{"current_steps": 5, "total_steps": 3882, "loss": 3.0388, "lr": 2.564102564102564e-06, "epoch": 0.0025753283543651817, "percentage": 0.13, "elapsed_time": "0:02:49", "remaining_time": "1 day, 12:31:05", "throughput": 345.02, "total_tokens": 58496}
{"current_steps": 10, "total_steps": 3882, "loss": 2.9831, "lr": 5.128205128205128e-06, "epoch": 0.0051506567087303634, "percentage": 0.26, "elapsed_time": "0:04:19", "remaining_time": "1 day, 3:52:53", "throughput": 451.18, "total_tokens": 116960}
{"current_steps": 15, "total_steps": 3882, "loss": 2.8696, "lr": 7.692307692307694e-06, "epoch": 0.007725985063095545, "percentage": 0.39, "elapsed_time": "0:05:47", "remaining_time": "1 day, 0:54:38", "throughput": 504.36, "total_tokens": 175448}
{"current_steps": 20, "total_steps": 3882, "loss": 2.6316, "lr": 1.0256410256410256e-05, "epoch": 0.010301313417460727, "percentage": 0.52, "elapsed_time": "0:07:17", "remaining_time": "23:26:52", "throughput": 535.17, "total_tokens": 233944}
{"current_steps": 25, "total_steps": 3882, "loss": 1.9707, "lr": 1.282051282051282e-05, "epoch": 0.012876641771825908, "percentage": 0.64, "elapsed_time": "0:08:45", "remaining_time": "22:30:52", "throughput": 556.6, "total_tokens": 292416}
{"current_steps": 30, "total_steps": 3882, "loss": 1.3782, "lr": 1.5384615384615387e-05, "epoch": 0.01545197012619109, "percentage": 0.77, "elapsed_time": "0:10:14", "remaining_time": "21:55:33", "throughput": 570.81, "total_tokens": 350904}
{"current_steps": 35, "total_steps": 3882, "loss": 1.0628, "lr": 1.794871794871795e-05, "epoch": 0.018027298480556272, "percentage": 0.9, "elapsed_time": "0:11:42", "remaining_time": "21:27:25", "throughput": 582.52, "total_tokens": 409384}
{"current_steps": 40, "total_steps": 3882, "loss": 0.9789, "lr": 2.0512820512820512e-05, "epoch": 0.020602626834921454, "percentage": 1.03, "elapsed_time": "0:13:11", "remaining_time": "21:07:29", "throughput": 590.91, "total_tokens": 467864}
{"current_steps": 45, "total_steps": 3882, "loss": 0.9327, "lr": 2.307692307692308e-05, "epoch": 0.023177955189286635, "percentage": 1.16, "elapsed_time": "0:14:39", "remaining_time": "20:50:06", "throughput": 598.39, "total_tokens": 526384}
{"current_steps": 50, "total_steps": 3882, "loss": 0.9233, "lr": 2.564102564102564e-05, "epoch": 0.025753283543651816, "percentage": 1.29, "elapsed_time": "0:16:09", "remaining_time": "20:38:40", "throughput": 603.11, "total_tokens": 584856}
{"current_steps": 50, "total_steps": 3882, "eval_loss": 0.9281821846961975, "epoch": 0.025753283543651816, "percentage": 1.29, "elapsed_time": "0:16:57", "remaining_time": "21:40:17", "throughput": 574.53, "total_tokens": 584856}
{"current_steps": 55, "total_steps": 3882, "loss": 0.897, "lr": 2.8205128205128207e-05, "epoch": 0.028328611898016998, "percentage": 1.42, "elapsed_time": "0:18:34", "remaining_time": "21:32:15", "throughput": 577.35, "total_tokens": 643344}
{"current_steps": 60, "total_steps": 3882, "loss": 0.9169, "lr": 3.0769230769230774e-05, "epoch": 0.03090394025238218, "percentage": 1.55, "elapsed_time": "0:20:03", "remaining_time": "21:17:49", "throughput": 583.09, "total_tokens": 701808}
{"current_steps": 65, "total_steps": 3882, "loss": 0.9019, "lr": 3.3333333333333335e-05, "epoch": 0.03347926860674736, "percentage": 1.67, "elapsed_time": "0:21:32", "remaining_time": "21:04:47", "throughput": 588.34, "total_tokens": 760304}
{"current_steps": 70, "total_steps": 3882, "loss": 0.8996, "lr": 3.58974358974359e-05, "epoch": 0.036054596961112545, "percentage": 1.8, "elapsed_time": "0:23:02", "remaining_time": "20:55:11", "throughput": 592.04, "total_tokens": 818760}
{"current_steps": 75, "total_steps": 3882, "loss": 0.9073, "lr": 3.846153846153846e-05, "epoch": 0.03862992531547772, "percentage": 1.93, "elapsed_time": "0:24:32", "remaining_time": "20:46:08", "throughput": 595.56, "total_tokens": 877256}
{"current_steps": 80, "total_steps": 3882, "loss": 0.9081, "lr": 4.1025641025641023e-05, "epoch": 0.04120525366984291, "percentage": 2.06, "elapsed_time": "0:26:03", "remaining_time": "20:38:41", "throughput": 598.36, "total_tokens": 935752}
{"current_steps": 85, "total_steps": 3882, "loss": 0.906, "lr": 4.358974358974359e-05, "epoch": 0.043780582024208085, "percentage": 2.19, "elapsed_time": "0:27:32", "remaining_time": "20:30:19", "throughput": 601.63, "total_tokens": 994216}
{"current_steps": 90, "total_steps": 3882, "loss": 0.8952, "lr": 4.615384615384616e-05, "epoch": 0.04635591037857327, "percentage": 2.32, "elapsed_time": "0:29:02", "remaining_time": "20:23:48", "throughput": 604.04, "total_tokens": 1052704}
{"current_steps": 95, "total_steps": 3882, "loss": 0.8996, "lr": 4.871794871794872e-05, "epoch": 0.04893123873293845, "percentage": 2.45, "elapsed_time": "0:30:31", "remaining_time": "20:17:00", "throughput": 606.61, "total_tokens": 1111176}
{"current_steps": 100, "total_steps": 3882, "loss": 0.9024, "lr": 5.128205128205128e-05, "epoch": 0.05150656708730363, "percentage": 2.58, "elapsed_time": "0:32:01", "remaining_time": "20:10:52", "throughput": 608.88, "total_tokens": 1169664}
{"current_steps": 100, "total_steps": 3882, "eval_loss": 0.911374032497406, "epoch": 0.05150656708730363, "percentage": 2.58, "elapsed_time": "0:32:20", "remaining_time": "20:23:12", "throughput": 602.74, "total_tokens": 1169664}
{"current_steps": 105, "total_steps": 3882, "loss": 0.9142, "lr": 5.384615384615385e-05, "epoch": 0.05408189544166881, "percentage": 2.7, "elapsed_time": "0:33:57", "remaining_time": "20:21:47", "throughput": 602.63, "total_tokens": 1228112}
{"current_steps": 110, "total_steps": 3882, "loss": 0.9054, "lr": 5.6410256410256414e-05, "epoch": 0.056657223796033995, "percentage": 2.83, "elapsed_time": "0:35:26", "remaining_time": "20:15:28", "throughput": 604.96, "total_tokens": 1286608}
{"current_steps": 115, "total_steps": 3882, "loss": 0.8997, "lr": 5.897435897435898e-05, "epoch": 0.05923255215039917, "percentage": 2.96, "elapsed_time": "0:36:56", "remaining_time": "20:10:10", "throughput": 606.8, "total_tokens": 1345072}
{"current_steps": 120, "total_steps": 3882, "loss": 0.8988, "lr": 6.153846153846155e-05, "epoch": 0.06180788050476436, "percentage": 3.09, "elapsed_time": "0:38:25", "remaining_time": "20:04:43", "throughput": 608.73, "total_tokens": 1403544}
{"current_steps": 125, "total_steps": 3882, "loss": 0.9087, "lr": 6.410256410256412e-05, "epoch": 0.06438320885912954, "percentage": 3.22, "elapsed_time": "0:39:56", "remaining_time": "20:00:26", "throughput": 610.09, "total_tokens": 1462024}
{"current_steps": 130, "total_steps": 3882, "loss": 0.8961, "lr": 6.666666666666667e-05, "epoch": 0.06695853721349472, "percentage": 3.35, "elapsed_time": "0:41:25", "remaining_time": "19:55:28", "throughput": 611.82, "total_tokens": 1520528}
{"current_steps": 135, "total_steps": 3882, "loss": 0.903, "lr": 6.923076923076924e-05, "epoch": 0.0695338655678599, "percentage": 3.48, "elapsed_time": "0:42:56", "remaining_time": "19:51:42", "throughput": 612.94, "total_tokens": 1579024}
{"current_steps": 140, "total_steps": 3882, "loss": 0.899, "lr": 7.17948717948718e-05, "epoch": 0.07210919392222509, "percentage": 3.61, "elapsed_time": "0:44:25", "remaining_time": "19:47:14", "throughput": 614.42, "total_tokens": 1637504}
{"current_steps": 145, "total_steps": 3882, "loss": 0.9071, "lr": 7.435897435897436e-05, "epoch": 0.07468452227659027, "percentage": 3.74, "elapsed_time": "0:45:55", "remaining_time": "19:43:28", "throughput": 615.57, "total_tokens": 1696024}
{"current_steps": 150, "total_steps": 3882, "loss": 0.9045, "lr": 7.692307692307693e-05, "epoch": 0.07725985063095545, "percentage": 3.86, "elapsed_time": "0:47:24", "remaining_time": "19:39:19", "throughput": 616.91, "total_tokens": 1754512}
{"current_steps": 150, "total_steps": 3882, "eval_loss": 0.8934853076934814, "epoch": 0.07725985063095545, "percentage": 3.86, "elapsed_time": "0:47:43", "remaining_time": "19:47:33", "throughput": 612.63, "total_tokens": 1754512}
{"current_steps": 155, "total_steps": 3882, "loss": 0.8992, "lr": 7.948717948717948e-05, "epoch": 0.07983517898532062, "percentage": 3.99, "elapsed_time": "0:49:20", "remaining_time": "19:46:21", "throughput": 612.43, "total_tokens": 1812976}
{"current_steps": 160, "total_steps": 3882, "loss": 0.8965, "lr": 8.205128205128205e-05, "epoch": 0.08241050733968582, "percentage": 4.12, "elapsed_time": "0:50:48", "remaining_time": "19:42:01", "throughput": 613.85, "total_tokens": 1871464}
{"current_steps": 165, "total_steps": 3882, "loss": 0.9094, "lr": 8.461538461538461e-05, "epoch": 0.08498583569405099, "percentage": 4.25, "elapsed_time": "0:52:18", "remaining_time": "19:38:32", "throughput": 614.83, "total_tokens": 1929928}
{"current_steps": 170, "total_steps": 3882, "loss": 0.903, "lr": 8.717948717948718e-05, "epoch": 0.08756116404841617, "percentage": 4.38, "elapsed_time": "0:53:47", "remaining_time": "19:34:27", "throughput": 616.14, "total_tokens": 1988432}
{"current_steps": 175, "total_steps": 3882, "loss": 0.902, "lr": 8.974358974358975e-05, "epoch": 0.09013649240278135, "percentage": 4.51, "elapsed_time": "0:55:16", "remaining_time": "19:30:56", "throughput": 617.16, "total_tokens": 2046920}
{"current_steps": 180, "total_steps": 3882, "loss": 0.9006, "lr": 9.230769230769232e-05, "epoch": 0.09271182075714654, "percentage": 4.64, "elapsed_time": "0:56:45", "remaining_time": "19:27:28", "throughput": 618.16, "total_tokens": 2105392}
{"current_steps": 185, "total_steps": 3882, "loss": 0.9042, "lr": 9.487179487179487e-05, "epoch": 0.09528714911151172, "percentage": 4.77, "elapsed_time": "0:58:15", "remaining_time": "19:24:13", "throughput": 619.04, "total_tokens": 2163872}
{"current_steps": 190, "total_steps": 3882, "loss": 0.9096, "lr": 9.743589743589744e-05, "epoch": 0.0978624774658769, "percentage": 4.89, "elapsed_time": "0:59:43", "remaining_time": "19:20:39", "throughput": 620.11, "total_tokens": 2222352}
{"current_steps": 195, "total_steps": 3882, "loss": 0.9037, "lr": 0.0001, "epoch": 0.10043780582024209, "percentage": 5.02, "elapsed_time": "1:01:11", "remaining_time": "19:17:04", "throughput": 621.18, "total_tokens": 2280800}
{"current_steps": 200, "total_steps": 3882, "loss": 0.904, "lr": 9.999954623308172e-05, "epoch": 0.10301313417460727, "percentage": 5.15, "elapsed_time": "1:02:41", "remaining_time": "19:14:00", "throughput": 621.98, "total_tokens": 2339304}
{"current_steps": 200, "total_steps": 3882, "eval_loss": 0.8980139493942261, "epoch": 0.10301313417460727, "percentage": 5.15, "elapsed_time": "1:03:00", "remaining_time": "19:19:56", "throughput": 618.8, "total_tokens": 2339304}
{"current_steps": 205, "total_steps": 3882, "loss": 0.9029, "lr": 9.999818494056303e-05, "epoch": 0.10558846252897244, "percentage": 5.28, "elapsed_time": "1:04:33", "remaining_time": "19:17:54", "throughput": 619.05, "total_tokens": 2397808}
{"current_steps": 210, "total_steps": 3882, "loss": 0.9005, "lr": 9.99959161471523e-05, "epoch": 0.10816379088333762, "percentage": 5.41, "elapsed_time": "1:06:02", "remaining_time": "19:14:52", "throughput": 619.83, "total_tokens": 2456288}
{"current_steps": 215, "total_steps": 3882, "loss": 0.9096, "lr": 9.99927398940297e-05, "epoch": 0.11073911923770281, "percentage": 5.54, "elapsed_time": "1:07:30", "remaining_time": "19:11:25", "throughput": 620.85, "total_tokens": 2514784}
{"current_steps": 220, "total_steps": 3882, "loss": 0.9036, "lr": 9.998865623884635e-05, "epoch": 0.11331444759206799, "percentage": 5.67, "elapsed_time": "1:08:59", "remaining_time": "19:08:31", "throughput": 621.56, "total_tokens": 2573240}
{"current_steps": 225, "total_steps": 3882, "loss": 0.901, "lr": 9.998366525572336e-05, "epoch": 0.11588977594643317, "percentage": 5.8, "elapsed_time": "1:10:27", "remaining_time": "19:05:14", "throughput": 622.48, "total_tokens": 2631672}
{"current_steps": 230, "total_steps": 3882, "loss": 0.9018, "lr": 9.997776703525046e-05, "epoch": 0.11846510430079835, "percentage": 5.92, "elapsed_time": "1:11:56", "remaining_time": "19:02:26", "throughput": 623.14, "total_tokens": 2690112}
{"current_steps": 235, "total_steps": 3882, "loss": 0.8934, "lr": 9.997096168448432e-05, "epoch": 0.12104043265516354, "percentage": 6.05, "elapsed_time": "1:13:24", "remaining_time": "18:59:11", "throughput": 624.07, "total_tokens": 2748608}
{"current_steps": 240, "total_steps": 3882, "loss": 0.8876, "lr": 9.996324932694668e-05, "epoch": 0.12361576100952872, "percentage": 6.18, "elapsed_time": "1:14:53", "remaining_time": "18:56:31", "throughput": 624.67, "total_tokens": 2807080}
{"current_steps": 245, "total_steps": 3882, "loss": 0.9084, "lr": 9.995463010262206e-05, "epoch": 0.1261910893638939, "percentage": 6.31, "elapsed_time": "1:16:21", "remaining_time": "18:53:29", "throughput": 625.48, "total_tokens": 2865520}
{"current_steps": 250, "total_steps": 3882, "loss": 0.9106, "lr": 9.994510416795519e-05, "epoch": 0.12876641771825909, "percentage": 6.44, "elapsed_time": "1:17:50", "remaining_time": "18:50:56", "throughput": 626.03, "total_tokens": 2924016}
{"current_steps": 250, "total_steps": 3882, "eval_loss": 0.8958488702774048, "epoch": 0.12876641771825909, "percentage": 6.44, "elapsed_time": "1:18:10", "remaining_time": "18:55:39", "throughput": 623.43, "total_tokens": 2924016}
{"current_steps": 255, "total_steps": 3882, "loss": 0.9012, "lr": 9.993467169584824e-05, "epoch": 0.13134174607262425, "percentage": 6.57, "elapsed_time": "1:19:44", "remaining_time": "18:54:06", "throughput": 623.43, "total_tokens": 2982520}
{"current_steps": 260, "total_steps": 3882, "loss": 0.9069, "lr": 9.992333287565765e-05, "epoch": 0.13391707442698944, "percentage": 6.7, "elapsed_time": "1:21:14", "remaining_time": "18:51:41", "throughput": 623.9, "total_tokens": 3041008}
{"current_steps": 265, "total_steps": 3882, "loss": 0.8918, "lr": 9.991108791319066e-05, "epoch": 0.13649240278135463, "percentage": 6.83, "elapsed_time": "1:22:42", "remaining_time": "18:48:57", "throughput": 624.54, "total_tokens": 3099464}
{"current_steps": 270, "total_steps": 3882, "loss": 0.8996, "lr": 9.989793703070163e-05, "epoch": 0.1390677311357198, "percentage": 6.96, "elapsed_time": "1:24:13", "remaining_time": "18:46:38", "throughput": 624.96, "total_tokens": 3157944}
{"current_steps": 275, "total_steps": 3882, "loss": 0.9009, "lr": 9.988388046688799e-05, "epoch": 0.141643059490085, "percentage": 7.08, "elapsed_time": "1:25:41", "remaining_time": "18:44:03", "throughput": 625.53, "total_tokens": 3216448}
{"current_steps": 280, "total_steps": 3882, "loss": 0.9059, "lr": 9.986891847688587e-05, "epoch": 0.14421838784445018, "percentage": 7.21, "elapsed_time": "1:27:11", "remaining_time": "18:41:42", "throughput": 625.97, "total_tokens": 3274928}
{"current_steps": 285, "total_steps": 3882, "loss": 0.8939, "lr": 9.985305133226553e-05, "epoch": 0.14679371619881534, "percentage": 7.34, "elapsed_time": "1:28:41", "remaining_time": "18:39:25", "throughput": 626.38, "total_tokens": 3333408}
{"current_steps": 290, "total_steps": 3882, "loss": 0.8899, "lr": 9.983627932102638e-05, "epoch": 0.14936904455318054, "percentage": 7.47, "elapsed_time": "1:30:10", "remaining_time": "18:37:01", "throughput": 626.86, "total_tokens": 3391896}
{"current_steps": 295, "total_steps": 3882, "loss": 0.9092, "lr": 9.981860274759173e-05, "epoch": 0.1519443729075457, "percentage": 7.6, "elapsed_time": "1:31:41", "remaining_time": "18:34:51", "throughput": 627.2, "total_tokens": 3450392}
{"current_steps": 300, "total_steps": 3882, "loss": 0.8901, "lr": 9.980002193280342e-05, "epoch": 0.1545197012619109, "percentage": 7.73, "elapsed_time": "1:33:10", "remaining_time": "18:32:25", "throughput": 627.7, "total_tokens": 3508888}
{"current_steps": 300, "total_steps": 3882, "eval_loss": 0.8932263255119324, "epoch": 0.1545197012619109, "percentage": 7.73, "elapsed_time": "1:33:29", "remaining_time": "18:36:21", "throughput": 625.48, "total_tokens": 3508888}
{"current_steps": 305, "total_steps": 3882, "loss": 0.9042, "lr": 9.978053721391578e-05, "epoch": 0.15709502961627608, "percentage": 7.86, "elapsed_time": "1:35:05", "remaining_time": "18:35:12", "throughput": 625.26, "total_tokens": 3567368}
{"current_steps": 310, "total_steps": 3882, "loss": 0.9007, "lr": 9.976014894458963e-05, "epoch": 0.15967035797064125, "percentage": 7.99, "elapsed_time": "1:36:33", "remaining_time": "18:32:40", "throughput": 625.81, "total_tokens": 3625848}
{"current_steps": 315, "total_steps": 3882, "loss": 0.9036, "lr": 9.973885749488589e-05, "epoch": 0.16224568632500644, "percentage": 8.11, "elapsed_time": "1:38:03", "remaining_time": "18:30:22", "throughput": 626.22, "total_tokens": 3684336}
{"current_steps": 320, "total_steps": 3882, "loss": 0.8936, "lr": 9.971666325125874e-05, "epoch": 0.16482101467937163, "percentage": 8.24, "elapsed_time": "1:39:30", "remaining_time": "18:27:44", "throughput": 626.83, "total_tokens": 3742800}
{"current_steps": 325, "total_steps": 3882, "loss": 0.8989, "lr": 9.969356661654876e-05, "epoch": 0.1673963430337368, "percentage": 8.37, "elapsed_time": "1:41:00", "remaining_time": "18:25:29", "throughput": 627.23, "total_tokens": 3801280}
{"current_steps": 330, "total_steps": 3882, "loss": 0.8983, "lr": 9.966956800997546e-05, "epoch": 0.16997167138810199, "percentage": 8.5, "elapsed_time": "1:42:28", "remaining_time": "18:22:55", "throughput": 627.81, "total_tokens": 3859792}
{"current_steps": 335, "total_steps": 3882, "loss": 0.9038, "lr": 9.964466786712984e-05, "epoch": 0.17254699974246718, "percentage": 8.63, "elapsed_time": "1:43:57", "remaining_time": "18:20:45", "throughput": 628.16, "total_tokens": 3918272}
{"current_steps": 340, "total_steps": 3882, "loss": 0.8947, "lr": 9.961886663996629e-05, "epoch": 0.17512232809683234, "percentage": 8.76, "elapsed_time": "1:45:26", "remaining_time": "18:18:25", "throughput": 628.61, "total_tokens": 3976760}
{"current_steps": 345, "total_steps": 3882, "loss": 0.9179, "lr": 9.959216479679458e-05, "epoch": 0.17769765645119753, "percentage": 8.89, "elapsed_time": "1:46:56", "remaining_time": "18:16:25", "throughput": 628.86, "total_tokens": 4035240}
{"current_steps": 350, "total_steps": 3882, "loss": 0.9059, "lr": 9.956456282227122e-05, "epoch": 0.1802729848055627, "percentage": 9.02, "elapsed_time": "1:48:25", "remaining_time": "18:14:06", "throughput": 629.29, "total_tokens": 4093688}
{"current_steps": 350, "total_steps": 3882, "eval_loss": 0.8960411548614502, "epoch": 0.1802729848055627, "percentage": 9.02, "elapsed_time": "1:48:45", "remaining_time": "18:17:29", "throughput": 627.36, "total_tokens": 4093688}
{"current_steps": 355, "total_steps": 3882, "loss": 0.8795, "lr": 9.953606121739074e-05, "epoch": 0.1828483131599279, "percentage": 9.14, "elapsed_time": "1:50:19", "remaining_time": "18:16:10", "throughput": 627.22, "total_tokens": 4152160}
{"current_steps": 360, "total_steps": 3882, "loss": 0.9143, "lr": 9.950666049947653e-05, "epoch": 0.18542364151429308, "percentage": 9.27, "elapsed_time": "1:51:48", "remaining_time": "18:13:48", "throughput": 627.69, "total_tokens": 4210648}
{"current_steps": 365, "total_steps": 3882, "loss": 0.9164, "lr": 9.947636120217155e-05, "epoch": 0.18799896986865824, "percentage": 9.4, "elapsed_time": "1:53:17", "remaining_time": "18:11:38", "throughput": 628.04, "total_tokens": 4269136}
{"current_steps": 370, "total_steps": 3882, "loss": 0.9061, "lr": 9.944516387542852e-05, "epoch": 0.19057429822302344, "percentage": 9.53, "elapsed_time": "1:54:46", "remaining_time": "18:09:27", "throughput": 628.41, "total_tokens": 4327664}
{"current_steps": 375, "total_steps": 3882, "loss": 0.8873, "lr": 9.941306908550005e-05, "epoch": 0.19314962657738863, "percentage": 9.66, "elapsed_time": "1:56:15", "remaining_time": "18:07:13", "throughput": 628.81, "total_tokens": 4386120}
{"current_steps": 380, "total_steps": 3882, "loss": 0.9038, "lr": 9.938007741492828e-05, "epoch": 0.1957249549317538, "percentage": 9.79, "elapsed_time": "1:57:45", "remaining_time": "18:05:10", "throughput": 629.08, "total_tokens": 4444560}
{"current_steps": 385, "total_steps": 3882, "loss": 0.9116, "lr": 9.934618946253437e-05, "epoch": 0.19830028328611898, "percentage": 9.92, "elapsed_time": "1:59:14", "remaining_time": "18:03:01", "throughput": 629.43, "total_tokens": 4503016}
{"current_steps": 390, "total_steps": 3882, "loss": 0.9023, "lr": 9.931140584340761e-05, "epoch": 0.20087561164048418, "percentage": 10.05, "elapsed_time": "2:00:43", "remaining_time": "18:00:58", "throughput": 629.72, "total_tokens": 4561496}
{"current_steps": 395, "total_steps": 3882, "loss": 0.8901, "lr": 9.92757271888942e-05, "epoch": 0.20345093999484934, "percentage": 10.18, "elapsed_time": "2:02:11", "remaining_time": "17:58:43", "throughput": 630.12, "total_tokens": 4619944}
{"current_steps": 400, "total_steps": 3882, "loss": 0.9033, "lr": 9.923915414658587e-05, "epoch": 0.20602626834921453, "percentage": 10.3, "elapsed_time": "2:03:41", "remaining_time": "17:56:44", "throughput": 630.38, "total_tokens": 4678384}
{"current_steps": 400, "total_steps": 3882, "eval_loss": 0.906301736831665, "epoch": 0.20602626834921453, "percentage": 10.3, "elapsed_time": "2:04:01", "remaining_time": "17:59:36", "throughput": 628.7, "total_tokens": 4678384}
{"current_steps": 405, "total_steps": 3882, "loss": 0.8951, "lr": 9.920168738030807e-05, "epoch": 0.2086015967035797, "percentage": 10.43, "elapsed_time": "2:05:35", "remaining_time": "17:58:10", "throughput": 628.64, "total_tokens": 4736904}
{"current_steps": 410, "total_steps": 3882, "loss": 0.9131, "lr": 9.916332757010799e-05, "epoch": 0.2111769250579449, "percentage": 10.56, "elapsed_time": "2:07:05", "remaining_time": "17:56:11", "throughput": 628.9, "total_tokens": 4795376}
{"current_steps": 415, "total_steps": 3882, "loss": 0.8923, "lr": 9.912407541224213e-05, "epoch": 0.21375225341231008, "percentage": 10.69, "elapsed_time": "2:08:33", "remaining_time": "17:54:00", "throughput": 629.27, "total_tokens": 4853880}
{"current_steps": 420, "total_steps": 3882, "loss": 0.9026, "lr": 9.908393161916374e-05, "epoch": 0.21632758176667524, "percentage": 10.82, "elapsed_time": "2:10:07", "remaining_time": "17:52:35", "throughput": 629.19, "total_tokens": 4912360}
{"current_steps": 425, "total_steps": 3882, "loss": 0.905, "lr": 9.904289691950979e-05, "epoch": 0.21890291012104043, "percentage": 10.95, "elapsed_time": "2:11:39", "remaining_time": "17:50:54", "throughput": 629.27, "total_tokens": 4970872}
{"current_steps": 430, "total_steps": 3882, "loss": 0.8941, "lr": 9.900097205808789e-05, "epoch": 0.22147823847540563, "percentage": 11.08, "elapsed_time": "2:13:11", "remaining_time": "17:49:11", "throughput": 629.36, "total_tokens": 5029304}
{"current_steps": 435, "total_steps": 3882, "loss": 0.9031, "lr": 9.895815779586262e-05, "epoch": 0.2240535668297708, "percentage": 11.21, "elapsed_time": "2:14:40", "remaining_time": "17:47:10", "throughput": 629.65, "total_tokens": 5087800}
{"current_steps": 440, "total_steps": 3882, "loss": 0.8975, "lr": 9.891445490994182e-05, "epoch": 0.22662889518413598, "percentage": 11.33, "elapsed_time": "2:16:12", "remaining_time": "17:45:27", "throughput": 629.74, "total_tokens": 5146312}
{"current_steps": 445, "total_steps": 3882, "loss": 0.901, "lr": 9.886986419356246e-05, "epoch": 0.22920422353850115, "percentage": 11.46, "elapsed_time": "2:17:43", "remaining_time": "17:43:40", "throughput": 629.89, "total_tokens": 5204800}
{"current_steps": 450, "total_steps": 3882, "loss": 0.9062, "lr": 9.88243864560762e-05, "epoch": 0.23177955189286634, "percentage": 11.59, "elapsed_time": "2:19:17", "remaining_time": "17:42:16", "throughput": 629.8, "total_tokens": 5263304}
{"current_steps": 450, "total_steps": 3882, "eval_loss": 0.9007609486579895, "epoch": 0.23177955189286634, "percentage": 11.59, "elapsed_time": "2:19:37", "remaining_time": "17:44:55", "throughput": 628.23, "total_tokens": 5263304}
{"current_steps": 455, "total_steps": 3882, "loss": 0.8891, "lr": 9.877802252293474e-05, "epoch": 0.23435488024723153, "percentage": 11.72, "elapsed_time": "2:21:18", "remaining_time": "17:44:17", "throughput": 627.69, "total_tokens": 5321760}
{"current_steps": 460, "total_steps": 3882, "loss": 0.9026, "lr": 9.873077323567488e-05, "epoch": 0.2369302086015967, "percentage": 11.85, "elapsed_time": "2:22:49", "remaining_time": "17:42:31", "throughput": 627.82, "total_tokens": 5380224}
{"current_steps": 465, "total_steps": 3882, "loss": 0.8707, "lr": 9.868263945190312e-05, "epoch": 0.23950553695596188, "percentage": 11.98, "elapsed_time": "2:24:21", "remaining_time": "17:40:47", "throughput": 627.93, "total_tokens": 5438704}
{"current_steps": 470, "total_steps": 3882, "loss": 0.9051, "lr": 9.863362204528024e-05, "epoch": 0.24208086531032708, "percentage": 12.11, "elapsed_time": "2:25:54", "remaining_time": "17:39:10", "throughput": 627.96, "total_tokens": 5497208}
{"current_steps": 475, "total_steps": 3882, "loss": 0.8711, "lr": 9.858372190550533e-05, "epoch": 0.24465619366469224, "percentage": 12.24, "elapsed_time": "2:27:28", "remaining_time": "17:37:43", "throughput": 627.9, "total_tokens": 5555704}
{"current_steps": 480, "total_steps": 3882, "loss": 0.885, "lr": 9.853293993829969e-05, "epoch": 0.24723152201905743, "percentage": 12.36, "elapsed_time": "2:29:00", "remaining_time": "17:36:05", "throughput": 627.95, "total_tokens": 5614160}
{"current_steps": 485, "total_steps": 3882, "loss": 0.8615, "lr": 9.848127706539039e-05, "epoch": 0.24980685037342262, "percentage": 12.49, "elapsed_time": "2:30:31", "remaining_time": "17:34:16", "throughput": 628.11, "total_tokens": 5672640}
{"current_steps": 490, "total_steps": 3882, "loss": 0.9057, "lr": 9.842873422449354e-05, "epoch": 0.2523821787277878, "percentage": 12.62, "elapsed_time": "2:32:01", "remaining_time": "17:32:21", "throughput": 628.32, "total_tokens": 5731072}
{"current_steps": 495, "total_steps": 3882, "loss": 0.8818, "lr": 9.837531236929726e-05, "epoch": 0.254957507082153, "percentage": 12.75, "elapsed_time": "2:33:31", "remaining_time": "17:30:30", "throughput": 628.5, "total_tokens": 5789544}
{"current_steps": 500, "total_steps": 3882, "loss": 0.8576, "lr": 9.832101246944439e-05, "epoch": 0.25753283543651817, "percentage": 12.88, "elapsed_time": "2:35:02", "remaining_time": "17:28:38", "throughput": 628.68, "total_tokens": 5848048}
{"current_steps": 500, "total_steps": 3882, "eval_loss": 0.8268976211547852, "epoch": 0.25753283543651817, "percentage": 12.88, "elapsed_time": "2:35:21", "remaining_time": "17:30:51", "throughput": 627.36, "total_tokens": 5848048}
{"current_steps": 505, "total_steps": 3882, "loss": 0.8566, "lr": 9.826583551051483e-05, "epoch": 0.26010816379088336, "percentage": 13.01, "elapsed_time": "2:36:58", "remaining_time": "17:29:43", "throughput": 627.12, "total_tokens": 5906512}
{"current_steps": 510, "total_steps": 3882, "loss": 0.8365, "lr": 9.820978249400773e-05, "epoch": 0.2626834921452485, "percentage": 13.14, "elapsed_time": "2:38:27", "remaining_time": "17:27:38", "throughput": 627.43, "total_tokens": 5965024}
{"current_steps": 515, "total_steps": 3882, "loss": 0.8882, "lr": 9.81528544373233e-05, "epoch": 0.2652588204996137, "percentage": 13.27, "elapsed_time": "2:39:56", "remaining_time": "17:25:40", "throughput": 627.67, "total_tokens": 6023496}
{"current_steps": 520, "total_steps": 3882, "loss": 0.8799, "lr": 9.809505237374426e-05, "epoch": 0.2678341488539789, "percentage": 13.4, "elapsed_time": "2:41:25", "remaining_time": "17:23:40", "throughput": 627.94, "total_tokens": 6082000}
{"current_steps": 525, "total_steps": 3882, "loss": 0.8758, "lr": 9.80363773524172e-05, "epoch": 0.2704094772083441, "percentage": 13.52, "elapsed_time": "2:42:59", "remaining_time": "17:22:15", "throughput": 627.87, "total_tokens": 6140480}
{"current_steps": 530, "total_steps": 3882, "loss": 0.8644, "lr": 9.797683043833345e-05, "epoch": 0.27298480556270927, "percentage": 13.65, "elapsed_time": "2:44:32", "remaining_time": "17:20:39", "throughput": 627.89, "total_tokens": 6198968}
{"current_steps": 535, "total_steps": 3882, "loss": 0.8453, "lr": 9.791641271230982e-05, "epoch": 0.2755601339170744, "percentage": 13.78, "elapsed_time": "2:46:06", "remaining_time": "17:19:13", "throughput": 627.82, "total_tokens": 6257464}
{"current_steps": 540, "total_steps": 3882, "loss": 0.8533, "lr": 9.78551252709689e-05, "epoch": 0.2781354622714396, "percentage": 13.91, "elapsed_time": "2:47:37", "remaining_time": "17:17:24", "throughput": 627.98, "total_tokens": 6315944}
{"current_steps": 545, "total_steps": 3882, "loss": 0.8575, "lr": 9.779296922671923e-05, "epoch": 0.2807107906258048, "percentage": 14.04, "elapsed_time": "2:49:07", "remaining_time": "17:15:32", "throughput": 628.18, "total_tokens": 6374408}
{"current_steps": 550, "total_steps": 3882, "loss": 0.8666, "lr": 9.77299457077351e-05, "epoch": 0.28328611898017, "percentage": 14.17, "elapsed_time": "2:50:37", "remaining_time": "17:13:38", "throughput": 628.39, "total_tokens": 6432936}
{"current_steps": 550, "total_steps": 3882, "eval_loss": 0.7909801602363586, "epoch": 0.28328611898017, "percentage": 14.17, "elapsed_time": "2:50:57", "remaining_time": "17:15:38", "throughput": 627.17, "total_tokens": 6432936}
{"current_steps": 555, "total_steps": 3882, "loss": 0.8264, "lr": 9.7666055857936e-05, "epoch": 0.28586144733453517, "percentage": 14.3, "elapsed_time": "2:52:32", "remaining_time": "17:14:16", "throughput": 627.06, "total_tokens": 6491400}
{"current_steps": 560, "total_steps": 3882, "loss": 0.8456, "lr": 9.760130083696595e-05, "epoch": 0.28843677568890036, "percentage": 14.43, "elapsed_time": "2:54:01", "remaining_time": "17:12:22", "throughput": 627.27, "total_tokens": 6549872}
{"current_steps": 565, "total_steps": 3882, "loss": 0.8032, "lr": 9.75356818201724e-05, "epoch": 0.2910121040432655, "percentage": 14.55, "elapsed_time": "2:55:30", "remaining_time": "17:10:19", "throughput": 627.57, "total_tokens": 6608296}
{"current_steps": 570, "total_steps": 3882, "loss": 0.8081, "lr": 9.746919999858492e-05, "epoch": 0.2935874323976307, "percentage": 14.68, "elapsed_time": "2:57:00", "remaining_time": "17:08:28", "throughput": 627.75, "total_tokens": 6666768}
{"current_steps": 575, "total_steps": 3882, "loss": 0.8398, "lr": 9.740185657889357e-05, "epoch": 0.2961627607519959, "percentage": 14.81, "elapsed_time": "2:58:28", "remaining_time": "17:06:28", "throughput": 628.02, "total_tokens": 6725248}
{"current_steps": 580, "total_steps": 3882, "loss": 0.8908, "lr": 9.733365278342696e-05, "epoch": 0.29873808910636107, "percentage": 14.94, "elapsed_time": "2:59:58", "remaining_time": "17:04:37", "throughput": 628.2, "total_tokens": 6783680}
{"current_steps": 585, "total_steps": 3882, "loss": 0.8248, "lr": 9.726458985013017e-05, "epoch": 0.30131341746072626, "percentage": 15.07, "elapsed_time": "3:01:26", "remaining_time": "17:02:35", "throughput": 628.49, "total_tokens": 6842144}
{"current_steps": 590, "total_steps": 3882, "loss": 0.829, "lr": 9.719466903254215e-05, "epoch": 0.3038887458150914, "percentage": 15.2, "elapsed_time": "3:02:56", "remaining_time": "17:00:43", "throughput": 628.69, "total_tokens": 6900656}
{"current_steps": 595, "total_steps": 3882, "loss": 0.8269, "lr": 9.712389159977307e-05, "epoch": 0.3064640741694566, "percentage": 15.33, "elapsed_time": "3:04:24", "remaining_time": "16:58:45", "throughput": 628.95, "total_tokens": 6959128}
{"current_steps": 600, "total_steps": 3882, "loss": 0.7997, "lr": 9.705225883648121e-05, "epoch": 0.3090394025238218, "percentage": 15.46, "elapsed_time": "3:05:54", "remaining_time": "16:56:56", "throughput": 629.11, "total_tokens": 7017576}
{"current_steps": 600, "total_steps": 3882, "eval_loss": 0.787663459777832, "epoch": 0.3090394025238218, "percentage": 15.46, "elapsed_time": "3:06:14", "remaining_time": "16:58:44", "throughput": 628.0, "total_tokens": 7017576}
{"current_steps": 605, "total_steps": 3882, "loss": 0.8925, "lr": 9.697977204284973e-05, "epoch": 0.311614730878187, "percentage": 15.58, "elapsed_time": "3:07:48", "remaining_time": "16:57:17", "throughput": 627.94, "total_tokens": 7076032}
{"current_steps": 610, "total_steps": 3882, "loss": 0.8159, "lr": 9.690643253456297e-05, "epoch": 0.31419005923255217, "percentage": 15.71, "elapsed_time": "3:09:18", "remaining_time": "16:55:24", "throughput": 628.15, "total_tokens": 7134536}
{"current_steps": 615, "total_steps": 3882, "loss": 0.826, "lr": 9.683224164278264e-05, "epoch": 0.31676538758691736, "percentage": 15.84, "elapsed_time": "3:10:46", "remaining_time": "16:53:28", "throughput": 628.38, "total_tokens": 7193032}
{"current_steps": 620, "total_steps": 3882, "loss": 0.8187, "lr": 9.675720071412365e-05, "epoch": 0.3193407159412825, "percentage": 15.97, "elapsed_time": "3:12:18", "remaining_time": "16:51:46", "throughput": 628.48, "total_tokens": 7251568}
{"current_steps": 625, "total_steps": 3882, "loss": 0.8524, "lr": 9.66813111106296e-05, "epoch": 0.3219160442956477, "percentage": 16.1, "elapsed_time": "3:13:47", "remaining_time": "16:49:51", "throughput": 628.7, "total_tokens": 7310072}
{"current_steps": 630, "total_steps": 3882, "loss": 0.7966, "lr": 9.660457420974819e-05, "epoch": 0.3244913726500129, "percentage": 16.23, "elapsed_time": "3:15:15", "remaining_time": "16:47:55", "throughput": 628.95, "total_tokens": 7368560}
{"current_steps": 635, "total_steps": 3882, "loss": 0.799, "lr": 9.652699140430608e-05, "epoch": 0.32706670100437807, "percentage": 16.36, "elapsed_time": "3:16:45", "remaining_time": "16:46:03", "throughput": 629.14, "total_tokens": 7427040}
{"current_steps": 640, "total_steps": 3882, "loss": 0.8477, "lr": 9.644856410248369e-05, "epoch": 0.32964202935874326, "percentage": 16.49, "elapsed_time": "3:18:13", "remaining_time": "16:44:09", "throughput": 629.37, "total_tokens": 7485552}
{"current_steps": 645, "total_steps": 3882, "loss": 0.7867, "lr": 9.636929372778963e-05, "epoch": 0.3322173577131084, "percentage": 16.62, "elapsed_time": "3:19:43", "remaining_time": "16:42:18", "throughput": 629.56, "total_tokens": 7544040}
{"current_steps": 650, "total_steps": 3882, "loss": 0.8367, "lr": 9.628918171903485e-05, "epoch": 0.3347926860674736, "percentage": 16.74, "elapsed_time": "3:21:11", "remaining_time": "16:40:22", "throughput": 629.8, "total_tokens": 7602512}
{"current_steps": 650, "total_steps": 3882, "eval_loss": 0.7940558791160583, "epoch": 0.3347926860674736, "percentage": 16.74, "elapsed_time": "3:21:31", "remaining_time": "16:42:00", "throughput": 628.77, "total_tokens": 7602512}
{"current_steps": 655, "total_steps": 3882, "loss": 0.8131, "lr": 9.620822953030652e-05, "epoch": 0.3373680144218388, "percentage": 16.87, "elapsed_time": "3:23:05", "remaining_time": "16:40:34", "throughput": 628.7, "total_tokens": 7660968}
{"current_steps": 660, "total_steps": 3882, "loss": 0.8348, "lr": 9.612643863094163e-05, "epoch": 0.33994334277620397, "percentage": 17.0, "elapsed_time": "3:24:34", "remaining_time": "16:38:40", "throughput": 628.92, "total_tokens": 7719448}
{"current_steps": 665, "total_steps": 3882, "loss": 0.8289, "lr": 9.604381050550038e-05, "epoch": 0.34251867113056916, "percentage": 17.13, "elapsed_time": "3:26:03", "remaining_time": "16:36:50", "throughput": 629.1, "total_tokens": 7777928}
{"current_steps": 670, "total_steps": 3882, "loss": 0.7758, "lr": 9.596034665373916e-05, "epoch": 0.34509399948493436, "percentage": 17.26, "elapsed_time": "3:27:31", "remaining_time": "16:34:53", "throughput": 629.35, "total_tokens": 7836424}
{"current_steps": 675, "total_steps": 3882, "loss": 0.8189, "lr": 9.587604859058334e-05, "epoch": 0.3476693278392995, "percentage": 17.39, "elapsed_time": "3:29:01", "remaining_time": "16:33:07", "throughput": 629.49, "total_tokens": 7894904}
{"current_steps": 680, "total_steps": 3882, "loss": 0.8221, "lr": 9.579091784609984e-05, "epoch": 0.3502446561936647, "percentage": 17.52, "elapsed_time": "3:30:29", "remaining_time": "16:31:10", "throughput": 629.75, "total_tokens": 7953432}
{"current_steps": 685, "total_steps": 3882, "loss": 0.8378, "lr": 9.570495596546926e-05, "epoch": 0.3528199845480299, "percentage": 17.65, "elapsed_time": "3:31:59", "remaining_time": "16:29:21", "throughput": 629.91, "total_tokens": 8011888}
{"current_steps": 690, "total_steps": 3882, "loss": 0.7529, "lr": 9.561816450895793e-05, "epoch": 0.35539531290239507, "percentage": 17.77, "elapsed_time": "3:33:26", "remaining_time": "16:27:25", "throughput": 630.16, "total_tokens": 8070344}
{"current_steps": 695, "total_steps": 3882, "loss": 0.7311, "lr": 9.55305450518895e-05, "epoch": 0.35797064125676026, "percentage": 17.9, "elapsed_time": "3:34:56", "remaining_time": "16:25:39", "throughput": 630.3, "total_tokens": 8128816}
{"current_steps": 700, "total_steps": 3882, "loss": 0.774, "lr": 9.544209918461642e-05, "epoch": 0.3605459696111254, "percentage": 18.03, "elapsed_time": "3:36:24", "remaining_time": "16:23:44", "throughput": 630.54, "total_tokens": 8187320}
{"current_steps": 700, "total_steps": 3882, "eval_loss": 0.7318872809410095, "epoch": 0.3605459696111254, "percentage": 18.03, "elapsed_time": "3:36:44", "remaining_time": "16:25:14", "throughput": 629.58, "total_tokens": 8187320}
{"current_steps": 705, "total_steps": 3882, "loss": 0.765, "lr": 9.535282851249103e-05, "epoch": 0.3631212979654906, "percentage": 18.16, "elapsed_time": "3:38:19", "remaining_time": "16:23:51", "throughput": 629.47, "total_tokens": 8245776}
{"current_steps": 710, "total_steps": 3882, "loss": 0.7287, "lr": 9.526273465583646e-05, "epoch": 0.3656966263198558, "percentage": 18.29, "elapsed_time": "3:39:48", "remaining_time": "16:21:58", "throughput": 629.68, "total_tokens": 8304280}
{"current_steps": 715, "total_steps": 3882, "loss": 0.758, "lr": 9.517181924991716e-05, "epoch": 0.36827195467422097, "percentage": 18.42, "elapsed_time": "3:41:17", "remaining_time": "16:20:12", "throughput": 629.83, "total_tokens": 8362728}
{"current_steps": 720, "total_steps": 3882, "loss": 0.795, "lr": 9.508008394490926e-05, "epoch": 0.37084728302858616, "percentage": 18.55, "elapsed_time": "3:42:47", "remaining_time": "16:18:23", "throughput": 630.0, "total_tokens": 8421224}
{"current_steps": 725, "total_steps": 3882, "loss": 0.6901, "lr": 9.498753040587066e-05, "epoch": 0.37342261138295135, "percentage": 18.68, "elapsed_time": "3:44:16", "remaining_time": "16:16:37", "throughput": 630.15, "total_tokens": 8479720}
{"current_steps": 730, "total_steps": 3882, "loss": 0.7618, "lr": 9.48941603127107e-05, "epoch": 0.3759979397373165, "percentage": 18.8, "elapsed_time": "3:45:45", "remaining_time": "16:14:47", "throughput": 630.33, "total_tokens": 8538192}
{"current_steps": 735, "total_steps": 3882, "loss": 0.7481, "lr": 9.479997536015977e-05, "epoch": 0.3785732680916817, "percentage": 18.93, "elapsed_time": "3:47:14", "remaining_time": "16:12:58", "throughput": 630.5, "total_tokens": 8596664}
{"current_steps": 740, "total_steps": 3882, "loss": 0.746, "lr": 9.47049772577385e-05, "epoch": 0.3811485964460469, "percentage": 19.06, "elapsed_time": "3:48:44", "remaining_time": "16:11:11", "throughput": 630.65, "total_tokens": 8655128}
{"current_steps": 745, "total_steps": 3882, "loss": 0.812, "lr": 9.460916772972672e-05, "epoch": 0.38372392480041206, "percentage": 19.19, "elapsed_time": "3:50:13", "remaining_time": "16:09:23", "throughput": 630.83, "total_tokens": 8713624}
{"current_steps": 750, "total_steps": 3882, "loss": 0.6751, "lr": 9.451254851513222e-05, "epoch": 0.38629925315477726, "percentage": 19.32, "elapsed_time": "3:51:43", "remaining_time": "16:07:42", "throughput": 630.91, "total_tokens": 8772104}
{"current_steps": 750, "total_steps": 3882, "eval_loss": 0.732211709022522, "epoch": 0.38629925315477726, "percentage": 19.32, "elapsed_time": "3:52:03", "remaining_time": "16:09:05", "throughput": 630.01, "total_tokens": 8772104}
{"current_steps": 755, "total_steps": 3882, "loss": 0.7772, "lr": 9.441512136765911e-05, "epoch": 0.3888745815091424, "percentage": 19.45, "elapsed_time": "3:53:37", "remaining_time": "16:07:35", "throughput": 629.98, "total_tokens": 8830568}
{"current_steps": 760, "total_steps": 3882, "loss": 0.7114, "lr": 9.431688805567607e-05, "epoch": 0.3914499098635076, "percentage": 19.58, "elapsed_time": "3:55:06", "remaining_time": "16:05:47", "throughput": 630.14, "total_tokens": 8889072}
{"current_steps": 765, "total_steps": 3882, "loss": 0.8463, "lr": 9.421785036218417e-05, "epoch": 0.3940252382178728, "percentage": 19.71, "elapsed_time": "3:56:34", "remaining_time": "16:03:54", "throughput": 630.37, "total_tokens": 8947568}
{"current_steps": 770, "total_steps": 3882, "loss": 0.7822, "lr": 9.411801008478459e-05, "epoch": 0.39660056657223797, "percentage": 19.84, "elapsed_time": "3:58:03", "remaining_time": "16:02:09", "throughput": 630.5, "total_tokens": 9006056}
{"current_steps": 775, "total_steps": 3882, "loss": 0.7628, "lr": 9.401736903564592e-05, "epoch": 0.39917589492660316, "percentage": 19.96, "elapsed_time": "3:59:31", "remaining_time": "16:00:16", "throughput": 630.73, "total_tokens": 9064592}
{"current_steps": 780, "total_steps": 3882, "loss": 0.7589, "lr": 9.39159290414713e-05, "epoch": 0.40175122328096835, "percentage": 20.09, "elapsed_time": "4:01:01", "remaining_time": "15:58:31", "throughput": 630.86, "total_tokens": 9123096}
{"current_steps": 785, "total_steps": 3882, "loss": 0.7564, "lr": 9.381369194346527e-05, "epoch": 0.4043265516353335, "percentage": 20.22, "elapsed_time": "4:02:29", "remaining_time": "15:56:39", "throughput": 631.08, "total_tokens": 9181576}
{"current_steps": 790, "total_steps": 3882, "loss": 0.6934, "lr": 9.371065959730039e-05, "epoch": 0.4069018799896987, "percentage": 20.35, "elapsed_time": "4:03:58", "remaining_time": "15:54:54", "throughput": 631.21, "total_tokens": 9240048}
{"current_steps": 795, "total_steps": 3882, "loss": 0.7314, "lr": 9.36068338730834e-05, "epoch": 0.40947720834406387, "percentage": 20.48, "elapsed_time": "4:05:26", "remaining_time": "15:53:02", "throughput": 631.42, "total_tokens": 9298528}
{"current_steps": 800, "total_steps": 3882, "loss": 0.6911, "lr": 9.35022166553215e-05, "epoch": 0.41205253669842906, "percentage": 20.61, "elapsed_time": "4:06:55", "remaining_time": "15:51:17", "throughput": 631.56, "total_tokens": 9357016}
{"current_steps": 800, "total_steps": 3882, "eval_loss": 0.7180347442626953, "epoch": 0.41205253669842906, "percentage": 20.61, "elapsed_time": "4:07:14", "remaining_time": "15:52:31", "throughput": 630.74, "total_tokens": 9357016}
{"current_steps": 805, "total_steps": 3882, "loss": 0.7086, "lr": 9.339680984288799e-05, "epoch": 0.41462786505279425, "percentage": 20.74, "elapsed_time": "4:08:48", "remaining_time": "15:51:02", "throughput": 630.7, "total_tokens": 9415480}
{"current_steps": 810, "total_steps": 3882, "loss": 0.7726, "lr": 9.329061534898783e-05, "epoch": 0.4172031934071594, "percentage": 20.87, "elapsed_time": "4:10:17", "remaining_time": "15:49:15", "throughput": 630.86, "total_tokens": 9473928}
{"current_steps": 815, "total_steps": 3882, "loss": 0.7286, "lr": 9.318363510112296e-05, "epoch": 0.4197785217615246, "percentage": 20.99, "elapsed_time": "4:11:45", "remaining_time": "15:47:25", "throughput": 631.05, "total_tokens": 9532408}
{"current_steps": 820, "total_steps": 3882, "loss": 0.7515, "lr": 9.307587104105729e-05, "epoch": 0.4223538501158898, "percentage": 21.12, "elapsed_time": "4:13:14", "remaining_time": "15:45:37", "throughput": 631.22, "total_tokens": 9590920}
{"current_steps": 825, "total_steps": 3882, "loss": 0.7344, "lr": 9.296732512478139e-05, "epoch": 0.42492917847025496, "percentage": 21.25, "elapsed_time": "4:14:42", "remaining_time": "15:43:49", "throughput": 631.39, "total_tokens": 9649400}
{"current_steps": 830, "total_steps": 3882, "loss": 0.6954, "lr": 9.285799932247714e-05, "epoch": 0.42750450682462016, "percentage": 21.38, "elapsed_time": "4:16:11", "remaining_time": "15:42:01", "throughput": 631.56, "total_tokens": 9707888}
{"current_steps": 835, "total_steps": 3882, "loss": 0.7312, "lr": 9.274789561848183e-05, "epoch": 0.43007983517898535, "percentage": 21.51, "elapsed_time": "4:17:39", "remaining_time": "15:40:13", "throughput": 631.73, "total_tokens": 9766384}
{"current_steps": 840, "total_steps": 3882, "loss": 0.647, "lr": 9.263701601125218e-05, "epoch": 0.4326551635333505, "percentage": 21.64, "elapsed_time": "4:19:07", "remaining_time": "15:38:24", "throughput": 631.92, "total_tokens": 9824896}
{"current_steps": 845, "total_steps": 3882, "loss": 0.7273, "lr": 9.252536251332813e-05, "epoch": 0.4352304918877157, "percentage": 21.77, "elapsed_time": "4:20:36", "remaining_time": "15:36:40", "throughput": 632.05, "total_tokens": 9883408}
{"current_steps": 850, "total_steps": 3882, "loss": 0.7455, "lr": 9.24129371512962e-05, "epoch": 0.43780582024208087, "percentage": 21.9, "elapsed_time": "4:22:04", "remaining_time": "15:34:49", "throughput": 632.26, "total_tokens": 9941896}
{"current_steps": 850, "total_steps": 3882, "eval_loss": 0.7039459347724915, "epoch": 0.43780582024208087, "percentage": 21.9, "elapsed_time": "4:22:24", "remaining_time": "15:36:00", "throughput": 631.47, "total_tokens": 9941896}
{"current_steps": 855, "total_steps": 3882, "loss": 0.6829, "lr": 9.22997419657528e-05, "epoch": 0.44038114859644606, "percentage": 22.02, "elapsed_time": "4:23:58", "remaining_time": "15:34:34", "throughput": 631.38, "total_tokens": 10000336}
{"current_steps": 860, "total_steps": 3882, "loss": 0.6971, "lr": 9.218577901126713e-05, "epoch": 0.44295647695081125, "percentage": 22.15, "elapsed_time": "4:25:26", "remaining_time": "15:32:45", "throughput": 631.58, "total_tokens": 10058816}
{"current_steps": 865, "total_steps": 3882, "loss": 0.7239, "lr": 9.207105035634397e-05, "epoch": 0.4455318053051764, "percentage": 22.28, "elapsed_time": "4:26:56", "remaining_time": "15:31:02", "throughput": 631.69, "total_tokens": 10117320}
{"current_steps": 870, "total_steps": 3882, "loss": 0.7113, "lr": 9.195555808338603e-05, "epoch": 0.4481071336595416, "percentage": 22.41, "elapsed_time": "4:28:23", "remaining_time": "15:29:12", "throughput": 631.89, "total_tokens": 10175824}
{"current_steps": 875, "total_steps": 3882, "loss": 0.685, "lr": 9.183930428865622e-05, "epoch": 0.45068246201390677, "percentage": 22.54, "elapsed_time": "4:29:53", "remaining_time": "15:27:28", "throughput": 632.02, "total_tokens": 10234288}
{"current_steps": 880, "total_steps": 3882, "loss": 0.6804, "lr": 9.17222910822396e-05, "epoch": 0.45325779036827196, "percentage": 22.67, "elapsed_time": "4:31:20", "remaining_time": "15:25:40", "throughput": 632.2, "total_tokens": 10292736}
{"current_steps": 885, "total_steps": 3882, "loss": 0.7056, "lr": 9.160452058800504e-05, "epoch": 0.45583311872263715, "percentage": 22.8, "elapsed_time": "4:32:50", "remaining_time": "15:23:57", "throughput": 632.31, "total_tokens": 10351224}
{"current_steps": 890, "total_steps": 3882, "loss": 0.7234, "lr": 9.148599494356671e-05, "epoch": 0.4584084470770023, "percentage": 22.93, "elapsed_time": "4:34:17", "remaining_time": "15:22:08", "throughput": 632.51, "total_tokens": 10409736}
{"current_steps": 895, "total_steps": 3882, "loss": 0.7505, "lr": 9.136671630024527e-05, "epoch": 0.4609837754313675, "percentage": 23.06, "elapsed_time": "4:35:46", "remaining_time": "15:20:22", "throughput": 632.66, "total_tokens": 10468240}
{"current_steps": 900, "total_steps": 3882, "loss": 0.7378, "lr": 9.124668682302882e-05, "epoch": 0.4635591037857327, "percentage": 23.18, "elapsed_time": "4:37:14", "remaining_time": "15:18:35", "throughput": 632.82, "total_tokens": 10526712}
{"current_steps": 900, "total_steps": 3882, "eval_loss": 0.7197856903076172, "epoch": 0.4635591037857327, "percentage": 23.18, "elapsed_time": "4:37:34", "remaining_time": "15:19:40", "throughput": 632.07, "total_tokens": 10526712}
{"current_steps": 905, "total_steps": 3882, "loss": 0.6794, "lr": 9.112590869053359e-05, "epoch": 0.46613443214009787, "percentage": 23.31, "elapsed_time": "4:39:07", "remaining_time": "15:18:11", "throughput": 632.04, "total_tokens": 10585232}
{"current_steps": 910, "total_steps": 3882, "loss": 0.6817, "lr": 9.100438409496444e-05, "epoch": 0.46870976049446306, "percentage": 23.44, "elapsed_time": "4:40:36", "remaining_time": "15:16:26", "throughput": 632.19, "total_tokens": 10643728}
{"current_steps": 915, "total_steps": 3882, "loss": 0.6503, "lr": 9.088211524207497e-05, "epoch": 0.47128508884882825, "percentage": 23.57, "elapsed_time": "4:42:04", "remaining_time": "15:14:39", "throughput": 632.36, "total_tokens": 10702240}
{"current_steps": 920, "total_steps": 3882, "loss": 0.6903, "lr": 9.075910435112766e-05, "epoch": 0.4738604172031934, "percentage": 23.7, "elapsed_time": "4:43:33", "remaining_time": "15:12:54", "throughput": 632.49, "total_tokens": 10760656}
{"current_steps": 925, "total_steps": 3882, "loss": 0.6611, "lr": 9.063535365485341e-05, "epoch": 0.4764357455575586, "percentage": 23.83, "elapsed_time": "4:45:01", "remaining_time": "15:11:09", "throughput": 632.64, "total_tokens": 10819128}
{"current_steps": 930, "total_steps": 3882, "loss": 0.6361, "lr": 9.051086539941108e-05, "epoch": 0.47901107391192377, "percentage": 23.96, "elapsed_time": "4:46:30", "remaining_time": "15:09:25", "throughput": 632.77, "total_tokens": 10877600}
{"current_steps": 935, "total_steps": 3882, "loss": 0.7006, "lr": 9.038564184434676e-05, "epoch": 0.48158640226628896, "percentage": 24.09, "elapsed_time": "4:47:58", "remaining_time": "15:07:38", "throughput": 632.94, "total_tokens": 10936088}
{"current_steps": 940, "total_steps": 3882, "loss": 0.7012, "lr": 9.025968526255275e-05, "epoch": 0.48416173062065415, "percentage": 24.21, "elapsed_time": "4:49:27", "remaining_time": "15:05:57", "throughput": 633.04, "total_tokens": 10994560}
{"current_steps": 945, "total_steps": 3882, "loss": 0.6968, "lr": 9.013299794022622e-05, "epoch": 0.4867370589750193, "percentage": 24.34, "elapsed_time": "4:50:55", "remaining_time": "15:04:10", "throughput": 633.21, "total_tokens": 11053016}
{"current_steps": 950, "total_steps": 3882, "loss": 0.6825, "lr": 9.00055821768278e-05, "epoch": 0.4893123873293845, "percentage": 24.47, "elapsed_time": "4:52:24", "remaining_time": "15:02:28", "throughput": 633.32, "total_tokens": 11111520}
{"current_steps": 950, "total_steps": 3882, "eval_loss": 0.6830747723579407, "epoch": 0.4893123873293845, "percentage": 24.47, "elapsed_time": "4:52:44", "remaining_time": "15:03:29", "throughput": 632.61, "total_tokens": 11111520}
{"current_steps": 955, "total_steps": 3882, "loss": 0.6957, "lr": 8.987744028503981e-05, "epoch": 0.49188771568374967, "percentage": 24.6, "elapsed_time": "4:54:17", "remaining_time": "15:01:58", "throughput": 632.6, "total_tokens": 11170016}
{"current_steps": 960, "total_steps": 3882, "loss": 0.6531, "lr": 8.974857459072435e-05, "epoch": 0.49446304403811486, "percentage": 24.73, "elapsed_time": "4:55:46", "remaining_time": "15:00:16", "throughput": 632.71, "total_tokens": 11228496}
{"current_steps": 965, "total_steps": 3882, "loss": 0.6201, "lr": 8.961898743288094e-05, "epoch": 0.49703837239248005, "percentage": 24.86, "elapsed_time": "4:57:14", "remaining_time": "14:58:29", "throughput": 632.88, "total_tokens": 11286928}
{"current_steps": 970, "total_steps": 3882, "loss": 0.5677, "lr": 8.948868116360421e-05, "epoch": 0.49961370074684525, "percentage": 24.99, "elapsed_time": "4:58:42", "remaining_time": "14:56:45", "throughput": 633.01, "total_tokens": 11345400}
{"current_steps": 975, "total_steps": 3882, "loss": 0.5763, "lr": 8.935765814804112e-05, "epoch": 0.5021890291012104, "percentage": 25.12, "elapsed_time": "5:00:10", "remaining_time": "14:54:59", "throughput": 633.18, "total_tokens": 11403912}
{"current_steps": 980, "total_steps": 3882, "loss": 0.6348, "lr": 8.922592076434804e-05, "epoch": 0.5047643574555756, "percentage": 25.24, "elapsed_time": "5:01:39", "remaining_time": "14:53:16", "throughput": 633.3, "total_tokens": 11462344}
{"current_steps": 985, "total_steps": 3882, "loss": 0.6541, "lr": 8.90934714036477e-05, "epoch": 0.5073396858099408, "percentage": 25.37, "elapsed_time": "5:03:07", "remaining_time": "14:51:31", "throughput": 633.45, "total_tokens": 11520808}
{"current_steps": 990, "total_steps": 3882, "loss": 0.7012, "lr": 8.896031246998558e-05, "epoch": 0.509915014164306, "percentage": 25.5, "elapsed_time": "5:04:35", "remaining_time": "14:49:45", "throughput": 633.61, "total_tokens": 11579248}
{"current_steps": 995, "total_steps": 3882, "loss": 0.6508, "lr": 8.882644638028646e-05, "epoch": 0.5124903425186711, "percentage": 25.63, "elapsed_time": "5:06:03", "remaining_time": "14:48:01", "throughput": 633.74, "total_tokens": 11637712}
{"current_steps": 1000, "total_steps": 3882, "loss": 0.5971, "lr": 8.869187556431046e-05, "epoch": 0.5150656708730363, "percentage": 25.76, "elapsed_time": "5:07:31", "remaining_time": "14:46:17", "throughput": 633.88, "total_tokens": 11696200}
{"current_steps": 1000, "total_steps": 3882, "eval_loss": 0.7078786492347717, "epoch": 0.5150656708730363, "percentage": 25.76, "elapsed_time": "5:07:51", "remaining_time": "14:47:14", "throughput": 633.21, "total_tokens": 11696200}
{"current_steps": 1005, "total_steps": 3882, "loss": 0.6959, "lr": 8.855660246460895e-05, "epoch": 0.5176409992274015, "percentage": 25.89, "elapsed_time": "5:09:25", "remaining_time": "14:45:47", "throughput": 633.15, "total_tokens": 11754720}
{"current_steps": 1010, "total_steps": 3882, "loss": 0.6918, "lr": 8.842062953648023e-05, "epoch": 0.5202163275817667, "percentage": 26.02, "elapsed_time": "5:10:53", "remaining_time": "14:44:02", "throughput": 633.3, "total_tokens": 11813216}
{"current_steps": 1015, "total_steps": 3882, "loss": 0.7269, "lr": 8.828395924792497e-05, "epoch": 0.5227916559361319, "percentage": 26.15, "elapsed_time": "5:12:21", "remaining_time": "14:42:19", "throughput": 633.43, "total_tokens": 11871712}
{"current_steps": 1020, "total_steps": 3882, "loss": 0.7026, "lr": 8.814659407960141e-05, "epoch": 0.525366984290497, "percentage": 26.28, "elapsed_time": "5:13:49", "remaining_time": "14:40:33", "throughput": 633.58, "total_tokens": 11930200}
{"current_steps": 1025, "total_steps": 3882, "loss": 0.6467, "lr": 8.800853652478028e-05, "epoch": 0.5279423126448622, "percentage": 26.4, "elapsed_time": "5:15:18", "remaining_time": "14:38:52", "throughput": 633.7, "total_tokens": 11988704}
{"current_steps": 1030, "total_steps": 3882, "loss": 0.6155, "lr": 8.786978908929966e-05, "epoch": 0.5305176409992274, "percentage": 26.53, "elapsed_time": "5:16:46", "remaining_time": "14:37:07", "throughput": 633.85, "total_tokens": 12047176}
{"current_steps": 1035, "total_steps": 3882, "loss": 0.7849, "lr": 8.773035429151937e-05, "epoch": 0.5330929693535926, "percentage": 26.66, "elapsed_time": "5:18:15", "remaining_time": "14:35:25", "throughput": 633.97, "total_tokens": 12105680}
{"current_steps": 1040, "total_steps": 3882, "loss": 0.6341, "lr": 8.759023466227538e-05, "epoch": 0.5356682977079578, "percentage": 26.79, "elapsed_time": "5:19:42", "remaining_time": "14:33:39", "throughput": 634.13, "total_tokens": 12164208}
{"current_steps": 1045, "total_steps": 3882, "loss": 0.7189, "lr": 8.744943274483376e-05, "epoch": 0.5382436260623229, "percentage": 26.92, "elapsed_time": "5:21:11", "remaining_time": "14:32:00", "throughput": 634.22, "total_tokens": 12222672}
{"current_steps": 1050, "total_steps": 3882, "loss": 0.6914, "lr": 8.730795109484461e-05, "epoch": 0.5408189544166881, "percentage": 27.05, "elapsed_time": "5:22:39", "remaining_time": "14:30:14", "throughput": 634.37, "total_tokens": 12281072}
{"current_steps": 1050, "total_steps": 3882, "eval_loss": 0.6824291348457336, "epoch": 0.5408189544166881, "percentage": 27.05, "elapsed_time": "5:22:59", "remaining_time": "14:31:08", "throughput": 633.73, "total_tokens": 12281072}
{"current_steps": 1055, "total_steps": 3882, "loss": 0.6567, "lr": 8.716579228029562e-05, "epoch": 0.5433942827710533, "percentage": 27.18, "elapsed_time": "5:24:32", "remaining_time": "14:29:39", "throughput": 633.68, "total_tokens": 12339544}
{"current_steps": 1060, "total_steps": 3882, "loss": 0.6155, "lr": 8.702295888146548e-05, "epoch": 0.5459696111254185, "percentage": 27.31, "elapsed_time": "5:26:00", "remaining_time": "14:27:56", "throughput": 633.81, "total_tokens": 12397992}
{"current_steps": 1065, "total_steps": 3882, "loss": 0.6029, "lr": 8.687945349087703e-05, "epoch": 0.5485449394797837, "percentage": 27.43, "elapsed_time": "5:27:29", "remaining_time": "14:26:13", "throughput": 633.94, "total_tokens": 12456480}
{"current_steps": 1070, "total_steps": 3882, "loss": 0.6129, "lr": 8.673527871325022e-05, "epoch": 0.5511202678341488, "percentage": 27.56, "elapsed_time": "5:28:56", "remaining_time": "14:24:28", "throughput": 634.1, "total_tokens": 12515000}
{"current_steps": 1075, "total_steps": 3882, "loss": 0.6825, "lr": 8.659043716545485e-05, "epoch": 0.553695596188514, "percentage": 27.69, "elapsed_time": "5:30:24", "remaining_time": "14:22:45", "throughput": 634.24, "total_tokens": 12573504}
{"current_steps": 1080, "total_steps": 3882, "loss": 0.6828, "lr": 8.644493147646302e-05, "epoch": 0.5562709245428792, "percentage": 27.82, "elapsed_time": "5:31:52", "remaining_time": "14:21:00", "throughput": 634.39, "total_tokens": 12632008}
{"current_steps": 1085, "total_steps": 3882, "loss": 0.6286, "lr": 8.629876428730145e-05, "epoch": 0.5588462528972444, "percentage": 27.95, "elapsed_time": "5:33:19", "remaining_time": "14:19:16", "throughput": 634.54, "total_tokens": 12690520}
{"current_steps": 1090, "total_steps": 3882, "loss": 0.6361, "lr": 8.615193825100355e-05, "epoch": 0.5614215812516096, "percentage": 28.08, "elapsed_time": "5:34:47", "remaining_time": "14:17:32", "throughput": 634.69, "total_tokens": 12749032}
{"current_steps": 1095, "total_steps": 3882, "loss": 0.6087, "lr": 8.600445603256123e-05, "epoch": 0.5639969096059748, "percentage": 28.21, "elapsed_time": "5:36:14", "remaining_time": "14:15:47", "throughput": 634.84, "total_tokens": 12807504}
{"current_steps": 1100, "total_steps": 3882, "loss": 0.5825, "lr": 8.585632030887658e-05, "epoch": 0.56657223796034, "percentage": 28.34, "elapsed_time": "5:37:42", "remaining_time": "14:14:05", "throughput": 634.96, "total_tokens": 12865992}
{"current_steps": 1100, "total_steps": 3882, "eval_loss": 0.6431913375854492, "epoch": 0.56657223796034, "percentage": 28.34, "elapsed_time": "5:38:01", "remaining_time": "14:14:54", "throughput": 634.36, "total_tokens": 12865992}
{"current_steps": 1105, "total_steps": 3882, "loss": 0.6113, "lr": 8.57075337687132e-05, "epoch": 0.5691475663147051, "percentage": 28.46, "elapsed_time": "5:39:34", "remaining_time": "14:13:24", "throughput": 634.34, "total_tokens": 12924448}
{"current_steps": 1110, "total_steps": 3882, "loss": 0.562, "lr": 8.55580991126475e-05, "epoch": 0.5717228946690703, "percentage": 28.59, "elapsed_time": "5:41:02", "remaining_time": "14:11:41", "throughput": 634.47, "total_tokens": 12982912}
{"current_steps": 1115, "total_steps": 3882, "loss": 0.6124, "lr": 8.540801905301963e-05, "epoch": 0.5742982230234355, "percentage": 28.72, "elapsed_time": "5:42:30", "remaining_time": "14:09:57", "throughput": 634.62, "total_tokens": 13041424}
{"current_steps": 1120, "total_steps": 3882, "loss": 0.6788, "lr": 8.525729631388421e-05, "epoch": 0.5768735513778007, "percentage": 28.85, "elapsed_time": "5:43:59", "remaining_time": "14:08:17", "throughput": 634.71, "total_tokens": 13099888}
{"current_steps": 1125, "total_steps": 3882, "loss": 0.5904, "lr": 8.510593363096097e-05, "epoch": 0.5794488797321659, "percentage": 28.98, "elapsed_time": "5:45:26", "remaining_time": "14:06:34", "throughput": 634.85, "total_tokens": 13158344}
{"current_steps": 1130, "total_steps": 3882, "loss": 0.5433, "lr": 8.495393375158504e-05, "epoch": 0.582024208086531, "percentage": 29.11, "elapsed_time": "5:46:55", "remaining_time": "14:04:54", "throughput": 634.95, "total_tokens": 13216840}
{"current_steps": 1135, "total_steps": 3882, "loss": 0.5937, "lr": 8.480129943465709e-05, "epoch": 0.5845995364408962, "percentage": 29.24, "elapsed_time": "5:48:22", "remaining_time": "14:03:09", "throughput": 635.11, "total_tokens": 13275328}
{"current_steps": 1140, "total_steps": 3882, "loss": 0.5785, "lr": 8.464803345059324e-05, "epoch": 0.5871748647952614, "percentage": 29.37, "elapsed_time": "5:49:50", "remaining_time": "14:01:28", "throughput": 635.22, "total_tokens": 13333784}
{"current_steps": 1145, "total_steps": 3882, "loss": 0.5472, "lr": 8.449413858127487e-05, "epoch": 0.5897501931496266, "percentage": 29.5, "elapsed_time": "5:51:17", "remaining_time": "13:59:44", "throughput": 635.37, "total_tokens": 13392280}
{"current_steps": 1150, "total_steps": 3882, "loss": 0.5228, "lr": 8.433961761999796e-05, "epoch": 0.5923255215039918, "percentage": 29.62, "elapsed_time": "5:52:45", "remaining_time": "13:58:03", "throughput": 635.49, "total_tokens": 13450720}
{"current_steps": 1150, "total_steps": 3882, "eval_loss": 0.6229755282402039, "epoch": 0.5923255215039918, "percentage": 29.62, "elapsed_time": "5:53:05", "remaining_time": "13:58:49", "throughput": 634.91, "total_tokens": 13450720}
{"current_steps": 1155, "total_steps": 3882, "loss": 0.5654, "lr": 8.418447337142254e-05, "epoch": 0.5949008498583569, "percentage": 29.75, "elapsed_time": "5:54:37", "remaining_time": "13:57:18", "throughput": 634.89, "total_tokens": 13509200}
{"current_steps": 1160, "total_steps": 3882, "loss": 0.5074, "lr": 8.402870865152172e-05, "epoch": 0.5974761782127221, "percentage": 29.88, "elapsed_time": "5:56:06", "remaining_time": "13:55:38", "throughput": 634.99, "total_tokens": 13567656}
{"current_steps": 1165, "total_steps": 3882, "loss": 0.6436, "lr": 8.387232628753056e-05, "epoch": 0.6000515065670873, "percentage": 30.01, "elapsed_time": "5:57:34", "remaining_time": "13:53:55", "throughput": 635.12, "total_tokens": 13626136}
{"current_steps": 1170, "total_steps": 3882, "loss": 0.55, "lr": 8.371532911789482e-05, "epoch": 0.6026268349214525, "percentage": 30.14, "elapsed_time": "5:59:02", "remaining_time": "13:52:14", "throughput": 635.24, "total_tokens": 13684608}
{"current_steps": 1175, "total_steps": 3882, "loss": 0.5399, "lr": 8.355771999221937e-05, "epoch": 0.6052021632758177, "percentage": 30.27, "elapsed_time": "6:00:29", "remaining_time": "13:50:30", "throughput": 635.38, "total_tokens": 13743080}
{"current_steps": 1180, "total_steps": 3882, "loss": 0.5402, "lr": 8.339950177121647e-05, "epoch": 0.6077774916301828, "percentage": 30.4, "elapsed_time": "6:01:58", "remaining_time": "13:48:50", "throughput": 635.49, "total_tokens": 13801552}
{"current_steps": 1185, "total_steps": 3882, "loss": 0.5559, "lr": 8.324067732665393e-05, "epoch": 0.610352819984548, "percentage": 30.53, "elapsed_time": "6:03:25", "remaining_time": "13:47:08", "throughput": 635.62, "total_tokens": 13860064}
{"current_steps": 1190, "total_steps": 3882, "loss": 0.5619, "lr": 8.308124954130289e-05, "epoch": 0.6129281483389132, "percentage": 30.65, "elapsed_time": "6:04:52", "remaining_time": "13:45:25", "throughput": 635.76, "total_tokens": 13918552}
{"current_steps": 1195, "total_steps": 3882, "loss": 0.5933, "lr": 8.292122130888558e-05, "epoch": 0.6155034766932784, "percentage": 30.78, "elapsed_time": "6:06:21", "remaining_time": "13:43:45", "throughput": 635.87, "total_tokens": 13977056}
{"current_steps": 1200, "total_steps": 3882, "loss": 0.5078, "lr": 8.276059553402265e-05, "epoch": 0.6180788050476436, "percentage": 30.91, "elapsed_time": "6:07:48", "remaining_time": "13:42:03", "throughput": 636.0, "total_tokens": 14035544}
{"current_steps": 1200, "total_steps": 3882, "eval_loss": 0.6184359192848206, "epoch": 0.6180788050476436, "percentage": 30.91, "elapsed_time": "6:08:07", "remaining_time": "13:42:46", "throughput": 635.44, "total_tokens": 14035544}
{"current_steps": 1205, "total_steps": 3882, "loss": 0.5109, "lr": 8.259937513218066e-05, "epoch": 0.6206541334020087, "percentage": 31.04, "elapsed_time": "6:09:41", "remaining_time": "13:41:17", "throughput": 635.4, "total_tokens": 14094024}
{"current_steps": 1210, "total_steps": 3882, "loss": 0.4738, "lr": 8.243756302961898e-05, "epoch": 0.623229461756374, "percentage": 31.17, "elapsed_time": "6:11:08", "remaining_time": "13:39:35", "throughput": 635.53, "total_tokens": 14152504}
{"current_steps": 1215, "total_steps": 3882, "loss": 0.5615, "lr": 8.227516216333679e-05, "epoch": 0.6258047901107391, "percentage": 31.3, "elapsed_time": "6:12:37", "remaining_time": "13:37:56", "throughput": 635.62, "total_tokens": 14210992}
{"current_steps": 1220, "total_steps": 3882, "loss": 0.5584, "lr": 8.211217548101973e-05, "epoch": 0.6283801184651043, "percentage": 31.43, "elapsed_time": "6:14:06", "remaining_time": "13:36:16", "throughput": 635.72, "total_tokens": 14269488}
{"current_steps": 1225, "total_steps": 3882, "loss": 0.4856, "lr": 8.194860594098635e-05, "epoch": 0.6309554468194695, "percentage": 31.56, "elapsed_time": "6:15:35", "remaining_time": "13:34:38", "throughput": 635.8, "total_tokens": 14327968}
{"current_steps": 1230, "total_steps": 3882, "loss": 0.5378, "lr": 8.17844565121345e-05, "epoch": 0.6335307751738347, "percentage": 31.68, "elapsed_time": "6:17:04", "remaining_time": "13:32:59", "throughput": 635.89, "total_tokens": 14386448}
{"current_steps": 1235, "total_steps": 3882, "loss": 0.4484, "lr": 8.161973017388744e-05, "epoch": 0.6361061035281999, "percentage": 31.81, "elapsed_time": "6:18:33", "remaining_time": "13:31:23", "throughput": 635.95, "total_tokens": 14444912}
{"current_steps": 1240, "total_steps": 3882, "loss": 0.4772, "lr": 8.145442991613963e-05, "epoch": 0.638681431882565, "percentage": 31.94, "elapsed_time": "6:20:02", "remaining_time": "13:29:44", "throughput": 636.04, "total_tokens": 14503392}
{"current_steps": 1245, "total_steps": 3882, "loss": 0.5807, "lr": 8.128855873920265e-05, "epoch": 0.6412567602369302, "percentage": 32.07, "elapsed_time": "6:21:31", "remaining_time": "13:28:06", "throughput": 636.12, "total_tokens": 14561872}
{"current_steps": 1250, "total_steps": 3882, "loss": 0.5268, "lr": 8.112211965375059e-05, "epoch": 0.6438320885912954, "percentage": 32.2, "elapsed_time": "6:23:00", "remaining_time": "13:26:27", "throughput": 636.21, "total_tokens": 14620336}
{"current_steps": 1250, "total_steps": 3882, "eval_loss": 0.5496931672096252, "epoch": 0.6438320885912954, "percentage": 32.2, "elapsed_time": "6:23:19", "remaining_time": "13:27:08", "throughput": 635.67, "total_tokens": 14620336}
{"current_steps": 1255, "total_steps": 3882, "loss": 0.6012, "lr": 8.095511568076548e-05, "epoch": 0.6464074169456606, "percentage": 32.33, "elapsed_time": "6:24:54", "remaining_time": "13:25:41", "throughput": 635.6, "total_tokens": 14678792}
{"current_steps": 1260, "total_steps": 3882, "loss": 0.5358, "lr": 8.078754985148247e-05, "epoch": 0.6489827453000258, "percentage": 32.46, "elapsed_time": "6:26:23", "remaining_time": "13:24:02", "throughput": 635.69, "total_tokens": 14737272}
{"current_steps": 1265, "total_steps": 3882, "loss": 0.4676, "lr": 8.061942520733474e-05, "epoch": 0.6515580736543909, "percentage": 32.59, "elapsed_time": "6:27:52", "remaining_time": "13:22:25", "throughput": 635.76, "total_tokens": 14795784}
{"current_steps": 1270, "total_steps": 3882, "loss": 0.489, "lr": 8.045074479989838e-05, "epoch": 0.6541334020087561, "percentage": 32.72, "elapsed_time": "6:29:21", "remaining_time": "13:20:47", "throughput": 635.84, "total_tokens": 14854272}
{"current_steps": 1275, "total_steps": 3882, "loss": 0.4505, "lr": 8.02815116908369e-05, "epoch": 0.6567087303631213, "percentage": 32.84, "elapsed_time": "6:30:50", "remaining_time": "13:19:10", "throughput": 635.91, "total_tokens": 14912720}
{"current_steps": 1280, "total_steps": 3882, "loss": 0.4987, "lr": 8.011172895184579e-05, "epoch": 0.6592840587174865, "percentage": 32.97, "elapsed_time": "6:32:19", "remaining_time": "13:17:32", "throughput": 635.99, "total_tokens": 14971192}
{"current_steps": 1285, "total_steps": 3882, "loss": 0.5156, "lr": 7.994139966459664e-05, "epoch": 0.6618593870718517, "percentage": 33.1, "elapsed_time": "6:33:49", "remaining_time": "13:15:54", "throughput": 636.06, "total_tokens": 15029656}
{"current_steps": 1290, "total_steps": 3882, "loss": 0.5266, "lr": 7.977052692068127e-05, "epoch": 0.6644347154262168, "percentage": 33.23, "elapsed_time": "6:35:18", "remaining_time": "13:14:17", "throughput": 636.14, "total_tokens": 15088144}
{"current_steps": 1295, "total_steps": 3882, "loss": 0.4502, "lr": 7.959911382155566e-05, "epoch": 0.667010043780582, "percentage": 33.36, "elapsed_time": "6:36:47", "remaining_time": "13:12:39", "throughput": 636.22, "total_tokens": 15146600}
{"current_steps": 1300, "total_steps": 3882, "loss": 0.4578, "lr": 7.942716347848353e-05, "epoch": 0.6695853721349472, "percentage": 33.49, "elapsed_time": "6:38:17", "remaining_time": "13:11:03", "throughput": 636.27, "total_tokens": 15205064}
{"current_steps": 1300, "total_steps": 3882, "eval_loss": 0.49471279978752136, "epoch": 0.6695853721349472, "percentage": 33.49, "elapsed_time": "6:38:36", "remaining_time": "13:11:42", "throughput": 635.75, "total_tokens": 15205064}
{"current_steps": 1305, "total_steps": 3882, "loss": 0.4589, "lr": 7.925467901247996e-05, "epoch": 0.6721607004893124, "percentage": 33.62, "elapsed_time": "6:40:10", "remaining_time": "13:10:13", "throughput": 635.71, "total_tokens": 15263560}
{"current_steps": 1310, "total_steps": 3882, "loss": 0.457, "lr": 7.908166355425475e-05, "epoch": 0.6747360288436776, "percentage": 33.75, "elapsed_time": "6:41:39", "remaining_time": "13:08:35", "throughput": 635.79, "total_tokens": 15322016}
{"current_steps": 1315, "total_steps": 3882, "loss": 0.51, "lr": 7.890812024415555e-05, "epoch": 0.6773113571980427, "percentage": 33.87, "elapsed_time": "6:43:06", "remaining_time": "13:06:54", "throughput": 635.91, "total_tokens": 15380504}
{"current_steps": 1320, "total_steps": 3882, "loss": 0.4994, "lr": 7.873405223211087e-05, "epoch": 0.6798866855524079, "percentage": 34.0, "elapsed_time": "6:44:35", "remaining_time": "13:05:16", "throughput": 635.99, "total_tokens": 15438944}
{"current_steps": 1325, "total_steps": 3882, "loss": 0.4501, "lr": 7.855946267757295e-05, "epoch": 0.6824620139067731, "percentage": 34.13, "elapsed_time": "6:46:03", "remaining_time": "13:03:37", "throughput": 636.09, "total_tokens": 15497384}
{"current_steps": 1330, "total_steps": 3882, "loss": 0.4807, "lr": 7.838435474946034e-05, "epoch": 0.6850373422611383, "percentage": 34.26, "elapsed_time": "6:47:33", "remaining_time": "13:02:00", "throughput": 636.15, "total_tokens": 15555856}
{"current_steps": 1335, "total_steps": 3882, "loss": 0.5112, "lr": 7.820873162610044e-05, "epoch": 0.6876126706155035, "percentage": 34.39, "elapsed_time": "6:49:01", "remaining_time": "13:00:22", "throughput": 636.24, "total_tokens": 15614368}
{"current_steps": 1340, "total_steps": 3882, "loss": 0.4825, "lr": 7.803259649517178e-05, "epoch": 0.6901879989698687, "percentage": 34.52, "elapsed_time": "6:50:30", "remaining_time": "12:58:44", "throughput": 636.31, "total_tokens": 15672864}
{"current_steps": 1345, "total_steps": 3882, "loss": 0.5147, "lr": 7.78559525536462e-05, "epoch": 0.6927633273242338, "percentage": 34.65, "elapsed_time": "6:51:59", "remaining_time": "12:57:06", "throughput": 636.4, "total_tokens": 15731376}
{"current_steps": 1350, "total_steps": 3882, "loss": 0.4702, "lr": 7.767880300773074e-05, "epoch": 0.695338655678599, "percentage": 34.78, "elapsed_time": "6:53:28", "remaining_time": "12:55:30", "throughput": 636.46, "total_tokens": 15789848}
{"current_steps": 1350, "total_steps": 3882, "eval_loss": 0.5247787237167358, "epoch": 0.695338655678599, "percentage": 34.78, "elapsed_time": "6:53:48", "remaining_time": "12:56:06", "throughput": 635.96, "total_tokens": 15789848}
{"current_steps": 1355, "total_steps": 3882, "loss": 0.4601, "lr": 7.750115107280959e-05, "epoch": 0.6979139840329642, "percentage": 34.9, "elapsed_time": "6:55:22", "remaining_time": "12:54:38", "throughput": 635.92, "total_tokens": 15848328}
{"current_steps": 1360, "total_steps": 3882, "loss": 0.4704, "lr": 7.732299997338557e-05, "epoch": 0.7004893123873294, "percentage": 35.03, "elapsed_time": "6:56:51", "remaining_time": "12:53:01", "throughput": 635.98, "total_tokens": 15906824}
{"current_steps": 1365, "total_steps": 3882, "loss": 0.4937, "lr": 7.714435294302168e-05, "epoch": 0.7030646407416946, "percentage": 35.16, "elapsed_time": "6:58:20", "remaining_time": "12:51:23", "throughput": 636.06, "total_tokens": 15965312}
{"current_steps": 1370, "total_steps": 3882, "loss": 0.4635, "lr": 7.696521322428245e-05, "epoch": 0.7056399690960597, "percentage": 35.29, "elapsed_time": "6:59:48", "remaining_time": "12:49:45", "throughput": 636.15, "total_tokens": 16023824}
{"current_steps": 1375, "total_steps": 3882, "loss": 0.4421, "lr": 7.678558406867498e-05, "epoch": 0.7082152974504249, "percentage": 35.42, "elapsed_time": "7:01:18", "remaining_time": "12:48:09", "throughput": 636.2, "total_tokens": 16082280}
{"current_steps": 1380, "total_steps": 3882, "loss": 0.4823, "lr": 7.660546873659e-05, "epoch": 0.7107906258047901, "percentage": 35.55, "elapsed_time": "7:02:47", "remaining_time": "12:46:31", "throughput": 636.29, "total_tokens": 16140800}
{"current_steps": 1385, "total_steps": 3882, "loss": 0.4122, "lr": 7.642487049724271e-05, "epoch": 0.7133659541591553, "percentage": 35.68, "elapsed_time": "7:04:16", "remaining_time": "12:44:54", "throughput": 636.36, "total_tokens": 16199320}
{"current_steps": 1390, "total_steps": 3882, "loss": 0.4372, "lr": 7.624379262861335e-05, "epoch": 0.7159412825135205, "percentage": 35.81, "elapsed_time": "7:05:43", "remaining_time": "12:43:14", "throughput": 636.47, "total_tokens": 16257768}
{"current_steps": 1395, "total_steps": 3882, "loss": 0.4899, "lr": 7.606223841738775e-05, "epoch": 0.7185166108678857, "percentage": 35.94, "elapsed_time": "7:07:11", "remaining_time": "12:41:35", "throughput": 636.57, "total_tokens": 16316264}
{"current_steps": 1400, "total_steps": 3882, "loss": 0.4294, "lr": 7.588021115889777e-05, "epoch": 0.7210919392222508, "percentage": 36.06, "elapsed_time": "7:08:40", "remaining_time": "12:39:57", "throughput": 636.66, "total_tokens": 16374784}
{"current_steps": 1400, "total_steps": 3882, "eval_loss": 0.47318556904792786, "epoch": 0.7210919392222508, "percentage": 36.06, "elapsed_time": "7:08:59", "remaining_time": "12:40:32", "throughput": 636.18, "total_tokens": 16374784}
{"current_steps": 1405, "total_steps": 3882, "loss": 0.444, "lr": 7.569771415706126e-05, "epoch": 0.723667267576616, "percentage": 36.19, "elapsed_time": "7:10:32", "remaining_time": "12:39:02", "throughput": 636.14, "total_tokens": 16433288}
{"current_steps": 1410, "total_steps": 3882, "loss": 0.4971, "lr": 7.551475072432237e-05, "epoch": 0.7262425959309812, "percentage": 36.32, "elapsed_time": "7:12:00", "remaining_time": "12:37:24", "throughput": 636.24, "total_tokens": 16491784}
{"current_steps": 1415, "total_steps": 3882, "loss": 0.4284, "lr": 7.533132418159119e-05, "epoch": 0.7288179242853464, "percentage": 36.45, "elapsed_time": "7:13:28", "remaining_time": "12:35:45", "throughput": 636.34, "total_tokens": 16550272}
{"current_steps": 1420, "total_steps": 3882, "loss": 0.414, "lr": 7.514743785818361e-05, "epoch": 0.7313932526397116, "percentage": 36.58, "elapsed_time": "7:14:56", "remaining_time": "12:34:06", "throughput": 636.43, "total_tokens": 16608704}
{"current_steps": 1425, "total_steps": 3882, "loss": 0.3715, "lr": 7.496309509176082e-05, "epoch": 0.7339685809940767, "percentage": 36.71, "elapsed_time": "7:16:25", "remaining_time": "12:32:29", "throughput": 636.51, "total_tokens": 16667200}
{"current_steps": 1430, "total_steps": 3882, "loss": 0.4723, "lr": 7.477829922826883e-05, "epoch": 0.7365439093484419, "percentage": 36.84, "elapsed_time": "7:17:53", "remaining_time": "12:30:50", "throughput": 636.61, "total_tokens": 16725704}
{"current_steps": 1435, "total_steps": 3882, "loss": 0.4357, "lr": 7.459305362187756e-05, "epoch": 0.7391192377028071, "percentage": 36.97, "elapsed_time": "7:19:21", "remaining_time": "12:29:12", "throughput": 636.69, "total_tokens": 16784192}
{"current_steps": 1440, "total_steps": 3882, "loss": 0.4402, "lr": 7.440736163492016e-05, "epoch": 0.7416945660571723, "percentage": 37.09, "elapsed_time": "7:20:49", "remaining_time": "12:27:33", "throughput": 636.79, "total_tokens": 16842704}
{"current_steps": 1445, "total_steps": 3882, "loss": 0.4457, "lr": 7.422122663783187e-05, "epoch": 0.7442698944115375, "percentage": 37.22, "elapsed_time": "7:22:18", "remaining_time": "12:25:56", "throughput": 636.87, "total_tokens": 16901176}
{"current_steps": 1450, "total_steps": 3882, "loss": 0.4353, "lr": 7.403465200908883e-05, "epoch": 0.7468452227659027, "percentage": 37.35, "elapsed_time": "7:23:45", "remaining_time": "12:24:17", "throughput": 636.97, "total_tokens": 16959632}
{"current_steps": 1450, "total_steps": 3882, "eval_loss": 0.43503889441490173, "epoch": 0.7468452227659027, "percentage": 37.35, "elapsed_time": "7:24:05", "remaining_time": "12:24:50", "throughput": 636.5, "total_tokens": 16959632}
{"current_steps": 1455, "total_steps": 3882, "loss": 0.4183, "lr": 7.38476411351468e-05, "epoch": 0.7494205511202678, "percentage": 37.48, "elapsed_time": "7:25:39", "remaining_time": "12:23:23", "throughput": 636.43, "total_tokens": 17018072}
{"current_steps": 1460, "total_steps": 3882, "loss": 0.4632, "lr": 7.366019741037973e-05, "epoch": 0.751995879474633, "percentage": 37.61, "elapsed_time": "7:27:09", "remaining_time": "12:21:47", "throughput": 636.49, "total_tokens": 17076560}
{"current_steps": 1465, "total_steps": 3882, "loss": 0.4512, "lr": 7.347232423701804e-05, "epoch": 0.7545712078289982, "percentage": 37.74, "elapsed_time": "7:28:37", "remaining_time": "12:20:09", "throughput": 636.57, "total_tokens": 17135016}
{"current_steps": 1470, "total_steps": 3882, "loss": 0.4121, "lr": 7.328402502508701e-05, "epoch": 0.7571465361833634, "percentage": 37.87, "elapsed_time": "7:30:06", "remaining_time": "12:18:33", "throughput": 636.64, "total_tokens": 17193496}
{"current_steps": 1475, "total_steps": 3882, "loss": 0.5234, "lr": 7.309530319234472e-05, "epoch": 0.7597218645377286, "percentage": 38.0, "elapsed_time": "7:31:35", "remaining_time": "12:16:55", "throughput": 636.72, "total_tokens": 17251944}
{"current_steps": 1480, "total_steps": 3882, "loss": 0.3895, "lr": 7.290616216422017e-05, "epoch": 0.7622971928920937, "percentage": 38.12, "elapsed_time": "7:33:03", "remaining_time": "12:15:18", "throughput": 636.79, "total_tokens": 17310456}
{"current_steps": 1485, "total_steps": 3882, "loss": 0.3641, "lr": 7.271660537375102e-05, "epoch": 0.7648725212464589, "percentage": 38.25, "elapsed_time": "7:34:32", "remaining_time": "12:13:42", "throughput": 636.86, "total_tokens": 17368968}
{"current_steps": 1490, "total_steps": 3882, "loss": 0.3867, "lr": 7.25266362615213e-05, "epoch": 0.7674478496008241, "percentage": 38.38, "elapsed_time": "7:36:01", "remaining_time": "12:12:05", "throughput": 636.92, "total_tokens": 17427424}
{"current_steps": 1495, "total_steps": 3882, "loss": 0.3214, "lr": 7.233625827559893e-05, "epoch": 0.7700231779551893, "percentage": 38.51, "elapsed_time": "7:37:30", "remaining_time": "12:10:28", "throughput": 637.01, "total_tokens": 17485936}
{"current_steps": 1500, "total_steps": 3882, "loss": 0.3369, "lr": 7.21454748714732e-05, "epoch": 0.7725985063095545, "percentage": 38.64, "elapsed_time": "7:38:59", "remaining_time": "12:08:52", "throughput": 637.07, "total_tokens": 17544440}
{"current_steps": 1500, "total_steps": 3882, "eval_loss": 0.3963810205459595, "epoch": 0.7725985063095545, "percentage": 38.64, "elapsed_time": "7:39:18", "remaining_time": "12:09:22", "throughput": 636.62, "total_tokens": 17544440}
{"current_steps": 1505, "total_steps": 3882, "loss": 0.3397, "lr": 7.195428951199204e-05, "epoch": 0.7751738346639196, "percentage": 38.77, "elapsed_time": "7:40:52", "remaining_time": "12:07:54", "throughput": 636.58, "total_tokens": 17602952}
{"current_steps": 1510, "total_steps": 3882, "loss": 0.5039, "lr": 7.176270566729904e-05, "epoch": 0.7777491630182848, "percentage": 38.9, "elapsed_time": "7:42:21", "remaining_time": "12:06:18", "throughput": 636.64, "total_tokens": 17661440}
{"current_steps": 1515, "total_steps": 3882, "loss": 0.426, "lr": 7.157072681477069e-05, "epoch": 0.78032449137265, "percentage": 39.03, "elapsed_time": "7:43:50", "remaining_time": "12:04:41", "throughput": 636.72, "total_tokens": 17719928}
{"current_steps": 1520, "total_steps": 3882, "loss": 0.3587, "lr": 7.137835643895305e-05, "epoch": 0.7828998197270152, "percentage": 39.16, "elapsed_time": "7:45:19", "remaining_time": "12:03:05", "throughput": 636.78, "total_tokens": 17778424}
{"current_steps": 1525, "total_steps": 3882, "loss": 0.4849, "lr": 7.118559803149865e-05, "epoch": 0.7854751480813804, "percentage": 39.28, "elapsed_time": "7:46:48", "remaining_time": "12:01:28", "throughput": 636.85, "total_tokens": 17836936}
{"current_steps": 1530, "total_steps": 3882, "loss": 0.3795, "lr": 7.099245509110299e-05, "epoch": 0.7880504764357456, "percentage": 39.41, "elapsed_time": "7:48:17", "remaining_time": "11:59:53", "throughput": 636.9, "total_tokens": 17895392}
{"current_steps": 1535, "total_steps": 3882, "loss": 0.4206, "lr": 7.079893112344118e-05, "epoch": 0.7906258047901107, "percentage": 39.54, "elapsed_time": "7:49:45", "remaining_time": "11:58:15", "throughput": 636.98, "total_tokens": 17953872}
{"current_steps": 1540, "total_steps": 3882, "loss": 0.3567, "lr": 7.060502964110418e-05, "epoch": 0.7932011331444759, "percentage": 39.67, "elapsed_time": "7:51:16", "remaining_time": "11:56:42", "throughput": 637.01, "total_tokens": 18012320}
{"current_steps": 1545, "total_steps": 3882, "loss": 0.4172, "lr": 7.041075416353513e-05, "epoch": 0.7957764614988411, "percentage": 39.8, "elapsed_time": "7:52:48", "remaining_time": "11:55:10", "throughput": 637.01, "total_tokens": 18070792}
{"current_steps": 1550, "total_steps": 3882, "loss": 0.4666, "lr": 7.02161082169654e-05, "epoch": 0.7983517898532063, "percentage": 39.93, "elapsed_time": "7:54:15", "remaining_time": "11:53:32", "throughput": 637.1, "total_tokens": 18129304}
{"current_steps": 1550, "total_steps": 3882, "eval_loss": 0.42660555243492126, "epoch": 0.7983517898532063, "percentage": 39.93, "elapsed_time": "7:54:35", "remaining_time": "11:54:01", "throughput": 636.67, "total_tokens": 18129304}
{"current_steps": 1555, "total_steps": 3882, "loss": 0.4305, "lr": 7.002109533435066e-05, "epoch": 0.8009271182075715, "percentage": 40.06, "elapsed_time": "7:56:09", "remaining_time": "11:52:33", "throughput": 636.61, "total_tokens": 18187728}
{"current_steps": 1560, "total_steps": 3882, "loss": 0.4529, "lr": 6.982571905530669e-05, "epoch": 0.8035024465619367, "percentage": 40.19, "elapsed_time": "7:57:38", "remaining_time": "11:50:56", "throughput": 636.69, "total_tokens": 18246192}
{"current_steps": 1565, "total_steps": 3882, "loss": 0.4569, "lr": 6.962998292604517e-05, "epoch": 0.8060777749163018, "percentage": 40.31, "elapsed_time": "7:59:07", "remaining_time": "11:49:21", "throughput": 636.74, "total_tokens": 18304632}
{"current_steps": 1570, "total_steps": 3882, "loss": 0.3936, "lr": 6.943389049930931e-05, "epoch": 0.808653103270667, "percentage": 40.44, "elapsed_time": "8:00:35", "remaining_time": "11:47:44", "throughput": 636.81, "total_tokens": 18363136}
{"current_steps": 1575, "total_steps": 3882, "loss": 0.4083, "lr": 6.923744533430937e-05, "epoch": 0.8112284316250322, "percentage": 40.57, "elapsed_time": "8:02:05", "remaining_time": "11:46:08", "throughput": 636.87, "total_tokens": 18421592}
{"current_steps": 1580, "total_steps": 3882, "loss": 0.4564, "lr": 6.904065099665803e-05, "epoch": 0.8138037599793974, "percentage": 40.7, "elapsed_time": "8:03:34", "remaining_time": "11:44:32", "throughput": 636.94, "total_tokens": 18480104}
{"current_steps": 1585, "total_steps": 3882, "loss": 0.3928, "lr": 6.884351105830568e-05, "epoch": 0.8163790883337626, "percentage": 40.83, "elapsed_time": "8:05:03", "remaining_time": "11:42:56", "throughput": 636.99, "total_tokens": 18538600}
{"current_steps": 1590, "total_steps": 3882, "loss": 0.4278, "lr": 6.864602909747563e-05, "epoch": 0.8189544166881277, "percentage": 40.96, "elapsed_time": "8:06:31", "remaining_time": "11:41:20", "throughput": 637.07, "total_tokens": 18597104}
{"current_steps": 1595, "total_steps": 3882, "loss": 0.3956, "lr": 6.84482086985991e-05, "epoch": 0.8215297450424929, "percentage": 41.09, "elapsed_time": "8:08:00", "remaining_time": "11:39:44", "throughput": 637.13, "total_tokens": 18655584}
{"current_steps": 1600, "total_steps": 3882, "loss": 0.3834, "lr": 6.825005345225019e-05, "epoch": 0.8241050733968581, "percentage": 41.22, "elapsed_time": "8:09:29", "remaining_time": "11:38:07", "throughput": 637.2, "total_tokens": 18714072}
{"current_steps": 1600, "total_steps": 3882, "eval_loss": 0.44774264097213745, "epoch": 0.8241050733968581, "percentage": 41.22, "elapsed_time": "8:09:48", "remaining_time": "11:38:35", "throughput": 636.78, "total_tokens": 18714072}
{"current_steps": 1605, "total_steps": 3882, "loss": 0.4136, "lr": 6.805156695508075e-05, "epoch": 0.8266804017512233, "percentage": 41.34, "elapsed_time": "8:11:23", "remaining_time": "11:37:08", "throughput": 636.71, "total_tokens": 18772552}
{"current_steps": 1610, "total_steps": 3882, "loss": 0.3523, "lr": 6.7852752809755e-05, "epoch": 0.8292557301055885, "percentage": 41.47, "elapsed_time": "8:12:52", "remaining_time": "11:35:31", "throughput": 636.78, "total_tokens": 18830992}
{"current_steps": 1615, "total_steps": 3882, "loss": 0.2983, "lr": 6.765361462488424e-05, "epoch": 0.8318310584599536, "percentage": 41.6, "elapsed_time": "8:14:21", "remaining_time": "11:33:56", "throughput": 636.83, "total_tokens": 18889480}
{"current_steps": 1620, "total_steps": 3882, "loss": 0.3987, "lr": 6.745415601496127e-05, "epoch": 0.8344063868143188, "percentage": 41.73, "elapsed_time": "8:15:49", "remaining_time": "11:32:19", "throughput": 636.91, "total_tokens": 18947936}
{"current_steps": 1625, "total_steps": 3882, "loss": 0.5012, "lr": 6.725438060029485e-05, "epoch": 0.836981715168684, "percentage": 41.86, "elapsed_time": "8:17:19", "remaining_time": "11:30:45", "throughput": 636.95, "total_tokens": 19006432}
{"current_steps": 1630, "total_steps": 3882, "loss": 0.36, "lr": 6.705429200694396e-05, "epoch": 0.8395570435230492, "percentage": 41.99, "elapsed_time": "8:18:48", "remaining_time": "11:29:08", "throughput": 637.02, "total_tokens": 19064920}
{"current_steps": 1635, "total_steps": 3882, "loss": 0.3816, "lr": 6.685389386665197e-05, "epoch": 0.8421323718774144, "percentage": 42.12, "elapsed_time": "8:20:17", "remaining_time": "11:27:33", "throughput": 637.07, "total_tokens": 19123376}
{"current_steps": 1640, "total_steps": 3882, "loss": 0.3503, "lr": 6.665318981678072e-05, "epoch": 0.8447077002317795, "percentage": 42.25, "elapsed_time": "8:21:46", "remaining_time": "11:25:57", "throughput": 637.13, "total_tokens": 19181864}
{"current_steps": 1645, "total_steps": 3882, "loss": 0.3644, "lr": 6.645218350024456e-05, "epoch": 0.8472830285861447, "percentage": 42.38, "elapsed_time": "8:23:14", "remaining_time": "11:24:20", "throughput": 637.21, "total_tokens": 19240352}
{"current_steps": 1650, "total_steps": 3882, "loss": 0.475, "lr": 6.625087856544416e-05, "epoch": 0.8498583569405099, "percentage": 42.5, "elapsed_time": "8:24:42", "remaining_time": "11:22:43", "throughput": 637.3, "total_tokens": 19298848}
{"current_steps": 1650, "total_steps": 3882, "eval_loss": 0.35129043459892273, "epoch": 0.8498583569405099, "percentage": 42.5, "elapsed_time": "8:25:01", "remaining_time": "11:23:09", "throughput": 636.9, "total_tokens": 19298848}
{"current_steps": 1655, "total_steps": 3882, "loss": 0.3414, "lr": 6.604927866620031e-05, "epoch": 0.8524336852948751, "percentage": 42.63, "elapsed_time": "8:26:34", "remaining_time": "11:21:39", "throughput": 636.87, "total_tokens": 19357304}
{"current_steps": 1660, "total_steps": 3882, "loss": 0.4728, "lr": 6.584738746168762e-05, "epoch": 0.8550090136492403, "percentage": 42.76, "elapsed_time": "8:28:02", "remaining_time": "11:20:02", "throughput": 636.95, "total_tokens": 19415800}
{"current_steps": 1665, "total_steps": 3882, "loss": 0.4264, "lr": 6.564520861636808e-05, "epoch": 0.8575843420036054, "percentage": 42.89, "elapsed_time": "8:29:31", "remaining_time": "11:18:26", "throughput": 637.01, "total_tokens": 19474256}
{"current_steps": 1670, "total_steps": 3882, "loss": 0.4038, "lr": 6.544274579992453e-05, "epoch": 0.8601596703579707, "percentage": 43.02, "elapsed_time": "8:31:00", "remaining_time": "11:16:51", "throughput": 637.07, "total_tokens": 19532736}
{"current_steps": 1675, "total_steps": 3882, "loss": 0.3241, "lr": 6.524000268719411e-05, "epoch": 0.8627349987123358, "percentage": 43.15, "elapsed_time": "8:32:29", "remaining_time": "11:15:15", "throughput": 637.13, "total_tokens": 19591184}
{"current_steps": 1680, "total_steps": 3882, "loss": 0.3248, "lr": 6.503698295810154e-05, "epoch": 0.865310327066701, "percentage": 43.28, "elapsed_time": "8:33:58", "remaining_time": "11:13:39", "throughput": 637.19, "total_tokens": 19649656}
{"current_steps": 1685, "total_steps": 3882, "loss": 0.4295, "lr": 6.483369029759229e-05, "epoch": 0.8678856554210662, "percentage": 43.41, "elapsed_time": "8:35:26", "remaining_time": "11:12:03", "throughput": 637.26, "total_tokens": 19708144}
{"current_steps": 1690, "total_steps": 3882, "loss": 0.3785, "lr": 6.463012839556569e-05, "epoch": 0.8704609837754314, "percentage": 43.53, "elapsed_time": "8:36:55", "remaining_time": "11:10:28", "throughput": 637.3, "total_tokens": 19766592}
{"current_steps": 1695, "total_steps": 3882, "loss": 0.3376, "lr": 6.442630094680805e-05, "epoch": 0.8730363121297966, "percentage": 43.66, "elapsed_time": "8:38:24", "remaining_time": "11:08:52", "throughput": 637.37, "total_tokens": 19825064}
{"current_steps": 1700, "total_steps": 3882, "loss": 0.3752, "lr": 6.42222116509255e-05, "epoch": 0.8756116404841617, "percentage": 43.79, "elapsed_time": "8:39:53", "remaining_time": "11:07:17", "throughput": 637.42, "total_tokens": 19883504}
{"current_steps": 1700, "total_steps": 3882, "eval_loss": 0.34382957220077515, "epoch": 0.8756116404841617, "percentage": 43.79, "elapsed_time": "8:40:12", "remaining_time": "11:07:42", "throughput": 637.03, "total_tokens": 19883504}
{"current_steps": 1705, "total_steps": 3882, "loss": 0.3453, "lr": 6.401786421227686e-05, "epoch": 0.8781869688385269, "percentage": 43.92, "elapsed_time": "8:41:46", "remaining_time": "11:06:13", "throughput": 636.99, "total_tokens": 19942000}
{"current_steps": 1710, "total_steps": 3882, "loss": 0.3278, "lr": 6.381326233990644e-05, "epoch": 0.8807622971928921, "percentage": 44.05, "elapsed_time": "8:43:16", "remaining_time": "11:04:38", "throughput": 637.03, "total_tokens": 20000440}
{"current_steps": 1715, "total_steps": 3882, "loss": 0.3695, "lr": 6.360840974747666e-05, "epoch": 0.8833376255472573, "percentage": 44.18, "elapsed_time": "8:44:44", "remaining_time": "11:03:02", "throughput": 637.1, "total_tokens": 20058920}
{"current_steps": 1720, "total_steps": 3882, "loss": 0.4362, "lr": 6.340331015320073e-05, "epoch": 0.8859129539016225, "percentage": 44.31, "elapsed_time": "8:46:13", "remaining_time": "11:01:27", "throughput": 637.16, "total_tokens": 20117416}
{"current_steps": 1725, "total_steps": 3882, "loss": 0.3178, "lr": 6.319796727977509e-05, "epoch": 0.8884882822559876, "percentage": 44.44, "elapsed_time": "8:47:42", "remaining_time": "10:59:51", "throughput": 637.23, "total_tokens": 20175936}
{"current_steps": 1730, "total_steps": 3882, "loss": 0.3679, "lr": 6.299238485431183e-05, "epoch": 0.8910636106103528, "percentage": 44.56, "elapsed_time": "8:49:10", "remaining_time": "10:58:16", "throughput": 637.28, "total_tokens": 20234368}
{"current_steps": 1735, "total_steps": 3882, "loss": 0.3529, "lr": 6.27865666082711e-05, "epoch": 0.893638938964718, "percentage": 44.69, "elapsed_time": "8:50:39", "remaining_time": "10:56:40", "throughput": 637.35, "total_tokens": 20292856}
{"current_steps": 1740, "total_steps": 3882, "loss": 0.281, "lr": 6.258051627739337e-05, "epoch": 0.8962142673190832, "percentage": 44.82, "elapsed_time": "8:52:08", "remaining_time": "10:55:04", "throughput": 637.41, "total_tokens": 20351272}
{"current_steps": 1745, "total_steps": 3882, "loss": 0.3148, "lr": 6.237423760163156e-05, "epoch": 0.8987895956734484, "percentage": 44.95, "elapsed_time": "8:53:36", "remaining_time": "10:53:29", "throughput": 637.47, "total_tokens": 20409744}
{"current_steps": 1750, "total_steps": 3882, "loss": 0.3233, "lr": 6.216773432508325e-05, "epoch": 0.9013649240278135, "percentage": 45.08, "elapsed_time": "8:55:06", "remaining_time": "10:51:54", "throughput": 637.52, "total_tokens": 20468200}
{"current_steps": 1750, "total_steps": 3882, "eval_loss": 0.3325226604938507, "epoch": 0.9013649240278135, "percentage": 45.08, "elapsed_time": "8:55:25", "remaining_time": "10:52:18", "throughput": 637.13, "total_tokens": 20468200}
{"current_steps": 1755, "total_steps": 3882, "loss": 0.3731, "lr": 6.196101019592264e-05, "epoch": 0.9039402523821787, "percentage": 45.21, "elapsed_time": "8:56:59", "remaining_time": "10:50:49", "throughput": 637.08, "total_tokens": 20526696}
{"current_steps": 1760, "total_steps": 3882, "loss": 0.3092, "lr": 6.175406896633258e-05, "epoch": 0.9065155807365439, "percentage": 45.34, "elapsed_time": "8:58:28", "remaining_time": "10:49:14", "throughput": 637.14, "total_tokens": 20585160}
{"current_steps": 1765, "total_steps": 3882, "loss": 0.311, "lr": 6.154691439243644e-05, "epoch": 0.9090909090909091, "percentage": 45.47, "elapsed_time": "8:59:57", "remaining_time": "10:47:38", "throughput": 637.2, "total_tokens": 20643672}
{"current_steps": 1770, "total_steps": 3882, "loss": 0.3085, "lr": 6.133955023422991e-05, "epoch": 0.9116662374452743, "percentage": 45.6, "elapsed_time": "9:01:26", "remaining_time": "10:46:03", "throughput": 637.26, "total_tokens": 20702152}
{"current_steps": 1775, "total_steps": 3882, "loss": 0.3236, "lr": 6.11319802555128e-05, "epoch": 0.9142415657996394, "percentage": 45.72, "elapsed_time": "9:02:54", "remaining_time": "10:44:27", "throughput": 637.32, "total_tokens": 20760664}
{"current_steps": 1780, "total_steps": 3882, "loss": 0.4067, "lr": 6.092420822382069e-05, "epoch": 0.9168168941540046, "percentage": 45.85, "elapsed_time": "9:04:23", "remaining_time": "10:42:52", "throughput": 637.37, "total_tokens": 20819144}
{"current_steps": 1785, "total_steps": 3882, "loss": 0.3228, "lr": 6.071623791035657e-05, "epoch": 0.9193922225083698, "percentage": 45.98, "elapsed_time": "9:05:52", "remaining_time": "10:41:17", "throughput": 637.43, "total_tokens": 20877624}
{"current_steps": 1790, "total_steps": 3882, "loss": 0.2672, "lr": 6.050807308992234e-05, "epoch": 0.921967550862735, "percentage": 46.11, "elapsed_time": "9:07:21", "remaining_time": "10:39:41", "throughput": 637.5, "total_tokens": 20936112}
{"current_steps": 1795, "total_steps": 3882, "loss": 0.3308, "lr": 6.02997175408504e-05, "epoch": 0.9245428792171002, "percentage": 46.24, "elapsed_time": "9:08:50", "remaining_time": "10:38:07", "throughput": 637.55, "total_tokens": 20994608}
{"current_steps": 1800, "total_steps": 3882, "loss": 0.3279, "lr": 6.009117504493493e-05, "epoch": 0.9271182075714653, "percentage": 46.37, "elapsed_time": "9:10:18", "remaining_time": "10:36:31", "throughput": 637.61, "total_tokens": 21053080}
{"current_steps": 1800, "total_steps": 3882, "eval_loss": 0.3501794636249542, "epoch": 0.9271182075714653, "percentage": 46.37, "elapsed_time": "9:10:37", "remaining_time": "10:36:53", "throughput": 637.24, "total_tokens": 21053080}
{"current_steps": 1805, "total_steps": 3882, "loss": 0.3107, "lr": 5.98824493873634e-05, "epoch": 0.9296935359258306, "percentage": 46.5, "elapsed_time": "9:12:12", "remaining_time": "10:35:25", "throughput": 637.19, "total_tokens": 21111592}
{"current_steps": 1810, "total_steps": 3882, "loss": 0.3453, "lr": 5.9673544356647706e-05, "epoch": 0.9322688642801957, "percentage": 46.63, "elapsed_time": "9:13:40", "remaining_time": "10:33:49", "throughput": 637.25, "total_tokens": 21170024}
{"current_steps": 1815, "total_steps": 3882, "loss": 0.3607, "lr": 5.946446374455555e-05, "epoch": 0.9348441926345609, "percentage": 46.75, "elapsed_time": "9:15:09", "remaining_time": "10:32:14", "throughput": 637.31, "total_tokens": 21228480}
{"current_steps": 1820, "total_steps": 3882, "loss": 0.3375, "lr": 5.9255211346041526e-05, "epoch": 0.9374195209889261, "percentage": 46.88, "elapsed_time": "9:16:38", "remaining_time": "10:30:39", "throughput": 637.36, "total_tokens": 21286944}
{"current_steps": 1825, "total_steps": 3882, "loss": 0.2521, "lr": 5.9045790959178296e-05, "epoch": 0.9399948493432912, "percentage": 47.01, "elapsed_time": "9:18:08", "remaining_time": "10:29:05", "throughput": 637.4, "total_tokens": 21345456}
{"current_steps": 1830, "total_steps": 3882, "loss": 0.3855, "lr": 5.883620638508756e-05, "epoch": 0.9425701776976565, "percentage": 47.14, "elapsed_time": "9:19:36", "remaining_time": "10:27:30", "throughput": 637.46, "total_tokens": 21403896}
{"current_steps": 1835, "total_steps": 3882, "loss": 0.2947, "lr": 5.8626461427871204e-05, "epoch": 0.9451455060520216, "percentage": 47.27, "elapsed_time": "9:21:05", "remaining_time": "10:25:55", "throughput": 637.51, "total_tokens": 21462360}
{"current_steps": 1840, "total_steps": 3882, "loss": 0.4505, "lr": 5.841655989454213e-05, "epoch": 0.9477208344063868, "percentage": 47.4, "elapsed_time": "9:22:34", "remaining_time": "10:24:19", "throughput": 637.57, "total_tokens": 21520864}
{"current_steps": 1845, "total_steps": 3882, "loss": 0.3758, "lr": 5.820650559495523e-05, "epoch": 0.950296162760752, "percentage": 47.53, "elapsed_time": "9:24:03", "remaining_time": "10:22:45", "throughput": 637.63, "total_tokens": 21579376}
{"current_steps": 1850, "total_steps": 3882, "loss": 0.3221, "lr": 5.7996302341738164e-05, "epoch": 0.9528714911151172, "percentage": 47.66, "elapsed_time": "9:25:32", "remaining_time": "10:21:10", "throughput": 637.67, "total_tokens": 21637848}
{"current_steps": 1850, "total_steps": 3882, "eval_loss": 0.2935050129890442, "epoch": 0.9528714911151172, "percentage": 47.66, "elapsed_time": "9:25:51", "remaining_time": "10:21:32", "throughput": 637.31, "total_tokens": 21637848}
{"current_steps": 1855, "total_steps": 3882, "loss": 0.4205, "lr": 5.778595395022226e-05, "epoch": 0.9554468194694824, "percentage": 47.78, "elapsed_time": "9:27:26", "remaining_time": "10:20:03", "throughput": 637.26, "total_tokens": 21696328}
{"current_steps": 1860, "total_steps": 3882, "loss": 0.3075, "lr": 5.757546423837314e-05, "epoch": 0.9580221478238475, "percentage": 47.91, "elapsed_time": "9:28:55", "remaining_time": "10:18:28", "throughput": 637.31, "total_tokens": 21754816}
{"current_steps": 1865, "total_steps": 3882, "loss": 0.2604, "lr": 5.736483702672155e-05, "epoch": 0.9605974761782127, "percentage": 48.04, "elapsed_time": "9:30:24", "remaining_time": "10:16:53", "throughput": 637.37, "total_tokens": 21813296}
{"current_steps": 1870, "total_steps": 3882, "loss": 0.342, "lr": 5.7154076138293914e-05, "epoch": 0.9631728045325779, "percentage": 48.17, "elapsed_time": "9:31:52", "remaining_time": "10:15:18", "throughput": 637.42, "total_tokens": 21871800}
{"current_steps": 1875, "total_steps": 3882, "loss": 0.411, "lr": 5.694318539854297e-05, "epoch": 0.9657481328869431, "percentage": 48.3, "elapsed_time": "9:33:21", "remaining_time": "10:13:43", "throughput": 637.48, "total_tokens": 21930272}
{"current_steps": 1880, "total_steps": 3882, "loss": 0.2606, "lr": 5.673216863527836e-05, "epoch": 0.9683234612413083, "percentage": 48.43, "elapsed_time": "9:34:50", "remaining_time": "10:12:09", "throughput": 637.52, "total_tokens": 21988736}
{"current_steps": 1885, "total_steps": 3882, "loss": 0.2647, "lr": 5.652102967859715e-05, "epoch": 0.9708987895956734, "percentage": 48.56, "elapsed_time": "9:36:19", "remaining_time": "10:10:34", "throughput": 637.58, "total_tokens": 22047200}
{"current_steps": 1890, "total_steps": 3882, "loss": 0.2961, "lr": 5.6309772360814295e-05, "epoch": 0.9734741179500386, "percentage": 48.69, "elapsed_time": "9:37:49", "remaining_time": "10:09:00", "throughput": 637.61, "total_tokens": 22105664}
{"current_steps": 1895, "total_steps": 3882, "loss": 0.2776, "lr": 5.6098400516393065e-05, "epoch": 0.9760494463044038, "percentage": 48.82, "elapsed_time": "9:39:18", "remaining_time": "10:07:25", "throughput": 637.67, "total_tokens": 22164184}
{"current_steps": 1900, "total_steps": 3882, "loss": 0.3781, "lr": 5.5886917981875485e-05, "epoch": 0.978624774658769, "percentage": 48.94, "elapsed_time": "9:40:47", "remaining_time": "10:05:51", "throughput": 637.71, "total_tokens": 22222632}
{"current_steps": 1900, "total_steps": 3882, "eval_loss": 0.297338604927063, "epoch": 0.978624774658769, "percentage": 48.94, "elapsed_time": "9:41:07", "remaining_time": "10:06:11", "throughput": 637.35, "total_tokens": 22222632}
{"current_steps": 1905, "total_steps": 3882, "loss": 0.2966, "lr": 5.567532859581267e-05, "epoch": 0.9812001030131342, "percentage": 49.07, "elapsed_time": "9:42:40", "remaining_time": "10:04:41", "throughput": 637.32, "total_tokens": 22281080}
{"current_steps": 1910, "total_steps": 3882, "loss": 0.304, "lr": 5.5463636198695166e-05, "epoch": 0.9837754313674993, "percentage": 49.2, "elapsed_time": "9:44:08", "remaining_time": "10:03:06", "throughput": 637.39, "total_tokens": 22339552}
{"current_steps": 1915, "total_steps": 3882, "loss": 0.3534, "lr": 5.5251844632883244e-05, "epoch": 0.9863507597218646, "percentage": 49.33, "elapsed_time": "9:45:36", "remaining_time": "10:01:30", "throughput": 637.45, "total_tokens": 22398056}
{"current_steps": 1920, "total_steps": 3882, "loss": 0.319, "lr": 5.503995774253714e-05, "epoch": 0.9889260880762297, "percentage": 49.46, "elapsed_time": "9:47:05", "remaining_time": "9:59:55", "throughput": 637.52, "total_tokens": 22456520}
{"current_steps": 1925, "total_steps": 3882, "loss": 0.312, "lr": 5.482797937354731e-05, "epoch": 0.9915014164305949, "percentage": 49.59, "elapsed_time": "9:48:32", "remaining_time": "9:58:19", "throughput": 637.59, "total_tokens": 22515024}
{"current_steps": 1930, "total_steps": 3882, "loss": 0.3001, "lr": 5.461591337346461e-05, "epoch": 0.9940767447849601, "percentage": 49.72, "elapsed_time": "9:50:01", "remaining_time": "9:56:44", "throughput": 637.65, "total_tokens": 22573480}
{"current_steps": 1935, "total_steps": 3882, "loss": 0.2938, "lr": 5.4403763591430436e-05, "epoch": 0.9966520731393252, "percentage": 49.85, "elapsed_time": "9:51:29", "remaining_time": "9:55:09", "throughput": 637.7, "total_tokens": 22631960}
{"current_steps": 1940, "total_steps": 3882, "loss": 0.3429, "lr": 5.419153387810693e-05, "epoch": 0.9992274014936905, "percentage": 49.97, "elapsed_time": "9:52:58", "remaining_time": "9:53:35", "throughput": 637.75, "total_tokens": 22690432}
{"current_steps": 1945, "total_steps": 3882, "loss": 0.292, "lr": 5.397922808560698e-05, "epoch": 1.001545197012619, "percentage": 50.1, "elapsed_time": "9:54:18", "remaining_time": "9:51:51", "throughput": 637.81, "total_tokens": 22743048}
{"current_steps": 1950, "total_steps": 3882, "loss": 0.2845, "lr": 5.376685006742441e-05, "epoch": 1.0041205253669843, "percentage": 50.23, "elapsed_time": "9:55:46", "remaining_time": "9:50:16", "throughput": 637.86, "total_tokens": 22801512}
{"current_steps": 1950, "total_steps": 3882, "eval_loss": 0.2472737729549408, "epoch": 1.0041205253669843, "percentage": 50.23, "elapsed_time": "9:56:06", "remaining_time": "9:50:35", "throughput": 637.52, "total_tokens": 22801512}
{"current_steps": 1955, "total_steps": 3882, "loss": 0.278, "lr": 5.355440367836396e-05, "epoch": 1.0066958537213495, "percentage": 50.36, "elapsed_time": "9:57:40", "remaining_time": "9:49:06", "throughput": 637.48, "total_tokens": 22859952}
{"current_steps": 1960, "total_steps": 3882, "loss": 0.3433, "lr": 5.334189277447138e-05, "epoch": 1.0092711820757148, "percentage": 50.49, "elapsed_time": "9:59:07", "remaining_time": "9:47:31", "throughput": 637.54, "total_tokens": 22918440}
{"current_steps": 1965, "total_steps": 3882, "loss": 0.2929, "lr": 5.312932121296339e-05, "epoch": 1.0118465104300798, "percentage": 50.62, "elapsed_time": "10:00:36", "remaining_time": "9:45:56", "throughput": 637.6, "total_tokens": 22976944}
{"current_steps": 1970, "total_steps": 3882, "loss": 0.2804, "lr": 5.291669285215766e-05, "epoch": 1.014421838784445, "percentage": 50.75, "elapsed_time": "10:02:04", "remaining_time": "9:44:21", "throughput": 637.66, "total_tokens": 23035408}
{"current_steps": 1975, "total_steps": 3882, "loss": 0.2583, "lr": 5.270401155140284e-05, "epoch": 1.0169971671388103, "percentage": 50.88, "elapsed_time": "10:03:33", "remaining_time": "9:42:46", "throughput": 637.71, "total_tokens": 23093912}
{"current_steps": 1980, "total_steps": 3882, "loss": 0.3232, "lr": 5.2491281171008476e-05, "epoch": 1.0195724954931753, "percentage": 51.0, "elapsed_time": "10:05:01", "remaining_time": "9:41:11", "throughput": 637.78, "total_tokens": 23152344}
{"current_steps": 1985, "total_steps": 3882, "loss": 0.2935, "lr": 5.227850557217494e-05, "epoch": 1.0221478238475405, "percentage": 51.13, "elapsed_time": "10:06:30", "remaining_time": "9:39:37", "throughput": 637.83, "total_tokens": 23210800}
{"current_steps": 1990, "total_steps": 3882, "loss": 0.2677, "lr": 5.2065688616923314e-05, "epoch": 1.0247231522019058, "percentage": 51.26, "elapsed_time": "10:07:58", "remaining_time": "9:38:02", "throughput": 637.89, "total_tokens": 23269304}
{"current_steps": 1995, "total_steps": 3882, "loss": 0.2697, "lr": 5.185283416802539e-05, "epoch": 1.0272984805562708, "percentage": 51.39, "elapsed_time": "10:09:27", "remaining_time": "9:36:28", "throughput": 637.93, "total_tokens": 23327800}
{"current_steps": 2000, "total_steps": 3882, "loss": 0.2272, "lr": 5.1639946088933444e-05, "epoch": 1.029873808910636, "percentage": 51.52, "elapsed_time": "10:10:55", "remaining_time": "9:34:53", "throughput": 637.99, "total_tokens": 23386232}
{"current_steps": 2000, "total_steps": 3882, "eval_loss": 0.2834003269672394, "epoch": 1.029873808910636, "percentage": 51.52, "elapsed_time": "10:11:15", "remaining_time": "9:35:11", "throughput": 637.66, "total_tokens": 23386232}
{"current_steps": 2005, "total_steps": 3882, "loss": 0.2455, "lr": 5.1427028243710174e-05, "epoch": 1.0324491372650013, "percentage": 51.65, "elapsed_time": "10:12:49", "remaining_time": "9:33:41", "throughput": 637.62, "total_tokens": 23444712}
{"current_steps": 2010, "total_steps": 3882, "loss": 0.268, "lr": 5.121408449695856e-05, "epoch": 1.0350244656193666, "percentage": 51.78, "elapsed_time": "10:14:18", "remaining_time": "9:32:07", "throughput": 637.66, "total_tokens": 23503192}
{"current_steps": 2015, "total_steps": 3882, "loss": 0.2721, "lr": 5.100111871375173e-05, "epoch": 1.0375997939737316, "percentage": 51.91, "elapsed_time": "10:15:46", "remaining_time": "9:30:32", "throughput": 637.72, "total_tokens": 23561688}
{"current_steps": 2020, "total_steps": 3882, "loss": 0.3521, "lr": 5.078813475956276e-05, "epoch": 1.0401751223280968, "percentage": 52.04, "elapsed_time": "10:17:15", "remaining_time": "9:28:58", "throughput": 637.78, "total_tokens": 23620160}
{"current_steps": 2025, "total_steps": 3882, "loss": 0.3624, "lr": 5.057513650019452e-05, "epoch": 1.042750450682462, "percentage": 52.16, "elapsed_time": "10:18:43", "remaining_time": "9:27:23", "throughput": 637.83, "total_tokens": 23678624}
{"current_steps": 2030, "total_steps": 3882, "loss": 0.2742, "lr": 5.0362127801709614e-05, "epoch": 1.045325779036827, "percentage": 52.29, "elapsed_time": "10:20:12", "remaining_time": "9:25:49", "throughput": 637.88, "total_tokens": 23737064}
{"current_steps": 2035, "total_steps": 3882, "loss": 0.3377, "lr": 5.014911253036004e-05, "epoch": 1.0479011073911924, "percentage": 52.42, "elapsed_time": "10:21:40", "remaining_time": "9:24:14", "throughput": 637.93, "total_tokens": 23795568}
{"current_steps": 2040, "total_steps": 3882, "loss": 0.3059, "lr": 4.993609455251713e-05, "epoch": 1.0504764357455576, "percentage": 52.55, "elapsed_time": "10:23:09", "remaining_time": "9:22:40", "throughput": 637.99, "total_tokens": 23854040}
{"current_steps": 2045, "total_steps": 3882, "loss": 0.2534, "lr": 4.972307773460133e-05, "epoch": 1.0530517640999228, "percentage": 52.68, "elapsed_time": "10:24:37", "remaining_time": "9:21:05", "throughput": 638.06, "total_tokens": 23912520}
{"current_steps": 2050, "total_steps": 3882, "loss": 0.2924, "lr": 4.9510065943012076e-05, "epoch": 1.0556270924542879, "percentage": 52.81, "elapsed_time": "10:26:05", "remaining_time": "9:19:30", "throughput": 638.11, "total_tokens": 23971048}
{"current_steps": 2050, "total_steps": 3882, "eval_loss": 0.2703675329685211, "epoch": 1.0556270924542879, "percentage": 52.81, "elapsed_time": "10:26:25", "remaining_time": "9:19:48", "throughput": 637.78, "total_tokens": 23971048}
{"current_steps": 2055, "total_steps": 3882, "loss": 0.217, "lr": 4.929706304405748e-05, "epoch": 1.0582024208086531, "percentage": 52.94, "elapsed_time": "10:27:58", "remaining_time": "9:18:18", "throughput": 637.75, "total_tokens": 24029544}
{"current_steps": 2060, "total_steps": 3882, "loss": 0.2132, "lr": 4.9084072903884345e-05, "epoch": 1.0607777491630184, "percentage": 53.07, "elapsed_time": "10:29:27", "remaining_time": "9:16:44", "throughput": 637.8, "total_tokens": 24088008}
{"current_steps": 2065, "total_steps": 3882, "loss": 0.2438, "lr": 4.887109938840783e-05, "epoch": 1.0633530775173834, "percentage": 53.19, "elapsed_time": "10:30:55", "remaining_time": "9:15:08", "throughput": 637.86, "total_tokens": 24146480}
{"current_steps": 2070, "total_steps": 3882, "loss": 0.2623, "lr": 4.8658146363241406e-05, "epoch": 1.0659284058717486, "percentage": 53.32, "elapsed_time": "10:32:23", "remaining_time": "9:13:34", "throughput": 637.92, "total_tokens": 24204984}
{"current_steps": 2075, "total_steps": 3882, "loss": 0.2041, "lr": 4.844521769362654e-05, "epoch": 1.0685037342261139, "percentage": 53.45, "elapsed_time": "10:33:51", "remaining_time": "9:11:59", "throughput": 637.99, "total_tokens": 24263456}
{"current_steps": 2080, "total_steps": 3882, "loss": 0.246, "lr": 4.823231724436271e-05, "epoch": 1.071079062580479, "percentage": 53.58, "elapsed_time": "10:35:19", "remaining_time": "9:10:25", "throughput": 638.04, "total_tokens": 24321960}
{"current_steps": 2085, "total_steps": 3882, "loss": 0.2857, "lr": 4.801944887973714e-05, "epoch": 1.0736543909348442, "percentage": 53.71, "elapsed_time": "10:36:47", "remaining_time": "9:08:50", "throughput": 638.1, "total_tokens": 24380464}
{"current_steps": 2090, "total_steps": 3882, "loss": 0.2577, "lr": 4.7806616463454715e-05, "epoch": 1.0762297192892094, "percentage": 53.84, "elapsed_time": "10:38:16", "remaining_time": "9:07:15", "throughput": 638.16, "total_tokens": 24438976}
{"current_steps": 2095, "total_steps": 3882, "loss": 0.2523, "lr": 4.759382385856779e-05, "epoch": 1.0788050476435747, "percentage": 53.97, "elapsed_time": "10:39:45", "remaining_time": "9:05:41", "throughput": 638.2, "total_tokens": 24497448}
{"current_steps": 2100, "total_steps": 3882, "loss": 0.2805, "lr": 4.738107492740619e-05, "epoch": 1.0813803759979397, "percentage": 54.1, "elapsed_time": "10:41:13", "remaining_time": "9:04:07", "throughput": 638.26, "total_tokens": 24555904}
{"current_steps": 2100, "total_steps": 3882, "eval_loss": 0.3205481469631195, "epoch": 1.0813803759979397, "percentage": 54.1, "elapsed_time": "10:41:32", "remaining_time": "9:04:23", "throughput": 637.94, "total_tokens": 24555904}
{"current_steps": 2105, "total_steps": 3882, "loss": 0.2942, "lr": 4.716837353150695e-05, "epoch": 1.083955704352305, "percentage": 54.22, "elapsed_time": "10:43:07", "remaining_time": "9:02:54", "throughput": 637.89, "total_tokens": 24614392}
{"current_steps": 2110, "total_steps": 3882, "loss": 0.2357, "lr": 4.695572353154438e-05, "epoch": 1.0865310327066702, "percentage": 54.35, "elapsed_time": "10:44:36", "remaining_time": "9:01:21", "throughput": 637.92, "total_tokens": 24672848}
{"current_steps": 2115, "total_steps": 3882, "loss": 0.2476, "lr": 4.674312878725985e-05, "epoch": 1.0891063610610352, "percentage": 54.48, "elapsed_time": "10:46:06", "remaining_time": "8:59:48", "throughput": 637.95, "total_tokens": 24731344}
{"current_steps": 2120, "total_steps": 3882, "loss": 0.3875, "lr": 4.653059315739188e-05, "epoch": 1.0916816894154004, "percentage": 54.61, "elapsed_time": "10:47:35", "remaining_time": "8:58:14", "throughput": 637.99, "total_tokens": 24789832}
{"current_steps": 2125, "total_steps": 3882, "loss": 0.3313, "lr": 4.631812049960595e-05, "epoch": 1.0942570177697657, "percentage": 54.74, "elapsed_time": "10:49:04", "remaining_time": "8:56:40", "throughput": 638.04, "total_tokens": 24848264}
{"current_steps": 2130, "total_steps": 3882, "loss": 0.2385, "lr": 4.61057146704246e-05, "epoch": 1.0968323461241307, "percentage": 54.87, "elapsed_time": "10:50:33", "remaining_time": "8:55:06", "throughput": 638.08, "total_tokens": 24906744}
{"current_steps": 2135, "total_steps": 3882, "loss": 0.2478, "lr": 4.589337952515736e-05, "epoch": 1.099407674478496, "percentage": 55.0, "elapsed_time": "10:52:03", "remaining_time": "8:53:33", "throughput": 638.12, "total_tokens": 24965256}
{"current_steps": 2140, "total_steps": 3882, "loss": 0.214, "lr": 4.5681118917830835e-05, "epoch": 1.1019830028328612, "percentage": 55.13, "elapsed_time": "10:53:31", "remaining_time": "8:51:59", "throughput": 638.17, "total_tokens": 25023768}
{"current_steps": 2145, "total_steps": 3882, "loss": 0.2806, "lr": 4.546893670111866e-05, "epoch": 1.1045583311872265, "percentage": 55.26, "elapsed_time": "10:55:01", "remaining_time": "8:50:25", "throughput": 638.2, "total_tokens": 25082248}
{"current_steps": 2150, "total_steps": 3882, "loss": 0.2536, "lr": 4.525683672627168e-05, "epoch": 1.1071336595415915, "percentage": 55.38, "elapsed_time": "10:56:29", "remaining_time": "8:48:51", "throughput": 638.25, "total_tokens": 25140752}
{"current_steps": 2150, "total_steps": 3882, "eval_loss": 0.30806976556777954, "epoch": 1.1071336595415915, "percentage": 55.38, "elapsed_time": "10:56:49", "remaining_time": "8:49:07", "throughput": 637.94, "total_tokens": 25140752}
{"current_steps": 2155, "total_steps": 3882, "loss": 0.2318, "lr": 4.504482284304795e-05, "epoch": 1.1097089878959567, "percentage": 55.51, "elapsed_time": "10:58:24", "remaining_time": "8:47:38", "throughput": 637.89, "total_tokens": 25199208}
{"current_steps": 2160, "total_steps": 3882, "loss": 0.2371, "lr": 4.4832898899642974e-05, "epoch": 1.112284316250322, "percentage": 55.64, "elapsed_time": "10:59:52", "remaining_time": "8:46:04", "throughput": 637.94, "total_tokens": 25257680}
{"current_steps": 2165, "total_steps": 3882, "loss": 0.2517, "lr": 4.462106874261969e-05, "epoch": 1.114859644604687, "percentage": 55.77, "elapsed_time": "11:01:22", "remaining_time": "8:44:30", "throughput": 637.97, "total_tokens": 25316200}
{"current_steps": 2170, "total_steps": 3882, "loss": 0.3029, "lr": 4.440933621683882e-05, "epoch": 1.1174349729590523, "percentage": 55.9, "elapsed_time": "11:02:50", "remaining_time": "8:42:56", "throughput": 638.02, "total_tokens": 25374656}
{"current_steps": 2175, "total_steps": 3882, "loss": 0.3005, "lr": 4.419770516538899e-05, "epoch": 1.1200103013134175, "percentage": 56.03, "elapsed_time": "11:04:19", "remaining_time": "8:41:23", "throughput": 638.07, "total_tokens": 25433168}
{"current_steps": 2180, "total_steps": 3882, "loss": 0.3059, "lr": 4.398617942951703e-05, "epoch": 1.1225856296677827, "percentage": 56.16, "elapsed_time": "11:05:48", "remaining_time": "8:39:49", "throughput": 638.11, "total_tokens": 25491664}
{"current_steps": 2185, "total_steps": 3882, "loss": 0.2385, "lr": 4.3774762848558124e-05, "epoch": 1.1251609580221478, "percentage": 56.29, "elapsed_time": "11:07:17", "remaining_time": "8:38:15", "throughput": 638.15, "total_tokens": 25550128}
{"current_steps": 2190, "total_steps": 3882, "loss": 0.2795, "lr": 4.356345925986637e-05, "epoch": 1.127736286376513, "percentage": 56.41, "elapsed_time": "11:08:46", "remaining_time": "8:36:41", "throughput": 638.2, "total_tokens": 25608624}
{"current_steps": 2195, "total_steps": 3882, "loss": 0.2622, "lr": 4.335227249874484e-05, "epoch": 1.1303116147308783, "percentage": 56.54, "elapsed_time": "11:10:15", "remaining_time": "8:35:08", "throughput": 638.24, "total_tokens": 25667128}
{"current_steps": 2200, "total_steps": 3882, "loss": 0.3184, "lr": 4.314120639837619e-05, "epoch": 1.1328869430852433, "percentage": 56.67, "elapsed_time": "11:11:44", "remaining_time": "8:33:34", "throughput": 638.28, "total_tokens": 25725560}
{"current_steps": 2200, "total_steps": 3882, "eval_loss": 0.24919146299362183, "epoch": 1.1328869430852433, "percentage": 56.67, "elapsed_time": "11:12:03", "remaining_time": "8:33:49", "throughput": 637.98, "total_tokens": 25725560}
{"current_steps": 2205, "total_steps": 3882, "loss": 0.3121, "lr": 4.2930264789752974e-05, "epoch": 1.1354622714396085, "percentage": 56.8, "elapsed_time": "11:13:37", "remaining_time": "8:32:19", "throughput": 637.94, "total_tokens": 25784048}
{"current_steps": 2210, "total_steps": 3882, "loss": 0.2448, "lr": 4.271945150160815e-05, "epoch": 1.1380375997939738, "percentage": 56.93, "elapsed_time": "11:15:06", "remaining_time": "8:30:45", "throughput": 637.99, "total_tokens": 25842512}
{"current_steps": 2215, "total_steps": 3882, "loss": 0.3141, "lr": 4.250877036034552e-05, "epoch": 1.140612928148339, "percentage": 57.06, "elapsed_time": "11:16:34", "remaining_time": "8:29:11", "throughput": 638.04, "total_tokens": 25900992}
{"current_steps": 2220, "total_steps": 3882, "loss": 0.2589, "lr": 4.2298225189970424e-05, "epoch": 1.143188256502704, "percentage": 57.19, "elapsed_time": "11:18:03", "remaining_time": "8:27:37", "throughput": 638.09, "total_tokens": 25959448}
{"current_steps": 2225, "total_steps": 3882, "loss": 0.3071, "lr": 4.2087819812020155e-05, "epoch": 1.1457635848570693, "percentage": 57.32, "elapsed_time": "11:19:30", "remaining_time": "8:26:02", "throughput": 638.16, "total_tokens": 26017936}
{"current_steps": 2230, "total_steps": 3882, "loss": 0.1913, "lr": 4.1877558045494735e-05, "epoch": 1.1483389132114346, "percentage": 57.44, "elapsed_time": "11:20:58", "remaining_time": "8:24:28", "throughput": 638.22, "total_tokens": 26076440}
{"current_steps": 2235, "total_steps": 3882, "loss": 0.1998, "lr": 4.166744370678748e-05, "epoch": 1.1509142415657996, "percentage": 57.57, "elapsed_time": "11:22:24", "remaining_time": "8:22:52", "throughput": 638.29, "total_tokens": 26134896}
{"current_steps": 2240, "total_steps": 3882, "loss": 0.2382, "lr": 4.1457480609615866e-05, "epoch": 1.1534895699201648, "percentage": 57.7, "elapsed_time": "11:23:53", "remaining_time": "8:21:19", "throughput": 638.33, "total_tokens": 26193368}
{"current_steps": 2245, "total_steps": 3882, "loss": 0.231, "lr": 4.124767256495217e-05, "epoch": 1.15606489827453, "percentage": 57.83, "elapsed_time": "11:25:26", "remaining_time": "8:19:48", "throughput": 638.32, "total_tokens": 26251848}
{"current_steps": 2250, "total_steps": 3882, "loss": 0.273, "lr": 4.1038023380954415e-05, "epoch": 1.158640226628895, "percentage": 57.96, "elapsed_time": "11:26:59", "remaining_time": "8:18:18", "throughput": 638.29, "total_tokens": 26310336}
{"current_steps": 2250, "total_steps": 3882, "eval_loss": 0.22014015913009644, "epoch": 1.158640226628895, "percentage": 57.96, "elapsed_time": "11:27:20", "remaining_time": "8:18:32", "throughput": 637.98, "total_tokens": 26310336}
{"current_steps": 2255, "total_steps": 3882, "loss": 0.2986, "lr": 4.0828536862897125e-05, "epoch": 1.1612155549832603, "percentage": 58.09, "elapsed_time": "11:28:57", "remaining_time": "8:17:05", "throughput": 637.89, "total_tokens": 26368840}
{"current_steps": 2260, "total_steps": 3882, "loss": 0.263, "lr": 4.061921681310241e-05, "epoch": 1.1637908833376256, "percentage": 58.22, "elapsed_time": "11:30:30", "remaining_time": "8:15:34", "throughput": 637.88, "total_tokens": 26427280}
{"current_steps": 2265, "total_steps": 3882, "loss": 0.2282, "lr": 4.04100670308708e-05, "epoch": 1.1663662116919906, "percentage": 58.35, "elapsed_time": "11:32:01", "remaining_time": "8:14:02", "throughput": 637.88, "total_tokens": 26485760}
{"current_steps": 2270, "total_steps": 3882, "loss": 0.2311, "lr": 4.0201091312412394e-05, "epoch": 1.1689415400463559, "percentage": 58.48, "elapsed_time": "11:33:32", "remaining_time": "8:12:30", "throughput": 637.89, "total_tokens": 26544224}
{"current_steps": 2275, "total_steps": 3882, "loss": 0.2939, "lr": 3.999229345077789e-05, "epoch": 1.1715168684007211, "percentage": 58.6, "elapsed_time": "11:35:02", "remaining_time": "8:10:57", "throughput": 637.91, "total_tokens": 26602712}
{"current_steps": 2280, "total_steps": 3882, "loss": 0.2165, "lr": 3.978367723578981e-05, "epoch": 1.1740921967550864, "percentage": 58.73, "elapsed_time": "11:36:34", "remaining_time": "8:09:25", "throughput": 637.92, "total_tokens": 26661160}
{"current_steps": 2285, "total_steps": 3882, "loss": 0.1823, "lr": 3.957524645397359e-05, "epoch": 1.1766675251094514, "percentage": 58.86, "elapsed_time": "11:38:06", "remaining_time": "8:07:54", "throughput": 637.91, "total_tokens": 26719648}
{"current_steps": 2290, "total_steps": 3882, "loss": 0.2827, "lr": 3.936700488848899e-05, "epoch": 1.1792428534638166, "percentage": 58.99, "elapsed_time": "11:39:38", "remaining_time": "8:06:22", "throughput": 637.91, "total_tokens": 26778112}
{"current_steps": 2295, "total_steps": 3882, "loss": 0.2897, "lr": 3.915895631906135e-05, "epoch": 1.1818181818181819, "percentage": 59.12, "elapsed_time": "11:41:09", "remaining_time": "8:04:51", "throughput": 637.91, "total_tokens": 26836592}
{"current_steps": 2300, "total_steps": 3882, "loss": 0.2903, "lr": 3.895110452191301e-05, "epoch": 1.184393510172547, "percentage": 59.25, "elapsed_time": "11:42:39", "remaining_time": "8:03:18", "throughput": 637.93, "total_tokens": 26895096}
{"current_steps": 2300, "total_steps": 3882, "eval_loss": 0.2940039336681366, "epoch": 1.184393510172547, "percentage": 59.25, "elapsed_time": "11:43:00", "remaining_time": "8:03:32", "throughput": 637.63, "total_tokens": 26895096}
{"current_steps": 2305, "total_steps": 3882, "loss": 0.2475, "lr": 3.874345326969471e-05, "epoch": 1.1869688385269122, "percentage": 59.38, "elapsed_time": "11:44:36", "remaining_time": "8:02:04", "throughput": 637.56, "total_tokens": 26953560}
{"current_steps": 2310, "total_steps": 3882, "loss": 0.2921, "lr": 3.853600633141725e-05, "epoch": 1.1895441668812774, "percentage": 59.51, "elapsed_time": "11:46:06", "remaining_time": "8:00:31", "throughput": 637.57, "total_tokens": 27012024}
{"current_steps": 2315, "total_steps": 3882, "loss": 0.2065, "lr": 3.8328767472382896e-05, "epoch": 1.1921194952356426, "percentage": 59.63, "elapsed_time": "11:47:37", "remaining_time": "7:58:58", "throughput": 637.6, "total_tokens": 27070520}
{"current_steps": 2320, "total_steps": 3882, "loss": 0.2232, "lr": 3.812174045411724e-05, "epoch": 1.1946948235900077, "percentage": 59.76, "elapsed_time": "11:49:07", "remaining_time": "7:57:26", "throughput": 637.62, "total_tokens": 27128984}
{"current_steps": 2325, "total_steps": 3882, "loss": 0.266, "lr": 3.791492903430071e-05, "epoch": 1.197270151944373, "percentage": 59.89, "elapsed_time": "11:50:38", "remaining_time": "7:55:53", "throughput": 637.63, "total_tokens": 27187432}
{"current_steps": 2330, "total_steps": 3882, "loss": 0.2451, "lr": 3.7708336966700574e-05, "epoch": 1.1998454802987382, "percentage": 60.02, "elapsed_time": "11:52:08", "remaining_time": "7:54:21", "throughput": 637.65, "total_tokens": 27245920}
{"current_steps": 2335, "total_steps": 3882, "loss": 0.2031, "lr": 3.7501968001102634e-05, "epoch": 1.2024208086531032, "percentage": 60.15, "elapsed_time": "11:53:41", "remaining_time": "7:52:50", "throughput": 637.64, "total_tokens": 27304384}
{"current_steps": 2340, "total_steps": 3882, "loss": 0.2605, "lr": 3.7295825883243306e-05, "epoch": 1.2049961370074684, "percentage": 60.28, "elapsed_time": "11:55:11", "remaining_time": "7:51:17", "throughput": 637.66, "total_tokens": 27362864}
{"current_steps": 2345, "total_steps": 3882, "loss": 0.2293, "lr": 3.708991435474151e-05, "epoch": 1.2075714653618337, "percentage": 60.41, "elapsed_time": "11:56:43", "remaining_time": "7:49:45", "throughput": 637.66, "total_tokens": 27421360}
{"current_steps": 2350, "total_steps": 3882, "loss": 0.2757, "lr": 3.688423715303082e-05, "epoch": 1.210146793716199, "percentage": 60.54, "elapsed_time": "11:58:14", "remaining_time": "7:48:14", "throughput": 637.66, "total_tokens": 27479840}
{"current_steps": 2350, "total_steps": 3882, "eval_loss": 0.26213815808296204, "epoch": 1.210146793716199, "percentage": 60.54, "elapsed_time": "11:58:34", "remaining_time": "7:48:26", "throughput": 637.37, "total_tokens": 27479840}
{"current_steps": 2355, "total_steps": 3882, "loss": 0.242, "lr": 3.6678798011291674e-05, "epoch": 1.212722122070564, "percentage": 60.66, "elapsed_time": "12:00:11", "remaining_time": "7:46:58", "throughput": 637.29, "total_tokens": 27538344}
{"current_steps": 2360, "total_steps": 3882, "loss": 0.2451, "lr": 3.647360065838348e-05, "epoch": 1.2152974504249292, "percentage": 60.79, "elapsed_time": "12:01:42", "remaining_time": "7:45:26", "throughput": 637.3, "total_tokens": 27596808}
{"current_steps": 2365, "total_steps": 3882, "loss": 0.2237, "lr": 3.6268648818777105e-05, "epoch": 1.2178727787792945, "percentage": 60.92, "elapsed_time": "12:03:13", "remaining_time": "7:43:54", "throughput": 637.32, "total_tokens": 27655272}
{"current_steps": 2370, "total_steps": 3882, "loss": 0.2033, "lr": 3.606394621248709e-05, "epoch": 1.2204481071336595, "percentage": 61.05, "elapsed_time": "12:04:45", "remaining_time": "7:42:22", "throughput": 637.31, "total_tokens": 27713784}
{"current_steps": 2375, "total_steps": 3882, "loss": 0.2741, "lr": 3.585949655500429e-05, "epoch": 1.2230234354880247, "percentage": 61.18, "elapsed_time": "12:06:16", "remaining_time": "7:40:50", "throughput": 637.32, "total_tokens": 27772248}
{"current_steps": 2380, "total_steps": 3882, "loss": 0.2677, "lr": 3.5655303557228335e-05, "epoch": 1.22559876384239, "percentage": 61.31, "elapsed_time": "12:07:47", "remaining_time": "7:39:18", "throughput": 637.33, "total_tokens": 27830704}
{"current_steps": 2400, "total_steps": 3882, "loss": 0.2766, "lr": 3.484117221672465e-05, "epoch": 1.2359000772598505, "percentage": 61.82, "elapsed_time": "12:13:57", "remaining_time": "7:33:12", "throughput": 637.29, "total_tokens": 28064552}
{"current_steps": 2400, "total_steps": 3882, "eval_loss": 0.2361450344324112, "epoch": 1.2359000772598505, "percentage": 61.82, "elapsed_time": "12:14:17", "remaining_time": "7:33:25", "throughput": 637.01, "total_tokens": 28064552}
{"current_steps": 2415, "total_steps": 3882, "loss": 0.2383, "lr": 3.42334497649294e-05, "epoch": 1.2436260623229463, "percentage": 62.21, "elapsed_time": "12:18:55", "remaining_time": "7:28:51", "throughput": 636.96, "total_tokens": 28239920}
{"current_steps": 2420, "total_steps": 3882, "loss": 0.2074, "lr": 3.403144306406466e-05, "epoch": 1.2462013906773113, "percentage": 62.34, "elapsed_time": "12:20:27", "remaining_time": "7:27:19", "throughput": 636.96, "total_tokens": 28298432}
{"current_steps": 2425, "total_steps": 3882, "loss": 0.2202, "lr": 3.382972620331475e-05, "epoch": 1.2487767190316765, "percentage": 62.47, "elapsed_time": "12:21:58", "remaining_time": "7:25:47", "throughput": 636.97, "total_tokens": 28356904}
{"current_steps": 2430, "total_steps": 3882, "loss": 0.1756, "lr": 3.362830284397716e-05, "epoch": 1.2513520473860418, "percentage": 62.6, "elapsed_time": "12:23:29", "remaining_time": "7:24:15", "throughput": 636.98, "total_tokens": 28415384}
{"current_steps": 2435, "total_steps": 3882, "loss": 0.2564, "lr": 3.342717664202223e-05, "epoch": 1.2539273757404068, "percentage": 62.73, "elapsed_time": "12:25:01", "remaining_time": "7:22:43", "throughput": 636.98, "total_tokens": 28473848}
{"current_steps": 2440, "total_steps": 3882, "loss": 0.242, "lr": 3.322635124802658e-05, "epoch": 1.256502704094772, "percentage": 62.85, "elapsed_time": "12:26:32", "remaining_time": "7:21:11", "throughput": 637.0, "total_tokens": 28532312}
{"current_steps": 2445, "total_steps": 3882, "loss": 0.3455, "lr": 3.3025830307107035e-05, "epoch": 1.2590780324491373, "percentage": 62.98, "elapsed_time": "12:28:02", "remaining_time": "7:19:38", "throughput": 637.02, "total_tokens": 28590784}
{"current_steps": 2450, "total_steps": 3882, "loss": 0.3076, "lr": 3.2825617458854376e-05, "epoch": 1.2616533608035025, "percentage": 63.11, "elapsed_time": "12:29:32", "remaining_time": "7:18:05", "throughput": 637.04, "total_tokens": 28649256}
{"current_steps": 2450, "total_steps": 3882, "eval_loss": 0.2371988147497177, "epoch": 1.2616533608035025, "percentage": 63.11, "elapsed_time": "12:29:51", "remaining_time": "7:18:17", "throughput": 636.76, "total_tokens": 28649256}
{"current_steps": 2460, "total_steps": 3882, "loss": 0.3022, "lr": 3.242613057068641e-05, "epoch": 1.2668040175122328, "percentage": 63.37, "elapsed_time": "12:32:57", "remaining_time": "7:15:14", "throughput": 636.74, "total_tokens": 28766232}
{"current_steps": 2465, "total_steps": 3882, "loss": 0.2756, "lr": 3.222686378172847e-05, "epoch": 1.269379345866598, "percentage": 63.5, "elapsed_time": "12:34:26", "remaining_time": "7:13:41", "throughput": 636.78, "total_tokens": 28824688}
{"current_steps": 2470, "total_steps": 3882, "loss": 0.2395, "lr": 3.2027919587220516e-05, "epoch": 1.271954674220963, "percentage": 63.63, "elapsed_time": "12:35:54", "remaining_time": "7:12:07", "throughput": 636.83, "total_tokens": 28883176}
{"current_steps": 2475, "total_steps": 3882, "loss": 0.2204, "lr": 3.1829301598134355e-05, "epoch": 1.2745300025753283, "percentage": 63.76, "elapsed_time": "12:37:24", "remaining_time": "7:10:34", "throughput": 636.86, "total_tokens": 28941592}
{"current_steps": 2480, "total_steps": 3882, "loss": 0.2477, "lr": 3.1631013419520855e-05, "epoch": 1.2771053309296936, "percentage": 63.88, "elapsed_time": "12:38:52", "remaining_time": "7:09:00", "throughput": 636.91, "total_tokens": 29000064}
{"current_steps": 2485, "total_steps": 3882, "loss": 0.2645, "lr": 3.143305865044467e-05, "epoch": 1.2796806592840588, "percentage": 64.01, "elapsed_time": "12:40:21", "remaining_time": "7:07:27", "throughput": 636.95, "total_tokens": 29058496}
{"current_steps": 2490, "total_steps": 3882, "loss": 0.2184, "lr": 3.123544088391881e-05, "epoch": 1.2822559876384239, "percentage": 64.14, "elapsed_time": "12:41:50", "remaining_time": "7:05:53", "throughput": 636.99, "total_tokens": 29116992}
{"current_steps": 2495, "total_steps": 3882, "loss": 0.1857, "lr": 3.10381637068395e-05, "epoch": 1.284831315992789, "percentage": 64.27, "elapsed_time": "12:43:19", "remaining_time": "7:04:20", "throughput": 637.03, "total_tokens": 29175496}
{"current_steps": 2500, "total_steps": 3882, "loss": 0.257, "lr": 3.084123069992096e-05, "epoch": 1.2874066443471541, "percentage": 64.4, "elapsed_time": "12:44:47", "remaining_time": "7:02:46", "throughput": 637.08, "total_tokens": 29233968}
{"current_steps": 2500, "total_steps": 3882, "eval_loss": 0.2488754242658615, "epoch": 1.2874066443471541, "percentage": 64.4, "elapsed_time": "12:45:07", "remaining_time": "7:02:57", "throughput": 636.81, "total_tokens": 29233968}
{"current_steps": 2505, "total_steps": 3882, "loss": 0.1617, "lr": 3.064464543763057e-05, "epoch": 1.2899819727015194, "percentage": 64.53, "elapsed_time": "12:46:41", "remaining_time": "7:01:27", "throughput": 636.77, "total_tokens": 29292464}
{"current_steps": 2510, "total_steps": 3882, "loss": 0.2068, "lr": 3.0448411488123862e-05, "epoch": 1.2925573010558846, "percentage": 64.66, "elapsed_time": "12:48:10", "remaining_time": "6:59:53", "throughput": 636.82, "total_tokens": 29350952}
{"current_steps": 2515, "total_steps": 3882, "loss": 0.2188, "lr": 3.0252532413179825e-05, "epoch": 1.2951326294102499, "percentage": 64.79, "elapsed_time": "12:49:38", "remaining_time": "6:58:19", "throughput": 636.86, "total_tokens": 29409432}
{"current_steps": 2520, "total_steps": 3882, "loss": 0.1948, "lr": 3.0057011768136223e-05, "epoch": 1.2977079577646151, "percentage": 64.91, "elapsed_time": "12:51:06", "remaining_time": "6:56:46", "throughput": 636.91, "total_tokens": 29467880}
{"current_steps": 2525, "total_steps": 3882, "loss": 0.2366, "lr": 2.9861853101825094e-05, "epoch": 1.3002832861189801, "percentage": 65.04, "elapsed_time": "12:52:36", "remaining_time": "6:55:13", "throughput": 636.94, "total_tokens": 29526384}
{"current_steps": 2530, "total_steps": 3882, "loss": 0.2601, "lr": 2.966705995650827e-05, "epoch": 1.3028586144733454, "percentage": 65.17, "elapsed_time": "12:54:04", "remaining_time": "6:53:39", "throughput": 636.99, "total_tokens": 29584880}
{"current_steps": 2535, "total_steps": 3882, "loss": 0.1913, "lr": 2.9472635867813185e-05, "epoch": 1.3054339428277104, "percentage": 65.3, "elapsed_time": "12:55:33", "remaining_time": "6:52:06", "throughput": 637.03, "total_tokens": 29643344}
{"current_steps": 2540, "total_steps": 3882, "loss": 0.2, "lr": 2.927858436466861e-05, "epoch": 1.3080092711820757, "percentage": 65.43, "elapsed_time": "12:57:02", "remaining_time": "6:50:32", "throughput": 637.07, "total_tokens": 29701864}
{"current_steps": 2545, "total_steps": 3882, "loss": 0.2503, "lr": 2.9084908969240664e-05, "epoch": 1.310584599536441, "percentage": 65.56, "elapsed_time": "12:58:30", "remaining_time": "6:48:59", "throughput": 637.12, "total_tokens": 29760376}
{"current_steps": 2550, "total_steps": 3882, "loss": 0.2192, "lr": 2.8891613196868816e-05, "epoch": 1.3131599278908062, "percentage": 65.69, "elapsed_time": "12:59:59", "remaining_time": "6:47:25", "throughput": 637.16, "total_tokens": 29818856}
{"current_steps": 2550, "total_steps": 3882, "eval_loss": 0.24317465722560883, "epoch": 1.3131599278908062, "percentage": 65.69, "elapsed_time": "13:00:19", "remaining_time": "6:47:36", "throughput": 636.89, "total_tokens": 29818856}
{"current_steps": 2555, "total_steps": 3882, "loss": 0.2231, "lr": 2.8698700556002178e-05, "epoch": 1.3157352562451712, "percentage": 65.82, "elapsed_time": "13:01:52", "remaining_time": "6:46:05", "throughput": 636.87, "total_tokens": 29877288}
{"current_steps": 2560, "total_steps": 3882, "loss": 0.3007, "lr": 2.8506174548135695e-05, "epoch": 1.3183105845995364, "percentage": 65.95, "elapsed_time": "13:03:21", "remaining_time": "6:44:31", "throughput": 636.91, "total_tokens": 29935776}
{"current_steps": 2565, "total_steps": 3882, "loss": 0.1895, "lr": 2.8314038667746732e-05, "epoch": 1.3208859129539017, "percentage": 66.07, "elapsed_time": "13:04:50", "remaining_time": "6:42:58", "throughput": 636.95, "total_tokens": 29994232}
{"current_steps": 2570, "total_steps": 3882, "loss": 0.2728, "lr": 2.81222964022315e-05, "epoch": 1.3234612413082667, "percentage": 66.2, "elapsed_time": "13:06:18", "remaining_time": "6:41:24", "throughput": 637.0, "total_tokens": 30052760}
{"current_steps": 2575, "total_steps": 3882, "loss": 0.2827, "lr": 2.793095123184193e-05, "epoch": 1.326036569662632, "percentage": 66.33, "elapsed_time": "13:07:47", "remaining_time": "6:39:51", "throughput": 637.04, "total_tokens": 30111200}
{"current_steps": 2580, "total_steps": 3882, "loss": 0.2191, "lr": 2.77400066296223e-05, "epoch": 1.3286118980169972, "percentage": 66.46, "elapsed_time": "13:09:15", "remaining_time": "6:38:18", "throughput": 637.09, "total_tokens": 30169680}
{"current_steps": 2585, "total_steps": 3882, "loss": 0.2317, "lr": 2.7549466061346386e-05, "epoch": 1.3311872263713624, "percentage": 66.59, "elapsed_time": "13:10:43", "remaining_time": "6:36:44", "throughput": 637.14, "total_tokens": 30228176}
{"current_steps": 2590, "total_steps": 3882, "loss": 0.2293, "lr": 2.7359332985454443e-05, "epoch": 1.3337625547257275, "percentage": 66.72, "elapsed_time": "13:12:12", "remaining_time": "6:35:11", "throughput": 637.18, "total_tokens": 30286656}
{"current_steps": 2895, "total_steps": 3882, "loss": 0.1844, "lr": 1.6663955420970667e-05, "epoch": 1.4908575843420036, "percentage": 74.57, "elapsed_time": "14:44:29", "remaining_time": "5:01:33", "throughput": 637.92, "total_tokens": 33853936}
{"current_steps": 2900, "total_steps": 3882, "loss": 0.2015, "lr": 1.650549436531442e-05, "epoch": 1.4934329126963688, "percentage": 74.7, "elapsed_time": "14:45:57", "remaining_time": "5:00:00", "throughput": 637.96, "total_tokens": 33912440}
{"current_steps": 2900, "total_steps": 3882, "eval_loss": 0.23923428356647491, "epoch": 1.4934329126963688, "percentage": 74.7, "elapsed_time": "14:46:16", "remaining_time": "5:00:06", "throughput": 637.73, "total_tokens": 33912440}
{"current_steps": 2905, "total_steps": 3882, "loss": 0.2244, "lr": 1.6347641257602236e-05, "epoch": 1.496008241050734, "percentage": 74.83, "elapsed_time": "14:47:50", "remaining_time": "4:58:35", "throughput": 637.7, "total_tokens": 33970888}
{"current_steps": 2910, "total_steps": 3882, "loss": 0.1864, "lr": 1.6190398962974856e-05, "epoch": 1.498583569405099, "percentage": 74.96, "elapsed_time": "14:49:18", "remaining_time": "4:57:02", "throughput": 637.75, "total_tokens": 34029376}
{"current_steps": 2915, "total_steps": 3882, "loss": 0.1969, "lr": 1.6033770335486305e-05, "epoch": 1.5011588977594643, "percentage": 75.09, "elapsed_time": "14:50:46", "remaining_time": "4:55:29", "throughput": 637.8, "total_tokens": 34087864}
{"current_steps": 2920, "total_steps": 3882, "loss": 0.2449, "lr": 1.5877758218052207e-05, "epoch": 1.5037342261138296, "percentage": 75.22, "elapsed_time": "14:52:14", "remaining_time": "4:53:57", "throughput": 637.83, "total_tokens": 34146296}
{"current_steps": 2925, "total_steps": 3882, "loss": 0.155, "lr": 1.572236544239803e-05, "epoch": 1.5063095544681948, "percentage": 75.35, "elapsed_time": "14:53:42", "remaining_time": "4:52:24", "throughput": 637.88, "total_tokens": 34204800}
{"current_steps": 2930, "total_steps": 3882, "loss": 0.2031, "lr": 1.5567594829007835e-05, "epoch": 1.5088848828225598, "percentage": 75.48, "elapsed_time": "14:55:10", "remaining_time": "4:50:51", "throughput": 637.93, "total_tokens": 34263296}
{"current_steps": 2935, "total_steps": 3882, "loss": 0.1917, "lr": 1.541344918707299e-05, "epoch": 1.511460211176925, "percentage": 75.61, "elapsed_time": "14:56:38", "remaining_time": "4:49:18", "throughput": 637.97, "total_tokens": 34321792}
{"current_steps": 2940, "total_steps": 3882, "loss": 0.1609, "lr": 1.525993131444123e-05, "epoch": 1.51403553953129, "percentage": 75.73, "elapsed_time": "14:58:06", "remaining_time": "4:47:45", "throughput": 638.01, "total_tokens": 34380264}
{"current_steps": 2945, "total_steps": 3882, "loss": 0.2217, "lr": 1.5107043997565795e-05, "epoch": 1.5166108678856554, "percentage": 75.86, "elapsed_time": "14:59:34", "remaining_time": "4:46:12", "throughput": 638.06, "total_tokens": 34438752}
{"current_steps": 2950, "total_steps": 3882, "loss": 0.2069, "lr": 1.495479001145499e-05, "epoch": 1.5191861962400206, "percentage": 75.99, "elapsed_time": "15:01:02", "remaining_time": "4:44:39", "throughput": 638.1, "total_tokens": 34497216}
{"current_steps": 2950, "total_steps": 3882, "eval_loss": 0.2638954222202301, "epoch": 1.5191861962400206, "percentage": 75.99, "elapsed_time": "15:01:21", "remaining_time": "4:44:46", "throughput": 637.87, "total_tokens": 34497216}
{"current_steps": 2955, "total_steps": 3882, "loss": 0.2811, "lr": 1.4803172119621661e-05, "epoch": 1.5217615245943859, "percentage": 76.12, "elapsed_time": "15:02:54", "remaining_time": "4:43:14", "throughput": 637.86, "total_tokens": 34555704}
{"current_steps": 2960, "total_steps": 3882, "loss": 0.2056, "lr": 1.4652193074033188e-05, "epoch": 1.524336852948751, "percentage": 76.25, "elapsed_time": "15:04:22", "remaining_time": "4:41:42", "throughput": 637.9, "total_tokens": 34614168}
{"current_steps": 2965, "total_steps": 3882, "loss": 0.2097, "lr": 1.4501855615061376e-05, "epoch": 1.5269121813031161, "percentage": 76.38, "elapsed_time": "15:05:49", "remaining_time": "4:40:09", "throughput": 637.95, "total_tokens": 34672624}
{"current_steps": 2970, "total_steps": 3882, "loss": 0.1913, "lr": 1.4352162471432872e-05, "epoch": 1.5294875096574814, "percentage": 76.51, "elapsed_time": "15:07:18", "remaining_time": "4:38:36", "throughput": 637.99, "total_tokens": 34731056}
{"current_steps": 2975, "total_steps": 3882, "loss": 0.1996, "lr": 1.420311636017953e-05, "epoch": 1.5320628380118464, "percentage": 76.64, "elapsed_time": "15:08:45", "remaining_time": "4:37:03", "throughput": 638.04, "total_tokens": 34789552}
{"current_steps": 2980, "total_steps": 3882, "loss": 0.2102, "lr": 1.4054719986589148e-05, "epoch": 1.5346381663662116, "percentage": 76.76, "elapsed_time": "15:10:14", "remaining_time": "4:35:30", "throughput": 638.08, "total_tokens": 34848064}
{"current_steps": 2985, "total_steps": 3882, "loss": 0.232, "lr": 1.39069760441563e-05, "epoch": 1.537213494720577, "percentage": 76.89, "elapsed_time": "15:11:41", "remaining_time": "4:33:58", "throughput": 638.12, "total_tokens": 34906592}
{"current_steps": 2990, "total_steps": 3882, "loss": 0.2109, "lr": 1.3759887214533557e-05, "epoch": 1.5397888230749421, "percentage": 77.02, "elapsed_time": "15:13:09", "remaining_time": "4:32:25", "throughput": 638.17, "total_tokens": 34965064}
{"current_steps": 2995, "total_steps": 3882, "loss": 0.1989, "lr": 1.3613456167482685e-05, "epoch": 1.5423641514293074, "percentage": 77.15, "elapsed_time": "15:14:38", "remaining_time": "4:30:52", "throughput": 638.2, "total_tokens": 35023552}
{"current_steps": 3000, "total_steps": 3882, "loss": 0.2027, "lr": 1.3467685560826315e-05, "epoch": 1.5449394797836724, "percentage": 77.28, "elapsed_time": "15:16:06", "remaining_time": "4:29:20", "throughput": 638.25, "total_tokens": 35082056}
{"current_steps": 3000, "total_steps": 3882, "eval_loss": 0.2371213436126709, "epoch": 1.5449394797836724, "percentage": 77.28, "elapsed_time": "15:16:26", "remaining_time": "4:29:25", "throughput": 638.02, "total_tokens": 35082056}
{"current_steps": 3005, "total_steps": 3882, "loss": 0.1383, "lr": 1.3322578040399558e-05, "epoch": 1.5475148081380374, "percentage": 77.41, "elapsed_time": "15:17:59", "remaining_time": "4:27:54", "throughput": 637.99, "total_tokens": 35140568}
{"current_steps": 3010, "total_steps": 3882, "loss": 0.2006, "lr": 1.3178136240002164e-05, "epoch": 1.5500901364924027, "percentage": 77.54, "elapsed_time": "15:19:27", "remaining_time": "4:26:22", "throughput": 638.04, "total_tokens": 35199064}
{"current_steps": 3015, "total_steps": 3882, "loss": 0.251, "lr": 1.3034362781350524e-05, "epoch": 1.552665464846768, "percentage": 77.67, "elapsed_time": "15:20:56", "remaining_time": "4:24:49", "throughput": 638.07, "total_tokens": 35257576}
{"current_steps": 3020, "total_steps": 3882, "loss": 0.1584, "lr": 1.289126027403023e-05, "epoch": 1.5552407932011332, "percentage": 77.79, "elapsed_time": "15:22:26", "remaining_time": "4:23:17", "throughput": 638.09, "total_tokens": 35316064}
{"current_steps": 3025, "total_steps": 3882, "loss": 0.2173, "lr": 1.2748831315448606e-05, "epoch": 1.5578161215554984, "percentage": 77.92, "elapsed_time": "15:23:56", "remaining_time": "4:21:45", "throughput": 638.11, "total_tokens": 35374504}
{"current_steps": 3030, "total_steps": 3882, "loss": 0.2518, "lr": 1.260707849078766e-05, "epoch": 1.5603914499098637, "percentage": 78.05, "elapsed_time": "15:25:26", "remaining_time": "4:20:13", "throughput": 638.13, "total_tokens": 35432984}
{"current_steps": 3035, "total_steps": 3882, "loss": 0.1936, "lr": 1.2466004372957068e-05, "epoch": 1.5629667782642287, "percentage": 78.18, "elapsed_time": "15:26:57", "remaining_time": "4:18:41", "throughput": 638.13, "total_tokens": 35491488}
{"current_steps": 3040, "total_steps": 3882, "loss": 0.1677, "lr": 1.2325611522547547e-05, "epoch": 1.5655421066185937, "percentage": 78.31, "elapsed_time": "15:28:27", "remaining_time": "4:17:09", "throughput": 638.15, "total_tokens": 35549984}
{"current_steps": 3045, "total_steps": 3882, "loss": 0.2021, "lr": 1.218590248778434e-05, "epoch": 1.568117434972959, "percentage": 78.44, "elapsed_time": "15:29:56", "remaining_time": "4:15:37", "throughput": 638.18, "total_tokens": 35608488}
{"current_steps": 3050, "total_steps": 3882, "loss": 0.1925, "lr": 1.2046879804480993e-05, "epoch": 1.5706927633273242, "percentage": 78.57, "elapsed_time": "15:31:25", "remaining_time": "4:14:04", "throughput": 638.22, "total_tokens": 35666976}
{"current_steps": 3050, "total_steps": 3882, "eval_loss": 0.24840803444385529, "epoch": 1.5706927633273242, "percentage": 78.57, "elapsed_time": "15:31:45", "remaining_time": "4:14:10", "throughput": 637.99, "total_tokens": 35666976}
{"current_steps": 3055, "total_steps": 3882, "loss": 0.1686, "lr": 1.1908545995993253e-05, "epoch": 1.5732680916816895, "percentage": 78.7, "elapsed_time": "15:33:20", "remaining_time": "4:12:39", "throughput": 637.95, "total_tokens": 35725456}
{"current_steps": 3060, "total_steps": 3882, "loss": 0.2366, "lr": 1.1770903573173398e-05, "epoch": 1.5758434200360547, "percentage": 78.83, "elapsed_time": "15:34:48", "remaining_time": "4:11:06", "throughput": 637.99, "total_tokens": 35783912}
{"current_steps": 3065, "total_steps": 3882, "loss": 0.2029, "lr": 1.1633955034324507e-05, "epoch": 1.5784187483904197, "percentage": 78.95, "elapsed_time": "15:36:17", "remaining_time": "4:09:34", "throughput": 638.02, "total_tokens": 35842376}
{"current_steps": 3070, "total_steps": 3882, "loss": 0.2116, "lr": 1.1497702865155274e-05, "epoch": 1.580994076744785, "percentage": 79.08, "elapsed_time": "15:37:45", "remaining_time": "4:08:02", "throughput": 638.06, "total_tokens": 35900824}
{"current_steps": 3075, "total_steps": 3882, "loss": 0.1676, "lr": 1.1362149538734745e-05, "epoch": 1.58356940509915, "percentage": 79.21, "elapsed_time": "15:39:14", "remaining_time": "4:06:29", "throughput": 638.1, "total_tokens": 35959264}
{"current_steps": 3080, "total_steps": 3882, "loss": 0.1811, "lr": 1.1227297515447543e-05, "epoch": 1.5861447334535153, "percentage": 79.34, "elapsed_time": "15:40:43", "remaining_time": "4:04:57", "throughput": 638.12, "total_tokens": 36017760}
{"current_steps": 3085, "total_steps": 3882, "loss": 0.1298, "lr": 1.1093149242949152e-05, "epoch": 1.5887200618078805, "percentage": 79.47, "elapsed_time": "15:42:13", "remaining_time": "4:03:25", "throughput": 638.13, "total_tokens": 36076280}
{"current_steps": 3090, "total_steps": 3882, "loss": 0.1979, "lr": 1.0959707156121506e-05, "epoch": 1.5912953901622457, "percentage": 79.6, "elapsed_time": "15:43:44", "remaining_time": "4:01:53", "throughput": 638.15, "total_tokens": 36134784}
{"current_steps": 3095, "total_steps": 3882, "loss": 0.2539, "lr": 1.0826973677028767e-05, "epoch": 1.593870718516611, "percentage": 79.73, "elapsed_time": "15:45:18", "remaining_time": "4:00:22", "throughput": 638.12, "total_tokens": 36193288}
{"current_steps": 3100, "total_steps": 3882, "loss": 0.2139, "lr": 1.0694951214873427e-05, "epoch": 1.596446046870976, "percentage": 79.86, "elapsed_time": "15:46:53", "remaining_time": "3:58:51", "throughput": 638.08, "total_tokens": 36251744}
{"current_steps": 3100, "total_steps": 3882, "eval_loss": 0.2746882140636444, "epoch": 1.596446046870976, "percentage": 79.86, "elapsed_time": "15:47:14", "remaining_time": "3:58:56", "throughput": 637.85, "total_tokens": 36251744}
{"current_steps": 3105, "total_steps": 3882, "loss": 0.1658, "lr": 1.0563642165952497e-05, "epoch": 1.5990213752253413, "percentage": 79.98, "elapsed_time": "15:48:54", "remaining_time": "3:57:27", "throughput": 637.76, "total_tokens": 36310200}
{"current_steps": 3110, "total_steps": 3882, "loss": 0.2308, "lr": 1.043304891361408e-05, "epoch": 1.6015967035797063, "percentage": 80.11, "elapsed_time": "15:50:29", "remaining_time": "3:55:56", "throughput": 637.71, "total_tokens": 36368688}
{"current_steps": 3115, "total_steps": 3882, "loss": 0.1838, "lr": 1.030317382821409e-05, "epoch": 1.6041720319340715, "percentage": 80.24, "elapsed_time": "15:52:05", "remaining_time": "3:54:25", "throughput": 637.67, "total_tokens": 36427192}
{"current_steps": 3120, "total_steps": 3882, "loss": 0.1604, "lr": 1.0174019267073225e-05, "epoch": 1.6067473602884368, "percentage": 80.37, "elapsed_time": "15:53:42", "remaining_time": "3:52:55", "throughput": 637.61, "total_tokens": 36485656}
{"current_steps": 3125, "total_steps": 3882, "loss": 0.2069, "lr": 1.0045587574434157e-05, "epoch": 1.609322688642802, "percentage": 80.5, "elapsed_time": "15:55:19", "remaining_time": "3:51:24", "throughput": 637.56, "total_tokens": 36544128}
{"current_steps": 3130, "total_steps": 3882, "loss": 0.2161, "lr": 9.917881081419033e-06, "epoch": 1.6118980169971673, "percentage": 80.63, "elapsed_time": "15:56:54", "remaining_time": "3:49:54", "throughput": 637.51, "total_tokens": 36602568}
{"current_steps": 3135, "total_steps": 3882, "loss": 0.2202, "lr": 9.790902105987103e-06, "epoch": 1.6144733453515323, "percentage": 80.76, "elapsed_time": "15:58:29", "remaining_time": "3:48:23", "throughput": 637.48, "total_tokens": 36661064}
{"current_steps": 3140, "total_steps": 3882, "loss": 0.1867, "lr": 9.66465295289274e-06, "epoch": 1.6170486737058976, "percentage": 80.89, "elapsed_time": "16:00:04", "remaining_time": "3:46:52", "throughput": 637.45, "total_tokens": 36719544}
{"current_steps": 3145, "total_steps": 3882, "loss": 0.1811, "lr": 9.539135913643465e-06, "epoch": 1.6196240020602626, "percentage": 81.01, "elapsed_time": "16:01:38", "remaining_time": "3:45:21", "throughput": 637.42, "total_tokens": 36778064}
{"current_steps": 3150, "total_steps": 3882, "loss": 0.204, "lr": 9.414353266458509e-06, "epoch": 1.6221993304146278, "percentage": 81.14, "elapsed_time": "16:03:14", "remaining_time": "3:43:50", "throughput": 637.37, "total_tokens": 36836560}
{"current_steps": 3150, "total_steps": 3882, "eval_loss": 0.2422773241996765, "epoch": 1.6221993304146278, "percentage": 81.14, "elapsed_time": "16:03:34", "remaining_time": "3:43:54", "throughput": 637.15, "total_tokens": 36836560}
{"current_steps": 3155, "total_steps": 3882, "loss": 0.211, "lr": 9.29030727622735e-06, "epoch": 1.624774658768993, "percentage": 81.27, "elapsed_time": "16:05:14", "remaining_time": "3:42:25", "throughput": 637.05, "total_tokens": 36895016}
{"current_steps": 3160, "total_steps": 3882, "loss": 0.1973, "lr": 9.167000194468684e-06, "epoch": 1.6273499871233583, "percentage": 81.4, "elapsed_time": "16:06:49", "remaining_time": "3:40:54", "throughput": 637.02, "total_tokens": 36953528}
{"current_steps": 3165, "total_steps": 3882, "loss": 0.2209, "lr": 9.044434259289452e-06, "epoch": 1.6299253154777236, "percentage": 81.53, "elapsed_time": "16:08:24", "remaining_time": "3:39:23", "throughput": 636.99, "total_tokens": 37012056}
{"current_steps": 3170, "total_steps": 3882, "loss": 0.1527, "lr": 8.922611695344352e-06, "epoch": 1.6325006438320886, "percentage": 81.66, "elapsed_time": "16:09:59", "remaining_time": "3:37:51", "throughput": 636.96, "total_tokens": 37070528}
{"current_steps": 3175, "total_steps": 3882, "loss": 0.189, "lr": 8.801534713795339e-06, "epoch": 1.6350759721864536, "percentage": 81.79, "elapsed_time": "16:11:32", "remaining_time": "3:36:20", "throughput": 636.94, "total_tokens": 37129008}
{"current_steps": 3180, "total_steps": 3882, "loss": 0.1712, "lr": 8.681205512271574e-06, "epoch": 1.6376513005408189, "percentage": 81.92, "elapsed_time": "16:13:05", "remaining_time": "3:34:48", "throughput": 636.93, "total_tokens": 37187472}
{"current_steps": 3185, "total_steps": 3882, "loss": 0.1751, "lr": 8.561626274829498e-06, "epoch": 1.6402266288951841, "percentage": 82.05, "elapsed_time": "16:14:39", "remaining_time": "3:33:17", "throughput": 636.9, "total_tokens": 37245936}
{"current_steps": 3190, "total_steps": 3882, "loss": 0.1963, "lr": 8.442799171913201e-06, "epoch": 1.6428019572495494, "percentage": 82.17, "elapsed_time": "16:16:12", "remaining_time": "3:31:46", "throughput": 636.89, "total_tokens": 37304424}
{"current_steps": 3195, "total_steps": 3882, "loss": 0.1133, "lr": 8.324726360314995e-06, "epoch": 1.6453772856039146, "percentage": 82.3, "elapsed_time": "16:17:46", "remaining_time": "3:30:14", "throughput": 636.87, "total_tokens": 37362888}
{"current_steps": 3200, "total_steps": 3882, "loss": 0.1851, "lr": 8.207409983136332e-06, "epoch": 1.6479526139582796, "percentage": 82.43, "elapsed_time": "16:19:18", "remaining_time": "3:28:43", "throughput": 636.86, "total_tokens": 37421416}
{"current_steps": 3200, "total_steps": 3882, "eval_loss": 0.22857078909873962, "epoch": 1.6479526139582796, "percentage": 82.43, "elapsed_time": "16:19:39", "remaining_time": "3:28:47", "throughput": 636.64, "total_tokens": 37421416}
{"current_steps": 3205, "total_steps": 3882, "loss": 0.2208, "lr": 8.090852169748848e-06, "epoch": 1.6505279423126449, "percentage": 82.56, "elapsed_time": "16:21:19", "remaining_time": "3:27:17", "throughput": 636.56, "total_tokens": 37479848}
{"current_steps": 3210, "total_steps": 3882, "loss": 0.2252, "lr": 7.975055035755713e-06, "epoch": 1.65310327066701, "percentage": 82.69, "elapsed_time": "16:22:52", "remaining_time": "3:25:45", "throughput": 636.54, "total_tokens": 37538312}
{"current_steps": 3215, "total_steps": 3882, "loss": 0.254, "lr": 7.860020682953284e-06, "epoch": 1.6556785990213752, "percentage": 82.82, "elapsed_time": "16:24:27", "remaining_time": "3:24:14", "throughput": 636.51, "total_tokens": 37596800}
{"current_steps": 3220, "total_steps": 3882, "loss": 0.1741, "lr": 7.745751199292923e-06, "epoch": 1.6582539273757404, "percentage": 82.95, "elapsed_time": "16:26:00", "remaining_time": "3:22:42", "throughput": 636.49, "total_tokens": 37655312}
{"current_steps": 3225, "total_steps": 3882, "loss": 0.2133, "lr": 7.632248658843088e-06, "epoch": 1.6608292557301056, "percentage": 83.08, "elapsed_time": "16:27:34", "remaining_time": "3:21:11", "throughput": 636.48, "total_tokens": 37713800}
{"current_steps": 3230, "total_steps": 3882, "loss": 0.1509, "lr": 7.51951512175168e-06, "epoch": 1.663404584084471, "percentage": 83.2, "elapsed_time": "16:29:06", "remaining_time": "3:19:39", "throughput": 636.47, "total_tokens": 37772280}
{"current_steps": 3235, "total_steps": 3882, "loss": 0.2229, "lr": 7.407552634208714e-06, "epoch": 1.665979912438836, "percentage": 83.33, "elapsed_time": "16:30:40", "remaining_time": "3:18:08", "throughput": 636.45, "total_tokens": 37830800}
{"current_steps": 3240, "total_steps": 3882, "loss": 0.1795, "lr": 7.2963632284090756e-06, "epoch": 1.6685552407932012, "percentage": 83.46, "elapsed_time": "16:32:13", "remaining_time": "3:16:36", "throughput": 636.43, "total_tokens": 37889280}
{"current_steps": 3245, "total_steps": 3882, "loss": 0.2316, "lr": 7.185948922515762e-06, "epoch": 1.6711305691475662, "percentage": 83.59, "elapsed_time": "16:33:47", "remaining_time": "3:15:04", "throughput": 636.42, "total_tokens": 37947728}
{"current_steps": 3250, "total_steps": 3882, "loss": 0.2072, "lr": 7.076311720623108e-06, "epoch": 1.6737058975019314, "percentage": 83.72, "elapsed_time": "16:35:20", "remaining_time": "3:13:33", "throughput": 636.4, "total_tokens": 38006200}
{"current_steps": 3250, "total_steps": 3882, "eval_loss": 0.2405671775341034, "epoch": 1.6737058975019314, "percentage": 83.72, "elapsed_time": "16:35:41", "remaining_time": "3:13:37", "throughput": 636.18, "total_tokens": 38006200}
{"current_steps": 3255, "total_steps": 3882, "loss": 0.2139, "lr": 6.967453612720543e-06, "epoch": 1.6762812258562967, "percentage": 83.85, "elapsed_time": "16:37:21", "remaining_time": "3:12:07", "throughput": 636.1, "total_tokens": 38064672}
{"current_steps": 3260, "total_steps": 3882, "loss": 0.2038, "lr": 6.859376574656395e-06, "epoch": 1.678856554210662, "percentage": 83.98, "elapsed_time": "16:38:54", "remaining_time": "3:10:35", "throughput": 636.08, "total_tokens": 38123192}
{"current_steps": 3265, "total_steps": 3882, "loss": 0.232, "lr": 6.7520825681020476e-06, "epoch": 1.6814318825650272, "percentage": 84.11, "elapsed_time": "16:40:27", "remaining_time": "3:09:03", "throughput": 636.07, "total_tokens": 38181696}
{"current_steps": 3270, "total_steps": 3882, "loss": 0.1632, "lr": 6.6455735405163145e-06, "epoch": 1.6840072109193922, "percentage": 84.23, "elapsed_time": "16:42:01", "remaining_time": "3:07:32", "throughput": 636.05, "total_tokens": 38240216}
{"current_steps": 3275, "total_steps": 3882, "loss": 0.1423, "lr": 6.539851425110138e-06, "epoch": 1.6865825392737575, "percentage": 84.36, "elapsed_time": "16:43:34", "remaining_time": "3:06:00", "throughput": 636.04, "total_tokens": 38298688}
{"current_steps": 3280, "total_steps": 3882, "loss": 0.1673, "lr": 6.434918140811447e-06, "epoch": 1.6891578676281225, "percentage": 84.49, "elapsed_time": "16:45:08", "remaining_time": "3:04:28", "throughput": 636.02, "total_tokens": 38357128}
{"current_steps": 3285, "total_steps": 3882, "loss": 0.137, "lr": 6.330775592230365e-06, "epoch": 1.6917331959824877, "percentage": 84.62, "elapsed_time": "16:46:41", "remaining_time": "3:02:57", "throughput": 636.0, "total_tokens": 38415600}
{"current_steps": 3290, "total_steps": 3882, "loss": 0.241, "lr": 6.2274256696246314e-06, "epoch": 1.694308524336853, "percentage": 84.75, "elapsed_time": "16:48:15", "remaining_time": "3:01:25", "throughput": 635.98, "total_tokens": 38474112}
{"current_steps": 3295, "total_steps": 3882, "loss": 0.1795, "lr": 6.124870248865289e-06, "epoch": 1.6968838526912182, "percentage": 84.88, "elapsed_time": "16:49:49", "remaining_time": "2:59:53", "throughput": 635.97, "total_tokens": 38532624}
{"current_steps": 3300, "total_steps": 3882, "loss": 0.2145, "lr": 6.023111191402603e-06, "epoch": 1.6994591810455835, "percentage": 85.01, "elapsed_time": "16:51:22", "remaining_time": "2:58:22", "throughput": 635.95, "total_tokens": 38591128}
{"current_steps": 3300, "total_steps": 3882, "eval_loss": 0.2692144513130188, "epoch": 1.6994591810455835, "percentage": 85.01, "elapsed_time": "16:51:42", "remaining_time": "2:58:25", "throughput": 635.74, "total_tokens": 38591128}
{"current_steps": 3305, "total_steps": 3882, "loss": 0.1577, "lr": 5.922150344232358e-06, "epoch": 1.7020345093999485, "percentage": 85.14, "elapsed_time": "16:53:21", "remaining_time": "2:56:55", "throughput": 635.67, "total_tokens": 38649640}
{"current_steps": 3310, "total_steps": 3882, "loss": 0.1865, "lr": 5.821989539862227e-06, "epoch": 1.7046098377543135, "percentage": 85.27, "elapsed_time": "16:54:55", "remaining_time": "2:55:23", "throughput": 635.65, "total_tokens": 38708112}
{"current_steps": 3315, "total_steps": 3882, "loss": 0.2041, "lr": 5.722630596278617e-06, "epoch": 1.7071851661086788, "percentage": 85.39, "elapsed_time": "16:56:28", "remaining_time": "2:53:51", "throughput": 635.64, "total_tokens": 38766624}
{"current_steps": 3320, "total_steps": 3882, "loss": 0.2379, "lr": 5.624075316913568e-06, "epoch": 1.709760494463044, "percentage": 85.52, "elapsed_time": "16:58:02", "remaining_time": "2:52:19", "throughput": 635.62, "total_tokens": 38825104}
{"current_steps": 3325, "total_steps": 3882, "loss": 0.1446, "lr": 5.526325490612106e-06, "epoch": 1.7123358228174093, "percentage": 85.65, "elapsed_time": "16:59:35", "remaining_time": "2:50:48", "throughput": 635.6, "total_tokens": 38883560}
{"current_steps": 3330, "total_steps": 3882, "loss": 0.2214, "lr": 5.429382891599732e-06, "epoch": 1.7149111511717745, "percentage": 85.78, "elapsed_time": "17:01:09", "remaining_time": "2:49:16", "throughput": 635.58, "total_tokens": 38942040}
{"current_steps": 3335, "total_steps": 3882, "loss": 0.1735, "lr": 5.333249279450231e-06, "epoch": 1.7174864795261395, "percentage": 85.91, "elapsed_time": "17:02:43", "remaining_time": "2:47:44", "throughput": 635.57, "total_tokens": 39000488}
{"current_steps": 3340, "total_steps": 3882, "loss": 0.1996, "lr": 5.237926399053699e-06, "epoch": 1.7200618078805048, "percentage": 86.04, "elapsed_time": "17:04:15", "remaining_time": "2:46:12", "throughput": 635.56, "total_tokens": 39058960}
{"current_steps": 3345, "total_steps": 3882, "loss": 0.2478, "lr": 5.143415980584948e-06, "epoch": 1.7226371362348698, "percentage": 86.17, "elapsed_time": "17:05:49", "remaining_time": "2:44:41", "throughput": 635.55, "total_tokens": 39117416}
{"current_steps": 3350, "total_steps": 3882, "loss": 0.2158, "lr": 5.049719739472009e-06, "epoch": 1.725212464589235, "percentage": 86.3, "elapsed_time": "17:07:22", "remaining_time": "2:43:09", "throughput": 635.53, "total_tokens": 39175888}
{"current_steps": 3350, "total_steps": 3882, "eval_loss": 0.2447366714477539, "epoch": 1.725212464589235, "percentage": 86.3, "elapsed_time": "17:07:42", "remaining_time": "2:43:12", "throughput": 635.33, "total_tokens": 39175888}
{"current_steps": 3355, "total_steps": 3882, "loss": 0.1795, "lr": 4.956839376365069e-06, "epoch": 1.7277877929436003, "percentage": 86.42, "elapsed_time": "17:09:22", "remaining_time": "2:41:41", "throughput": 635.25, "total_tokens": 39234336}
{"current_steps": 3360, "total_steps": 3882, "loss": 0.1964, "lr": 4.864776577105584e-06, "epoch": 1.7303631212979655, "percentage": 86.55, "elapsed_time": "17:10:55", "remaining_time": "2:40:09", "throughput": 635.24, "total_tokens": 39292800}
{"current_steps": 3365, "total_steps": 3882, "loss": 0.1558, "lr": 4.773533012695663e-06, "epoch": 1.7329384496523308, "percentage": 86.68, "elapsed_time": "17:12:28", "remaining_time": "2:38:37", "throughput": 635.22, "total_tokens": 39351272}
{"current_steps": 3370, "total_steps": 3882, "loss": 0.187, "lr": 4.683110339267732e-06, "epoch": 1.7355137780066958, "percentage": 86.81, "elapsed_time": "17:14:02", "remaining_time": "2:37:05", "throughput": 635.21, "total_tokens": 39409736}
{"current_steps": 3375, "total_steps": 3882, "loss": 0.1945, "lr": 4.5935101980545295e-06, "epoch": 1.738089106361061, "percentage": 86.94, "elapsed_time": "17:15:35", "remaining_time": "2:35:34", "throughput": 635.2, "total_tokens": 39468176}
{"current_steps": 3380, "total_steps": 3882, "loss": 0.1475, "lr": 4.5047342153592245e-06, "epoch": 1.740664434715426, "percentage": 87.07, "elapsed_time": "17:17:08", "remaining_time": "2:34:02", "throughput": 635.19, "total_tokens": 39526688}
{"current_steps": 3385, "total_steps": 3882, "loss": 0.1334, "lr": 4.416784002526003e-06, "epoch": 1.7432397630697913, "percentage": 87.2, "elapsed_time": "17:18:41", "remaining_time": "2:32:30", "throughput": 635.18, "total_tokens": 39585152}
{"current_steps": 3390, "total_steps": 3882, "loss": 0.1953, "lr": 4.329661155910736e-06, "epoch": 1.7458150914241566, "percentage": 87.33, "elapsed_time": "17:20:14", "remaining_time": "2:30:58", "throughput": 635.16, "total_tokens": 39643640}
{"current_steps": 3395, "total_steps": 3882, "loss": 0.2158, "lr": 4.243367256852049e-06, "epoch": 1.7483904197785218, "percentage": 87.45, "elapsed_time": "17:21:48", "remaining_time": "2:29:26", "throughput": 635.15, "total_tokens": 39702144}
{"current_steps": 3400, "total_steps": 3882, "loss": 0.1488, "lr": 4.15790387164261e-06, "epoch": 1.750965748132887, "percentage": 87.58, "elapsed_time": "17:23:21", "remaining_time": "2:27:54", "throughput": 635.14, "total_tokens": 39760664}
{"current_steps": 3400, "total_steps": 3882, "eval_loss": 0.22250184416770935, "epoch": 1.750965748132887, "percentage": 87.58, "elapsed_time": "17:23:42", "remaining_time": "2:27:57", "throughput": 634.93, "total_tokens": 39760664}
{"current_steps": 3405, "total_steps": 3882, "loss": 0.1891, "lr": 4.073272551500701e-06, "epoch": 1.753541076487252, "percentage": 87.71, "elapsed_time": "17:25:22", "remaining_time": "2:26:26", "throughput": 634.85, "total_tokens": 39819128}
{"current_steps": 3401, "total_steps": 3400, "epoch": 1.751995879474633, "percentage": 100.03, "elapsed_time": "0:01:53", "remaining_time": "0:00:00", "throughput": 351097.84, "total_tokens": 39772368}