 2.814814814814815, "percentage": 56.3, "elapsed_time": "1:32:30", "remaining_time": "1:11:49"}
+{"current_steps": 305, "total_steps": 540, "loss": 0.0124, "lr": 4.7415479349715275e-05, "epoch": 2.824074074074074, "percentage": 56.48, "elapsed_time": "1:32:48", "remaining_time": "1:11:30"}
+{"current_steps": 305, "total_steps": 540, "eval_loss": 0.015797268599271774, "epoch": 2.824074074074074, "percentage": 56.48, "elapsed_time": "1:32:57", "remaining_time": "1:11:37"}
+{"current_steps": 306, "total_steps": 540, "loss": 0.0154, "lr": 4.709275855447621e-05, "epoch": 2.8333333333333335, "percentage": 56.67, "elapsed_time": "1:33:14", "remaining_time": "1:11:18"}
+{"current_steps": 307, "total_steps": 540, "loss": 0.0138, "lr": 4.677015923980011e-05, "epoch": 2.8425925925925926, "percentage": 56.85, "elapsed_time": "1:33:29", "remaining_time": "1:10:57"}
+{"current_steps": 308, "total_steps": 540, "loss": 0.0124, "lr": 4.6447694885663514e-05, "epoch": 2.851851851851852, "percentage": 57.04, "elapsed_time": "1:33:45", "remaining_time": "1:10:37"}
+{"current_steps": 309, "total_steps": 540, "loss": 0.0155, "lr": 4.612537896640346e-05, "epoch": 2.861111111111111, "percentage": 57.22, "elapsed_time": "1:34:02", "remaining_time": "1:10:17"}
+{"current_steps": 310, "total_steps": 540, "loss": 0.0132, "lr": 4.5803224950154656e-05, "epoch": 2.8703703703703702, "percentage": 57.41, "elapsed_time": "1:34:18", "remaining_time": "1:09:58"}
+{"current_steps": 310, "total_steps": 540, "eval_loss": 0.015400240197777748, "epoch": 2.8703703703703702, "percentage": 57.41, "elapsed_time": "1:34:27", "remaining_time": "1:10:04"}
+{"current_steps": 311, "total_steps": 540, "loss": 0.0139, "lr": 4.54812462982866e-05, "epoch": 2.8796296296296298, "percentage": 57.59, "elapsed_time": "1:34:43", "remaining_time": "1:09:44"}
+{"current_steps": 312, "total_steps": 540, "loss": 0.0133, "lr": 4.515945646484105e-05, "epoch": 2.888888888888889, "percentage": 57.78, "elapsed_time": "1:35:00", "remaining_time": "1:09:25"}
+{"current_steps": 313, "total_steps": 540, "loss": 0.0126, "lr": 4.4837868895969936e-05, "epoch": 2.898148148148148, "percentage": 57.96, "elapsed_time": "1:35:16", "remaining_time": "1:09:05"}
+{"current_steps": 314, "total_steps": 540, "loss": 0.0106, "lr": 4.451649702937342e-05, "epoch": 2.9074074074074074, "percentage": 58.15, "elapsed_time": "1:35:32", "remaining_time": "1:08:45"}
+{"current_steps": 315, "total_steps": 540, "loss": 0.0146, "lr": 4.4195354293738484e-05, "epoch": 2.9166666666666665, "percentage": 58.33, "elapsed_time": "1:35:48", "remaining_time": "1:08:26"}
+{"current_steps": 315, "total_steps": 540, "eval_loss": 0.015166966244578362, "epoch": 2.9166666666666665, "percentage": 58.33, "elapsed_time": "1:35:57", "remaining_time": "1:08:32"}
+{"current_steps": 316, "total_steps": 540, "loss": 0.0124, "lr": 4.387445410817774e-05, "epoch": 2.925925925925926, "percentage": 58.52, "elapsed_time": "1:36:15", "remaining_time": "1:08:13"}
+{"current_steps": 317, "total_steps": 540, "loss": 0.0119, "lr": 4.355380988166867e-05, "epoch": 2.935185185185185, "percentage": 58.7, "elapsed_time": "1:36:30", "remaining_time": "1:07:53"}
+{"current_steps": 318, "total_steps": 540, "loss": 0.0144, "lr": 4.323343501249346e-05, "epoch": 2.9444444444444446, "percentage": 58.89, "elapsed_time": "1:36:46", "remaining_time": "1:07:33"}
+{"current_steps": 319, "total_steps": 540, "loss": 0.0117, "lr": 4.2913342887678985e-05, "epoch": 2.9537037037037037, "percentage": 59.07, "elapsed_time": "1:37:01", "remaining_time": "1:07:13"}
+{"current_steps": 320, "total_steps": 540, "loss": 0.014, "lr": 4.259354688243757e-05, "epoch": 2.962962962962963, "percentage": 59.26, "elapsed_time": "1:37:18", "remaining_time": "1:06:53"}
+{"current_steps": 320, "total_steps": 540, "eval_loss": 0.014957955107092857, "epoch": 2.962962962962963, "percentage": 59.26, "elapsed_time": "1:37:27", "remaining_time": "1:07:00"}
+{"current_steps": 321, "total_steps": 540, "loss": 0.0121, "lr": 4.227406035960798e-05, "epoch": 2.9722222222222223, "percentage": 59.44, "elapsed_time": "1:37:43", "remaining_time": "1:06:40"}
+{"current_steps": 322, "total_steps": 540, "loss": 0.0119, "lr": 4.195489666909713e-05, "epoch": 2.9814814814814814, "percentage": 59.63, "elapsed_time": "1:38:00", "remaining_time": "1:06:20"}
+{"current_steps": 323, "total_steps": 540, "loss": 0.0136, "lr": 4.1636069147322246e-05, "epoch": 2.9907407407407405, "percentage": 59.81, "elapsed_time": "1:38:15", "remaining_time": "1:06:00"}
+{"current_steps": 324, "total_steps": 540, "loss": 0.0137, "lr": 4.131759111665349e-05, "epoch": 3.0, "percentage": 60.0, "elapsed_time": "1:38:31", "remaining_time": "1:05:40"}
+{"current_steps": 325, "total_steps": 540, "loss": 0.0122, "lr": 4.099947588485744e-05, "epoch": 3.009259259259259, "percentage": 60.19, "elapsed_time": "1:38:46", "remaining_time": "1:05:20"}
+{"current_steps": 325, "total_steps": 540, "eval_loss": 0.015089023858308792, "epoch": 3.009259259259259, "percentage": 60.19, "elapsed_time": "1:38:56", "remaining_time": "1:05:26"}
+{"current_steps": 326, "total_steps": 540, "loss": 0.0095, "lr": 4.06817367445409e-05, "epoch": 3.0185185185185186, "percentage": 60.37, "elapsed_time": "1:39:12", "remaining_time": "1:05:07"}
+{"current_steps": 327, "total_steps": 540, "loss": 0.0134, "lr": 4.036438697259551e-05, "epoch": 3.0277777777777777, "percentage": 60.56, "elapsed_time": "1:39:28", "remaining_time": "1:04:47"}
+{"current_steps": 328, "total_steps": 540, "loss": 0.0122, "lr": 4.004743982964298e-05, "epoch": 3.037037037037037, "percentage": 60.74, "elapsed_time": "1:39:44", "remaining_time": "1:04:28"}
+{"current_steps": 329, "total_steps": 540, "loss": 0.0101, "lr": 3.97309085594809e-05, "epoch": 3.0462962962962963, "percentage": 60.93, "elapsed_time": "1:40:00", "remaining_time": "1:04:08"}
+{"current_steps": 330, "total_steps": 540, "loss": 0.0118, "lr": 3.941480638852948e-05, "epoch": 3.0555555555555554, "percentage": 61.11, "elapsed_time": "1:40:16", "remaining_time": "1:03:48"}
+{"current_steps": 330, "total_steps": 540, "eval_loss": 0.015503546223044395, "epoch": 3.0555555555555554, "percentage": 61.11, "elapsed_time": "1:40:25", "remaining_time": "1:03:54"}
+{"current_steps": 331, "total_steps": 540, "loss": 0.0109, "lr": 3.909914652527871e-05, "epoch": 3.064814814814815, "percentage": 61.3, "elapsed_time": "1:40:40", "remaining_time": "1:03:34"}
+{"current_steps": 332, "total_steps": 540, "loss": 0.0123, "lr": 3.878394215973663e-05, "epoch": 3.074074074074074, "percentage": 61.48, "elapsed_time": "1:40:55", "remaining_time": "1:03:13"}
+{"current_steps": 333, "total_steps": 540, "loss": 0.0122, "lr": 3.846920646287799e-05, "epoch": 3.0833333333333335, "percentage": 61.67, "elapsed_time": "1:41:11", "remaining_time": "1:02:54"}
+{"current_steps": 334, "total_steps": 540, "loss": 0.0125, "lr": 3.815495258609404e-05, "epoch": 3.0925925925925926, "percentage": 61.85, "elapsed_time": "1:41:27", "remaining_time": "1:02:34"}
+{"current_steps": 335, "total_steps": 540, "loss": 0.0108, "lr": 3.784119366064293e-05, "epoch": 3.1018518518518516, "percentage": 62.04, "elapsed_time": "1:41:43", "remaining_time": "1:02:15"}
+{"current_steps": 335, "total_steps": 540, "eval_loss": 0.0155374426394701, "epoch": 3.1018518518518516, "percentage": 62.04, "elapsed_time": "1:41:52", "remaining_time": "1:02:20"}
+{"current_steps": 336, "total_steps": 540, "loss": 0.0114, "lr": 3.752794279710094e-05, "epoch": 3.111111111111111, "percentage": 62.22, "elapsed_time": "1:42:09", "remaining_time": "1:02:01"}
+{"current_steps": 337, "total_steps": 540, "loss": 0.0101, "lr": 3.721521308481482e-05, "epoch": 3.1203703703703702, "percentage": 62.41, "elapsed_time": "1:42:26", "remaining_time": "1:01:42"}
+{"current_steps": 338, "total_steps": 540, "loss": 0.0107, "lr": 3.6903017591354706e-05, "epoch": 3.1296296296296298, "percentage": 62.59, "elapsed_time": "1:42:41", "remaining_time": "1:01:22"}
+{"current_steps": 339, "total_steps": 540, "loss": 0.012, "lr": 3.6591369361968124e-05, "epoch": 3.138888888888889, "percentage": 62.78, "elapsed_time": "1:42:57", "remaining_time": "1:01:03"}
+{"current_steps": 340, "total_steps": 540, "loss": 0.0103, "lr": 3.628028141903493e-05, "epoch": 3.148148148148148, "percentage": 62.96, "elapsed_time": "1:43:13", "remaining_time": "1:00:43"}
+{"current_steps": 340, "total_steps": 540, "eval_loss": 0.01546421181410551, "epoch": 3.148148148148148, "percentage": 62.96, "elapsed_time": "1:43:22", "remaining_time": "1:00:48"}
+{"current_steps": 341, "total_steps": 540, "loss": 0.0116, "lr": 3.596976676152306e-05, "epoch": 3.1574074074074074, "percentage": 63.15, "elapsed_time": "1:43:38", "remaining_time": "1:00:28"}
+{"current_steps": 342, "total_steps": 540, "loss": 0.0108, "lr": 3.5659838364445505e-05, "epoch": 3.1666666666666665, "percentage": 63.33, "elapsed_time": "1:43:54", "remaining_time": "1:00:09"}
+{"current_steps": 343, "total_steps": 540, "loss": 0.0108, "lr": 3.535050917831797e-05, "epoch": 3.175925925925926, "percentage": 63.52, "elapsed_time": "1:44:11", "remaining_time": "0:59:50"}
+{"current_steps": 344, "total_steps": 540, "loss": 0.0094, "lr": 3.5041792128617927e-05, "epoch": 3.185185185185185, "percentage": 63.7, "elapsed_time": "1:44:26", "remaining_time": "0:59:30"}
+{"current_steps": 345, "total_steps": 540, "loss": 0.0099, "lr": 3.473370011524435e-05, "epoch": 3.1944444444444446, "percentage": 63.89, "elapsed_time": "1:44:41", "remaining_time": "0:59:10"}
+{"current_steps": 345, "total_steps": 540, "eval_loss": 0.015372861176729202, "epoch": 3.1944444444444446, "percentage": 63.89, "elapsed_time": "1:44:51", "remaining_time": "0:59:15"}
+{"current_steps": 346, "total_steps": 540, "loss": 0.0096, "lr": 3.442624601197877e-05, "epoch": 3.2037037037037037, "percentage": 64.07, "elapsed_time": "1:45:06", "remaining_time": "0:58:55"}
+{"current_steps": 347, "total_steps": 540, "loss": 0.0094, "lr": 3.4119442665947344e-05, "epoch": 3.212962962962963, "percentage": 64.26, "elapsed_time": "1:45:21", "remaining_time": "0:58:35"}
+{"current_steps": 348, "total_steps": 540, "loss": 0.011, "lr": 3.381330289708396e-05, "epoch": 3.2222222222222223, "percentage": 64.44, "elapsed_time": "1:45:36", "remaining_time": "0:58:15"}
+{"current_steps": 349, "total_steps": 540, "loss": 0.0105, "lr": 3.350783949759462e-05, "epoch": 3.2314814814814814, "percentage": 64.63, "elapsed_time": "1:45:51", "remaining_time": "0:57:56"}
+{"current_steps": 350, "total_steps": 540, "loss": 0.0115, "lr": 3.3203065231422904e-05, "epoch": 3.240740740740741, "percentage": 64.81, "elapsed_time": "1:46:06", "remaining_time": "0:57:36"}
+{"current_steps": 350, "total_steps": 540, "eval_loss": 0.015474287793040276, "epoch": 3.240740740740741, "percentage": 64.81, "elapsed_time": "1:46:15", "remaining_time": "0:57:41"}
+{"current_steps": 351, "total_steps": 540, "loss": 0.014, "lr": 3.289899283371657e-05, "epoch": 3.25, "percentage": 65.0, "elapsed_time": "1:46:39", "remaining_time": "0:57:25"}
+{"current_steps": 352, "total_steps": 540, "loss": 0.0132, "lr": 3.2595635010295475e-05, "epoch": 3.259259259259259, "percentage": 65.19, "elapsed_time": "1:46:55", "remaining_time": "0:57:06"}
+{"current_steps": 353, "total_steps": 540, "loss": 0.0093, "lr": 3.2293004437120624e-05, "epoch": 3.2685185185185186, "percentage": 65.37, "elapsed_time": "1:47:11", "remaining_time": "0:56:47"}
+{"current_steps": 354, "total_steps": 540, "loss": 0.0107, "lr": 3.199111375976449e-05, "epoch": 3.2777777777777777, "percentage": 65.56, "elapsed_time": "1:47:28", "remaining_time": "0:56:28"}
+{"current_steps": 355, "total_steps": 540, "loss": 0.0099, "lr": 3.1689975592882603e-05, "epoch": 3.287037037037037, "percentage": 65.74, "elapsed_time": "1:47:44", "remaining_time": "0:56:08"}
+{"current_steps": 355, "total_steps": 540, "eval_loss": 0.015444349497556686, "epoch": 3.287037037037037, "percentage": 65.74, "elapsed_time": "1:47:53", "remaining_time": "0:56:13"}
+{"current_steps": 356, "total_steps": 540, "loss": 0.0118, "lr": 3.1389602519686515e-05, "epoch": 3.2962962962962963, "percentage": 65.93, "elapsed_time": "1:48:10", "remaining_time": "0:55:54"}
+{"current_steps": 357, "total_steps": 540, "loss": 0.0121, "lr": 3.109000709141788e-05, "epoch": 3.3055555555555554, "percentage": 66.11, "elapsed_time": "1:48:25", "remaining_time": "0:55:34"}
+{"current_steps": 358, "total_steps": 540, "loss": 0.0099, "lr": 3.079120182682412e-05, "epoch": 3.314814814814815, "percentage": 66.3, "elapsed_time": "1:48:39", "remaining_time": "0:55:14"}
+{"current_steps": 359, "total_steps": 540, "loss": 0.0119, "lr": 3.049319921163526e-05, "epoch": 3.324074074074074, "percentage": 66.48, "elapsed_time": "1:48:56", "remaining_time": "0:54:55"}
+{"current_steps": 360, "total_steps": 540, "loss": 0.0129, "lr": 3.019601169804216e-05, "epoch": 3.3333333333333335, "percentage": 66.67, "elapsed_time": "1:49:13", "remaining_time": "0:54:36"}
+{"current_steps": 360, "total_steps": 540, "eval_loss": 0.0157760102301836, "epoch": 3.3333333333333335, "percentage": 66.67, "elapsed_time": "1:49:23", "remaining_time": "0:54:41"}
+{"current_steps": 361, "total_steps": 540, "loss": 0.011, "lr": 2.9899651704176325e-05, "epoch": 3.3425925925925926, "percentage": 66.85, "elapsed_time": "1:49:38", "remaining_time": "0:54:21"}
+{"current_steps": 362, "total_steps": 540, "loss": 0.0109, "lr": 2.9604131613590824e-05, "epoch": 3.351851851851852, "percentage": 67.04, "elapsed_time": "1:49:54", "remaining_time": "0:54:02"}
+{"current_steps": 363, "total_steps": 540, "loss": 0.0122, "lr": 2.9309463774743046e-05, "epoch": 3.361111111111111, "percentage": 67.22, "elapsed_time": "1:50:11", "remaining_time": "0:53:43"}
+{"current_steps": 364, "total_steps": 540, "loss": 0.0113, "lr": 2.901566050047855e-05, "epoch": 3.3703703703703702, "percentage": 67.41, "elapsed_time": "1:50:26", "remaining_time": "0:53:24"}
+{"current_steps": 365, "total_steps": 540, "loss": 0.0105, "lr": 2.872273406751664e-05, "epoch": 3.3796296296296298, "percentage": 67.59, "elapsed_time": "1:50:43", "remaining_time": "0:53:05"}
+{"current_steps": 365, "total_steps": 540, "eval_loss": 0.015391937457025051, "epoch": 3.3796296296296298, "percentage": 67.59, "elapsed_time": "1:50:52", "remaining_time": "0:53:09"}
+{"current_steps": 366, "total_steps": 540, "loss": 0.0107, "lr": 2.8430696715937337e-05, "epoch": 3.388888888888889, "percentage": 67.78, "elapsed_time": "1:51:09", "remaining_time": "0:52:50"}
+{"current_steps": 367, "total_steps": 540, "loss": 0.0113, "lr": 2.8139560648669962e-05, "epoch": 3.398148148148148, "percentage": 67.96, "elapsed_time": "1:51:25", "remaining_time": "0:52:31"}
+{"current_steps": 368, "total_steps": 540, "loss": 0.012, "lr": 2.7849338030983257e-05, "epoch": 3.4074074074074074, "percentage": 68.15, "elapsed_time": "1:51:41", "remaining_time": "0:52:12"}
+{"current_steps": 369, "total_steps": 540, "loss": 0.01, "lr": 2.7560040989976892e-05, "epoch": 3.4166666666666665, "percentage": 68.33, "elapsed_time": "1:51:58", "remaining_time": "0:51:53"}
+{"current_steps": 370, "total_steps": 540, "loss": 0.0121, "lr": 2.7271681614074973e-05, "epoch": 3.425925925925926, "percentage": 68.52, "elapsed_time": "1:52:12", "remaining_time": "0:51:33"}
+{"current_steps": 370, "total_steps": 540, "eval_loss": 0.015503110364079475, "epoch": 3.425925925925926, "percentage": 68.52, "elapsed_time": "1:52:21", "remaining_time": "0:51:37"}
+{"current_steps": 371, "total_steps": 540, "loss": 0.0104, "lr": 2.6984271952520722e-05, "epoch": 3.435185185185185, "percentage": 68.7, "elapsed_time": "1:52:38", "remaining_time": "0:51:18"}
+{"current_steps": 372, "total_steps": 540, "loss": 0.0132, "lr": 2.6697824014873075e-05, "epoch": 3.4444444444444446, "percentage": 68.89, "elapsed_time": "1:52:54", "remaining_time": "0:50:59"}
+{"current_steps": 373, "total_steps": 540, "loss": 0.0085, "lr": 2.641234977050484e-05, "epoch": 3.4537037037037037, "percentage": 69.07, "elapsed_time": "1:53:09", "remaining_time": "0:50:39"}
+{"current_steps": 374, "total_steps": 540, "loss": 0.0096, "lr": 2.612786114810255e-05, "epoch": 3.462962962962963, "percentage": 69.26, "elapsed_time": "1:53:26", "remaining_time": "0:50:21"}
+{"current_steps": 375, "total_steps": 540, "loss": 0.0096, "lr": 2.5844370035168073e-05, "epoch": 3.4722222222222223, "percentage": 69.44, "elapsed_time": "1:53:41", "remaining_time": "0:50:01"}
+{"current_steps": 375, "total_steps": 540, "eval_loss": 0.015461472794413567, "epoch": 3.4722222222222223, "percentage": 69.44, "elapsed_time": "1:53:50", "remaining_time": "0:50:05"}
+{"current_steps": 376, "total_steps": 540, "loss": 0.0098, "lr": 2.5561888277521794e-05, "epoch": 3.4814814814814814, "percentage": 69.63, "elapsed_time": "1:54:05", "remaining_time": "0:49:45"}
+{"current_steps": 377, "total_steps": 540, "loss": 0.0114, "lr": 2.528042767880766e-05, "epoch": 3.490740740740741, "percentage": 69.81, "elapsed_time": "1:54:22", "remaining_time": "0:49:26"}
+{"current_steps": 378, "total_steps": 540, "loss": 0.0112, "lr": 2.500000000000001e-05, "epoch": 3.5, "percentage": 70.0, "elapsed_time": "1:54:38", "remaining_time": "0:49:08"}
+{"current_steps": 379, "total_steps": 540, "loss": 0.0121, "lr": 2.4720616958912053e-05, "epoch": 3.5092592592592595, "percentage": 70.19, "elapsed_time": "1:54:54", "remaining_time": "0:48:48"}
+{"current_steps": 380, "total_steps": 540, "loss": 0.0112, "lr": 2.4442290229706344e-05, "epoch": 3.5185185185185186, "percentage": 70.37, "elapsed_time": "1:55:10", "remaining_time": "0:48:29"}
+{"current_steps": 380, "total_steps": 540, "eval_loss": 0.015212837606668472, "epoch": 3.5185185185185186, "percentage": 70.37, "elapsed_time": "1:55:19", "remaining_time": "0:48:33"}
+{"current_steps": 381, "total_steps": 540, "loss": 0.0117, "lr": 2.4165031442406855e-05, "epoch": 3.5277777777777777, "percentage": 70.56, "elapsed_time": "1:55:36", "remaining_time": "0:48:14"}
+{"current_steps": 382, "total_steps": 540, "loss": 0.0091, "lr": 2.3888852182413085e-05, "epoch": 3.537037037037037, "percentage": 70.74, "elapsed_time": "1:55:53", "remaining_time": "0:47:56"}
+{"current_steps": 383, "total_steps": 540, "loss": 0.0108, "lr": 2.361376399001592e-05, "epoch": 3.5462962962962963, "percentage": 70.93, "elapsed_time": "1:56:09", "remaining_time": "0:47:36"}
+{"current_steps": 384, "total_steps": 540, "loss": 0.0093, "lr": 2.333977835991545e-05, "epoch": 3.5555555555555554, "percentage": 71.11, "elapsed_time": "1:56:25", "remaining_time": "0:47:17"}
+{"current_steps": 385, "total_steps": 540, "loss": 0.0118, "lr": 2.3066906740740623e-05, "epoch": 3.564814814814815, "percentage": 71.3, "elapsed_time": "1:56:42", "remaining_time": "0:46:59"}
+{"current_steps": 385, "total_steps": 540, "eval_loss": 0.01467986311763525, "epoch": 3.564814814814815, "percentage": 71.3, "elapsed_time": "1:56:51", "remaining_time": "0:47:02"}
+{"current_steps": 386, "total_steps": 540, "loss": 0.0086, "lr": 2.2795160534570864e-05, "epoch": 3.574074074074074, "percentage": 71.48, "elapsed_time": "1:57:07", "remaining_time": "0:46:43"}
+{"current_steps": 387, "total_steps": 540, "loss": 0.0128, "lr": 2.25245510964597e-05, "epoch": 3.5833333333333335, "percentage": 71.67, "elapsed_time": "1:57:25", "remaining_time": "0:46:25"}
+{"current_steps": 388, "total_steps": 540, "loss": 0.0121, "lr": 2.225508973396016e-05, "epoch": 3.5925925925925926, "percentage": 71.85, "elapsed_time": "1:57:41", "remaining_time": "0:46:06"}
+{"current_steps": 389, "total_steps": 540, "loss": 0.0108, "lr": 2.198678770665238e-05, "epoch": 3.601851851851852, "percentage": 72.04, "elapsed_time": "1:57:56", "remaining_time": "0:45:47"}
+{"current_steps": 390, "total_steps": 540, "loss": 0.0082, "lr": 2.171965622567308e-05, "epoch": 3.611111111111111, "percentage": 72.22, "elapsed_time": "1:58:12", "remaining_time": "0:45:27"}
+{"current_steps": 390, "total_steps": 540, "eval_loss": 0.014544774778187275, "epoch": 3.611111111111111, "percentage": 72.22, "elapsed_time": "1:58:21", "remaining_time": "0:45:31"}
+{"current_steps": 391, "total_steps": 540, "loss": 0.0092, "lr": 2.1453706453247087e-05, "epoch": 3.6203703703703702, "percentage": 72.41, "elapsed_time": "1:58:36", "remaining_time": "0:45:12"}
+{"current_steps": 392, "total_steps": 540, "loss": 0.0101, "lr": 2.1188949502220983e-05, "epoch": 3.6296296296296298, "percentage": 72.59, "elapsed_time": "1:58:53", "remaining_time": "0:44:53"}
+{"current_steps": 393, "total_steps": 540, "loss": 0.0111, "lr": 2.0925396435598664e-05, "epoch": 3.638888888888889, "percentage": 72.78, "elapsed_time": "1:59:10", "remaining_time": "0:44:34"}
+{"current_steps": 394, "total_steps": 540, "loss": 0.0091, "lr": 2.066305826607911e-05, "epoch": 3.648148148148148, "percentage": 72.96, "elapsed_time": "1:59:27", "remaining_time": "0:44:15"}
+{"current_steps": 395, "total_steps": 540, "loss": 0.0112, "lr": 2.0401945955596206e-05, "epoch": 3.6574074074074074, "percentage": 73.15, "elapsed_time": "1:59:44", "remaining_time": "0:43:57"}
+{"current_steps": 395, "total_steps": 540, "eval_loss": 0.01460795197635889, "epoch": 3.6574074074074074, "percentage": 73.15, "elapsed_time": "1:59:53", "remaining_time": "0:44:00"}
+{"current_steps": 396, "total_steps": 540, "loss": 0.01, "lr": 2.0142070414860704e-05, "epoch": 3.6666666666666665, "percentage": 73.33, "elapsed_time": "2:00:10", "remaining_time": "0:43:41"}
+{"current_steps": 397, "total_steps": 540, "loss": 0.0095, "lr": 1.9883442502904283e-05, "epoch": 3.675925925925926, "percentage": 73.52, "elapsed_time": "2:00:26", "remaining_time": "0:43:23"}
+{"current_steps": 398, "total_steps": 540, "loss": 0.0095, "lr": 1.9626073026625818e-05, "epoch": 3.685185185185185, "percentage": 73.7, "elapsed_time": "2:00:43", "remaining_time": "0:43:04"}
+{"current_steps": 399, "total_steps": 540, "loss": 0.0108, "lr": 1.936997274033986e-05, "epoch": 3.6944444444444446, "percentage": 73.89, "elapsed_time": "2:01:00", "remaining_time": "0:42:45"}
+{"current_steps": 400, "total_steps": 540, "loss": 0.0086, "lr": 1.9115152345327152e-05, "epoch": 3.7037037037037037, "percentage": 74.07, "elapsed_time": "2:01:16", "remaining_time": "0:42:26"}
+{"current_steps": 400, "total_steps": 540, "eval_loss": 0.014902754686772823, "epoch": 3.7037037037037037, "percentage": 74.07, "elapsed_time": "2:01:25", "remaining_time": "0:42:30"}
+{"current_steps": 401, "total_steps": 540, "loss": 0.0128, "lr": 1.8861622489387555e-05, "epoch": 3.712962962962963, "percentage": 74.26, "elapsed_time": "2:01:50", "remaining_time": "0:42:14"}
+{"current_steps": 402, "total_steps": 540, "loss": 0.0123, "lr": 1.8609393766395085e-05, "epoch": 3.7222222222222223, "percentage": 74.44, "elapsed_time": "2:02:05", "remaining_time": "0:41:54"}
+{"current_steps": 403, "total_steps": 540, "loss": 0.0114, "lr": 1.835847671585526e-05, "epoch": 3.7314814814814814, "percentage": 74.63, "elapsed_time": "2:02:22", "remaining_time": "0:41:36"}
+{"current_steps": 404, "total_steps": 540, "loss": 0.0099, "lr": 1.8108881822464696e-05, "epoch": 3.7407407407407405, "percentage": 74.81, "elapsed_time": "2:02:37", "remaining_time": "0:41:16"}
+{"current_steps": 405, "total_steps": 540, "loss": 0.0102, "lr": 1.7860619515673033e-05, "epoch": 3.75, "percentage": 75.0, "elapsed_time": "2:02:55", "remaining_time": "0:40:58"}
+{"current_steps": 405, "total_steps": 540, "eval_loss": 0.014981208369135857, "epoch": 3.75, "percentage": 75.0, "elapsed_time": "2:03:04", "remaining_time": "0:41:01"}
+{"current_steps": 406, "total_steps": 540, "loss": 0.012, "lr": 1.7613700169247056e-05, "epoch": 3.7592592592592595, "percentage": 75.19, "elapsed_time": "2:03:20", "remaining_time": "0:40:42"}
+{"current_steps": 407, "total_steps": 540, "loss": 0.0088, "lr": 1.7368134100837287e-05, "epoch": 3.7685185185185186, "percentage": 75.37, "elapsed_time": "2:03:35", "remaining_time": "0:40:23"}
+{"current_steps": 408, "total_steps": 540, "loss": 0.0119, "lr": 1.7123931571546827e-05, "epoch": 3.7777777777777777, "percentage": 75.56, "elapsed_time": "2:03:51", "remaining_time": "0:40:04"}
+{"current_steps": 409, "total_steps": 540, "loss": 0.011, "lr": 1.6881102785502616e-05, "epoch": 3.787037037037037, "percentage": 75.74, "elapsed_time": "2:04:07", "remaining_time": "0:39:45"}
+{"current_steps": 410, "total_steps": 540, "loss": 0.0116, "lr": 1.6639657889429018e-05, "epoch": 3.7962962962962963, "percentage": 75.93, "elapsed_time": "2:04:24", "remaining_time": "0:39:26"}
+{"current_steps": 410, "total_steps": 540, "eval_loss": 0.014859426766633987, "epoch": 3.7962962962962963, "percentage": 75.93, "elapsed_time": "2:04:33", "remaining_time": "0:39:29"}
+{"current_steps": 411, "total_steps": 540, "loss": 0.0106, "lr": 1.639960697222388e-05, "epoch": 3.8055555555555554, "percentage": 76.11, "elapsed_time": "2:04:51", "remaining_time": "0:39:11"}
+{"current_steps": 412, "total_steps": 540, "loss": 0.0113, "lr": 1.6160960064536908e-05, "epoch": 3.814814814814815, "percentage": 76.3, "elapsed_time": "2:05:08", "remaining_time": "0:38:52"}
+{"current_steps": 413, "total_steps": 540, "loss": 0.0115, "lr": 1.592372713835055e-05, "epoch": 3.824074074074074, "percentage": 76.48, "elapsed_time": "2:05:24", "remaining_time": "0:38:33"}
+{"current_steps": 414, "total_steps": 540, "loss": 0.0112, "lr": 1.5687918106563326e-05, "epoch": 3.8333333333333335, "percentage": 76.67, "elapsed_time": "2:05:42", "remaining_time": "0:38:15"}
+{"current_steps": 415, "total_steps": 540, "loss": 0.0126, "lr": 1.545354282257562e-05, "epoch": 3.8425925925925926, "percentage": 76.85, "elapsed_time": "2:05:59", "remaining_time": "0:37:56"}
+{"current_steps": 415, "total_steps": 540, "eval_loss": 0.014735485427081585, "epoch": 3.8425925925925926, "percentage": 76.85, "elapsed_time": "2:06:08", "remaining_time": "0:37:59"}
+{"current_steps": 416, "total_steps": 540, "loss": 0.0103, "lr": 1.52206110798779e-05, "epoch": 3.851851851851852, "percentage": 77.04, "elapsed_time": "2:06:24", "remaining_time": "0:37:40"}
+{"current_steps": 417, "total_steps": 540, "loss": 0.012, "lr": 1.4989132611641576e-05, "epoch": 3.861111111111111, "percentage": 77.22, "elapsed_time": "2:06:39", "remaining_time": "0:37:21"}
+{"current_steps": 418, "total_steps": 540, "loss": 0.0096, "lr": 1.4759117090312197e-05, "epoch": 3.8703703703703702, "percentage": 77.41, "elapsed_time": "2:06:56", "remaining_time": "0:37:02"}
+{"current_steps": 419, "total_steps": 540, "loss": 0.0094, "lr": 1.453057412720536e-05, "epoch": 3.8796296296296298, "percentage": 77.59, "elapsed_time": "2:07:11", "remaining_time": "0:36:43"}
+{"current_steps": 420, "total_steps": 540, "loss": 0.0112, "lr": 1.4303513272105057e-05, "epoch": 3.888888888888889, "percentage": 77.78, "elapsed_time": "2:07:27", "remaining_time": "0:36:25"}
+{"current_steps": 420, "total_steps": 540, "eval_loss": 0.014594363048672676, "epoch": 3.888888888888889, "percentage": 77.78, "elapsed_time": "2:07:36", "remaining_time": "0:36:27"}
+{"current_steps": 421, "total_steps": 540, "loss": 0.0093, "lr": 1.4077944012864636e-05, "epoch": 3.898148148148148, "percentage": 77.96, "elapsed_time": "2:07:52", "remaining_time": "0:36:08"}
+{"current_steps": 422, "total_steps": 540, "loss": 0.0102, "lr": 1.3853875775010355e-05, "epoch": 3.9074074074074074, "percentage": 78.15, "elapsed_time": "2:08:09", "remaining_time": "0:35:50"}
+{"current_steps": 423, "total_steps": 540, "loss": 0.0084, "lr": 1.3631317921347563e-05, "epoch": 3.9166666666666665, "percentage": 78.33, "elapsed_time": "2:08:26", "remaining_time": "0:35:31"}
+{"current_steps": 424, "total_steps": 540, "loss": 0.0122, "lr": 1.3410279751569399e-05, "epoch": 3.925925925925926, "percentage": 78.52, "elapsed_time": "2:08:42", "remaining_time": "0:35:12"}
+{"current_steps": 425, "total_steps": 540, "loss": 0.0107, "lr": 1.3190770501868243e-05, "epoch": 3.935185185185185, "percentage": 78.7, "elapsed_time": "2:08:58", "remaining_time": "0:34:53"}
+{"current_steps": 425, "total_steps": 540, "eval_loss": 0.014631365425884724, "epoch": 3.935185185185185, "percentage": 78.7, "elapsed_time": "2:09:07", "remaining_time": "0:34:56"}
+{"current_steps": 426, "total_steps": 540, "loss": 0.0096, "lr": 1.297279934454978e-05, "epoch": 3.9444444444444446, "percentage": 78.89, "elapsed_time": "2:09:23", "remaining_time": "0:34:37"}
+{"current_steps": 427, "total_steps": 540, "loss": 0.0124, "lr": 1.2756375387649716e-05, "epoch": 3.9537037037037037, "percentage": 79.07, "elapsed_time": "2:09:40", "remaining_time": "0:34:19"}
+{"current_steps": 428, "total_steps": 540, "loss": 0.0091, "lr": 1.25415076745532e-05, "epoch": 3.962962962962963, "percentage": 79.26, "elapsed_time": "2:09:57", "remaining_time": "0:34:00"}
+{"current_steps": 429, "total_steps": 540, "loss": 0.0105, "lr": 1.2328205183616965e-05, "epoch": 3.9722222222222223, "percentage": 79.44, "elapsed_time": "2:10:13", "remaining_time": "0:33:41"}
+{"current_steps": 430, "total_steps": 540, "loss": 0.0113, "lr": 1.2116476827794104e-05, "epoch": 3.9814814814814814, "percentage": 79.63, "elapsed_time": "2:10:30", "remaining_time": "0:33:23"}
+{"current_steps": 430, "total_steps": 540, "eval_loss": 0.01471536885946989, "epoch": 3.9814814814814814, "percentage": 79.63, "elapsed_time": "2:10:39", "remaining_time": "0:33:25"}
+{"current_steps": 431, "total_steps": 540, "loss": 0.0093, "lr": 1.1906331454261704e-05, "epoch": 3.9907407407407405, "percentage": 79.81, "elapsed_time": "2:10:55", "remaining_time": "0:33:06"}
+{"current_steps": 432, "total_steps": 540, "loss": 0.011, "lr": 1.1697777844051105e-05, "epoch": 4.0, "percentage": 80.0, "elapsed_time": "2:11:12", "remaining_time": "0:32:48"}
+{"current_steps": 433, "total_steps": 540, "loss": 0.0094, "lr": 1.1490824711681025e-05, "epoch": 4.0092592592592595, "percentage": 80.19, "elapsed_time": "2:11:28", "remaining_time": "0:32:29"}
+{"current_steps": 434, "total_steps": 540, "loss": 0.0093, "lr": 1.1285480704793377e-05, "epoch": 4.018518518518518, "percentage": 80.37, "elapsed_time": "2:11:43", "remaining_time": "0:32:10"}
+{"current_steps": 435, "total_steps": 540, "loss": 0.0091, "lr": 1.1081754403791999e-05, "epoch": 4.027777777777778, "percentage": 80.56, "elapsed_time": "2:11:59", "remaining_time": "0:31:51"}
+{"current_steps": 435, "total_steps": 540, "eval_loss": 0.014734329655766487, "epoch": 4.027777777777778, "percentage": 80.56, "elapsed_time": "2:12:08", "remaining_time": "0:31:53"}
+{"current_steps": 436, "total_steps": 540, "loss": 0.0067, "lr": 1.0879654321484012e-05, "epoch": 4.037037037037037, "percentage": 80.74, "elapsed_time": "2:12:24", "remaining_time": "0:31:35"}
+{"current_steps": 437, "total_steps": 540, "loss": 0.0108, "lr": 1.0679188902724191e-05, "epoch": 4.046296296296297, "percentage": 80.93, "elapsed_time": "2:12:40", "remaining_time": "0:31:16"}
+{"current_steps": 438, "total_steps": 540, "loss": 0.0088, "lr": 1.0480366524062042e-05, "epoch": 4.055555555555555, "percentage": 81.11, "elapsed_time": "2:12:56", "remaining_time": "0:30:57"}
+{"current_steps": 439, "total_steps": 540, "loss": 0.0103, "lr": 1.0283195493391823e-05, "epoch": 4.064814814814815, "percentage": 81.3, "elapsed_time": "2:13:13", "remaining_time": "0:30:39"}
+{"current_steps": 440, "total_steps": 540, "loss": 0.0094, "lr": 1.008768404960535e-05, "epoch": 4.074074074074074, "percentage": 81.48, "elapsed_time": "2:13:29", "remaining_time": "0:30:20"}
+{"current_steps": 440, "total_steps": 540, "eval_loss": 0.014965096488595009, "epoch": 4.074074074074074, "percentage": 81.48, "elapsed_time": "2:13:38", "remaining_time": "0:30:22"}
+{"current_steps": 441, "total_steps": 540, "loss": 0.0056, "lr": 9.893840362247809e-06, "epoch": 4.083333333333333, "percentage": 81.67, "elapsed_time": "2:13:54", "remaining_time": "0:30:03"}
+{"current_steps": 442, "total_steps": 540, "loss": 0.0089, "lr": 9.701672531176286e-06, "epoch": 4.092592592592593, "percentage": 81.85, "elapsed_time": "2:14:09", "remaining_time": "0:29:44"}
+{"current_steps": 443, "total_steps": 540, "loss": 0.0092, "lr": 9.511188586221376e-06, "epoch": 4.101851851851852, "percentage": 82.04, "elapsed_time": "2:14:24", "remaining_time": "0:29:25"}
+{"current_steps": 444, "total_steps": 540, "loss": 0.0104, "lr": 9.322396486851626e-06, "epoch": 4.111111111111111, "percentage": 82.22, "elapsed_time": "2:14:41", "remaining_time": "0:29:07"}
+{"current_steps": 445, "total_steps": 540, "loss": 0.0096, "lr": 9.135304121840976e-06, "epoch": 4.12037037037037, "percentage": 82.41, "elapsed_time": "2:14:58", "remaining_time": "0:28:48"}
+{"current_steps": 445, "total_steps": 540, "eval_loss": 0.014996801503002644, "epoch": 4.12037037037037, "percentage": 82.41, "elapsed_time": "2:15:07", "remaining_time": "0:28:50"}
+{"current_steps": 446, "total_steps": 540, "loss": 0.0109, "lr": 8.949919308939082e-06, "epoch": 4.12962962962963, "percentage": 82.59, "elapsed_time": "2:15:22", "remaining_time": "0:28:31"}
+{"current_steps": 447, "total_steps": 540, "loss": 0.0073, "lr": 8.766249794544662e-06, "epoch": 4.138888888888889, "percentage": 82.78, "elapsed_time": "2:15:38", "remaining_time": "0:28:13"}
+{"current_steps": 448, "total_steps": 540, "loss": 0.0105, "lr": 8.584303253381847e-06, "epoch": 4.148148148148148, "percentage": 82.96, "elapsed_time": "2:15:55", "remaining_time": "0:27:54"}
+{"current_steps": 449, "total_steps": 540, "loss": 0.0096, "lr": 8.404087288179424e-06, "epoch": 4.157407407407407, "percentage": 83.15, "elapsed_time": "2:16:12", "remaining_time": "0:27:36"}
+{"current_steps": 450, "total_steps": 540, "loss": 0.0091, "lr": 8.225609429353187e-06, "epoch": 4.166666666666667, "percentage": 83.33, "elapsed_time": "2:16:29", "remaining_time": "0:27:17"}
+{"current_steps": 450, "total_steps": 540, "eval_loss": 0.015186839736998081, "epoch": 4.166666666666667, "percentage": 83.33, "elapsed_time": "2:16:38", "remaining_time": "0:27:19"}
+{"current_steps": 451, "total_steps": 540, "loss": 0.0091, "lr": 8.048877134691268e-06, "epoch": 4.175925925925926, "percentage": 83.52, "elapsed_time": "2:17:02", "remaining_time": "0:27:02"}
+{"current_steps": 452, "total_steps": 540, "loss": 0.0092, "lr": 7.873897789042523e-06, "epoch": 4.185185185185185, "percentage": 83.7, "elapsed_time": "2:17:19", "remaining_time": "0:26:44"}
+{"current_steps": 453, "total_steps": 540, "loss": 0.0077, "lr": 7.700678704007947e-06, "epoch": 4.194444444444445, "percentage": 83.89, "elapsed_time": "2:17:36", "remaining_time": "0:26:25"}
+{"current_steps": 454, "total_steps": 540, "loss": 0.0077, "lr": 7.529227117635135e-06, "epoch": 4.203703703703703, "percentage": 84.07, "elapsed_time": "2:17:52", "remaining_time": "0:26:07"}
+{"current_steps": 455, "total_steps": 540, "loss": 0.0089, "lr": 7.35955019411585e-06, "epoch": 4.212962962962963, "percentage": 84.26, "elapsed_time": "2:18:07", "remaining_time": "0:25:48"}
+{"current_steps": 455, "total_steps": 540, "eval_loss": 0.015497377142310143, "epoch": 4.212962962962963, "percentage": 84.26, "elapsed_time": "2:18:16", "remaining_time": "0:25:49"}
+{"current_steps": 456, "total_steps": 540, "loss": 0.01, "lr": 7.191655023486682e-06, "epoch": 4.222222222222222, "percentage": 84.44, "elapsed_time": "2:18:34", "remaining_time": "0:25:31"}
+{"current_steps": 457, "total_steps": 540, "loss": 0.0105, "lr": 7.02554862133275e-06, "epoch": 4.231481481481482, "percentage": 84.63, "elapsed_time": "2:18:51", "remaining_time": "0:25:13"}
+{"current_steps": 458, "total_steps": 540, "loss": 0.009, "lr": 6.861237928494579e-06, "epoch": 4.2407407407407405, "percentage": 84.81, "elapsed_time": "2:19:07", "remaining_time": "0:24:54"}
+{"current_steps": 459, "total_steps": 540, "loss": 0.0102, "lr": 6.698729810778065e-06, "epoch": 4.25, "percentage": 85.0, "elapsed_time": "2:19:23", "remaining_time": "0:24:35"}
+{"current_steps": 460, "total_steps": 540, "loss": 0.0063, "lr": 6.53803105866761e-06, "epoch": 4.2592592592592595, "percentage": 85.19, "elapsed_time": "2:19:38", "remaining_time": "0:24:17"}
+{"current_steps": 460, "total_steps": 540, "eval_loss": 0.01563325710594654, "epoch": 4.2592592592592595, "percentage": 85.19, "elapsed_time": "2:19:47", "remaining_time": "0:24:18"}
+{"current_steps": 461, "total_steps": 540, "loss": 0.01, "lr": 6.379148387042316e-06, "epoch": 4.268518518518518, "percentage": 85.37, "elapsed_time": "2:20:03", "remaining_time": "0:24:00"}
+{"current_steps": 462, "total_steps": 540, "loss": 0.0072, "lr": 6.222088434895462e-06, "epoch": 4.277777777777778, "percentage": 85.56, "elapsed_time": "2:20:20", "remaining_time": "0:23:41"}
+{"current_steps": 463, "total_steps": 540, "loss": 0.0088, "lr": 6.066857765057055e-06, "epoch": 4.287037037037037, "percentage": 85.74, "elapsed_time": "2:20:38", "remaining_time": "0:23:23"}
+{"current_steps": 464, "total_steps": 540, "loss": 0.0085, "lr": 5.9134628639196e-06, "epoch": 4.296296296296296, "percentage": 85.93, "elapsed_time": "2:20:53", "remaining_time": "0:23:04"}
+{"current_steps": 465, "total_steps": 540, "loss": 0.0099, "lr": 5.7619101411671095e-06, "epoch": 4.305555555555555, "percentage": 86.11, "elapsed_time": "2:21:10", "remaining_time": "0:22:46"}
+{"current_steps": 465, "total_steps": 540, "eval_loss": 0.015693385154008865, "epoch": 4.305555555555555, "percentage": 86.11, "elapsed_time": "2:21:19", "remaining_time": "0:22:47"}
+{"current_steps": 466, "total_steps": 540, "loss": 0.0096, "lr": 5.6122059295072085e-06, "epoch": 4.314814814814815, "percentage": 86.3, "elapsed_time": "2:21:36", "remaining_time": "0:22:29"}
+{"current_steps": 467, "total_steps": 540, "loss": 0.0072, "lr": 5.464356484406535e-06, "epoch": 4.324074074074074, "percentage": 86.48, "elapsed_time": "2:21:53", "remaining_time": "0:22:10"}
+{"current_steps": 468, "total_steps": 540, "loss": 0.0079, "lr": 5.318367983829392e-06, "epoch": 4.333333333333333, "percentage": 86.67, "elapsed_time": "2:22:09", "remaining_time": "0:21:52"}
+{"current_steps": 469, "total_steps": 540, "loss": 0.0095, "lr": 5.174246527979531e-06, "epoch": 4.342592592592593, "percentage": 86.85, "elapsed_time": "2:22:27", "remaining_time": "0:21:33"}
+{"current_steps": 470, "total_steps": 540, "loss": 0.0085, "lr": 5.031998139045352e-06, "epoch": 4.351851851851852, "percentage": 87.04, "elapsed_time": "2:22:42", "remaining_time": "0:21:15"}
+{"current_steps": 470, "total_steps": 540, "eval_loss": 0.015615792945027351, "epoch": 4.351851851851852, "percentage": 87.04, "elapsed_time": "2:22:51", "remaining_time": "0:21:16"}
+{"current_steps": 471, "total_steps": 540, "loss": 0.009, "lr": 4.891628760948114e-06, "epoch": 4.361111111111111, "percentage": 87.22, "elapsed_time": "2:23:07", "remaining_time": "0:20:58"}
+{"current_steps": 472, "total_steps": 540, "loss": 0.0102, "lr": 4.7531442590937335e-06, "epoch": 4.37037037037037, "percentage": 87.41, "elapsed_time": "2:23:24", "remaining_time": "0:20:39"}
+{"current_steps": 473, "total_steps": 540, "loss": 0.0078, "lr": 4.616550420127563e-06, "epoch": 4.37962962962963, "percentage": 87.59, "elapsed_time": "2:23:39", "remaining_time": "0:20:20"}
+{"current_steps": 474, "total_steps": 540, "loss": 0.0086, "lr": 4.4818529516926726e-06, "epoch": 4.388888888888889, "percentage": 87.78, "elapsed_time": "2:23:56", "remaining_time": "0:20:02"}
+{"current_steps": 475, "total_steps": 540, "loss": 0.011, "lr": 4.349057482191299e-06, "epoch": 4.398148148148148, "percentage": 87.96, "elapsed_time": "2:24:12", "remaining_time": "0:19:44"}
+{"current_steps": 475, "total_steps": 540, "eval_loss": 0.015554042533040047, "epoch": 4.398148148148148, "percentage": 87.96, "elapsed_time": "2:24:21", "remaining_time": "0:19:45"}
+{"current_steps": 476, "total_steps": 540, "loss": 0.0108, "lr": 4.218169560549706e-06, "epoch": 4.407407407407407, "percentage": 88.15, "elapsed_time": "2:24:38", "remaining_time": "0:19:26"}
+{"current_steps": 477, "total_steps": 540, "loss": 0.0099, "lr": 4.089194655986306e-06, "epoch": 4.416666666666667, "percentage": 88.33, "elapsed_time": "2:24:56", "remaining_time": "0:19:08"}
+{"current_steps": 478, "total_steps": 540, "loss": 0.0095, "lr": 3.962138157783085e-06, "epoch": 4.425925925925926, "percentage": 88.52, "elapsed_time": "2:25:12", "remaining_time": "0:18:50"}
+{"current_steps": 479, "total_steps": 540, "loss": 0.0089, "lr": 3.837005375060482e-06, "epoch": 4.435185185185185, "percentage": 88.7, "elapsed_time": "2:25:28", "remaining_time": "0:18:31"}
+{"current_steps": 480, "total_steps": 540, "loss": 0.0067, "lr": 3.7138015365554833e-06, "epoch": 4.444444444444445, "percentage": 88.89, "elapsed_time": "2:25:43", "remaining_time": "0:18:12"}
+{"current_steps": 480, "total_steps": 540, "eval_loss": 0.01539613213390112, "epoch": 4.444444444444445, "percentage": 88.89, "elapsed_time": "2:25:53", "remaining_time": "0:18:14"}
+{"current_steps": 481, "total_steps": 540, "loss": 0.0087, "lr": 3.5925317904031587e-06, "epoch": 4.453703703703704, "percentage": 89.07, "elapsed_time": "2:26:08", "remaining_time": "0:17:55"}
+{"current_steps": 482, "total_steps": 540, "loss": 0.011, "lr": 3.4732012039215776e-06, "epoch": 4.462962962962963, "percentage": 89.26, "elapsed_time": "2:26:25", "remaining_time": "0:17:37"}
+{"current_steps": 483, "total_steps": 540, "loss": 0.0096, "lr": 3.3558147633999728e-06, "epoch": 4.472222222222222, "percentage": 89.44, "elapsed_time": "2:26:42", "remaining_time": "0:17:18"}
+{"current_steps": 484, "total_steps": 540, "loss": 0.0087, "lr": 3.2403773738905187e-06, "epoch": 4.481481481481482, "percentage": 89.63, "elapsed_time": "2:26:58", "remaining_time": "0:17:00"}
+{"current_steps": 485, "total_steps": 540, "loss": 0.0092, "lr": 3.126893859003249e-06, "epoch": 4.4907407407407405, "percentage": 89.81, "elapsed_time": "2:27:14", "remaining_time": "0:16:41"}
+{"current_steps": 485, "total_steps": 540, "eval_loss": 0.015287145972251892, "epoch": 4.4907407407407405, "percentage": 89.81, "elapsed_time": "2:27:23", "remaining_time": "0:16:42"}
+{"current_steps": 486, "total_steps": 540, "loss": 0.0086, "lr": 3.0153689607045845e-06, "epoch": 4.5, "percentage": 90.0, "elapsed_time": "2:27:39", "remaining_time": "0:16:24"}
+{"current_steps": 487, "total_steps": 540, "loss": 0.0068, "lr": 2.9058073391191375e-06, "epoch": 4.5092592592592595, "percentage": 90.19, "elapsed_time": "2:27:55", "remaining_time": "0:16:05"}
+{"current_steps": 488, "total_steps": 540, "loss": 0.0062, "lr": 2.798213572335001e-06, "epoch": 4.518518518518518, "percentage": 90.37, "elapsed_time": "2:28:11", "remaining_time": "0:15:47"}
+{"current_steps": 489, "total_steps": 540, "loss": 0.0092, "lr": 2.692592156212487e-06, "epoch": 4.527777777777778, "percentage": 90.56, "elapsed_time": "2:28:26", "remaining_time": "0:15:28"}
+{"current_steps": 490, "total_steps": 540, "loss": 0.0072, "lr": 2.5889475041961765e-06, "epoch": 4.537037037037037, "percentage": 90.74, "elapsed_time": "2:28:41", "remaining_time": "0:15:10"}
+{"current_steps": 490, "total_steps": 540, "eval_loss": 0.015211592428386211, "epoch": 4.537037037037037, "percentage": 90.74, "elapsed_time": "2:28:50", "remaining_time": "0:15:11"}
+{"current_steps": 491, "total_steps": 540, "loss": 0.0082, "lr": 2.4872839471306084e-06, "epoch": 4.546296296296296, "percentage": 90.93, "elapsed_time": "2:29:06", "remaining_time": "0:14:52"}
+{"current_steps": 492, "total_steps": 540, "loss": 0.008, "lr": 2.3876057330792346e-06, "epoch": 4.555555555555555, "percentage": 91.11, "elapsed_time": "2:29:22", "remaining_time": "0:14:34"}
+{"current_steps": 493, "total_steps": 540, "loss": 0.011, "lr": 2.2899170271469428e-06, "epoch": 4.564814814814815, "percentage": 91.3, "elapsed_time": "2:29:38", "remaining_time": "0:14:15"}
+{"current_steps": 494, "total_steps": 540, "loss": 0.0075, "lr": 2.1942219113060212e-06, "epoch": 4.574074074074074, "percentage": 91.48, "elapsed_time": "2:29:53", "remaining_time": "0:13:57"}
+{"current_steps": 495, "total_steps": 540, "loss": 0.0078, "lr": 2.100524384225555e-06, "epoch": 4.583333333333333, "percentage": 91.67, "elapsed_time": "2:30:10", "remaining_time": "0:13:39"}
+{"current_steps": 495, "total_steps": 540, "eval_loss": 0.015181516297161579, "epoch": 4.583333333333333, "percentage": 91.67, "elapsed_time": "2:30:19", "remaining_time": "0:13:39"}
+{"current_steps": 496, "total_steps": 540, "loss": 0.0062, "lr": 2.0088283611044036e-06, "epoch": 4.592592592592593, "percentage": 91.85, "elapsed_time": "2:30:34", "remaining_time": "0:13:21"}
+{"current_steps": 497, "total_steps": 540, "loss": 0.0088, "lr": 1.9191376735075427e-06, "epoch": 4.601851851851852, "percentage": 92.04, "elapsed_time": "2:30:48", "remaining_time": "0:13:02"}
+{"current_steps": 498, "total_steps": 540, "loss": 0.0089, "lr": 1.8314560692059835e-06, "epoch": 4.611111111111111, "percentage": 92.22, "elapsed_time": "2:31:04", "remaining_time": "0:12:44"}
+{"current_steps": 499, "total_steps": 540, "loss": 0.0086, "lr": 1.7457872120201779e-06, "epoch": 4.62037037037037, "percentage": 92.41, "elapsed_time": "2:31:19", "remaining_time": "0:12:26"}
+{"current_steps": 500, "total_steps": 540, "loss": 0.0091, "lr": 1.6621346816668992e-06, "epoch": 4.62962962962963, "percentage": 92.59, "elapsed_time": "2:31:35", "remaining_time": "0:12:07"}
+{"current_steps": 500, "total_steps": 540, "eval_loss": 0.015207822434604168, "epoch": 4.62962962962963, "percentage": 92.59, "elapsed_time": "2:31:44", "remaining_time": "0:12:08"}
+{"current_steps": 501, "total_steps": 540, "loss": 0.009, "lr": 1.5805019736097104e-06, "epoch": 4.638888888888889, "percentage": 92.78, "elapsed_time": "2:32:07", "remaining_time": "0:11:50"}
+{"current_steps": 502, "total_steps": 540, "loss": 0.0089, "lr": 1.5008924989128258e-06, "epoch": 4.648148148148148, "percentage": 92.96, "elapsed_time": "2:32:23", "remaining_time": "0:11:32"}
+{"current_steps": 503, "total_steps": 540, "loss": 0.0093, "lr": 1.4233095840986753e-06, "epoch": 4.657407407407407, "percentage": 93.15, "elapsed_time": "2:32:40", "remaining_time": "0:11:13"}
+{"current_steps": 504, "total_steps": 540, "loss": 0.0094, "lr": 1.3477564710088098e-06, "epoch": 4.666666666666667, "percentage": 93.33, "elapsed_time": "2:32:57", "remaining_time": "0:10:55"}
+{"current_steps": 505, "total_steps": 540, "loss": 0.007, "lr": 1.2742363166685034e-06, "epoch": 4.675925925925926, "percentage": 93.52, "elapsed_time": "2:33:11", "remaining_time": "0:10:37"}
+{"current_steps": 505, "total_steps": 540, "eval_loss": 0.015200878493487835, "epoch": 4.675925925925926, "percentage": 93.52, "elapsed_time": "2:33:21", "remaining_time": "0:10:37"}
+{"current_steps": 506, "total_steps": 540, "loss": 0.0074, "lr": 1.2027521931548214e-06, "epoch": 4.685185185185185, "percentage": 93.7, "elapsed_time": "2:33:37", "remaining_time": "0:10:19"}
+{"current_steps": 507, "total_steps": 540, "loss": 0.0093, "lr": 1.1333070874682216e-06, "epoch": 4.694444444444445, "percentage": 93.89, "elapsed_time": "2:33:53", "remaining_time": "0:10:01"}
+{"current_steps": 508, "total_steps": 540, "loss": 0.0084, "lr": 1.0659039014077944e-06, "epoch": 4.703703703703704, "percentage": 94.07, "elapsed_time": "2:34:10", "remaining_time": "0:09:42"}
+{"current_steps": 509, "total_steps": 540, "loss": 0.0088, "lr": 1.0005454514499414e-06, "epoch": 4.712962962962963, "percentage": 94.26, "elapsed_time": "2:34:25", "remaining_time": "0:09:24"}
+{"current_steps": 510, "total_steps": 540, "loss": 0.0064, "lr": 9.372344686307655e-07, "epoch": 4.722222222222222, "percentage": 94.44, "elapsed_time": "2:34:40", "remaining_time": "0:09:05"}
+{"current_steps": 510, "total_steps": 540, "eval_loss": 0.01521637849509716, "epoch": 4.722222222222222, "percentage": 94.44, "elapsed_time": "2:34:49", "remaining_time": "0:09:06"}
+{"current_steps": 511, "total_steps": 540, "loss": 0.0096, "lr": 8.759735984318895e-07, "epoch": 4.731481481481482, "percentage": 94.63, "elapsed_time": "2:35:05", "remaining_time": "0:08:48"}
+{"current_steps": 512, "total_steps": 540, "loss": 0.0077, "lr": 8.167654006699443e-07, "epoch": 4.7407407407407405, "percentage": 94.81, "elapsed_time": "2:35:21", "remaining_time": "0:08:29"}
+{"current_steps": 513, "total_steps": 540, "loss": 0.0072, "lr": 7.596123493895991e-07, "epoch": 4.75, "percentage": 95.0, "elapsed_time": "2:35:36", "remaining_time": "0:08:11"}
+{"current_steps": 514, "total_steps": 540, "loss": 0.0094, "lr": 7.04516832760177e-07, "epoch": 4.7592592592592595, "percentage": 95.19, "elapsed_time": "2:35:52", "remaining_time": "0:07:53"}
+{"current_steps": 515, "total_steps": 540, "loss": 0.0099, "lr": 6.514811529758747e-07, "epoch": 4.768518518518518, "percentage": 95.37, "elapsed_time": "2:36:08", "remaining_time": "0:07:34"}
+{"current_steps": 515, "total_steps": 540, "eval_loss": 0.01521516963839531, "epoch": 4.768518518518518, "percentage": 95.37, "elapsed_time": "2:36:18", "remaining_time": "0:07:35"}
+{"current_steps": 516, "total_steps": 540, "loss": 0.0086, "lr": 6.005075261595494e-07, "epoch": 4.777777777777778, "percentage": 95.56, "elapsed_time": "2:36:33", "remaining_time": "0:07:16"}
+{"current_steps": 517, "total_steps": 540, "loss": 0.0092, "lr": 5.515980822701439e-07, "epoch": 4.787037037037037, "percentage": 95.74, "elapsed_time": "2:36:49", "remaining_time": "0:06:58"}
+{"current_steps": 518, "total_steps": 540, "loss": 0.009, "lr": 5.047548650136513e-07, "epoch": 4.796296296296296, "percentage": 95.93, "elapsed_time": "2:37:06", "remaining_time": "0:06:40"}
+{"current_steps": 519, "total_steps": 540, "loss": 0.0092, "lr": 4.5997983175773417e-07, "epoch": 4.805555555555555, "percentage": 96.11, "elapsed_time": "2:37:23", "remaining_time": "0:06:22"}
+{"current_steps": 520, "total_steps": 540, "loss": 0.0088, "lr": 4.1727485344994486e-07, "epoch": 4.814814814814815, "percentage": 96.3, "elapsed_time": "2:37:39", "remaining_time": "0:06:03"}
+{"current_steps": 520, "total_steps": 540, "eval_loss": 0.015235532075166702, "epoch": 4.814814814814815, "percentage": 96.3, "elapsed_time": "2:37:49", "remaining_time": "0:06:04"}
+{"current_steps": 521, "total_steps": 540, "loss": 0.0086, "lr": 3.766417145395218e-07, "epoch": 4.824074074074074, "percentage": 96.48, "elapsed_time": "2:38:05", "remaining_time": "0:05:45"}
+{"current_steps": 522, "total_steps": 540, "loss": 0.0084, "lr": 3.380821129028489e-07, "epoch": 4.833333333333333, "percentage": 96.67, "elapsed_time": "2:38:21", "remaining_time": "0:05:27"}
+{"current_steps": 523, "total_steps": 540, "loss": 0.0103, "lr": 3.0159765977250673e-07, "epoch": 4.842592592592593, "percentage": 96.85, "elapsed_time": "2:38:39", "remaining_time": "0:05:09"}
+{"current_steps": 524, "total_steps": 540, "loss": 0.0084, "lr": 2.671898796699268e-07, "epoch": 4.851851851851852, "percentage": 97.04, "elapsed_time": "2:38:55", "remaining_time": "0:04:51"}
+{"current_steps": 525, "total_steps": 540, "loss": 0.0089, "lr": 2.3486021034170857e-07, "epoch": 4.861111111111111, "percentage": 97.22, "elapsed_time": "2:39:11", "remaining_time": "0:04:32"}
+{"current_steps": 525, "total_steps": 540, "eval_loss": 0.015216498635709286, "epoch": 4.861111111111111, "percentage": 97.22, "elapsed_time": "2:39:20", "remaining_time": "0:04:33"}
+{"current_steps": 526, "total_steps": 540, "loss": 0.0075, "lr": 2.0461000269953456e-07, "epoch": 4.87037037037037, "percentage": 97.41, "elapsed_time": "2:39:35", "remaining_time": "0:04:14"}
+{"current_steps": 527, "total_steps": 540, "loss": 0.0083, "lr": 1.7644052076371542e-07, "epoch": 4.87962962962963, "percentage": 97.59, "elapsed_time": "2:39:51", "remaining_time": "0:03:56"}
+{"current_steps": 528, "total_steps": 540, "loss": 0.009, "lr": 1.503529416103988e-07, "epoch": 4.888888888888889, "percentage": 97.78, "elapsed_time": "2:40:08", "remaining_time": "0:03:38"}
+{"current_steps": 529, "total_steps": 540, "loss": 0.0093, "lr": 1.2634835532233657e-07, "epoch": 4.898148148148148, "percentage": 97.96, "elapsed_time": "2:40:23", "remaining_time": "0:03:20"}
+{"current_steps": 530, "total_steps": 540, "loss": 0.0083, "lr": 1.044277649433989e-07, "epoch": 4.907407407407407, "percentage": 98.15, "elapsed_time": "2:40:40", "remaining_time": "0:03:01"}
+{"current_steps": 530, "total_steps": 540, "eval_loss": 0.015229844488203526, "epoch": 4.907407407407407, "percentage": 98.15, "elapsed_time": "2:40:49", "remaining_time": "0:03:02"}
+{"current_steps": 531, "total_steps": 540, "loss": 0.0084, "lr": 8.459208643659122e-08, "epoch": 4.916666666666667, "percentage": 98.33, "elapsed_time": "2:41:06", "remaining_time": "0:02:43"}
+{"current_steps": 532, "total_steps": 540, "loss": 0.009, "lr": 6.684214864584038e-08, "epoch": 4.925925925925926, "percentage": 98.52, "elapsed_time": "2:41:23", "remaining_time": "0:02:25"}
+{"current_steps": 533, "total_steps": 540, "loss": 0.0055, "lr": 5.11786932613223e-08, "epoch": 4.935185185185185, "percentage": 98.7, "elapsed_time": "2:41:37", "remaining_time": "0:02:07"}
+{"current_steps": 534, "total_steps": 540, "loss": 0.0093, "lr": 3.760237478849793e-08, "epoch": 4.944444444444445, "percentage": 98.89, "elapsed_time": "2:41:54", "remaining_time": "0:01:49"}
+{"current_steps": 535, "total_steps": 540, "loss": 0.0103, "lr": 2.6113760520735108e-08, "epoch": 4.953703703703704, "percentage": 99.07, "elapsed_time": "2:42:10", "remaining_time": "0:01:30"}
+{"current_steps": 535, "total_steps": 540, "eval_loss": 0.015256751328706741, "epoch": 4.953703703703704, "percentage": 99.07, "elapsed_time": "2:42:19", "remaining_time": "0:01:31"}
+{"current_steps": 536, "total_steps": 540, "loss": 0.0106, "lr": 1.6713330515627513e-08, "epoch": 4.962962962962963, "percentage": 99.26, "elapsed_time": "2:42:36", "remaining_time": "0:01:12"}
+{"current_steps": 537, "total_steps": 540, "loss": 0.0074, "lr": 9.401477574932926e-09, "epoch": 4.972222222222222, "percentage": 99.44, "elapsed_time": "2:42:51", "remaining_time": "0:00:54"}
+{"current_steps": 538, "total_steps": 540, "loss": 0.0082, "lr": 4.178507228136397e-09, "epoch": 4.981481481481482, "percentage": 99.63, "elapsed_time": "2:43:08", "remaining_time": "0:00:36"}
+{"current_steps": 539, "total_steps": 540, "loss": 0.0085, "lr": 1.0446377197104173e-09, "epoch": 4.9907407407407405, "percentage": 99.81, "elapsed_time": "2:43:24", "remaining_time": "0:00:18"}
+{"current_steps": 540, "total_steps": 540, "loss": 0.0092, "lr": 0.0, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "2:43:42", "remaining_time": "0:00:00"}
+{"current_steps": 540, "total_steps": 540, "eval_loss": 0.01526525616645813, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "2:43:51", "remaining_time": "0:00:00"}
+{"current_steps": 540, "total_steps": 540, "epoch": 5.0, "percentage": 100.0, "elapsed_time": "2:43:58", "remaining_time": "0:00:00"}