{"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}