| {"current_steps": 1, "total_steps": 775, "loss": 12.2736, "learning_rate": 5.000000000000001e-07, "epoch": 0.006436041834271922, "percentage": 0.13, "elapsed_time": "0:00:33", "remaining_time": "7:10:57", "throughput": "196.84", "total_tokens": 6576} | |
| {"current_steps": 2, "total_steps": 775, "loss": 12.5518, "learning_rate": 1.0000000000000002e-06, "epoch": 0.012872083668543845, "percentage": 0.26, "elapsed_time": "0:00:46", "remaining_time": "5:00:32", "throughput": "285.33", "total_tokens": 13312} | |
| {"current_steps": 3, "total_steps": 775, "loss": 11.8551, "learning_rate": 1.5e-06, "epoch": 0.019308125502815767, "percentage": 0.39, "elapsed_time": "0:00:59", "remaining_time": "4:16:38", "throughput": "333.43", "total_tokens": 19952} | |
| {"current_steps": 4, "total_steps": 775, "loss": 10.9045, "learning_rate": 2.0000000000000003e-06, "epoch": 0.02574416733708769, "percentage": 0.52, "elapsed_time": "0:01:13", "remaining_time": "3:54:41", "throughput": "364.66", "total_tokens": 26640} | |
| {"current_steps": 5, "total_steps": 775, "loss": 8.9845, "learning_rate": 2.5e-06, "epoch": 0.032180209171359615, "percentage": 0.65, "elapsed_time": "0:01:26", "remaining_time": "3:41:24", "throughput": "386.72", "total_tokens": 33360} | |
| {"current_steps": 6, "total_steps": 775, "loss": 6.35, "learning_rate": 3e-06, "epoch": 0.038616251005631534, "percentage": 0.77, "elapsed_time": "0:01:39", "remaining_time": "3:32:28", "throughput": "401.96", "total_tokens": 39984} | |
| {"current_steps": 7, "total_steps": 775, "loss": 5.4864, "learning_rate": 3.5e-06, "epoch": 0.04505229283990346, "percentage": 0.9, "elapsed_time": "0:01:52", "remaining_time": "3:26:03", "throughput": "412.61", "total_tokens": 46496} | |
| {"current_steps": 8, "total_steps": 775, "loss": 2.8013, "learning_rate": 4.000000000000001e-06, "epoch": 0.05148833467417538, "percentage": 1.03, "elapsed_time": "0:02:05", "remaining_time": "3:21:10", "throughput": "421.04", "total_tokens": 53008} | |
| {"current_steps": 9, "total_steps": 775, "loss": 0.9851, "learning_rate": 4.5e-06, "epoch": 0.057924376508447305, "percentage": 1.16, "elapsed_time": "0:02:19", "remaining_time": "3:17:20", "throughput": "430.24", "total_tokens": 59856} | |
| {"current_steps": 10, "total_steps": 775, "loss": 0.5088, "learning_rate": 5e-06, "epoch": 0.06436041834271923, "percentage": 1.29, "elapsed_time": "0:02:32", "remaining_time": "3:14:12", "throughput": "435.61", "total_tokens": 66352} | |
| {"current_steps": 11, "total_steps": 775, "loss": 2.022, "learning_rate": 4.99997891923933e-06, "epoch": 0.07079646017699115, "percentage": 1.42, "elapsed_time": "0:02:45", "remaining_time": "3:11:37", "throughput": "442.08", "total_tokens": 73184} | |
| {"current_steps": 12, "total_steps": 775, "loss": 0.5842, "learning_rate": 4.999915677312839e-06, "epoch": 0.07723250201126307, "percentage": 1.55, "elapsed_time": "0:02:58", "remaining_time": "3:09:24", "throughput": "447.95", "total_tokens": 80064} | |
| {"current_steps": 13, "total_steps": 775, "loss": 0.8538, "learning_rate": 4.999810275287077e-06, "epoch": 0.083668543845535, "percentage": 1.68, "elapsed_time": "0:03:11", "remaining_time": "3:07:31", "throughput": "451.61", "total_tokens": 86688} | |
| {"current_steps": 14, "total_steps": 775, "loss": 0.5843, "learning_rate": 4.9996627149396075e-06, "epoch": 0.09010458567980692, "percentage": 1.81, "elapsed_time": "0:03:25", "remaining_time": "3:05:51", "throughput": "454.14", "total_tokens": 93168} | |
| {"current_steps": 15, "total_steps": 775, "loss": 0.4456, "learning_rate": 4.999472998758979e-06, "epoch": 0.09654062751407884, "percentage": 1.94, "elapsed_time": "0:03:38", "remaining_time": "3:04:23", "throughput": "456.56", "total_tokens": 99696} | |
| {"current_steps": 16, "total_steps": 775, "loss": 0.3542, "learning_rate": 4.99924112994468e-06, "epoch": 0.10297666934835076, "percentage": 2.06, "elapsed_time": "0:03:51", "remaining_time": "3:03:05", "throughput": "457.81", "total_tokens": 106016} | |
| {"current_steps": 17, "total_steps": 775, "loss": 0.3416, "learning_rate": 4.998967112407087e-06, "epoch": 0.10941271118262269, "percentage": 2.19, "elapsed_time": "0:04:04", "remaining_time": "3:01:54", "throughput": "460.31", "total_tokens": 112672} | |
| {"current_steps": 18, "total_steps": 775, "loss": 0.2803, "learning_rate": 4.9986509507673986e-06, "epoch": 0.11584875301689461, "percentage": 2.32, "elapsed_time": "0:04:17", "remaining_time": "3:00:49", "throughput": "462.47", "total_tokens": 119312} | |
| {"current_steps": 19, "total_steps": 775, "loss": 0.2603, "learning_rate": 4.998292650357558e-06, "epoch": 0.12228479485116653, "percentage": 2.45, "elapsed_time": "0:04:31", "remaining_time": "2:59:50", "throughput": "464.66", "total_tokens": 126016} | |
| {"current_steps": 20, "total_steps": 775, "loss": 0.2456, "learning_rate": 4.99789221722016e-06, "epoch": 0.12872083668543846, "percentage": 2.58, "elapsed_time": "0:04:44", "remaining_time": "2:58:56", "throughput": "466.61", "total_tokens": 132704} | |
| {"current_steps": 21, "total_steps": 775, "loss": 0.2269, "learning_rate": 4.997449658108354e-06, "epoch": 0.13515687851971037, "percentage": 2.71, "elapsed_time": "0:04:57", "remaining_time": "2:58:05", "throughput": "468.64", "total_tokens": 139472} | |
| {"current_steps": 22, "total_steps": 775, "loss": 0.2896, "learning_rate": 4.996964980485725e-06, "epoch": 0.1415929203539823, "percentage": 2.84, "elapsed_time": "0:05:10", "remaining_time": "2:57:18", "throughput": "469.63", "total_tokens": 145968} | |
| {"current_steps": 23, "total_steps": 775, "loss": 0.1415, "learning_rate": 4.996438192526173e-06, "epoch": 0.14802896218825423, "percentage": 2.97, "elapsed_time": "0:05:24", "remaining_time": "2:56:34", "throughput": "469.52", "total_tokens": 152144} | |
| {"current_steps": 24, "total_steps": 775, "loss": 0.2228, "learning_rate": 4.995869303113768e-06, "epoch": 0.15446500402252614, "percentage": 3.1, "elapsed_time": "0:05:37", "remaining_time": "2:55:52", "throughput": "469.78", "total_tokens": 158432} | |
| {"current_steps": 25, "total_steps": 775, "loss": 0.1537, "learning_rate": 4.995258321842611e-06, "epoch": 0.16090104585679807, "percentage": 3.23, "elapsed_time": "0:05:50", "remaining_time": "2:55:13", "throughput": "469.88", "total_tokens": 164672} | |
| {"current_steps": 26, "total_steps": 775, "loss": 0.2146, "learning_rate": 4.994605259016658e-06, "epoch": 0.16733708769107, "percentage": 3.35, "elapsed_time": "0:06:03", "remaining_time": "2:54:36", "throughput": "469.94", "total_tokens": 170896} | |
| {"current_steps": 27, "total_steps": 775, "loss": 0.1688, "learning_rate": 4.993910125649561e-06, "epoch": 0.1737731295253419, "percentage": 3.48, "elapsed_time": "0:06:16", "remaining_time": "2:54:00", "throughput": "472.07", "total_tokens": 177904} | |
| {"current_steps": 28, "total_steps": 775, "loss": 0.1592, "learning_rate": 4.99317293346447e-06, "epoch": 0.18020917135961384, "percentage": 3.61, "elapsed_time": "0:06:30", "remaining_time": "2:53:26", "throughput": "473.37", "total_tokens": 184640} | |
| {"current_steps": 29, "total_steps": 775, "loss": 0.2333, "learning_rate": 4.992393694893844e-06, "epoch": 0.18664521319388577, "percentage": 3.74, "elapsed_time": "0:06:43", "remaining_time": "2:52:53", "throughput": "473.68", "total_tokens": 191008} | |
| {"current_steps": 30, "total_steps": 775, "loss": 0.1812, "learning_rate": 4.991572423079236e-06, "epoch": 0.19308125502815768, "percentage": 3.87, "elapsed_time": "0:06:56", "remaining_time": "2:52:22", "throughput": "474.37", "total_tokens": 197568} | |
| {"current_steps": 31, "total_steps": 775, "loss": 0.1925, "learning_rate": 4.990709131871074e-06, "epoch": 0.1995172968624296, "percentage": 4.0, "elapsed_time": "0:07:09", "remaining_time": "2:51:52", "throughput": "475.40", "total_tokens": 204272} | |
| {"current_steps": 32, "total_steps": 775, "loss": 0.1864, "learning_rate": 4.989803835828426e-06, "epoch": 0.20595333869670152, "percentage": 4.13, "elapsed_time": "0:07:22", "remaining_time": "2:51:23", "throughput": "476.28", "total_tokens": 210944} | |
| {"current_steps": 33, "total_steps": 775, "loss": 0.1848, "learning_rate": 4.988856550218755e-06, "epoch": 0.21238938053097345, "percentage": 4.26, "elapsed_time": "0:07:36", "remaining_time": "2:50:55", "throughput": "477.05", "total_tokens": 217584} | |
| {"current_steps": 34, "total_steps": 775, "loss": 0.1466, "learning_rate": 4.987867291017662e-06, "epoch": 0.21882542236524538, "percentage": 4.39, "elapsed_time": "0:07:49", "remaining_time": "2:50:28", "throughput": "477.42", "total_tokens": 224064} | |
| {"current_steps": 35, "total_steps": 775, "loss": 0.1481, "learning_rate": 4.986836074908616e-06, "epoch": 0.2252614641995173, "percentage": 4.52, "elapsed_time": "0:08:02", "remaining_time": "2:50:02", "throughput": "478.48", "total_tokens": 230880} | |
| {"current_steps": 36, "total_steps": 775, "loss": 0.149, "learning_rate": 4.985762919282674e-06, "epoch": 0.23169750603378922, "percentage": 4.65, "elapsed_time": "0:08:15", "remaining_time": "2:49:36", "throughput": "478.70", "total_tokens": 237312} | |
| {"current_steps": 37, "total_steps": 775, "loss": 0.1506, "learning_rate": 4.984647842238185e-06, "epoch": 0.23813354786806115, "percentage": 4.77, "elapsed_time": "0:08:28", "remaining_time": "2:49:11", "throughput": "478.72", "total_tokens": 243648} | |
| {"current_steps": 38, "total_steps": 775, "loss": 0.1709, "learning_rate": 4.983490862580486e-06, "epoch": 0.24456958970233306, "percentage": 4.9, "elapsed_time": "0:08:42", "remaining_time": "2:48:47", "throughput": "478.96", "total_tokens": 250096} | |
| {"current_steps": 39, "total_steps": 775, "loss": 0.0898, "learning_rate": 4.982291999821587e-06, "epoch": 0.251005631536605, "percentage": 5.03, "elapsed_time": "0:08:55", "remaining_time": "2:48:23", "throughput": "478.97", "total_tokens": 256432} | |
| {"current_steps": 40, "total_steps": 775, "loss": 0.1856, "learning_rate": 4.98105127417984e-06, "epoch": 0.2574416733708769, "percentage": 5.16, "elapsed_time": "0:09:08", "remaining_time": "2:47:59", "throughput": "479.60", "total_tokens": 263088} | |
| {"current_steps": 41, "total_steps": 775, "loss": 0.0947, "learning_rate": 4.979768706579595e-06, "epoch": 0.26387771520514886, "percentage": 5.29, "elapsed_time": "0:09:21", "remaining_time": "2:47:37", "throughput": "480.45", "total_tokens": 269904} | |
| {"current_steps": 42, "total_steps": 775, "loss": 0.1367, "learning_rate": 4.978444318650855e-06, "epoch": 0.27031375703942073, "percentage": 5.42, "elapsed_time": "0:09:34", "remaining_time": "2:47:14", "throughput": "481.04", "total_tokens": 276592} | |
| {"current_steps": 43, "total_steps": 775, "loss": 0.1645, "learning_rate": 4.977078132728901e-06, "epoch": 0.27674979887369267, "percentage": 5.55, "elapsed_time": "0:09:48", "remaining_time": "2:46:53", "throughput": "481.85", "total_tokens": 283424} | |
| {"current_steps": 44, "total_steps": 775, "loss": 0.1179, "learning_rate": 4.975670171853926e-06, "epoch": 0.2831858407079646, "percentage": 5.68, "elapsed_time": "0:10:01", "remaining_time": "2:46:31", "throughput": "482.49", "total_tokens": 290176} | |
| {"current_steps": 45, "total_steps": 775, "loss": 0.133, "learning_rate": 4.9742204597706386e-06, "epoch": 0.28962188254223653, "percentage": 5.81, "elapsed_time": "0:10:14", "remaining_time": "2:46:10", "throughput": "482.82", "total_tokens": 296752} | |
| {"current_steps": 46, "total_steps": 775, "loss": 0.083, "learning_rate": 4.972729020927866e-06, "epoch": 0.29605792437650846, "percentage": 5.94, "elapsed_time": "0:10:27", "remaining_time": "2:45:49", "throughput": "483.25", "total_tokens": 303392} | |
| {"current_steps": 47, "total_steps": 775, "loss": 0.1748, "learning_rate": 4.9711958804781385e-06, "epoch": 0.3024939662107804, "percentage": 6.06, "elapsed_time": "0:10:41", "remaining_time": "2:45:28", "throughput": "484.08", "total_tokens": 310304} | |
| {"current_steps": 48, "total_steps": 775, "loss": 0.1854, "learning_rate": 4.969621064277271e-06, "epoch": 0.3089300080450523, "percentage": 6.19, "elapsed_time": "0:10:54", "remaining_time": "2:45:08", "throughput": "485.23", "total_tokens": 317440} | |
| {"current_steps": 49, "total_steps": 775, "loss": 0.1232, "learning_rate": 4.968004598883923e-06, "epoch": 0.3153660498793242, "percentage": 6.32, "elapsed_time": "0:11:07", "remaining_time": "2:44:48", "throughput": "485.90", "total_tokens": 324304} | |
| {"current_steps": 50, "total_steps": 775, "loss": 0.1172, "learning_rate": 4.966346511559149e-06, "epoch": 0.32180209171359614, "percentage": 6.45, "elapsed_time": "0:11:20", "remaining_time": "2:44:29", "throughput": "486.06", "total_tokens": 330832} | |
| {"current_steps": 51, "total_steps": 775, "loss": 0.1404, "learning_rate": 4.964646830265944e-06, "epoch": 0.32823813354786807, "percentage": 6.58, "elapsed_time": "0:11:33", "remaining_time": "2:44:09", "throughput": "487.08", "total_tokens": 337952} | |
| {"current_steps": 52, "total_steps": 775, "loss": 0.137, "learning_rate": 4.962905583668766e-06, "epoch": 0.33467417538214, "percentage": 6.71, "elapsed_time": "0:11:47", "remaining_time": "2:43:50", "throughput": "487.08", "total_tokens": 344384} | |
| {"current_steps": 53, "total_steps": 775, "loss": 0.1191, "learning_rate": 4.961122801133059e-06, "epoch": 0.3411102172164119, "percentage": 6.84, "elapsed_time": "0:12:00", "remaining_time": "2:43:31", "throughput": "487.59", "total_tokens": 351184} | |
| {"current_steps": 54, "total_steps": 775, "loss": 0.1624, "learning_rate": 4.9592985127247525e-06, "epoch": 0.3475462590506838, "percentage": 6.97, "elapsed_time": "0:12:13", "remaining_time": "2:43:12", "throughput": "487.69", "total_tokens": 357696} | |
| {"current_steps": 55, "total_steps": 775, "loss": 0.1256, "learning_rate": 4.957432749209755e-06, "epoch": 0.35398230088495575, "percentage": 7.1, "elapsed_time": "0:12:26", "remaining_time": "2:42:54", "throughput": "488.00", "total_tokens": 364368} | |
| {"current_steps": 56, "total_steps": 775, "loss": 0.1274, "learning_rate": 4.955525542053438e-06, "epoch": 0.3604183427192277, "percentage": 7.23, "elapsed_time": "0:12:39", "remaining_time": "2:42:36", "throughput": "488.11", "total_tokens": 370896} | |
| {"current_steps": 57, "total_steps": 775, "loss": 0.174, "learning_rate": 4.953576923420105e-06, "epoch": 0.3668543845534996, "percentage": 7.35, "elapsed_time": "0:12:53", "remaining_time": "2:42:17", "throughput": "487.88", "total_tokens": 377168} | |
| {"current_steps": 58, "total_steps": 775, "loss": 0.0805, "learning_rate": 4.9515869261724444e-06, "epoch": 0.37329042638777155, "percentage": 7.48, "elapsed_time": "0:13:06", "remaining_time": "2:42:00", "throughput": "488.03", "total_tokens": 383728} | |
| {"current_steps": 59, "total_steps": 775, "loss": 0.1181, "learning_rate": 4.949555583870983e-06, "epoch": 0.3797264682220434, "percentage": 7.61, "elapsed_time": "0:13:19", "remaining_time": "2:41:42", "throughput": "488.37", "total_tokens": 390448} | |
| {"current_steps": 60, "total_steps": 775, "loss": 0.1613, "learning_rate": 4.9474829307735115e-06, "epoch": 0.38616251005631536, "percentage": 7.74, "elapsed_time": "0:13:32", "remaining_time": "2:41:24", "throughput": "488.45", "total_tokens": 396960} | |
| {"current_steps": 61, "total_steps": 775, "loss": 0.0885, "learning_rate": 4.9453690018345144e-06, "epoch": 0.3925985518905873, "percentage": 7.87, "elapsed_time": "0:13:45", "remaining_time": "2:41:07", "throughput": "488.78", "total_tokens": 403680} | |
| {"current_steps": 62, "total_steps": 775, "loss": 0.1526, "learning_rate": 4.943213832704575e-06, "epoch": 0.3990345937248592, "percentage": 8.0, "elapsed_time": "0:13:59", "remaining_time": "2:40:49", "throughput": "488.74", "total_tokens": 410096} | |
| {"current_steps": 63, "total_steps": 775, "loss": 0.1114, "learning_rate": 4.941017459729778e-06, "epoch": 0.40547063555913115, "percentage": 8.13, "elapsed_time": "0:14:12", "remaining_time": "2:40:32", "throughput": "488.87", "total_tokens": 416672} | |
| {"current_steps": 64, "total_steps": 775, "loss": 0.1139, "learning_rate": 4.938779919951092e-06, "epoch": 0.41190667739340303, "percentage": 8.26, "elapsed_time": "0:14:25", "remaining_time": "2:40:15", "throughput": "488.87", "total_tokens": 423136} | |
| {"current_steps": 65, "total_steps": 775, "loss": 0.0878, "learning_rate": 4.936501251103751e-06, "epoch": 0.41834271922767496, "percentage": 8.39, "elapsed_time": "0:14:38", "remaining_time": "2:39:58", "throughput": "489.21", "total_tokens": 429888} | |
| {"current_steps": 66, "total_steps": 775, "loss": 0.1047, "learning_rate": 4.934181491616613e-06, "epoch": 0.4247787610619469, "percentage": 8.52, "elapsed_time": "0:14:51", "remaining_time": "2:39:41", "throughput": "489.63", "total_tokens": 436720} | |
| {"current_steps": 67, "total_steps": 775, "loss": 0.1323, "learning_rate": 4.9318206806115125e-06, "epoch": 0.43121480289621883, "percentage": 8.65, "elapsed_time": "0:15:05", "remaining_time": "2:39:24", "throughput": "490.13", "total_tokens": 443648} | |
| {"current_steps": 68, "total_steps": 775, "loss": 0.095, "learning_rate": 4.929418857902603e-06, "epoch": 0.43765084473049076, "percentage": 8.77, "elapsed_time": "0:15:18", "remaining_time": "2:39:08", "throughput": "490.51", "total_tokens": 450464} | |
| {"current_steps": 69, "total_steps": 775, "loss": 0.2024, "learning_rate": 4.926976063995687e-06, "epoch": 0.4440868865647627, "percentage": 8.9, "elapsed_time": "0:15:31", "remaining_time": "2:38:51", "throughput": "490.90", "total_tokens": 457296} | |
| {"current_steps": 70, "total_steps": 775, "loss": 0.1614, "learning_rate": 4.9244923400875245e-06, "epoch": 0.4505229283990346, "percentage": 9.03, "elapsed_time": "0:15:44", "remaining_time": "2:38:34", "throughput": "491.13", "total_tokens": 463984} | |
| {"current_steps": 71, "total_steps": 775, "loss": 0.2073, "learning_rate": 4.921967728065147e-06, "epoch": 0.4569589702333065, "percentage": 9.16, "elapsed_time": "0:15:57", "remaining_time": "2:38:18", "throughput": "491.09", "total_tokens": 470432} | |
| {"current_steps": 72, "total_steps": 775, "loss": 0.108, "learning_rate": 4.91940227050515e-06, "epoch": 0.46339501206757844, "percentage": 9.29, "elapsed_time": "0:16:11", "remaining_time": "2:38:02", "throughput": "490.90", "total_tokens": 476736} | |
| {"current_steps": 73, "total_steps": 775, "loss": 0.1455, "learning_rate": 4.916796010672969e-06, "epoch": 0.46983105390185037, "percentage": 9.42, "elapsed_time": "0:16:24", "remaining_time": "2:37:46", "throughput": "490.57", "total_tokens": 482896} | |
| {"current_steps": 74, "total_steps": 775, "loss": 0.131, "learning_rate": 4.914148992522157e-06, "epoch": 0.4762670957361223, "percentage": 9.55, "elapsed_time": "0:16:37", "remaining_time": "2:37:30", "throughput": "490.69", "total_tokens": 489504} | |
| {"current_steps": 75, "total_steps": 775, "loss": 0.1416, "learning_rate": 4.911461260693639e-06, "epoch": 0.4827031375703942, "percentage": 9.68, "elapsed_time": "0:16:50", "remaining_time": "2:37:14", "throughput": "490.86", "total_tokens": 496160} | |
| {"current_steps": 76, "total_steps": 775, "loss": 0.1045, "learning_rate": 4.908732860514958e-06, "epoch": 0.4891391794046661, "percentage": 9.81, "elapsed_time": "0:17:04", "remaining_time": "2:36:58", "throughput": "490.75", "total_tokens": 502528} | |
| {"current_steps": 77, "total_steps": 775, "loss": 0.1466, "learning_rate": 4.905963837999518e-06, "epoch": 0.49557522123893805, "percentage": 9.94, "elapsed_time": "0:17:17", "remaining_time": "2:36:42", "throughput": "490.84", "total_tokens": 509104} | |
| {"current_steps": 78, "total_steps": 775, "loss": 0.1502, "learning_rate": 4.903154239845798e-06, "epoch": 0.50201126307321, "percentage": 10.06, "elapsed_time": "0:17:30", "remaining_time": "2:36:26", "throughput": "491.00", "total_tokens": 515760} | |
| {"current_steps": 79, "total_steps": 775, "loss": 0.1235, "learning_rate": 4.900304113436571e-06, "epoch": 0.5084473049074819, "percentage": 10.19, "elapsed_time": "0:17:43", "remaining_time": "2:36:10", "throughput": "491.08", "total_tokens": 522336} | |
| {"current_steps": 80, "total_steps": 775, "loss": 0.1022, "learning_rate": 4.897413506838103e-06, "epoch": 0.5148833467417538, "percentage": 10.32, "elapsed_time": "0:17:56", "remaining_time": "2:35:54", "throughput": "491.22", "total_tokens": 528960} | |
| {"current_steps": 81, "total_steps": 775, "loss": 0.0922, "learning_rate": 4.894482468799344e-06, "epoch": 0.5213193885760258, "percentage": 10.45, "elapsed_time": "0:18:10", "remaining_time": "2:35:39", "throughput": "491.65", "total_tokens": 535920} | |
| {"current_steps": 82, "total_steps": 775, "loss": 0.0699, "learning_rate": 4.891511048751102e-06, "epoch": 0.5277554304102977, "percentage": 10.58, "elapsed_time": "0:18:23", "remaining_time": "2:35:23", "throughput": "491.72", "total_tokens": 542496} | |
| {"current_steps": 83, "total_steps": 775, "loss": 0.1057, "learning_rate": 4.888499296805214e-06, "epoch": 0.5341914722445696, "percentage": 10.71, "elapsed_time": "0:18:36", "remaining_time": "2:35:08", "throughput": "491.50", "total_tokens": 548752} | |
| {"current_steps": 84, "total_steps": 775, "loss": 0.0793, "learning_rate": 4.8854472637536966e-06, "epoch": 0.5406275140788415, "percentage": 10.84, "elapsed_time": "0:18:49", "remaining_time": "2:34:53", "throughput": "491.90", "total_tokens": 555696} | |
| {"current_steps": 85, "total_steps": 775, "loss": 0.14, "learning_rate": 4.882355001067892e-06, "epoch": 0.5470635559131134, "percentage": 10.97, "elapsed_time": "0:19:02", "remaining_time": "2:34:37", "throughput": "491.90", "total_tokens": 562192} | |
| {"current_steps": 86, "total_steps": 775, "loss": 0.1538, "learning_rate": 4.8792225608976e-06, "epoch": 0.5534995977473853, "percentage": 11.1, "elapsed_time": "0:19:16", "remaining_time": "2:34:22", "throughput": "491.89", "total_tokens": 568672} | |
| {"current_steps": 87, "total_steps": 775, "loss": 0.1135, "learning_rate": 4.8760499960702005e-06, "epoch": 0.5599356395816573, "percentage": 11.23, "elapsed_time": "0:19:29", "remaining_time": "2:34:06", "throughput": "492.12", "total_tokens": 575440} | |
| {"current_steps": 88, "total_steps": 775, "loss": 0.1253, "learning_rate": 4.8728373600897535e-06, "epoch": 0.5663716814159292, "percentage": 11.35, "elapsed_time": "0:19:42", "remaining_time": "2:33:51", "throughput": "492.01", "total_tokens": 581808} | |
| {"current_steps": 89, "total_steps": 775, "loss": 0.1408, "learning_rate": 4.869584707136109e-06, "epoch": 0.5728077232502011, "percentage": 11.48, "elapsed_time": "0:19:55", "remaining_time": "2:33:36", "throughput": "492.23", "total_tokens": 588576} | |
| {"current_steps": 90, "total_steps": 775, "loss": 0.0916, "learning_rate": 4.8662920920639866e-06, "epoch": 0.5792437650844731, "percentage": 11.61, "elapsed_time": "0:20:08", "remaining_time": "2:33:21", "throughput": "492.43", "total_tokens": 595328} | |
| {"current_steps": 91, "total_steps": 775, "loss": 0.1016, "learning_rate": 4.86295957040205e-06, "epoch": 0.585679806918745, "percentage": 11.74, "elapsed_time": "0:20:22", "remaining_time": "2:33:06", "throughput": "492.41", "total_tokens": 601808} | |
| {"current_steps": 92, "total_steps": 775, "loss": 0.0936, "learning_rate": 4.8595871983519705e-06, "epoch": 0.5921158487530169, "percentage": 11.87, "elapsed_time": "0:20:35", "remaining_time": "2:32:51", "throughput": "492.48", "total_tokens": 608400} | |
| {"current_steps": 93, "total_steps": 775, "loss": 0.1557, "learning_rate": 4.856175032787485e-06, "epoch": 0.5985518905872889, "percentage": 12.0, "elapsed_time": "0:20:48", "remaining_time": "2:32:36", "throughput": "492.79", "total_tokens": 615296} | |
| {"current_steps": 94, "total_steps": 775, "loss": 0.0879, "learning_rate": 4.852723131253429e-06, "epoch": 0.6049879324215608, "percentage": 12.13, "elapsed_time": "0:21:01", "remaining_time": "2:32:21", "throughput": "492.85", "total_tokens": 621888} | |
| {"current_steps": 95, "total_steps": 775, "loss": 0.1399, "learning_rate": 4.849231551964771e-06, "epoch": 0.6114239742558326, "percentage": 12.26, "elapsed_time": "0:21:15", "remaining_time": "2:32:06", "throughput": "493.15", "total_tokens": 628768} | |
| {"current_steps": 96, "total_steps": 775, "loss": 0.1724, "learning_rate": 4.845700353805629e-06, "epoch": 0.6178600160901045, "percentage": 12.39, "elapsed_time": "0:21:28", "remaining_time": "2:31:51", "throughput": "492.97", "total_tokens": 635056} | |
| {"current_steps": 97, "total_steps": 775, "loss": 0.1018, "learning_rate": 4.842129596328277e-06, "epoch": 0.6242960579243765, "percentage": 12.52, "elapsed_time": "0:21:41", "remaining_time": "2:31:36", "throughput": "493.20", "total_tokens": 641872} | |
| {"current_steps": 98, "total_steps": 775, "loss": 0.0398, "learning_rate": 4.838519339752143e-06, "epoch": 0.6307320997586484, "percentage": 12.65, "elapsed_time": "0:21:54", "remaining_time": "2:31:21", "throughput": "493.48", "total_tokens": 648752} | |
| {"current_steps": 99, "total_steps": 775, "loss": 0.1094, "learning_rate": 4.834869644962789e-06, "epoch": 0.6371681415929203, "percentage": 12.77, "elapsed_time": "0:22:07", "remaining_time": "2:31:07", "throughput": "493.59", "total_tokens": 655424} | |
| {"current_steps": 100, "total_steps": 775, "loss": 0.1915, "learning_rate": 4.83118057351089e-06, "epoch": 0.6436041834271923, "percentage": 12.9, "elapsed_time": "0:22:21", "remaining_time": "2:30:52", "throughput": "493.80", "total_tokens": 662224} | |
| {"current_steps": 101, "total_steps": 775, "loss": 0.1518, "learning_rate": 4.827452187611192e-06, "epoch": 0.6500402252614642, "percentage": 13.03, "elapsed_time": "0:22:34", "remaining_time": "2:30:37", "throughput": "493.68", "total_tokens": 668576} | |
| {"current_steps": 102, "total_steps": 775, "loss": 0.141, "learning_rate": 4.823684550141464e-06, "epoch": 0.6564762670957361, "percentage": 13.16, "elapsed_time": "0:22:47", "remaining_time": "2:30:22", "throughput": "493.78", "total_tokens": 675232} | |
| {"current_steps": 103, "total_steps": 775, "loss": 0.1564, "learning_rate": 4.819877724641437e-06, "epoch": 0.6629123089300081, "percentage": 13.29, "elapsed_time": "0:23:00", "remaining_time": "2:30:08", "throughput": "493.85", "total_tokens": 681856} | |
| {"current_steps": 104, "total_steps": 775, "loss": 0.1267, "learning_rate": 4.8160317753117326e-06, "epoch": 0.66934835076428, "percentage": 13.42, "elapsed_time": "0:23:13", "remaining_time": "2:29:53", "throughput": "493.87", "total_tokens": 688416} | |
| {"current_steps": 105, "total_steps": 775, "loss": 0.228, "learning_rate": 4.81214676701278e-06, "epoch": 0.6757843925985519, "percentage": 13.55, "elapsed_time": "0:23:27", "remaining_time": "2:29:38", "throughput": "494.08", "total_tokens": 695248} | |
| {"current_steps": 106, "total_steps": 775, "loss": 0.1532, "learning_rate": 4.808222765263724e-06, "epoch": 0.6822204344328238, "percentage": 13.68, "elapsed_time": "0:23:40", "remaining_time": "2:29:24", "throughput": "494.21", "total_tokens": 701952} | |
| {"current_steps": 107, "total_steps": 775, "loss": 0.1482, "learning_rate": 4.8042598362413175e-06, "epoch": 0.6886564762670957, "percentage": 13.81, "elapsed_time": "0:23:53", "remaining_time": "2:29:09", "throughput": "494.13", "total_tokens": 708368} | |
| {"current_steps": 108, "total_steps": 775, "loss": 0.1074, "learning_rate": 4.800258046778809e-06, "epoch": 0.6950925181013676, "percentage": 13.94, "elapsed_time": "0:24:06", "remaining_time": "2:28:55", "throughput": "494.04", "total_tokens": 714768} | |
| {"current_steps": 109, "total_steps": 775, "loss": 0.1163, "learning_rate": 4.796217464364808e-06, "epoch": 0.7015285599356396, "percentage": 14.06, "elapsed_time": "0:24:19", "remaining_time": "2:28:40", "throughput": "494.26", "total_tokens": 721600} | |
| {"current_steps": 110, "total_steps": 775, "loss": 0.1871, "learning_rate": 4.792138157142158e-06, "epoch": 0.7079646017699115, "percentage": 14.19, "elapsed_time": "0:24:33", "remaining_time": "2:28:25", "throughput": "494.48", "total_tokens": 728448} | |
| {"current_steps": 111, "total_steps": 775, "loss": 0.092, "learning_rate": 4.788020193906776e-06, "epoch": 0.7144006436041834, "percentage": 14.32, "elapsed_time": "0:24:46", "remaining_time": "2:28:11", "throughput": "494.30", "total_tokens": 734720} | |
| {"current_steps": 112, "total_steps": 775, "loss": 0.0959, "learning_rate": 4.783863644106502e-06, "epoch": 0.7208366854384554, "percentage": 14.45, "elapsed_time": "0:24:59", "remaining_time": "2:27:56", "throughput": "494.28", "total_tokens": 741216} | |
| {"current_steps": 113, "total_steps": 775, "loss": 0.0853, "learning_rate": 4.779668577839921e-06, "epoch": 0.7272727272727273, "percentage": 14.58, "elapsed_time": "0:25:12", "remaining_time": "2:27:42", "throughput": "494.23", "total_tokens": 747664} | |
| {"current_steps": 114, "total_steps": 775, "loss": 0.1581, "learning_rate": 4.775435065855183e-06, "epoch": 0.7337087691069992, "percentage": 14.71, "elapsed_time": "0:25:25", "remaining_time": "2:27:28", "throughput": "494.42", "total_tokens": 754480} | |
| {"current_steps": 115, "total_steps": 775, "loss": 0.0907, "learning_rate": 4.771163179548809e-06, "epoch": 0.7401448109412712, "percentage": 14.84, "elapsed_time": "0:25:39", "remaining_time": "2:27:13", "throughput": "494.64", "total_tokens": 761344} | |
| {"current_steps": 116, "total_steps": 775, "loss": 0.0826, "learning_rate": 4.766852990964492e-06, "epoch": 0.7465808527755431, "percentage": 14.97, "elapsed_time": "0:25:52", "remaining_time": "2:26:59", "throughput": "494.82", "total_tokens": 768160} | |
| {"current_steps": 117, "total_steps": 775, "loss": 0.0979, "learning_rate": 4.762504572791873e-06, "epoch": 0.7530168946098149, "percentage": 15.1, "elapsed_time": "0:26:05", "remaining_time": "2:26:44", "throughput": "494.70", "total_tokens": 774496} | |
| {"current_steps": 118, "total_steps": 775, "loss": 0.1617, "learning_rate": 4.7581179983653224e-06, "epoch": 0.7594529364440868, "percentage": 15.23, "elapsed_time": "0:26:18", "remaining_time": "2:26:30", "throughput": "494.83", "total_tokens": 781232} | |
| {"current_steps": 119, "total_steps": 775, "loss": 0.1306, "learning_rate": 4.753693341662702e-06, "epoch": 0.7658889782783588, "percentage": 15.35, "elapsed_time": "0:26:32", "remaining_time": "2:26:16", "throughput": "495.01", "total_tokens": 788064} | |
| {"current_steps": 120, "total_steps": 775, "loss": 0.0955, "learning_rate": 4.749230677304114e-06, "epoch": 0.7723250201126307, "percentage": 15.48, "elapsed_time": "0:26:45", "remaining_time": "2:26:01", "throughput": "495.05", "total_tokens": 794656} | |
| {"current_steps": 121, "total_steps": 775, "loss": 0.0683, "learning_rate": 4.7447300805506455e-06, "epoch": 0.7787610619469026, "percentage": 15.61, "elapsed_time": "0:26:58", "remaining_time": "2:25:47", "throughput": "495.04", "total_tokens": 801184} | |
| {"current_steps": 122, "total_steps": 775, "loss": 0.1117, "learning_rate": 4.7401916273031e-06, "epoch": 0.7851971037811746, "percentage": 15.74, "elapsed_time": "0:27:11", "remaining_time": "2:25:33", "throughput": "495.21", "total_tokens": 808000} | |
| {"current_steps": 123, "total_steps": 775, "loss": 0.1115, "learning_rate": 4.7356153941007145e-06, "epoch": 0.7916331456154465, "percentage": 15.87, "elapsed_time": "0:27:24", "remaining_time": "2:25:18", "throughput": "495.26", "total_tokens": 814608} | |
| {"current_steps": 124, "total_steps": 775, "loss": 0.0945, "learning_rate": 4.73100145811987e-06, "epoch": 0.7980691874497184, "percentage": 16.0, "elapsed_time": "0:27:38", "remaining_time": "2:25:04", "throughput": "495.22", "total_tokens": 821072} | |
| {"current_steps": 125, "total_steps": 775, "loss": 0.1125, "learning_rate": 4.726349897172791e-06, "epoch": 0.8045052292839904, "percentage": 16.13, "elapsed_time": "0:27:51", "remaining_time": "2:24:50", "throughput": "495.35", "total_tokens": 827840} | |
| {"current_steps": 126, "total_steps": 775, "loss": 0.15, "learning_rate": 4.721660789706232e-06, "epoch": 0.8109412711182623, "percentage": 16.26, "elapsed_time": "0:28:04", "remaining_time": "2:24:36", "throughput": "495.64", "total_tokens": 834880} | |
| {"current_steps": 127, "total_steps": 775, "loss": 0.1015, "learning_rate": 4.716934214800155e-06, "epoch": 0.8173773129525342, "percentage": 16.39, "elapsed_time": "0:28:17", "remaining_time": "2:24:21", "throughput": "495.61", "total_tokens": 841360} | |
| {"current_steps": 128, "total_steps": 775, "loss": 0.1271, "learning_rate": 4.712170252166395e-06, "epoch": 0.8238133547868061, "percentage": 16.52, "elapsed_time": "0:28:30", "remaining_time": "2:24:07", "throughput": "495.60", "total_tokens": 847888} | |
| {"current_steps": 129, "total_steps": 775, "loss": 0.0762, "learning_rate": 4.707368982147318e-06, "epoch": 0.830249396621078, "percentage": 16.65, "elapsed_time": "0:28:44", "remaining_time": "2:23:53", "throughput": "495.87", "total_tokens": 854896} | |
| {"current_steps": 130, "total_steps": 775, "loss": 0.1196, "learning_rate": 4.702530485714462e-06, "epoch": 0.8366854384553499, "percentage": 16.77, "elapsed_time": "0:28:57", "remaining_time": "2:23:39", "throughput": "495.96", "total_tokens": 861600} | |
| {"current_steps": 131, "total_steps": 775, "loss": 0.0866, "learning_rate": 4.697654844467175e-06, "epoch": 0.8431214802896219, "percentage": 16.9, "elapsed_time": "0:29:10", "remaining_time": "2:23:25", "throughput": "496.03", "total_tokens": 868272} | |
| {"current_steps": 132, "total_steps": 775, "loss": 0.0718, "learning_rate": 4.69274214063124e-06, "epoch": 0.8495575221238938, "percentage": 17.03, "elapsed_time": "0:29:23", "remaining_time": "2:23:11", "throughput": "496.26", "total_tokens": 875232} | |
| {"current_steps": 133, "total_steps": 775, "loss": 0.0808, "learning_rate": 4.687792457057482e-06, "epoch": 0.8559935639581657, "percentage": 17.16, "elapsed_time": "0:29:36", "remaining_time": "2:22:57", "throughput": "496.44", "total_tokens": 882112} | |
| {"current_steps": 134, "total_steps": 775, "loss": 0.1069, "learning_rate": 4.682805877220378e-06, "epoch": 0.8624296057924377, "percentage": 17.29, "elapsed_time": "0:29:50", "remaining_time": "2:22:42", "throughput": "496.55", "total_tokens": 888848} | |
| {"current_steps": 135, "total_steps": 775, "loss": 0.0804, "learning_rate": 4.677782485216644e-06, "epoch": 0.8688656476267096, "percentage": 17.42, "elapsed_time": "0:30:03", "remaining_time": "2:22:28", "throughput": "496.40", "total_tokens": 895136} | |
| {"current_steps": 136, "total_steps": 775, "loss": 0.068, "learning_rate": 4.672722365763821e-06, "epoch": 0.8753016894609815, "percentage": 17.55, "elapsed_time": "0:30:16", "remaining_time": "2:22:14", "throughput": "496.32", "total_tokens": 901552} | |
| {"current_steps": 137, "total_steps": 775, "loss": 0.1193, "learning_rate": 4.667625604198842e-06, "epoch": 0.8817377312952535, "percentage": 17.68, "elapsed_time": "0:30:29", "remaining_time": "2:22:00", "throughput": "496.41", "total_tokens": 908272} | |
| {"current_steps": 138, "total_steps": 775, "loss": 0.1535, "learning_rate": 4.662492286476595e-06, "epoch": 0.8881737731295254, "percentage": 17.81, "elapsed_time": "0:30:42", "remaining_time": "2:21:46", "throughput": "496.47", "total_tokens": 914928} | |
| {"current_steps": 139, "total_steps": 775, "loss": 0.1303, "learning_rate": 4.657322499168475e-06, "epoch": 0.8946098149637972, "percentage": 17.94, "elapsed_time": "0:30:56", "remaining_time": "2:21:32", "throughput": "496.37", "total_tokens": 921296} | |
| {"current_steps": 140, "total_steps": 775, "loss": 0.0879, "learning_rate": 4.65211632946092e-06, "epoch": 0.9010458567980691, "percentage": 18.06, "elapsed_time": "0:31:09", "remaining_time": "2:21:18", "throughput": "496.30", "total_tokens": 927728} | |
| {"current_steps": 141, "total_steps": 775, "loss": 0.0986, "learning_rate": 4.646873865153945e-06, "epoch": 0.9074818986323411, "percentage": 18.19, "elapsed_time": "0:31:22", "remaining_time": "2:21:04", "throughput": "496.28", "total_tokens": 934240} | |
| {"current_steps": 142, "total_steps": 775, "loss": 0.1219, "learning_rate": 4.641595194659657e-06, "epoch": 0.913917940466613, "percentage": 18.32, "elapsed_time": "0:31:35", "remaining_time": "2:20:50", "throughput": "496.30", "total_tokens": 940832} | |
| {"current_steps": 143, "total_steps": 775, "loss": 0.1303, "learning_rate": 4.63628040700077e-06, "epoch": 0.9203539823008849, "percentage": 18.45, "elapsed_time": "0:31:48", "remaining_time": "2:20:36", "throughput": "496.54", "total_tokens": 947856} | |
| {"current_steps": 144, "total_steps": 775, "loss": 0.081, "learning_rate": 4.630929591809095e-06, "epoch": 0.9267900241351569, "percentage": 18.58, "elapsed_time": "0:32:02", "remaining_time": "2:20:22", "throughput": "496.41", "total_tokens": 954160} | |
| {"current_steps": 145, "total_steps": 775, "loss": 0.0979, "learning_rate": 4.625542839324036e-06, "epoch": 0.9332260659694288, "percentage": 18.71, "elapsed_time": "0:32:15", "remaining_time": "2:20:08", "throughput": "496.47", "total_tokens": 960848} | |
| {"current_steps": 146, "total_steps": 775, "loss": 0.1446, "learning_rate": 4.620120240391065e-06, "epoch": 0.9396621078037007, "percentage": 18.84, "elapsed_time": "0:32:28", "remaining_time": "2:19:54", "throughput": "496.49", "total_tokens": 967440} | |
| {"current_steps": 147, "total_steps": 775, "loss": 0.1267, "learning_rate": 4.614661886460191e-06, "epoch": 0.9460981496379727, "percentage": 18.97, "elapsed_time": "0:32:41", "remaining_time": "2:19:40", "throughput": "496.40", "total_tokens": 973808} | |
| {"current_steps": 148, "total_steps": 775, "loss": 0.1095, "learning_rate": 4.609167869584416e-06, "epoch": 0.9525341914722446, "percentage": 19.1, "elapsed_time": "0:32:54", "remaining_time": "2:19:26", "throughput": "496.39", "total_tokens": 980352} | |
| {"current_steps": 149, "total_steps": 775, "loss": 0.1252, "learning_rate": 4.6036382824181836e-06, "epoch": 0.9589702333065165, "percentage": 19.23, "elapsed_time": "0:33:08", "remaining_time": "2:19:13", "throughput": "496.48", "total_tokens": 987088} | |
| {"current_steps": 150, "total_steps": 775, "loss": 0.0645, "learning_rate": 4.598073218215817e-06, "epoch": 0.9654062751407884, "percentage": 19.35, "elapsed_time": "0:33:21", "remaining_time": "2:18:59", "throughput": "496.48", "total_tokens": 993648} | |
| {"current_steps": 151, "total_steps": 775, "loss": 0.0974, "learning_rate": 4.592472770829945e-06, "epoch": 0.9718423169750603, "percentage": 19.48, "elapsed_time": "0:33:34", "remaining_time": "2:18:45", "throughput": "496.51", "total_tokens": 1000272} | |
| {"current_steps": 152, "total_steps": 775, "loss": 0.0833, "learning_rate": 4.586837034709921e-06, "epoch": 0.9782783588093322, "percentage": 19.61, "elapsed_time": "0:33:47", "remaining_time": "2:18:31", "throughput": "496.55", "total_tokens": 1006912} | |
| {"current_steps": 153, "total_steps": 775, "loss": 0.0787, "learning_rate": 4.581166104900228e-06, "epoch": 0.9847144006436042, "percentage": 19.74, "elapsed_time": "0:34:01", "remaining_time": "2:18:17", "throughput": "496.48", "total_tokens": 1013328} | |
| {"current_steps": 154, "total_steps": 775, "loss": 0.0814, "learning_rate": 4.575460077038877e-06, "epoch": 0.9911504424778761, "percentage": 19.87, "elapsed_time": "0:34:14", "remaining_time": "2:18:03", "throughput": "496.59", "total_tokens": 1020128} | |
| {"current_steps": 155, "total_steps": 775, "loss": 0.0278, "learning_rate": 4.569719047355795e-06, "epoch": 0.997586484312148, "percentage": 20.0, "elapsed_time": "0:34:27", "remaining_time": "2:17:49", "throughput": "496.67", "total_tokens": 1026848} | |
| {"current_steps": 156, "total_steps": 775, "loss": 0.093, "learning_rate": 4.5639431126712e-06, "epoch": 1.00402252614642, "percentage": 20.13, "elapsed_time": "0:34:40", "remaining_time": "2:17:36", "throughput": "496.82", "total_tokens": 1033728} | |
| {"current_steps": 157, "total_steps": 775, "loss": 0.073, "learning_rate": 4.5581323703939685e-06, "epoch": 1.010458567980692, "percentage": 20.26, "elapsed_time": "0:34:53", "remaining_time": "2:17:22", "throughput": "497.03", "total_tokens": 1040720} | |
| {"current_steps": 158, "total_steps": 775, "loss": 0.0625, "learning_rate": 4.552286918519996e-06, "epoch": 1.0168946098149638, "percentage": 20.39, "elapsed_time": "0:35:07", "remaining_time": "2:17:08", "throughput": "496.97", "total_tokens": 1047168} | |
| {"current_steps": 159, "total_steps": 775, "loss": 0.0461, "learning_rate": 4.5464068556305375e-06, "epoch": 1.0233306516492358, "percentage": 20.52, "elapsed_time": "0:35:20", "remaining_time": "2:16:54", "throughput": "496.99", "total_tokens": 1053760} | |
| {"current_steps": 160, "total_steps": 775, "loss": 0.0318, "learning_rate": 4.540492280890555e-06, "epoch": 1.0297666934835077, "percentage": 20.65, "elapsed_time": "0:35:33", "remaining_time": "2:16:40", "throughput": "496.91", "total_tokens": 1060176} | |
| {"current_steps": 161, "total_steps": 775, "loss": 0.0068, "learning_rate": 4.534543294047033e-06, "epoch": 1.0362027353177796, "percentage": 20.77, "elapsed_time": "0:35:46", "remaining_time": "2:16:26", "throughput": "496.93", "total_tokens": 1066768} | |
| {"current_steps": 162, "total_steps": 775, "loss": 0.0462, "learning_rate": 4.528559995427309e-06, "epoch": 1.0426387771520516, "percentage": 20.9, "elapsed_time": "0:35:59", "remaining_time": "2:16:13", "throughput": "496.94", "total_tokens": 1073376} | |
| {"current_steps": 163, "total_steps": 775, "loss": 0.0487, "learning_rate": 4.522542485937369e-06, "epoch": 1.0490748189863235, "percentage": 21.03, "elapsed_time": "0:36:13", "remaining_time": "2:15:59", "throughput": "496.95", "total_tokens": 1079952} | |
| {"current_steps": 164, "total_steps": 775, "loss": 0.0664, "learning_rate": 4.516490867060156e-06, "epoch": 1.0555108608205954, "percentage": 21.16, "elapsed_time": "0:36:26", "remaining_time": "2:15:45", "throughput": "497.10", "total_tokens": 1086848} | |
| {"current_steps": 165, "total_steps": 775, "loss": 0.0347, "learning_rate": 4.5104052408538545e-06, "epoch": 1.0619469026548674, "percentage": 21.29, "elapsed_time": "0:36:39", "remaining_time": "2:15:31", "throughput": "497.06", "total_tokens": 1093328} | |
| {"current_steps": 166, "total_steps": 775, "loss": 0.0202, "learning_rate": 4.504285709950167e-06, "epoch": 1.068382944489139, "percentage": 21.42, "elapsed_time": "0:36:52", "remaining_time": "2:15:17", "throughput": "497.04", "total_tokens": 1099840} | |
| {"current_steps": 167, "total_steps": 775, "loss": 0.0573, "learning_rate": 4.498132377552587e-06, "epoch": 1.074818986323411, "percentage": 21.55, "elapsed_time": "0:37:06", "remaining_time": "2:15:04", "throughput": "497.09", "total_tokens": 1106528} | |
| {"current_steps": 168, "total_steps": 775, "loss": 0.0848, "learning_rate": 4.491945347434656e-06, "epoch": 1.081255028157683, "percentage": 21.68, "elapsed_time": "0:37:19", "remaining_time": "2:14:50", "throughput": "497.24", "total_tokens": 1113424} | |
| {"current_steps": 169, "total_steps": 775, "loss": 0.0464, "learning_rate": 4.485724723938215e-06, "epoch": 1.0876910699919549, "percentage": 21.81, "elapsed_time": "0:37:32", "remaining_time": "2:14:36", "throughput": "497.27", "total_tokens": 1120064} | |
| {"current_steps": 170, "total_steps": 775, "loss": 0.1004, "learning_rate": 4.479470611971646e-06, "epoch": 1.0941271118262268, "percentage": 21.94, "elapsed_time": "0:37:45", "remaining_time": "2:14:23", "throughput": "497.41", "total_tokens": 1126960} | |
| {"current_steps": 171, "total_steps": 775, "loss": 0.0842, "learning_rate": 4.473183117008096e-06, "epoch": 1.1005631536604987, "percentage": 22.06, "elapsed_time": "0:37:58", "remaining_time": "2:14:09", "throughput": "497.47", "total_tokens": 1133664} | |
| {"current_steps": 172, "total_steps": 775, "loss": 0.0452, "learning_rate": 4.4668623450837085e-06, "epoch": 1.1069991954947707, "percentage": 22.19, "elapsed_time": "0:38:12", "remaining_time": "2:13:55", "throughput": "497.39", "total_tokens": 1140048} | |
| {"current_steps": 173, "total_steps": 775, "loss": 0.0225, "learning_rate": 4.460508402795827e-06, "epoch": 1.1134352373290426, "percentage": 22.32, "elapsed_time": "0:38:25", "remaining_time": "2:13:41", "throughput": "497.32", "total_tokens": 1146448} | |
| {"current_steps": 174, "total_steps": 775, "loss": 0.0058, "learning_rate": 4.4541213973012005e-06, "epoch": 1.1198712791633145, "percentage": 22.45, "elapsed_time": "0:38:38", "remaining_time": "2:13:28", "throughput": "497.29", "total_tokens": 1152960} | |
| {"current_steps": 175, "total_steps": 775, "loss": 0.0565, "learning_rate": 4.447701436314176e-06, "epoch": 1.1263073209975865, "percentage": 22.58, "elapsed_time": "0:38:51", "remaining_time": "2:13:14", "throughput": "497.34", "total_tokens": 1159632} | |
| {"current_steps": 176, "total_steps": 775, "loss": 0.0591, "learning_rate": 4.441248628104884e-06, "epoch": 1.1327433628318584, "percentage": 22.71, "elapsed_time": "0:39:04", "remaining_time": "2:13:00", "throughput": "497.53", "total_tokens": 1166640} | |
| {"current_steps": 177, "total_steps": 775, "loss": 0.0488, "learning_rate": 4.434763081497407e-06, "epoch": 1.1391794046661303, "percentage": 22.84, "elapsed_time": "0:39:18", "remaining_time": "2:12:46", "throughput": "497.61", "total_tokens": 1173408} | |
| {"current_steps": 178, "total_steps": 775, "loss": 0.0318, "learning_rate": 4.428244905867952e-06, "epoch": 1.1456154465004023, "percentage": 22.97, "elapsed_time": "0:39:31", "remaining_time": "2:12:33", "throughput": "497.52", "total_tokens": 1179776} | |
| {"current_steps": 179, "total_steps": 775, "loss": 0.064, "learning_rate": 4.421694211142998e-06, "epoch": 1.1520514883346742, "percentage": 23.1, "elapsed_time": "0:39:44", "remaining_time": "2:12:19", "throughput": "497.68", "total_tokens": 1186720} | |
| {"current_steps": 180, "total_steps": 775, "loss": 0.0465, "learning_rate": 4.415111107797445e-06, "epoch": 1.1584875301689461, "percentage": 23.23, "elapsed_time": "0:39:57", "remaining_time": "2:12:05", "throughput": "497.73", "total_tokens": 1193408} | |
| {"current_steps": 181, "total_steps": 775, "loss": 0.0344, "learning_rate": 4.408495706852758e-06, "epoch": 1.164923572003218, "percentage": 23.35, "elapsed_time": "0:40:10", "remaining_time": "2:11:52", "throughput": "497.78", "total_tokens": 1200096} | |
| {"current_steps": 182, "total_steps": 775, "loss": 0.0747, "learning_rate": 4.401848119875081e-06, "epoch": 1.17135961383749, "percentage": 23.48, "elapsed_time": "0:40:24", "remaining_time": "2:11:38", "throughput": "497.86", "total_tokens": 1206848} | |
| {"current_steps": 183, "total_steps": 775, "loss": 0.073, "learning_rate": 4.395168458973368e-06, "epoch": 1.177795655671762, "percentage": 23.61, "elapsed_time": "0:40:37", "remaining_time": "2:11:24", "throughput": "497.95", "total_tokens": 1213632} | |
| {"current_steps": 184, "total_steps": 775, "loss": 0.0652, "learning_rate": 4.388456836797484e-06, "epoch": 1.1842316975060339, "percentage": 23.74, "elapsed_time": "0:40:50", "remaining_time": "2:11:10", "throughput": "498.00", "total_tokens": 1220336} | |
| {"current_steps": 185, "total_steps": 775, "loss": 0.0881, "learning_rate": 4.381713366536312e-06, "epoch": 1.1906677393403058, "percentage": 23.87, "elapsed_time": "0:41:03", "remaining_time": "2:10:57", "throughput": "497.93", "total_tokens": 1226736} | |
| {"current_steps": 186, "total_steps": 775, "loss": 0.0676, "learning_rate": 4.374938161915835e-06, "epoch": 1.1971037811745777, "percentage": 24.0, "elapsed_time": "0:41:16", "remaining_time": "2:10:43", "throughput": "498.02", "total_tokens": 1233536} | |
| {"current_steps": 187, "total_steps": 775, "loss": 0.0481, "learning_rate": 4.368131337197228e-06, "epoch": 1.2035398230088497, "percentage": 24.13, "elapsed_time": "0:41:30", "remaining_time": "2:10:29", "throughput": "497.99", "total_tokens": 1240032} | |
| {"current_steps": 188, "total_steps": 775, "loss": 0.0477, "learning_rate": 4.361293007174926e-06, "epoch": 1.2099758648431216, "percentage": 24.26, "elapsed_time": "0:41:43", "remaining_time": "2:10:16", "throughput": "498.15", "total_tokens": 1247008} | |
| {"current_steps": 189, "total_steps": 775, "loss": 0.0456, "learning_rate": 4.354423287174686e-06, "epoch": 1.2164119066773935, "percentage": 24.39, "elapsed_time": "0:41:56", "remaining_time": "2:10:02", "throughput": "498.32", "total_tokens": 1254032} | |
| {"current_steps": 190, "total_steps": 775, "loss": 0.133, "learning_rate": 4.3475222930516484e-06, "epoch": 1.2228479485116655, "percentage": 24.52, "elapsed_time": "0:42:09", "remaining_time": "2:09:48", "throughput": "498.52", "total_tokens": 1261104} | |
| {"current_steps": 191, "total_steps": 775, "loss": 0.0672, "learning_rate": 4.340590141188377e-06, "epoch": 1.2292839903459372, "percentage": 24.65, "elapsed_time": "0:42:22", "remaining_time": "2:09:35", "throughput": "498.51", "total_tokens": 1267680} | |
| {"current_steps": 192, "total_steps": 775, "loss": 0.0352, "learning_rate": 4.333626948492898e-06, "epoch": 1.235720032180209, "percentage": 24.77, "elapsed_time": "0:42:36", "remaining_time": "2:09:21", "throughput": "498.46", "total_tokens": 1274112} | |
| {"current_steps": 193, "total_steps": 775, "loss": 0.0361, "learning_rate": 4.326632832396733e-06, "epoch": 1.242156074014481, "percentage": 24.9, "elapsed_time": "0:42:49", "remaining_time": "2:09:07", "throughput": "498.39", "total_tokens": 1280528} | |
| {"current_steps": 194, "total_steps": 775, "loss": 0.0432, "learning_rate": 4.319607910852911e-06, "epoch": 1.248592115848753, "percentage": 25.03, "elapsed_time": "0:43:02", "remaining_time": "2:08:54", "throughput": "498.44", "total_tokens": 1287232} | |
| {"current_steps": 195, "total_steps": 775, "loss": 0.0398, "learning_rate": 4.3125523023339825e-06, "epoch": 1.255028157683025, "percentage": 25.16, "elapsed_time": "0:43:15", "remaining_time": "2:08:40", "throughput": "498.43", "total_tokens": 1293792} | |
| {"current_steps": 196, "total_steps": 775, "loss": 0.0473, "learning_rate": 4.305466125830023e-06, "epoch": 1.2614641995172968, "percentage": 25.29, "elapsed_time": "0:43:28", "remaining_time": "2:08:26", "throughput": "498.53", "total_tokens": 1300624} | |
| {"current_steps": 197, "total_steps": 775, "loss": 0.0394, "learning_rate": 4.2983495008466285e-06, "epoch": 1.2679002413515688, "percentage": 25.42, "elapsed_time": "0:43:42", "remaining_time": "2:08:13", "throughput": "498.65", "total_tokens": 1307520} | |
| {"current_steps": 198, "total_steps": 775, "loss": 0.0645, "learning_rate": 4.29120254740289e-06, "epoch": 1.2743362831858407, "percentage": 25.55, "elapsed_time": "0:43:55", "remaining_time": "2:07:59", "throughput": "498.55", "total_tokens": 1313824} | |
| {"current_steps": 199, "total_steps": 775, "loss": 0.0663, "learning_rate": 4.284025386029381e-06, "epoch": 1.2807723250201126, "percentage": 25.68, "elapsed_time": "0:44:08", "remaining_time": "2:07:46", "throughput": "498.37", "total_tokens": 1319952} | |
| {"current_steps": 200, "total_steps": 775, "loss": 0.1067, "learning_rate": 4.276818137766118e-06, "epoch": 1.2872083668543846, "percentage": 25.81, "elapsed_time": "0:44:21", "remaining_time": "2:07:32", "throughput": "498.30", "total_tokens": 1326352} | |
| {"current_steps": 201, "total_steps": 775, "loss": 0.0603, "learning_rate": 4.269580924160523e-06, "epoch": 1.2936444086886565, "percentage": 25.94, "elapsed_time": "0:44:34", "remaining_time": "2:07:18", "throughput": "498.30", "total_tokens": 1332912} | |
| {"current_steps": 202, "total_steps": 775, "loss": 0.0055, "learning_rate": 4.262313867265369e-06, "epoch": 1.3000804505229284, "percentage": 26.06, "elapsed_time": "0:44:48", "remaining_time": "2:07:05", "throughput": "498.44", "total_tokens": 1339872} | |
| {"current_steps": 203, "total_steps": 775, "loss": 0.0547, "learning_rate": 4.255017089636725e-06, "epoch": 1.3065164923572004, "percentage": 26.19, "elapsed_time": "0:45:01", "remaining_time": "2:06:51", "throughput": "498.36", "total_tokens": 1346240} | |
| {"current_steps": 204, "total_steps": 775, "loss": 0.0684, "learning_rate": 4.24769071433189e-06, "epoch": 1.3129525341914723, "percentage": 26.32, "elapsed_time": "0:45:14", "remaining_time": "2:06:37", "throughput": "498.47", "total_tokens": 1353104} | |
| {"current_steps": 205, "total_steps": 775, "loss": 0.0484, "learning_rate": 4.240334864907317e-06, "epoch": 1.3193885760257442, "percentage": 26.45, "elapsed_time": "0:45:27", "remaining_time": "2:06:24", "throughput": "498.46", "total_tokens": 1359664} | |
| {"current_steps": 206, "total_steps": 775, "loss": 0.0225, "learning_rate": 4.232949665416526e-06, "epoch": 1.3258246178600162, "percentage": 26.58, "elapsed_time": "0:45:40", "remaining_time": "2:06:10", "throughput": "498.41", "total_tokens": 1366112} | |
| {"current_steps": 207, "total_steps": 775, "loss": 0.0278, "learning_rate": 4.225535240408014e-06, "epoch": 1.332260659694288, "percentage": 26.71, "elapsed_time": "0:45:54", "remaining_time": "2:05:57", "throughput": "498.49", "total_tokens": 1372912} | |
| {"current_steps": 208, "total_steps": 775, "loss": 0.034, "learning_rate": 4.218091714923157e-06, "epoch": 1.33869670152856, "percentage": 26.84, "elapsed_time": "0:46:07", "remaining_time": "2:05:43", "throughput": "498.38", "total_tokens": 1379200} | |
| {"current_steps": 209, "total_steps": 775, "loss": 0.0453, "learning_rate": 4.210619214494099e-06, "epoch": 1.3451327433628317, "percentage": 26.97, "elapsed_time": "0:46:20", "remaining_time": "2:05:30", "throughput": "498.29", "total_tokens": 1385520} | |
| {"current_steps": 210, "total_steps": 775, "loss": 0.0564, "learning_rate": 4.203117865141635e-06, "epoch": 1.3515687851971037, "percentage": 27.1, "elapsed_time": "0:46:33", "remaining_time": "2:05:16", "throughput": "498.24", "total_tokens": 1391968} | |
| {"current_steps": 211, "total_steps": 775, "loss": 0.0318, "learning_rate": 4.195587793373085e-06, "epoch": 1.3580048270313756, "percentage": 27.23, "elapsed_time": "0:46:46", "remaining_time": "2:05:02", "throughput": "498.25", "total_tokens": 1398576} | |
| {"current_steps": 212, "total_steps": 775, "loss": 0.0575, "learning_rate": 4.188029126180161e-06, "epoch": 1.3644408688656475, "percentage": 27.35, "elapsed_time": "0:47:00", "remaining_time": "2:04:49", "throughput": "498.30", "total_tokens": 1405280} | |
| {"current_steps": 213, "total_steps": 775, "loss": 0.0448, "learning_rate": 4.180441991036827e-06, "epoch": 1.3708769106999195, "percentage": 27.48, "elapsed_time": "0:47:13", "remaining_time": "2:04:35", "throughput": "498.33", "total_tokens": 1411968} | |
| {"current_steps": 214, "total_steps": 775, "loss": 0.0493, "learning_rate": 4.172826515897146e-06, "epoch": 1.3773129525341914, "percentage": 27.61, "elapsed_time": "0:47:26", "remaining_time": "2:04:22", "throughput": "498.34", "total_tokens": 1418576} | |
| {"current_steps": 215, "total_steps": 775, "loss": 0.0478, "learning_rate": 4.165182829193126e-06, "epoch": 1.3837489943684633, "percentage": 27.74, "elapsed_time": "0:47:39", "remaining_time": "2:04:08", "throughput": "498.42", "total_tokens": 1425360} | |
| {"current_steps": 216, "total_steps": 775, "loss": 0.0887, "learning_rate": 4.15751105983255e-06, "epoch": 1.3901850362027353, "percentage": 27.87, "elapsed_time": "0:47:52", "remaining_time": "2:03:55", "throughput": "498.48", "total_tokens": 1432144} | |
| {"current_steps": 217, "total_steps": 775, "loss": 0.0634, "learning_rate": 4.149811337196808e-06, "epoch": 1.3966210780370072, "percentage": 28.0, "elapsed_time": "0:48:06", "remaining_time": "2:03:41", "throughput": "498.65", "total_tokens": 1439200} | |
| {"current_steps": 218, "total_steps": 775, "loss": 0.0349, "learning_rate": 4.142083791138703e-06, "epoch": 1.4030571198712791, "percentage": 28.13, "elapsed_time": "0:48:19", "remaining_time": "2:03:28", "throughput": "498.63", "total_tokens": 1445728} | |
| {"current_steps": 219, "total_steps": 775, "loss": 0.0459, "learning_rate": 4.134328551980279e-06, "epoch": 1.409493161705551, "percentage": 28.26, "elapsed_time": "0:48:32", "remaining_time": "2:03:14", "throughput": "498.66", "total_tokens": 1452384} | |
| {"current_steps": 220, "total_steps": 775, "loss": 0.0304, "learning_rate": 4.126545750510605e-06, "epoch": 1.415929203539823, "percentage": 28.39, "elapsed_time": "0:48:45", "remaining_time": "2:03:01", "throughput": "498.71", "total_tokens": 1459136} | |
| {"current_steps": 221, "total_steps": 775, "loss": 0.0658, "learning_rate": 4.118735517983584e-06, "epoch": 1.422365245374095, "percentage": 28.52, "elapsed_time": "0:48:59", "remaining_time": "2:02:47", "throughput": "498.68", "total_tokens": 1465632} | |
| {"current_steps": 222, "total_steps": 775, "loss": 0.087, "learning_rate": 4.110897986115729e-06, "epoch": 1.4288012872083669, "percentage": 28.65, "elapsed_time": "0:49:12", "remaining_time": "2:02:33", "throughput": "498.81", "total_tokens": 1472592} | |
| {"current_steps": 223, "total_steps": 775, "loss": 0.0952, "learning_rate": 4.1030332870839466e-06, "epoch": 1.4352373290426388, "percentage": 28.77, "elapsed_time": "0:49:25", "remaining_time": "2:02:20", "throughput": "498.80", "total_tokens": 1479168} | |
| {"current_steps": 224, "total_steps": 775, "loss": 0.0358, "learning_rate": 4.0951415535233065e-06, "epoch": 1.4416733708769107, "percentage": 28.9, "elapsed_time": "0:49:38", "remaining_time": "2:02:06", "throughput": "498.77", "total_tokens": 1485664} | |
| {"current_steps": 225, "total_steps": 775, "loss": 0.0527, "learning_rate": 4.087222918524807e-06, "epoch": 1.4481094127111827, "percentage": 29.03, "elapsed_time": "0:49:51", "remaining_time": "2:01:53", "throughput": "498.81", "total_tokens": 1492368} | |
| {"current_steps": 226, "total_steps": 775, "loss": 0.0452, "learning_rate": 4.079277515633127e-06, "epoch": 1.4545454545454546, "percentage": 29.16, "elapsed_time": "0:50:05", "remaining_time": "2:01:39", "throughput": "498.74", "total_tokens": 1498752} | |
| {"current_steps": 227, "total_steps": 775, "loss": 0.0313, "learning_rate": 4.0713054788443776e-06, "epoch": 1.4609814963797265, "percentage": 29.29, "elapsed_time": "0:50:18", "remaining_time": "2:01:26", "throughput": "498.73", "total_tokens": 1505296} | |
| {"current_steps": 228, "total_steps": 775, "loss": 0.0544, "learning_rate": 4.063306942603835e-06, "epoch": 1.4674175382139985, "percentage": 29.42, "elapsed_time": "0:50:31", "remaining_time": "2:01:12", "throughput": "498.63", "total_tokens": 1511584} | |
| {"current_steps": 229, "total_steps": 775, "loss": 0.0428, "learning_rate": 4.0552820418036855e-06, "epoch": 1.4738535800482704, "percentage": 29.55, "elapsed_time": "0:50:44", "remaining_time": "2:00:59", "throughput": "498.50", "total_tokens": 1517776} | |
| {"current_steps": 230, "total_steps": 775, "loss": 0.0183, "learning_rate": 4.0472309117807365e-06, "epoch": 1.4802896218825423, "percentage": 29.68, "elapsed_time": "0:50:57", "remaining_time": "2:00:45", "throughput": "498.52", "total_tokens": 1524416} | |
| {"current_steps": 231, "total_steps": 775, "loss": 0.0662, "learning_rate": 4.039153688314146e-06, "epoch": 1.4867256637168142, "percentage": 29.81, "elapsed_time": "0:51:11", "remaining_time": "2:00:32", "throughput": "498.47", "total_tokens": 1530864} | |
| {"current_steps": 232, "total_steps": 775, "loss": 0.0258, "learning_rate": 4.031050507623125e-06, "epoch": 1.4931617055510862, "percentage": 29.94, "elapsed_time": "0:51:24", "remaining_time": "2:00:18", "throughput": "498.40", "total_tokens": 1537216} | |
| {"current_steps": 233, "total_steps": 775, "loss": 0.0584, "learning_rate": 4.022921506364644e-06, "epoch": 1.4995977473853581, "percentage": 30.06, "elapsed_time": "0:51:37", "remaining_time": "2:00:05", "throughput": "498.41", "total_tokens": 1543824} | |
| {"current_steps": 234, "total_steps": 775, "loss": 0.0916, "learning_rate": 4.014766821631128e-06, "epoch": 1.50603378921963, "percentage": 30.19, "elapsed_time": "0:51:50", "remaining_time": "1:59:51", "throughput": "498.42", "total_tokens": 1550432} | |
| {"current_steps": 235, "total_steps": 775, "loss": 0.0397, "learning_rate": 4.006586590948141e-06, "epoch": 1.512469831053902, "percentage": 30.32, "elapsed_time": "0:52:03", "remaining_time": "1:59:38", "throughput": "498.38", "total_tokens": 1556912} | |
| {"current_steps": 236, "total_steps": 775, "loss": 0.0405, "learning_rate": 3.998380952272073e-06, "epoch": 1.518905872888174, "percentage": 30.45, "elapsed_time": "0:52:17", "remaining_time": "1:59:24", "throughput": "498.37", "total_tokens": 1563456} | |
| {"current_steps": 237, "total_steps": 775, "loss": 0.0645, "learning_rate": 3.990150043987806e-06, "epoch": 1.5253419147224458, "percentage": 30.58, "elapsed_time": "0:52:30", "remaining_time": "1:59:11", "throughput": "498.43", "total_tokens": 1570240} | |
| {"current_steps": 238, "total_steps": 775, "loss": 0.0389, "learning_rate": 3.981894004906388e-06, "epoch": 1.5317779565567178, "percentage": 30.71, "elapsed_time": "0:52:43", "remaining_time": "1:58:57", "throughput": "498.46", "total_tokens": 1576896} | |
| {"current_steps": 239, "total_steps": 775, "loss": 0.0341, "learning_rate": 3.973612974262685e-06, "epoch": 1.5382139983909895, "percentage": 30.84, "elapsed_time": "0:52:56", "remaining_time": "1:58:44", "throughput": "498.45", "total_tokens": 1583440} | |
| {"current_steps": 240, "total_steps": 775, "loss": 0.0625, "learning_rate": 3.965307091713037e-06, "epoch": 1.5446500402252614, "percentage": 30.97, "elapsed_time": "0:53:09", "remaining_time": "1:58:30", "throughput": "498.43", "total_tokens": 1589968} | |
| {"current_steps": 241, "total_steps": 775, "loss": 0.0651, "learning_rate": 3.956976497332903e-06, "epoch": 1.5510860820595334, "percentage": 31.1, "elapsed_time": "0:53:23", "remaining_time": "1:58:17", "throughput": "498.39", "total_tokens": 1596416} | |
| {"current_steps": 242, "total_steps": 775, "loss": 0.0391, "learning_rate": 3.948621331614495e-06, "epoch": 1.5575221238938053, "percentage": 31.23, "elapsed_time": "0:53:36", "remaining_time": "1:58:03", "throughput": "498.37", "total_tokens": 1602944} | |
| {"current_steps": 243, "total_steps": 775, "loss": 0.0486, "learning_rate": 3.9402417354644115e-06, "epoch": 1.5639581657280772, "percentage": 31.35, "elapsed_time": "0:53:49", "remaining_time": "1:57:50", "throughput": "498.40", "total_tokens": 1609632} | |
| {"current_steps": 244, "total_steps": 775, "loss": 0.1192, "learning_rate": 3.9318378502012636e-06, "epoch": 1.5703942075623492, "percentage": 31.48, "elapsed_time": "0:54:02", "remaining_time": "1:57:36", "throughput": "498.37", "total_tokens": 1616096} | |
| {"current_steps": 245, "total_steps": 775, "loss": 0.0679, "learning_rate": 3.923409817553284e-06, "epoch": 1.576830249396621, "percentage": 31.61, "elapsed_time": "0:54:15", "remaining_time": "1:57:23", "throughput": "498.42", "total_tokens": 1622848} | |
| {"current_steps": 246, "total_steps": 775, "loss": 0.0493, "learning_rate": 3.914957779655946e-06, "epoch": 1.583266291230893, "percentage": 31.74, "elapsed_time": "0:54:29", "remaining_time": "1:57:10", "throughput": "498.48", "total_tokens": 1629600} | |
| {"current_steps": 247, "total_steps": 775, "loss": 0.0456, "learning_rate": 3.906481879049559e-06, "epoch": 1.589702333065165, "percentage": 31.87, "elapsed_time": "0:54:42", "remaining_time": "1:56:56", "throughput": "498.48", "total_tokens": 1636192} | |
| {"current_steps": 248, "total_steps": 775, "loss": 0.0391, "learning_rate": 3.897982258676867e-06, "epoch": 1.5961383748994369, "percentage": 32.0, "elapsed_time": "0:54:55", "remaining_time": "1:56:43", "throughput": "498.50", "total_tokens": 1642832} | |
| {"current_steps": 249, "total_steps": 775, "loss": 0.0501, "learning_rate": 3.8894590618806435e-06, "epoch": 1.6025744167337088, "percentage": 32.13, "elapsed_time": "0:55:08", "remaining_time": "1:56:29", "throughput": "498.65", "total_tokens": 1649904} | |
| {"current_steps": 250, "total_steps": 775, "loss": 0.0397, "learning_rate": 3.880912432401265e-06, "epoch": 1.6090104585679805, "percentage": 32.26, "elapsed_time": "0:55:21", "remaining_time": "1:56:16", "throughput": "498.71", "total_tokens": 1656704} | |
| {"current_steps": 251, "total_steps": 775, "loss": 0.0846, "learning_rate": 3.872342514374291e-06, "epoch": 1.6154465004022525, "percentage": 32.39, "elapsed_time": "0:55:35", "remaining_time": "1:56:02", "throughput": "498.83", "total_tokens": 1663680} | |
| {"current_steps": 252, "total_steps": 775, "loss": 0.0443, "learning_rate": 3.863749452328035e-06, "epoch": 1.6218825422365244, "percentage": 32.52, "elapsed_time": "0:55:48", "remaining_time": "1:55:49", "throughput": "498.80", "total_tokens": 1670160} | |
| {"current_steps": 253, "total_steps": 775, "loss": 0.045, "learning_rate": 3.855133391181124e-06, "epoch": 1.6283185840707963, "percentage": 32.65, "elapsed_time": "0:56:01", "remaining_time": "1:55:35", "throughput": "498.82", "total_tokens": 1676832} | |
| {"current_steps": 254, "total_steps": 775, "loss": 0.0172, "learning_rate": 3.846494476240057e-06, "epoch": 1.6347546259050683, "percentage": 32.77, "elapsed_time": "0:56:14", "remaining_time": "1:55:22", "throughput": "498.90", "total_tokens": 1683664} | |
| {"current_steps": 255, "total_steps": 775, "loss": 0.0949, "learning_rate": 3.837832853196751e-06, "epoch": 1.6411906677393402, "percentage": 32.9, "elapsed_time": "0:56:27", "remaining_time": "1:55:08", "throughput": "498.89", "total_tokens": 1690208} | |
| {"current_steps": 256, "total_steps": 775, "loss": 0.0277, "learning_rate": 3.8291486681260904e-06, "epoch": 1.6476267095736121, "percentage": 33.03, "elapsed_time": "0:56:41", "remaining_time": "1:54:55", "throughput": "499.03", "total_tokens": 1697296} | |
| {"current_steps": 257, "total_steps": 775, "loss": 0.0247, "learning_rate": 3.820442067483455e-06, "epoch": 1.654062751407884, "percentage": 33.16, "elapsed_time": "0:56:54", "remaining_time": "1:54:41", "throughput": "498.92", "total_tokens": 1703504} | |
| {"current_steps": 258, "total_steps": 775, "loss": 0.0549, "learning_rate": 3.811713198102258e-06, "epoch": 1.660498793242156, "percentage": 33.29, "elapsed_time": "0:57:07", "remaining_time": "1:54:28", "throughput": "498.89", "total_tokens": 1710016} | |
| {"current_steps": 259, "total_steps": 775, "loss": 0.0342, "learning_rate": 3.802962207191463e-06, "epoch": 1.666934835076428, "percentage": 33.42, "elapsed_time": "0:57:20", "remaining_time": "1:54:15", "throughput": "499.00", "total_tokens": 1716960} | |
| {"current_steps": 260, "total_steps": 775, "loss": 0.0617, "learning_rate": 3.794189242333107e-06, "epoch": 1.6733708769106999, "percentage": 33.55, "elapsed_time": "0:57:34", "remaining_time": "1:54:01", "throughput": "498.99", "total_tokens": 1723504} | |
| {"current_steps": 261, "total_steps": 775, "loss": 0.0675, "learning_rate": 3.785394451479806e-06, "epoch": 1.6798069187449718, "percentage": 33.68, "elapsed_time": "0:57:47", "remaining_time": "1:53:48", "throughput": "499.00", "total_tokens": 1730160} | |
| {"current_steps": 262, "total_steps": 775, "loss": 0.1055, "learning_rate": 3.7765779829522674e-06, "epoch": 1.6862429605792437, "percentage": 33.81, "elapsed_time": "0:58:00", "remaining_time": "1:53:34", "throughput": "499.01", "total_tokens": 1736752} | |
| {"current_steps": 263, "total_steps": 775, "loss": 0.0355, "learning_rate": 3.7677399854367815e-06, "epoch": 1.6926790024135157, "percentage": 33.94, "elapsed_time": "0:58:13", "remaining_time": "1:53:21", "throughput": "499.00", "total_tokens": 1743328} | |
| {"current_steps": 264, "total_steps": 775, "loss": 0.0622, "learning_rate": 3.7588806079827147e-06, "epoch": 1.6991150442477876, "percentage": 34.06, "elapsed_time": "0:58:26", "remaining_time": "1:53:07", "throughput": "498.96", "total_tokens": 1749776} | |
| {"current_steps": 265, "total_steps": 775, "loss": 0.0297, "learning_rate": 3.7500000000000005e-06, "epoch": 1.7055510860820595, "percentage": 34.19, "elapsed_time": "0:58:40", "remaining_time": "1:52:54", "throughput": "499.01", "total_tokens": 1756512} | |
| {"current_steps": 266, "total_steps": 775, "loss": 0.0312, "learning_rate": 3.7410983112566166e-06, "epoch": 1.7119871279163315, "percentage": 34.32, "elapsed_time": "0:58:53", "remaining_time": "1:52:40", "throughput": "498.96", "total_tokens": 1762928} | |
| {"current_steps": 267, "total_steps": 775, "loss": 0.0811, "learning_rate": 3.7321756918760587e-06, "epoch": 1.7184231697506034, "percentage": 34.45, "elapsed_time": "0:59:06", "remaining_time": "1:52:27", "throughput": "498.92", "total_tokens": 1769392} | |
| {"current_steps": 268, "total_steps": 775, "loss": 0.067, "learning_rate": 3.7232322923348093e-06, "epoch": 1.7248592115848753, "percentage": 34.58, "elapsed_time": "0:59:19", "remaining_time": "1:52:14", "throughput": "498.94", "total_tokens": 1776032} | |
| {"current_steps": 269, "total_steps": 775, "loss": 0.0553, "learning_rate": 3.7142682634598016e-06, "epoch": 1.7312952534191473, "percentage": 34.71, "elapsed_time": "0:59:32", "remaining_time": "1:52:00", "throughput": "498.91", "total_tokens": 1782512} | |
| {"current_steps": 270, "total_steps": 775, "loss": 0.021, "learning_rate": 3.7052837564258728e-06, "epoch": 1.7377312952534192, "percentage": 34.84, "elapsed_time": "0:59:46", "remaining_time": "1:51:47", "throughput": "498.96", "total_tokens": 1789280} | |
| {"current_steps": 271, "total_steps": 775, "loss": 0.0589, "learning_rate": 3.6962789227532165e-06, "epoch": 1.7441673370876911, "percentage": 34.97, "elapsed_time": "0:59:59", "remaining_time": "1:51:33", "throughput": "498.91", "total_tokens": 1795696} | |
| {"current_steps": 272, "total_steps": 775, "loss": 0.0521, "learning_rate": 3.6872539143048287e-06, "epoch": 1.750603378921963, "percentage": 35.1, "elapsed_time": "1:00:12", "remaining_time": "1:51:20", "throughput": "498.96", "total_tokens": 1802448} | |
| {"current_steps": 273, "total_steps": 775, "loss": 0.0402, "learning_rate": 3.6782088832839436e-06, "epoch": 1.757039420756235, "percentage": 35.23, "elapsed_time": "1:00:25", "remaining_time": "1:51:06", "throughput": "499.02", "total_tokens": 1809264} | |
| {"current_steps": 274, "total_steps": 775, "loss": 0.0363, "learning_rate": 3.6691439822314672e-06, "epoch": 1.763475462590507, "percentage": 35.35, "elapsed_time": "1:00:38", "remaining_time": "1:50:53", "throughput": "499.01", "total_tokens": 1815808} | |
| {"current_steps": 275, "total_steps": 775, "loss": 0.0523, "learning_rate": 3.660059364023409e-06, "epoch": 1.7699115044247788, "percentage": 35.48, "elapsed_time": "1:00:52", "remaining_time": "1:50:40", "throughput": "498.99", "total_tokens": 1822352} | |
| {"current_steps": 276, "total_steps": 775, "loss": 0.0255, "learning_rate": 3.650955181868298e-06, "epoch": 1.7763475462590508, "percentage": 35.61, "elapsed_time": "1:01:05", "remaining_time": "1:50:26", "throughput": "499.02", "total_tokens": 1829056} | |
| {"current_steps": 277, "total_steps": 775, "loss": 0.1031, "learning_rate": 3.641831589304602e-06, "epoch": 1.7827835880933227, "percentage": 35.74, "elapsed_time": "1:01:18", "remaining_time": "1:50:13", "throughput": "499.04", "total_tokens": 1835696} | |
| {"current_steps": 278, "total_steps": 775, "loss": 0.0452, "learning_rate": 3.6326887401981386e-06, "epoch": 1.7892196299275946, "percentage": 35.87, "elapsed_time": "1:01:31", "remaining_time": "1:49:59", "throughput": "499.04", "total_tokens": 1842288} | |
| {"current_steps": 279, "total_steps": 775, "loss": 0.0537, "learning_rate": 3.6235267887394774e-06, "epoch": 1.7956556717618666, "percentage": 36.0, "elapsed_time": "1:01:44", "remaining_time": "1:49:46", "throughput": "499.06", "total_tokens": 1848960} | |
| {"current_steps": 280, "total_steps": 775, "loss": 0.0572, "learning_rate": 3.6143458894413463e-06, "epoch": 1.8020917135961385, "percentage": 36.13, "elapsed_time": "1:01:58", "remaining_time": "1:49:33", "throughput": "499.09", "total_tokens": 1855648} | |
| {"current_steps": 281, "total_steps": 775, "loss": 0.0298, "learning_rate": 3.6051461971360146e-06, "epoch": 1.8085277554304104, "percentage": 36.26, "elapsed_time": "1:02:11", "remaining_time": "1:49:19", "throughput": "499.07", "total_tokens": 1862160} | |
| {"current_steps": 282, "total_steps": 775, "loss": 0.037, "learning_rate": 3.595927866972694e-06, "epoch": 1.8149637972646824, "percentage": 36.39, "elapsed_time": "1:02:24", "remaining_time": "1:49:06", "throughput": "499.11", "total_tokens": 1868896} | |
| {"current_steps": 283, "total_steps": 775, "loss": 0.0783, "learning_rate": 3.586691054414913e-06, "epoch": 1.8213998390989543, "percentage": 36.52, "elapsed_time": "1:02:37", "remaining_time": "1:48:52", "throughput": "499.05", "total_tokens": 1875248} | |
| {"current_steps": 284, "total_steps": 775, "loss": 0.0436, "learning_rate": 3.577435915237899e-06, "epoch": 1.827835880933226, "percentage": 36.65, "elapsed_time": "1:02:50", "remaining_time": "1:48:39", "throughput": "499.02", "total_tokens": 1881728} | |
| {"current_steps": 285, "total_steps": 775, "loss": 0.0258, "learning_rate": 3.5681626055259526e-06, "epoch": 1.834271922767498, "percentage": 36.77, "elapsed_time": "1:03:04", "remaining_time": "1:48:25", "throughput": "499.04", "total_tokens": 1888384} | |
| {"current_steps": 286, "total_steps": 775, "loss": 0.0235, "learning_rate": 3.558871281669811e-06, "epoch": 1.8407079646017699, "percentage": 36.9, "elapsed_time": "1:03:17", "remaining_time": "1:48:12", "throughput": "499.01", "total_tokens": 1894864} | |
| {"current_steps": 287, "total_steps": 775, "loss": 0.0541, "learning_rate": 3.549562100364014e-06, "epoch": 1.8471440064360418, "percentage": 37.03, "elapsed_time": "1:03:30", "remaining_time": "1:47:59", "throughput": "499.06", "total_tokens": 1901680} | |
| {"current_steps": 288, "total_steps": 775, "loss": 0.0767, "learning_rate": 3.5402352186042602e-06, "epoch": 1.8535800482703138, "percentage": 37.16, "elapsed_time": "1:03:43", "remaining_time": "1:47:45", "throughput": "499.07", "total_tokens": 1908304} | |
| {"current_steps": 289, "total_steps": 775, "loss": 0.0558, "learning_rate": 3.530890793684759e-06, "epoch": 1.8600160901045857, "percentage": 37.29, "elapsed_time": "1:03:56", "remaining_time": "1:47:32", "throughput": "499.03", "total_tokens": 1914736} | |
| {"current_steps": 290, "total_steps": 775, "loss": 0.0483, "learning_rate": 3.521528983195579e-06, "epoch": 1.8664521319388576, "percentage": 37.42, "elapsed_time": "1:04:10", "remaining_time": "1:47:18", "throughput": "498.97", "total_tokens": 1921088} | |
| {"current_steps": 291, "total_steps": 775, "loss": 0.0389, "learning_rate": 3.512149945019989e-06, "epoch": 1.8728881737731295, "percentage": 37.55, "elapsed_time": "1:04:23", "remaining_time": "1:47:05", "throughput": "498.91", "total_tokens": 1927408} | |
| {"current_steps": 292, "total_steps": 775, "loss": 0.034, "learning_rate": 3.502753837331797e-06, "epoch": 1.8793242156074015, "percentage": 37.68, "elapsed_time": "1:04:36", "remaining_time": "1:46:52", "throughput": "498.95", "total_tokens": 1934160} | |
| {"current_steps": 293, "total_steps": 775, "loss": 0.0921, "learning_rate": 3.4933408185926805e-06, "epoch": 1.8857602574416734, "percentage": 37.81, "elapsed_time": "1:04:49", "remaining_time": "1:46:38", "throughput": "498.99", "total_tokens": 1940912} | |
| {"current_steps": 294, "total_steps": 775, "loss": 0.0661, "learning_rate": 3.4839110475495153e-06, "epoch": 1.8921962992759453, "percentage": 37.94, "elapsed_time": "1:05:02", "remaining_time": "1:46:25", "throughput": "498.99", "total_tokens": 1947488} | |
| {"current_steps": 295, "total_steps": 775, "loss": 0.0301, "learning_rate": 3.4744646832316985e-06, "epoch": 1.898632341110217, "percentage": 38.06, "elapsed_time": "1:05:16", "remaining_time": "1:46:11", "throughput": "498.97", "total_tokens": 1954000} | |
| {"current_steps": 296, "total_steps": 775, "loss": 0.0878, "learning_rate": 3.465001884948468e-06, "epoch": 1.905068382944489, "percentage": 38.19, "elapsed_time": "1:05:29", "remaining_time": "1:45:58", "throughput": "498.92", "total_tokens": 1960400} | |
| {"current_steps": 297, "total_steps": 775, "loss": 0.1126, "learning_rate": 3.45552281228621e-06, "epoch": 1.911504424778761, "percentage": 38.32, "elapsed_time": "1:05:42", "remaining_time": "1:45:45", "throughput": "499.11", "total_tokens": 1967728} | |
| {"current_steps": 298, "total_steps": 775, "loss": 0.0679, "learning_rate": 3.446027625105776e-06, "epoch": 1.9179404666130329, "percentage": 38.45, "elapsed_time": "1:05:55", "remaining_time": "1:45:31", "throughput": "499.05", "total_tokens": 1974096} | |
| {"current_steps": 299, "total_steps": 775, "loss": 0.031, "learning_rate": 3.436516483539781e-06, "epoch": 1.9243765084473048, "percentage": 38.58, "elapsed_time": "1:06:08", "remaining_time": "1:45:18", "throughput": "499.05", "total_tokens": 1980672} | |
| {"current_steps": 300, "total_steps": 775, "loss": 0.0687, "learning_rate": 3.4269895479899023e-06, "epoch": 1.9308125502815767, "percentage": 38.71, "elapsed_time": "1:06:22", "remaining_time": "1:45:04", "throughput": "499.01", "total_tokens": 1987104} | |
| {"current_steps": 301, "total_steps": 775, "loss": 0.0497, "learning_rate": 3.4174469791241805e-06, "epoch": 1.9372485921158487, "percentage": 38.84, "elapsed_time": "1:06:35", "remaining_time": "1:44:51", "throughput": "499.10", "total_tokens": 1994064} | |
| {"current_steps": 302, "total_steps": 775, "loss": 0.0829, "learning_rate": 3.4078889378743036e-06, "epoch": 1.9436846339501206, "percentage": 38.97, "elapsed_time": "1:06:48", "remaining_time": "1:44:38", "throughput": "499.20", "total_tokens": 2001056} | |
| {"current_steps": 303, "total_steps": 775, "loss": 0.0578, "learning_rate": 3.3983155854328942e-06, "epoch": 1.9501206757843925, "percentage": 39.1, "elapsed_time": "1:07:01", "remaining_time": "1:44:24", "throughput": "499.22", "total_tokens": 2007712} | |
| {"current_steps": 304, "total_steps": 775, "loss": 0.0398, "learning_rate": 3.388727083250795e-06, "epoch": 1.9565567176186645, "percentage": 39.23, "elapsed_time": "1:07:14", "remaining_time": "1:44:11", "throughput": "499.24", "total_tokens": 2014368} | |
| {"current_steps": 305, "total_steps": 775, "loss": 0.0754, "learning_rate": 3.379123593034342e-06, "epoch": 1.9629927594529364, "percentage": 39.35, "elapsed_time": "1:07:28", "remaining_time": "1:43:58", "throughput": "499.14", "total_tokens": 2020592} | |
| {"current_steps": 306, "total_steps": 775, "loss": 0.0395, "learning_rate": 3.369505276742638e-06, "epoch": 1.9694288012872083, "percentage": 39.48, "elapsed_time": "1:07:41", "remaining_time": "1:43:44", "throughput": "499.07", "total_tokens": 2026864} | |
| {"current_steps": 307, "total_steps": 775, "loss": 0.0575, "learning_rate": 3.359872296584821e-06, "epoch": 1.9758648431214803, "percentage": 39.61, "elapsed_time": "1:07:54", "remaining_time": "1:43:31", "throughput": "499.06", "total_tokens": 2033440} | |
| {"current_steps": 308, "total_steps": 775, "loss": 0.0472, "learning_rate": 3.350224815017331e-06, "epoch": 1.9823008849557522, "percentage": 39.74, "elapsed_time": "1:08:07", "remaining_time": "1:43:17", "throughput": "498.99", "total_tokens": 2039712} | |
| {"current_steps": 309, "total_steps": 775, "loss": 0.0498, "learning_rate": 3.3405629947411687e-06, "epoch": 1.9887369267900241, "percentage": 39.87, "elapsed_time": "1:08:20", "remaining_time": "1:43:04", "throughput": "499.05", "total_tokens": 2046576} | |
| {"current_steps": 310, "total_steps": 775, "loss": 0.0447, "learning_rate": 3.3308869986991493e-06, "epoch": 1.995172968624296, "percentage": 40.0, "elapsed_time": "1:08:34", "remaining_time": "1:42:51", "throughput": "499.07", "total_tokens": 2053248} | |
| {"current_steps": 311, "total_steps": 775, "loss": 0.0155, "learning_rate": 3.32119699007316e-06, "epoch": 2.001609010458568, "percentage": 40.13, "elapsed_time": "1:08:47", "remaining_time": "1:42:37", "throughput": "499.08", "total_tokens": 2059840} | |
| {"current_steps": 312, "total_steps": 775, "loss": 0.0183, "learning_rate": 3.311493132281402e-06, "epoch": 2.00804505229284, "percentage": 40.26, "elapsed_time": "1:09:00", "remaining_time": "1:42:24", "throughput": "499.07", "total_tokens": 2066384} | |
| {"current_steps": 313, "total_steps": 775, "loss": 0.0102, "learning_rate": 3.3017755889756382e-06, "epoch": 2.014481094127112, "percentage": 40.39, "elapsed_time": "1:09:13", "remaining_time": "1:42:11", "throughput": "499.09", "total_tokens": 2073088} | |
| {"current_steps": 314, "total_steps": 775, "loss": 0.0119, "learning_rate": 3.292044524038433e-06, "epoch": 2.020917135961384, "percentage": 40.52, "elapsed_time": "1:09:26", "remaining_time": "1:41:57", "throughput": "499.07", "total_tokens": 2079600} | |
| {"current_steps": 315, "total_steps": 775, "loss": 0.0037, "learning_rate": 3.2823001015803863e-06, "epoch": 2.0273531777956557, "percentage": 40.65, "elapsed_time": "1:09:40", "remaining_time": "1:41:44", "throughput": "499.05", "total_tokens": 2086080} | |
| {"current_steps": 316, "total_steps": 775, "loss": 0.0048, "learning_rate": 3.272542485937369e-06, "epoch": 2.0337892196299276, "percentage": 40.77, "elapsed_time": "1:09:53", "remaining_time": "1:41:30", "throughput": "499.07", "total_tokens": 2092768} | |
| {"current_steps": 317, "total_steps": 775, "loss": 0.004, "learning_rate": 3.2627718416677484e-06, "epoch": 2.0402252614641996, "percentage": 40.9, "elapsed_time": "1:10:06", "remaining_time": "1:41:17", "throughput": "499.06", "total_tokens": 2099296} | |
| {"current_steps": 318, "total_steps": 775, "loss": 0.0472, "learning_rate": 3.2529883335496163e-06, "epoch": 2.0466613032984715, "percentage": 41.03, "elapsed_time": "1:10:19", "remaining_time": "1:41:04", "throughput": "499.13", "total_tokens": 2106176} | |
| {"current_steps": 319, "total_steps": 775, "loss": 0.0008, "learning_rate": 3.243192126578007e-06, "epoch": 2.0530973451327434, "percentage": 41.16, "elapsed_time": "1:10:32", "remaining_time": "1:40:50", "throughput": "499.08", "total_tokens": 2112560} | |
| {"current_steps": 320, "total_steps": 775, "loss": 0.0332, "learning_rate": 3.2333833859621155e-06, "epoch": 2.0595333869670154, "percentage": 41.29, "elapsed_time": "1:10:46", "remaining_time": "1:40:37", "throughput": "499.11", "total_tokens": 2119296} | |
| {"current_steps": 321, "total_steps": 775, "loss": 0.0434, "learning_rate": 3.223562277122513e-06, "epoch": 2.0659694288012873, "percentage": 41.42, "elapsed_time": "1:10:59", "remaining_time": "1:40:24", "throughput": "499.05", "total_tokens": 2125632} | |
| {"current_steps": 322, "total_steps": 775, "loss": 0.0105, "learning_rate": 3.213728965688356e-06, "epoch": 2.0724054706355592, "percentage": 41.55, "elapsed_time": "1:11:12", "remaining_time": "1:40:10", "throughput": "499.03", "total_tokens": 2132096} | |
| {"current_steps": 323, "total_steps": 775, "loss": 0.0188, "learning_rate": 3.2038836174945907e-06, "epoch": 2.078841512469831, "percentage": 41.68, "elapsed_time": "1:11:25", "remaining_time": "1:39:57", "throughput": "498.94", "total_tokens": 2138336} | |
| {"current_steps": 324, "total_steps": 775, "loss": 0.0382, "learning_rate": 3.194026398579162e-06, "epoch": 2.085277554304103, "percentage": 41.81, "elapsed_time": "1:11:38", "remaining_time": "1:39:44", "throughput": "498.88", "total_tokens": 2144672} | |
| {"current_steps": 325, "total_steps": 775, "loss": 0.0002, "learning_rate": 3.184157475180208e-06, "epoch": 2.091713596138375, "percentage": 41.94, "elapsed_time": "1:11:52", "remaining_time": "1:39:30", "throughput": "498.87", "total_tokens": 2151216} | |
| {"current_steps": 326, "total_steps": 775, "loss": 0.0473, "learning_rate": 3.1742770137332567e-06, "epoch": 2.098149637972647, "percentage": 42.06, "elapsed_time": "1:12:05", "remaining_time": "1:39:17", "throughput": "498.92", "total_tokens": 2158000} | |
| {"current_steps": 327, "total_steps": 775, "loss": 0.0008, "learning_rate": 3.164385180868425e-06, "epoch": 2.104585679806919, "percentage": 42.19, "elapsed_time": "1:12:18", "remaining_time": "1:39:03", "throughput": "498.89", "total_tokens": 2164448} | |
| {"current_steps": 328, "total_steps": 775, "loss": 0.0123, "learning_rate": 3.1544821434076013e-06, "epoch": 2.111021721641191, "percentage": 42.32, "elapsed_time": "1:12:31", "remaining_time": "1:38:50", "throughput": "498.91", "total_tokens": 2171120} | |
| {"current_steps": 329, "total_steps": 775, "loss": 0.03, "learning_rate": 3.144568068361634e-06, "epoch": 2.1174577634754628, "percentage": 42.45, "elapsed_time": "1:12:44", "remaining_time": "1:38:37", "throughput": "498.90", "total_tokens": 2177648} | |
| {"current_steps": 330, "total_steps": 775, "loss": 0.0207, "learning_rate": 3.1346431229275197e-06, "epoch": 2.1238938053097347, "percentage": 42.58, "elapsed_time": "1:12:58", "remaining_time": "1:38:23", "throughput": "498.81", "total_tokens": 2183856} | |
| {"current_steps": 331, "total_steps": 775, "loss": 0.0172, "learning_rate": 3.124707474485577e-06, "epoch": 2.1303298471440066, "percentage": 42.71, "elapsed_time": "1:13:11", "remaining_time": "1:38:10", "throughput": "498.84", "total_tokens": 2190608} | |
| {"current_steps": 332, "total_steps": 775, "loss": 0.0115, "learning_rate": 3.1147612905966286e-06, "epoch": 2.136765888978278, "percentage": 42.84, "elapsed_time": "1:13:24", "remaining_time": "1:37:57", "throughput": "498.85", "total_tokens": 2197232} | |
| {"current_steps": 333, "total_steps": 775, "loss": 0.0217, "learning_rate": 3.1048047389991693e-06, "epoch": 2.14320193081255, "percentage": 42.97, "elapsed_time": "1:13:37", "remaining_time": "1:37:43", "throughput": "498.77", "total_tokens": 2203456} | |
| {"current_steps": 334, "total_steps": 775, "loss": 0.0035, "learning_rate": 3.094837987606547e-06, "epoch": 2.149637972646822, "percentage": 43.1, "elapsed_time": "1:13:50", "remaining_time": "1:37:30", "throughput": "498.73", "total_tokens": 2209856} | |
| {"current_steps": 335, "total_steps": 775, "loss": 0.0426, "learning_rate": 3.084861204504122e-06, "epoch": 2.156074014481094, "percentage": 43.23, "elapsed_time": "1:14:04", "remaining_time": "1:37:17", "throughput": "498.72", "total_tokens": 2216400} | |
| {"current_steps": 336, "total_steps": 775, "loss": 0.0382, "learning_rate": 3.0748745579464347e-06, "epoch": 2.162510056315366, "percentage": 43.35, "elapsed_time": "1:14:17", "remaining_time": "1:37:03", "throughput": "498.69", "total_tokens": 2222864} | |
| {"current_steps": 337, "total_steps": 775, "loss": 0.0082, "learning_rate": 3.0648782163543696e-06, "epoch": 2.168946098149638, "percentage": 43.48, "elapsed_time": "1:14:30", "remaining_time": "1:36:50", "throughput": "498.76", "total_tokens": 2229760} | |
| {"current_steps": 338, "total_steps": 775, "loss": 0.0338, "learning_rate": 3.0548723483123157e-06, "epoch": 2.1753821399839097, "percentage": 43.61, "elapsed_time": "1:14:43", "remaining_time": "1:36:37", "throughput": "498.76", "total_tokens": 2236368} | |
| {"current_steps": 339, "total_steps": 775, "loss": 0.0141, "learning_rate": 3.0448571225653195e-06, "epoch": 2.1818181818181817, "percentage": 43.74, "elapsed_time": "1:14:57", "remaining_time": "1:36:23", "throughput": "498.73", "total_tokens": 2242816} | |
| {"current_steps": 340, "total_steps": 775, "loss": 0.0287, "learning_rate": 3.0348327080162438e-06, "epoch": 2.1882542236524536, "percentage": 43.87, "elapsed_time": "1:15:10", "remaining_time": "1:36:10", "throughput": "498.75", "total_tokens": 2249488} | |
| {"current_steps": 341, "total_steps": 775, "loss": 0.0027, "learning_rate": 3.0247992737229147e-06, "epoch": 2.1946902654867255, "percentage": 44.0, "elapsed_time": "1:15:23", "remaining_time": "1:35:57", "throughput": "498.73", "total_tokens": 2255968} | |
| {"current_steps": 342, "total_steps": 775, "loss": 0.0026, "learning_rate": 3.014756988895275e-06, "epoch": 2.2011263073209975, "percentage": 44.13, "elapsed_time": "1:15:36", "remaining_time": "1:35:43", "throughput": "498.73", "total_tokens": 2262544} | |
| {"current_steps": 343, "total_steps": 775, "loss": 0.039, "learning_rate": 3.0047060228925256e-06, "epoch": 2.2075623491552694, "percentage": 44.26, "elapsed_time": "1:15:49", "remaining_time": "1:35:30", "throughput": "498.77", "total_tokens": 2269312} | |
| {"current_steps": 344, "total_steps": 775, "loss": 0.0154, "learning_rate": 2.994646545220275e-06, "epoch": 2.2139983909895413, "percentage": 44.39, "elapsed_time": "1:16:03", "remaining_time": "1:35:17", "throughput": "498.78", "total_tokens": 2275968} | |
| {"current_steps": 345, "total_steps": 775, "loss": 0.0231, "learning_rate": 2.9845787255276753e-06, "epoch": 2.2204344328238133, "percentage": 44.52, "elapsed_time": "1:16:16", "remaining_time": "1:35:03", "throughput": "498.87", "total_tokens": 2282976} | |
| {"current_steps": 346, "total_steps": 775, "loss": 0.04, "learning_rate": 2.9745027336045652e-06, "epoch": 2.226870474658085, "percentage": 44.65, "elapsed_time": "1:16:29", "remaining_time": "1:34:50", "throughput": "498.90", "total_tokens": 2289696} | |
| {"current_steps": 347, "total_steps": 775, "loss": 0.0282, "learning_rate": 2.964418739378603e-06, "epoch": 2.233306516492357, "percentage": 44.77, "elapsed_time": "1:16:42", "remaining_time": "1:34:37", "throughput": "498.90", "total_tokens": 2296272} | |
| {"current_steps": 348, "total_steps": 775, "loss": 0.0143, "learning_rate": 2.954326912912404e-06, "epoch": 2.239742558326629, "percentage": 44.9, "elapsed_time": "1:16:55", "remaining_time": "1:34:23", "throughput": "498.96", "total_tokens": 2303120} | |
| {"current_steps": 349, "total_steps": 775, "loss": 0.0194, "learning_rate": 2.9442274244006725e-06, "epoch": 2.246178600160901, "percentage": 45.03, "elapsed_time": "1:17:09", "remaining_time": "1:34:10", "throughput": "498.96", "total_tokens": 2309728} | |
| {"current_steps": 350, "total_steps": 775, "loss": 0.0051, "learning_rate": 2.9341204441673267e-06, "epoch": 2.252614641995173, "percentage": 45.16, "elapsed_time": "1:17:22", "remaining_time": "1:33:57", "throughput": "498.93", "total_tokens": 2316144} | |
| {"current_steps": 351, "total_steps": 775, "loss": 0.0162, "learning_rate": 2.924006142662632e-06, "epoch": 2.259050683829445, "percentage": 45.29, "elapsed_time": "1:17:35", "remaining_time": "1:33:43", "throughput": "498.94", "total_tokens": 2322768} | |
| {"current_steps": 352, "total_steps": 775, "loss": 0.0313, "learning_rate": 2.913884690460325e-06, "epoch": 2.265486725663717, "percentage": 45.42, "elapsed_time": "1:17:48", "remaining_time": "1:33:30", "throughput": "498.93", "total_tokens": 2329312} | |
| {"current_steps": 353, "total_steps": 775, "loss": 0.0041, "learning_rate": 2.903756258254734e-06, "epoch": 2.2719227674979887, "percentage": 45.55, "elapsed_time": "1:18:01", "remaining_time": "1:33:17", "throughput": "498.91", "total_tokens": 2335824} | |
| {"current_steps": 354, "total_steps": 775, "loss": 0.0321, "learning_rate": 2.8936210168579043e-06, "epoch": 2.2783588093322606, "percentage": 45.68, "elapsed_time": "1:18:15", "remaining_time": "1:33:03", "throughput": "498.88", "total_tokens": 2342272} | |
| {"current_steps": 355, "total_steps": 775, "loss": 0.0064, "learning_rate": 2.883479137196714e-06, "epoch": 2.2847948511665326, "percentage": 45.81, "elapsed_time": "1:18:28", "remaining_time": "1:32:50", "throughput": "498.92", "total_tokens": 2349056} | |
| {"current_steps": 356, "total_steps": 775, "loss": 0.0237, "learning_rate": 2.8733307903099926e-06, "epoch": 2.2912308930008045, "percentage": 45.94, "elapsed_time": "1:18:41", "remaining_time": "1:32:37", "throughput": "498.90", "total_tokens": 2355552} | |
| {"current_steps": 357, "total_steps": 775, "loss": 0.0152, "learning_rate": 2.8631761473456377e-06, "epoch": 2.2976669348350764, "percentage": 46.06, "elapsed_time": "1:18:54", "remaining_time": "1:32:23", "throughput": "498.83", "total_tokens": 2361808} | |
| {"current_steps": 358, "total_steps": 775, "loss": 0.0038, "learning_rate": 2.853015379557729e-06, "epoch": 2.3041029766693484, "percentage": 46.19, "elapsed_time": "1:19:07", "remaining_time": "1:32:10", "throughput": "498.81", "total_tokens": 2368288} | |
| {"current_steps": 359, "total_steps": 775, "loss": 0.0168, "learning_rate": 2.842848658303637e-06, "epoch": 2.3105390185036203, "percentage": 46.32, "elapsed_time": "1:19:21", "remaining_time": "1:31:57", "throughput": "498.83", "total_tokens": 2374960} | |
| {"current_steps": 360, "total_steps": 775, "loss": 0.0049, "learning_rate": 2.832676155041135e-06, "epoch": 2.3169750603378922, "percentage": 46.45, "elapsed_time": "1:19:34", "remaining_time": "1:31:43", "throughput": "498.88", "total_tokens": 2381776} | |
| {"current_steps": 361, "total_steps": 775, "loss": 0.0114, "learning_rate": 2.822498041325509e-06, "epoch": 2.323411102172164, "percentage": 46.58, "elapsed_time": "1:19:47", "remaining_time": "1:31:30", "throughput": "498.82", "total_tokens": 2388112} | |
| {"current_steps": 362, "total_steps": 775, "loss": 0.022, "learning_rate": 2.8123144888066623e-06, "epoch": 2.329847144006436, "percentage": 46.71, "elapsed_time": "1:20:00", "remaining_time": "1:31:17", "throughput": "498.83", "total_tokens": 2394736} | |
| {"current_steps": 363, "total_steps": 775, "loss": 0.0154, "learning_rate": 2.802125669226222e-06, "epoch": 2.336283185840708, "percentage": 46.84, "elapsed_time": "1:20:13", "remaining_time": "1:31:03", "throughput": "498.82", "total_tokens": 2401248} | |
| {"current_steps": 364, "total_steps": 775, "loss": 0.0204, "learning_rate": 2.7919317544146405e-06, "epoch": 2.34271922767498, "percentage": 46.97, "elapsed_time": "1:20:27", "remaining_time": "1:30:50", "throughput": "498.82", "total_tokens": 2407872} | |
| {"current_steps": 365, "total_steps": 775, "loss": 0.0334, "learning_rate": 2.7817329162883033e-06, "epoch": 2.349155269509252, "percentage": 47.1, "elapsed_time": "1:20:40", "remaining_time": "1:30:37", "throughput": "498.82", "total_tokens": 2414432} | |
| {"current_steps": 366, "total_steps": 775, "loss": 0.0132, "learning_rate": 2.7715293268466204e-06, "epoch": 2.355591311343524, "percentage": 47.23, "elapsed_time": "1:20:53", "remaining_time": "1:30:23", "throughput": "498.79", "total_tokens": 2420848} | |
| {"current_steps": 367, "total_steps": 775, "loss": 0.0291, "learning_rate": 2.761321158169134e-06, "epoch": 2.3620273531777958, "percentage": 47.35, "elapsed_time": "1:21:06", "remaining_time": "1:30:10", "throughput": "498.84", "total_tokens": 2427728} | |
| {"current_steps": 368, "total_steps": 775, "loss": 0.0089, "learning_rate": 2.7511085824126133e-06, "epoch": 2.3684633950120677, "percentage": 47.48, "elapsed_time": "1:21:19", "remaining_time": "1:29:57", "throughput": "498.96", "total_tokens": 2434880} | |
| {"current_steps": 369, "total_steps": 775, "loss": 0.0306, "learning_rate": 2.74089177180815e-06, "epoch": 2.3748994368463396, "percentage": 47.61, "elapsed_time": "1:21:33", "remaining_time": "1:29:43", "throughput": "498.90", "total_tokens": 2441168} | |
| {"current_steps": 370, "total_steps": 775, "loss": 0.0297, "learning_rate": 2.730670898658255e-06, "epoch": 2.3813354786806116, "percentage": 47.74, "elapsed_time": "1:21:46", "remaining_time": "1:29:30", "throughput": "498.93", "total_tokens": 2447920} | |
| {"current_steps": 371, "total_steps": 775, "loss": 0.0247, "learning_rate": 2.7204461353339546e-06, "epoch": 2.3877715205148835, "percentage": 47.87, "elapsed_time": "1:21:59", "remaining_time": "1:29:17", "throughput": "498.97", "total_tokens": 2454704} | |
| {"current_steps": 372, "total_steps": 775, "loss": 0.0234, "learning_rate": 2.7102176542718783e-06, "epoch": 2.3942075623491554, "percentage": 48.0, "elapsed_time": "1:22:12", "remaining_time": "1:29:03", "throughput": "498.96", "total_tokens": 2461216} | |
| {"current_steps": 373, "total_steps": 775, "loss": 0.0192, "learning_rate": 2.699985627971354e-06, "epoch": 2.4006436041834274, "percentage": 48.13, "elapsed_time": "1:22:25", "remaining_time": "1:28:50", "throughput": "499.00", "total_tokens": 2468032} | |
| {"current_steps": 374, "total_steps": 775, "loss": 0.0324, "learning_rate": 2.689750228991503e-06, "epoch": 2.4070796460176993, "percentage": 48.26, "elapsed_time": "1:22:39", "remaining_time": "1:28:37", "throughput": "498.99", "total_tokens": 2474544} | |
| {"current_steps": 375, "total_steps": 775, "loss": 0.0332, "learning_rate": 2.679511629948319e-06, "epoch": 2.4135156878519712, "percentage": 48.39, "elapsed_time": "1:22:52", "remaining_time": "1:28:23", "throughput": "499.02", "total_tokens": 2481312} | |
| {"current_steps": 376, "total_steps": 775, "loss": 0.0043, "learning_rate": 2.669270003511769e-06, "epoch": 2.419951729686243, "percentage": 48.52, "elapsed_time": "1:23:05", "remaining_time": "1:28:10", "throughput": "499.02", "total_tokens": 2487888} | |
| {"current_steps": 377, "total_steps": 775, "loss": 0.0197, "learning_rate": 2.6590255224028725e-06, "epoch": 2.426387771520515, "percentage": 48.65, "elapsed_time": "1:23:18", "remaining_time": "1:27:57", "throughput": "499.08", "total_tokens": 2494784} | |
| {"current_steps": 378, "total_steps": 775, "loss": 0.0366, "learning_rate": 2.648778359390794e-06, "epoch": 2.432823813354787, "percentage": 48.77, "elapsed_time": "1:23:31", "remaining_time": "1:27:43", "throughput": "499.15", "total_tokens": 2501712} | |
| {"current_steps": 379, "total_steps": 775, "loss": 0.0173, "learning_rate": 2.638528687289925e-06, "epoch": 2.439259855189059, "percentage": 48.9, "elapsed_time": "1:23:45", "remaining_time": "1:27:30", "throughput": "499.21", "total_tokens": 2508592} | |
| {"current_steps": 380, "total_steps": 775, "loss": 0.0102, "learning_rate": 2.6282766789569742e-06, "epoch": 2.445695897023331, "percentage": 49.03, "elapsed_time": "1:23:58", "remaining_time": "1:27:17", "throughput": "499.22", "total_tokens": 2515216} | |
| {"current_steps": 381, "total_steps": 775, "loss": 0.0361, "learning_rate": 2.618022507288049e-06, "epoch": 2.4521319388576024, "percentage": 49.16, "elapsed_time": "1:24:11", "remaining_time": "1:27:03", "throughput": "499.27", "total_tokens": 2522064} | |
| {"current_steps": 382, "total_steps": 775, "loss": 0.0292, "learning_rate": 2.6077663452157398e-06, "epoch": 2.4585679806918743, "percentage": 49.29, "elapsed_time": "1:24:24", "remaining_time": "1:26:50", "throughput": "499.26", "total_tokens": 2528608} | |
| {"current_steps": 383, "total_steps": 775, "loss": 0.0095, "learning_rate": 2.5975083657062043e-06, "epoch": 2.4650040225261463, "percentage": 49.42, "elapsed_time": "1:24:37", "remaining_time": "1:26:37", "throughput": "499.28", "total_tokens": 2535328} | |
| {"current_steps": 384, "total_steps": 775, "loss": 0.015, "learning_rate": 2.587248741756253e-06, "epoch": 2.471440064360418, "percentage": 49.55, "elapsed_time": "1:24:51", "remaining_time": "1:26:23", "throughput": "499.34", "total_tokens": 2542224} | |
| {"current_steps": 385, "total_steps": 775, "loss": 0.0276, "learning_rate": 2.576987646390426e-06, "epoch": 2.47787610619469, "percentage": 49.68, "elapsed_time": "1:25:04", "remaining_time": "1:26:10", "throughput": "499.37", "total_tokens": 2548976} | |
| {"current_steps": 386, "total_steps": 775, "loss": 0.0284, "learning_rate": 2.566725252658081e-06, "epoch": 2.484312148028962, "percentage": 49.81, "elapsed_time": "1:25:17", "remaining_time": "1:25:57", "throughput": "499.37", "total_tokens": 2555568} | |
| {"current_steps": 387, "total_steps": 775, "loss": 0.0366, "learning_rate": 2.5564617336304703e-06, "epoch": 2.490748189863234, "percentage": 49.94, "elapsed_time": "1:25:30", "remaining_time": "1:25:44", "throughput": "499.37", "total_tokens": 2562128} | |
| {"current_steps": 388, "total_steps": 775, "loss": 0.0322, "learning_rate": 2.546197262397825e-06, "epoch": 2.497184231697506, "percentage": 50.06, "elapsed_time": "1:25:43", "remaining_time": "1:25:30", "throughput": "499.35", "total_tokens": 2568640} | |
| {"current_steps": 389, "total_steps": 775, "loss": 0.0057, "learning_rate": 2.535932012066434e-06, "epoch": 2.503620273531778, "percentage": 50.19, "elapsed_time": "1:25:57", "remaining_time": "1:25:17", "throughput": "499.31", "total_tokens": 2575024} | |
| {"current_steps": 390, "total_steps": 775, "loss": 0.0054, "learning_rate": 2.525666155755725e-06, "epoch": 2.51005631536605, "percentage": 50.32, "elapsed_time": "1:26:10", "remaining_time": "1:25:04", "throughput": "499.29", "total_tokens": 2581520} | |
| {"current_steps": 391, "total_steps": 775, "loss": 0.0199, "learning_rate": 2.515399866595347e-06, "epoch": 2.5164923572003217, "percentage": 50.45, "elapsed_time": "1:26:23", "remaining_time": "1:24:50", "throughput": "499.37", "total_tokens": 2588528} | |
| {"current_steps": 392, "total_steps": 775, "loss": 0.0045, "learning_rate": 2.5051333177222476e-06, "epoch": 2.5229283990345936, "percentage": 50.58, "elapsed_time": "1:26:36", "remaining_time": "1:24:37", "throughput": "499.35", "total_tokens": 2594992} | |
| {"current_steps": 393, "total_steps": 775, "loss": 0.0283, "learning_rate": 2.4948666822777536e-06, "epoch": 2.5293644408688656, "percentage": 50.71, "elapsed_time": "1:26:49", "remaining_time": "1:24:24", "throughput": "499.35", "total_tokens": 2601568} | |
| {"current_steps": 394, "total_steps": 775, "loss": 0.0248, "learning_rate": 2.4846001334046537e-06, "epoch": 2.5358004827031375, "percentage": 50.84, "elapsed_time": "1:27:03", "remaining_time": "1:24:10", "throughput": "499.35", "total_tokens": 2608160} | |
| {"current_steps": 395, "total_steps": 775, "loss": 0.0132, "learning_rate": 2.474333844244276e-06, "epoch": 2.5422365245374094, "percentage": 50.97, "elapsed_time": "1:27:16", "remaining_time": "1:23:57", "throughput": "499.33", "total_tokens": 2614656} | |
| {"current_steps": 396, "total_steps": 775, "loss": 0.04, "learning_rate": 2.464067987933567e-06, "epoch": 2.5486725663716814, "percentage": 51.1, "elapsed_time": "1:27:29", "remaining_time": "1:23:44", "throughput": "499.40", "total_tokens": 2621600} | |
| {"current_steps": 397, "total_steps": 775, "loss": 0.0029, "learning_rate": 2.453802737602176e-06, "epoch": 2.5551086082059533, "percentage": 51.23, "elapsed_time": "1:27:42", "remaining_time": "1:23:30", "throughput": "499.36", "total_tokens": 2627984} | |
| {"current_steps": 398, "total_steps": 775, "loss": 0.0254, "learning_rate": 2.4435382663695305e-06, "epoch": 2.5615446500402252, "percentage": 51.35, "elapsed_time": "1:27:55", "remaining_time": "1:23:17", "throughput": "499.38", "total_tokens": 2634720} | |
| {"current_steps": 399, "total_steps": 775, "loss": 0.0108, "learning_rate": 2.4332747473419193e-06, "epoch": 2.567980691874497, "percentage": 51.48, "elapsed_time": "1:28:09", "remaining_time": "1:23:04", "throughput": "499.41", "total_tokens": 2641456} | |
| {"current_steps": 400, "total_steps": 775, "loss": 0.0269, "learning_rate": 2.4230123536095746e-06, "epoch": 2.574416733708769, "percentage": 51.61, "elapsed_time": "1:28:22", "remaining_time": "1:22:50", "throughput": "499.36", "total_tokens": 2647760} | |
| {"current_steps": 401, "total_steps": 775, "loss": 0.0111, "learning_rate": 2.4127512582437486e-06, "epoch": 2.580852775543041, "percentage": 51.74, "elapsed_time": "1:28:35", "remaining_time": "1:22:37", "throughput": "499.40", "total_tokens": 2654608} | |
| {"current_steps": 402, "total_steps": 775, "loss": 0.0222, "learning_rate": 2.4024916342937966e-06, "epoch": 2.587288817377313, "percentage": 51.87, "elapsed_time": "1:28:48", "remaining_time": "1:22:24", "throughput": "499.38", "total_tokens": 2661072} | |
| {"current_steps": 403, "total_steps": 775, "loss": 0.0101, "learning_rate": 2.392233654784262e-06, "epoch": 2.593724859211585, "percentage": 52.0, "elapsed_time": "1:29:01", "remaining_time": "1:22:11", "throughput": "499.39", "total_tokens": 2667712} | |
| {"current_steps": 404, "total_steps": 775, "loss": 0.0138, "learning_rate": 2.3819774927119523e-06, "epoch": 2.600160901045857, "percentage": 52.13, "elapsed_time": "1:29:15", "remaining_time": "1:21:57", "throughput": "499.42", "total_tokens": 2674496} | |
| {"current_steps": 405, "total_steps": 775, "loss": 0.0281, "learning_rate": 2.3717233210430258e-06, "epoch": 2.6065969428801288, "percentage": 52.26, "elapsed_time": "1:29:28", "remaining_time": "1:21:44", "throughput": "499.37", "total_tokens": 2680816} | |
| {"current_steps": 406, "total_steps": 775, "loss": 0.0042, "learning_rate": 2.3614713127100752e-06, "epoch": 2.6130329847144007, "percentage": 52.39, "elapsed_time": "1:29:41", "remaining_time": "1:21:31", "throughput": "499.42", "total_tokens": 2687632} | |
| {"current_steps": 407, "total_steps": 775, "loss": 0.0015, "learning_rate": 2.3512216406092066e-06, "epoch": 2.6194690265486726, "percentage": 52.52, "elapsed_time": "1:29:54", "remaining_time": "1:21:17", "throughput": "499.46", "total_tokens": 2694464} | |
| {"current_steps": 408, "total_steps": 775, "loss": 0.0279, "learning_rate": 2.340974477597128e-06, "epoch": 2.6259050683829446, "percentage": 52.65, "elapsed_time": "1:30:07", "remaining_time": "1:21:04", "throughput": "499.51", "total_tokens": 2701344} | |
| {"current_steps": 409, "total_steps": 775, "loss": 0.0399, "learning_rate": 2.3307299964882314e-06, "epoch": 2.6323411102172165, "percentage": 52.77, "elapsed_time": "1:30:21", "remaining_time": "1:20:51", "throughput": "499.44", "total_tokens": 2707536} | |
| {"current_steps": 410, "total_steps": 775, "loss": 0.0074, "learning_rate": 2.3204883700516813e-06, "epoch": 2.6387771520514884, "percentage": 52.9, "elapsed_time": "1:30:34", "remaining_time": "1:20:37", "throughput": "499.52", "total_tokens": 2714544} | |
| {"current_steps": 411, "total_steps": 775, "loss": 0.0078, "learning_rate": 2.310249771008498e-06, "epoch": 2.6452131938857604, "percentage": 53.03, "elapsed_time": "1:30:47", "remaining_time": "1:20:24", "throughput": "499.50", "total_tokens": 2721056} | |
| {"current_steps": 412, "total_steps": 775, "loss": 0.0406, "learning_rate": 2.3000143720286463e-06, "epoch": 2.6516492357200323, "percentage": 53.16, "elapsed_time": "1:31:00", "remaining_time": "1:20:11", "throughput": "499.50", "total_tokens": 2727664} | |
| {"current_steps": 413, "total_steps": 775, "loss": 0.0023, "learning_rate": 2.2897823457281225e-06, "epoch": 2.6580852775543042, "percentage": 53.29, "elapsed_time": "1:31:13", "remaining_time": "1:19:58", "throughput": "499.38", "total_tokens": 2733600} | |
| {"current_steps": 414, "total_steps": 775, "loss": 0.006, "learning_rate": 2.2795538646660462e-06, "epoch": 2.664521319388576, "percentage": 53.42, "elapsed_time": "1:31:27", "remaining_time": "1:19:44", "throughput": "499.42", "total_tokens": 2740400} | |
| {"current_steps": 415, "total_steps": 775, "loss": 0.0236, "learning_rate": 2.269329101341745e-06, "epoch": 2.670957361222848, "percentage": 53.55, "elapsed_time": "1:31:40", "remaining_time": "1:19:31", "throughput": "499.46", "total_tokens": 2747248} | |
| {"current_steps": 416, "total_steps": 775, "loss": 0.0136, "learning_rate": 2.2591082281918507e-06, "epoch": 2.67739340305712, "percentage": 53.68, "elapsed_time": "1:31:53", "remaining_time": "1:19:18", "throughput": "499.45", "total_tokens": 2753776} | |
| {"current_steps": 417, "total_steps": 775, "loss": 0.015, "learning_rate": 2.2488914175873876e-06, "epoch": 2.6838294448913915, "percentage": 53.81, "elapsed_time": "1:32:06", "remaining_time": "1:19:04", "throughput": "499.52", "total_tokens": 2760720} | |
| {"current_steps": 418, "total_steps": 775, "loss": 0.0483, "learning_rate": 2.238678841830867e-06, "epoch": 2.6902654867256635, "percentage": 53.94, "elapsed_time": "1:32:20", "remaining_time": "1:18:51", "throughput": "499.48", "total_tokens": 2767136} | |
| {"current_steps": 419, "total_steps": 775, "loss": 0.0014, "learning_rate": 2.2284706731533805e-06, "epoch": 2.6967015285599354, "percentage": 54.06, "elapsed_time": "1:32:33", "remaining_time": "1:18:38", "throughput": "499.47", "total_tokens": 2773680} | |
| {"current_steps": 420, "total_steps": 775, "loss": 0.0279, "learning_rate": 2.2182670837116975e-06, "epoch": 2.7031375703942073, "percentage": 54.19, "elapsed_time": "1:32:46", "remaining_time": "1:18:24", "throughput": "499.45", "total_tokens": 2780160} | |
| {"current_steps": 421, "total_steps": 775, "loss": 0.0705, "learning_rate": 2.20806824558536e-06, "epoch": 2.7095736122284793, "percentage": 54.32, "elapsed_time": "1:32:59", "remaining_time": "1:18:11", "throughput": "499.48", "total_tokens": 2786912} | |
| {"current_steps": 422, "total_steps": 775, "loss": 0.005, "learning_rate": 2.197874330773779e-06, "epoch": 2.716009654062751, "percentage": 54.45, "elapsed_time": "1:33:12", "remaining_time": "1:17:58", "throughput": "499.55", "total_tokens": 2793888} | |
| {"current_steps": 423, "total_steps": 775, "loss": 0.0453, "learning_rate": 2.1876855111933385e-06, "epoch": 2.722445695897023, "percentage": 54.58, "elapsed_time": "1:33:26", "remaining_time": "1:17:45", "throughput": "499.52", "total_tokens": 2800320} | |
| {"current_steps": 424, "total_steps": 775, "loss": 0.0095, "learning_rate": 2.1775019586744924e-06, "epoch": 2.728881737731295, "percentage": 54.71, "elapsed_time": "1:33:39", "remaining_time": "1:17:31", "throughput": "499.55", "total_tokens": 2807088} | |
| {"current_steps": 425, "total_steps": 775, "loss": 0.0095, "learning_rate": 2.167323844958867e-06, "epoch": 2.735317779565567, "percentage": 54.84, "elapsed_time": "1:33:52", "remaining_time": "1:17:18", "throughput": "499.48", "total_tokens": 2813312} | |
| {"current_steps": 426, "total_steps": 775, "loss": 0.0138, "learning_rate": 2.1571513416963647e-06, "epoch": 2.741753821399839, "percentage": 54.97, "elapsed_time": "1:34:05", "remaining_time": "1:17:05", "throughput": "499.49", "total_tokens": 2819936} | |
| {"current_steps": 427, "total_steps": 775, "loss": 0.0272, "learning_rate": 2.1469846204422724e-06, "epoch": 2.748189863234111, "percentage": 55.1, "elapsed_time": "1:34:18", "remaining_time": "1:16:51", "throughput": "499.43", "total_tokens": 2826224} | |
| {"current_steps": 428, "total_steps": 775, "loss": 0.0026, "learning_rate": 2.136823852654363e-06, "epoch": 2.754625905068383, "percentage": 55.23, "elapsed_time": "1:34:32", "remaining_time": "1:16:38", "throughput": "499.46", "total_tokens": 2832960} | |
| {"current_steps": 429, "total_steps": 775, "loss": 0.0016, "learning_rate": 2.126669209690008e-06, "epoch": 2.7610619469026547, "percentage": 55.35, "elapsed_time": "1:34:45", "remaining_time": "1:16:25", "throughput": "499.52", "total_tokens": 2839888} | |
| {"current_steps": 430, "total_steps": 775, "loss": 0.0633, "learning_rate": 2.1165208628032863e-06, "epoch": 2.7674979887369267, "percentage": 55.48, "elapsed_time": "1:34:58", "remaining_time": "1:16:12", "throughput": "499.56", "total_tokens": 2846688} | |
| {"current_steps": 431, "total_steps": 775, "loss": 0.0191, "learning_rate": 2.1063789831420957e-06, "epoch": 2.7739340305711986, "percentage": 55.61, "elapsed_time": "1:35:11", "remaining_time": "1:15:58", "throughput": "499.54", "total_tokens": 2853184} | |
| {"current_steps": 432, "total_steps": 775, "loss": 0.0075, "learning_rate": 2.096243741745266e-06, "epoch": 2.7803700724054705, "percentage": 55.74, "elapsed_time": "1:35:24", "remaining_time": "1:15:45", "throughput": "499.51", "total_tokens": 2859632} | |
| {"current_steps": 433, "total_steps": 775, "loss": 0.0371, "learning_rate": 2.086115309539675e-06, "epoch": 2.7868061142397424, "percentage": 55.87, "elapsed_time": "1:35:38", "remaining_time": "1:15:32", "throughput": "499.46", "total_tokens": 2865920} | |
| {"current_steps": 434, "total_steps": 775, "loss": 0.0275, "learning_rate": 2.0759938573373683e-06, "epoch": 2.7932421560740144, "percentage": 56.0, "elapsed_time": "1:35:51", "remaining_time": "1:15:18", "throughput": "499.45", "total_tokens": 2872464} | |
| {"current_steps": 435, "total_steps": 775, "loss": 0.0441, "learning_rate": 2.0658795558326745e-06, "epoch": 2.7996781979082863, "percentage": 56.13, "elapsed_time": "1:36:04", "remaining_time": "1:15:05", "throughput": "499.47", "total_tokens": 2879168} | |
| {"current_steps": 436, "total_steps": 775, "loss": 0.0074, "learning_rate": 2.0557725755993283e-06, "epoch": 2.8061142397425582, "percentage": 56.26, "elapsed_time": "1:36:17", "remaining_time": "1:14:52", "throughput": "499.43", "total_tokens": 2885520} | |
| {"current_steps": 437, "total_steps": 775, "loss": 0.0412, "learning_rate": 2.0456730870875964e-06, "epoch": 2.81255028157683, "percentage": 56.39, "elapsed_time": "1:36:30", "remaining_time": "1:14:38", "throughput": "499.47", "total_tokens": 2892368} | |
| {"current_steps": 438, "total_steps": 775, "loss": 0.007, "learning_rate": 2.035581260621398e-06, "epoch": 2.818986323411102, "percentage": 56.52, "elapsed_time": "1:36:44", "remaining_time": "1:14:25", "throughput": "499.41", "total_tokens": 2898640} | |
| {"current_steps": 439, "total_steps": 775, "loss": 0.0195, "learning_rate": 2.0254972663954356e-06, "epoch": 2.825422365245374, "percentage": 56.65, "elapsed_time": "1:36:57", "remaining_time": "1:14:12", "throughput": "499.43", "total_tokens": 2905312} | |
| {"current_steps": 440, "total_steps": 775, "loss": 0.0104, "learning_rate": 2.015421274472325e-06, "epoch": 2.831858407079646, "percentage": 56.77, "elapsed_time": "1:37:10", "remaining_time": "1:13:59", "throughput": "499.42", "total_tokens": 2911872} | |
| {"current_steps": 441, "total_steps": 775, "loss": 0.0196, "learning_rate": 2.005353454779726e-06, "epoch": 2.838294448913918, "percentage": 56.9, "elapsed_time": "1:37:23", "remaining_time": "1:13:45", "throughput": "499.43", "total_tokens": 2918496} | |
| {"current_steps": 442, "total_steps": 775, "loss": 0.0131, "learning_rate": 1.995293977107475e-06, "epoch": 2.84473049074819, "percentage": 57.03, "elapsed_time": "1:37:36", "remaining_time": "1:13:32", "throughput": "499.40", "total_tokens": 2924944} | |
| {"current_steps": 443, "total_steps": 775, "loss": 0.0072, "learning_rate": 1.9852430111047254e-06, "epoch": 2.8511665325824618, "percentage": 57.16, "elapsed_time": "1:37:50", "remaining_time": "1:13:19", "throughput": "499.39", "total_tokens": 2931440} | |
| {"current_steps": 444, "total_steps": 775, "loss": 0.0058, "learning_rate": 1.9752007262770857e-06, "epoch": 2.8576025744167337, "percentage": 57.29, "elapsed_time": "1:38:03", "remaining_time": "1:13:05", "throughput": "499.43", "total_tokens": 2938304} | |
| {"current_steps": 445, "total_steps": 775, "loss": 0.0242, "learning_rate": 1.965167291983757e-06, "epoch": 2.8640386162510056, "percentage": 57.42, "elapsed_time": "1:38:16", "remaining_time": "1:12:52", "throughput": "499.48", "total_tokens": 2945168} | |
| {"current_steps": 446, "total_steps": 775, "loss": 0.0323, "learning_rate": 1.955142877434681e-06, "epoch": 2.8704746580852776, "percentage": 57.55, "elapsed_time": "1:38:29", "remaining_time": "1:12:39", "throughput": "499.51", "total_tokens": 2951952} | |
| {"current_steps": 447, "total_steps": 775, "loss": 0.0186, "learning_rate": 1.9451276516876856e-06, "epoch": 2.8769106999195495, "percentage": 57.68, "elapsed_time": "1:38:42", "remaining_time": "1:12:26", "throughput": "499.49", "total_tokens": 2958432} | |
| {"current_steps": 448, "total_steps": 775, "loss": 0.0071, "learning_rate": 1.9351217836456316e-06, "epoch": 2.8833467417538214, "percentage": 57.81, "elapsed_time": "1:38:56", "remaining_time": "1:12:12", "throughput": "499.54", "total_tokens": 2965328} | |
| {"current_steps": 449, "total_steps": 775, "loss": 0.0177, "learning_rate": 1.9251254420535665e-06, "epoch": 2.8897827835880934, "percentage": 57.94, "elapsed_time": "1:39:09", "remaining_time": "1:11:59", "throughput": "499.53", "total_tokens": 2971872} | |
| {"current_steps": 450, "total_steps": 775, "loss": 0.044, "learning_rate": 1.9151387954958792e-06, "epoch": 2.8962188254223653, "percentage": 58.06, "elapsed_time": "1:39:22", "remaining_time": "1:11:46", "throughput": "499.59", "total_tokens": 2978784} | |
| {"current_steps": 451, "total_steps": 775, "loss": 0.0119, "learning_rate": 1.9051620123934538e-06, "epoch": 2.9026548672566372, "percentage": 58.19, "elapsed_time": "1:39:35", "remaining_time": "1:11:32", "throughput": "499.65", "total_tokens": 2985760} | |
| {"current_steps": 452, "total_steps": 775, "loss": 0.0121, "learning_rate": 1.895195261000831e-06, "epoch": 2.909090909090909, "percentage": 58.32, "elapsed_time": "1:39:48", "remaining_time": "1:11:19", "throughput": "499.65", "total_tokens": 2992352} | |
| {"current_steps": 453, "total_steps": 775, "loss": 0.0319, "learning_rate": 1.885238709403372e-06, "epoch": 2.915526950925181, "percentage": 58.45, "elapsed_time": "1:40:02", "remaining_time": "1:11:06", "throughput": "499.62", "total_tokens": 2998800} | |
| {"current_steps": 454, "total_steps": 775, "loss": 0.0024, "learning_rate": 1.8752925255144228e-06, "epoch": 2.921962992759453, "percentage": 58.58, "elapsed_time": "1:40:15", "remaining_time": "1:10:53", "throughput": "499.62", "total_tokens": 3005392} | |
| {"current_steps": 455, "total_steps": 775, "loss": 0.0102, "learning_rate": 1.8653568770724805e-06, "epoch": 2.928399034593725, "percentage": 58.71, "elapsed_time": "1:40:28", "remaining_time": "1:10:39", "throughput": "499.63", "total_tokens": 3012016} | |
| {"current_steps": 456, "total_steps": 775, "loss": 0.0419, "learning_rate": 1.8554319316383657e-06, "epoch": 2.934835076427997, "percentage": 58.84, "elapsed_time": "1:40:41", "remaining_time": "1:10:26", "throughput": "499.65", "total_tokens": 3018768} | |
| {"current_steps": 457, "total_steps": 775, "loss": 0.0109, "learning_rate": 1.8455178565923993e-06, "epoch": 2.941271118262269, "percentage": 58.97, "elapsed_time": "1:40:54", "remaining_time": "1:10:13", "throughput": "499.65", "total_tokens": 3025328} | |
| {"current_steps": 458, "total_steps": 775, "loss": 0.0092, "learning_rate": 1.8356148191315753e-06, "epoch": 2.9477071600965408, "percentage": 59.1, "elapsed_time": "1:41:08", "remaining_time": "1:10:00", "throughput": "499.67", "total_tokens": 3032080} | |
| {"current_steps": 459, "total_steps": 775, "loss": 0.0449, "learning_rate": 1.8257229862667437e-06, "epoch": 2.9541432019308127, "percentage": 59.23, "elapsed_time": "1:41:21", "remaining_time": "1:09:46", "throughput": "499.71", "total_tokens": 3038880} | |
| {"current_steps": 460, "total_steps": 775, "loss": 0.014, "learning_rate": 1.8158425248197931e-06, "epoch": 2.9605792437650846, "percentage": 59.35, "elapsed_time": "1:41:34", "remaining_time": "1:09:33", "throughput": "499.72", "total_tokens": 3045552} | |
| {"current_steps": 461, "total_steps": 775, "loss": 0.0035, "learning_rate": 1.8059736014208388e-06, "epoch": 2.9670152855993566, "percentage": 59.48, "elapsed_time": "1:41:47", "remaining_time": "1:09:20", "throughput": "499.74", "total_tokens": 3052288} | |
| {"current_steps": 462, "total_steps": 775, "loss": 0.016, "learning_rate": 1.7961163825054101e-06, "epoch": 2.9734513274336285, "percentage": 59.61, "elapsed_time": "1:42:00", "remaining_time": "1:09:06", "throughput": "499.72", "total_tokens": 3058768} | |
| {"current_steps": 463, "total_steps": 775, "loss": 0.0151, "learning_rate": 1.7862710343116451e-06, "epoch": 2.9798873692679004, "percentage": 59.74, "elapsed_time": "1:42:14", "remaining_time": "1:08:53", "throughput": "499.76", "total_tokens": 3065584} | |
| {"current_steps": 464, "total_steps": 775, "loss": 0.0039, "learning_rate": 1.7764377228774877e-06, "epoch": 2.9863234111021724, "percentage": 59.87, "elapsed_time": "1:42:27", "remaining_time": "1:08:40", "throughput": "499.79", "total_tokens": 3072368} | |
| {"current_steps": 465, "total_steps": 775, "loss": 0.0361, "learning_rate": 1.7666166140378853e-06, "epoch": 2.9927594529364443, "percentage": 60.0, "elapsed_time": "1:42:40", "remaining_time": "1:08:27", "throughput": "499.77", "total_tokens": 3078864} | |
| {"current_steps": 466, "total_steps": 775, "loss": 0.0609, "learning_rate": 1.7568078734219934e-06, "epoch": 2.9991954947707162, "percentage": 60.13, "elapsed_time": "1:42:53", "remaining_time": "1:08:13", "throughput": "499.80", "total_tokens": 3085664} | |
| {"current_steps": 467, "total_steps": 775, "loss": 0.0027, "learning_rate": 1.747011666450384e-06, "epoch": 3.0056315366049877, "percentage": 60.26, "elapsed_time": "1:43:06", "remaining_time": "1:08:00", "throughput": "499.69", "total_tokens": 3091568} | |
| {"current_steps": 468, "total_steps": 775, "loss": 0.0097, "learning_rate": 1.737228158332252e-06, "epoch": 3.0120675784392597, "percentage": 60.39, "elapsed_time": "1:43:20", "remaining_time": "1:07:47", "throughput": "499.75", "total_tokens": 3098544} | |
| {"current_steps": 469, "total_steps": 775, "loss": 0.0091, "learning_rate": 1.7274575140626318e-06, "epoch": 3.0185036202735316, "percentage": 60.52, "elapsed_time": "1:43:33", "remaining_time": "1:07:33", "throughput": "499.75", "total_tokens": 3105120} | |
| {"current_steps": 470, "total_steps": 775, "loss": 0.0026, "learning_rate": 1.7176998984196148e-06, "epoch": 3.0249396621078035, "percentage": 60.65, "elapsed_time": "1:43:46", "remaining_time": "1:07:20", "throughput": "499.72", "total_tokens": 3111552} | |
| {"current_steps": 471, "total_steps": 775, "loss": 0.0311, "learning_rate": 1.7079554759615685e-06, "epoch": 3.0313757039420755, "percentage": 60.77, "elapsed_time": "1:43:59", "remaining_time": "1:07:07", "throughput": "499.73", "total_tokens": 3118192} | |
| {"current_steps": 472, "total_steps": 775, "loss": 0.0014, "learning_rate": 1.6982244110243626e-06, "epoch": 3.0378117457763474, "percentage": 60.9, "elapsed_time": "1:44:13", "remaining_time": "1:06:54", "throughput": "499.70", "total_tokens": 3124640} | |
| {"current_steps": 473, "total_steps": 775, "loss": 0.0185, "learning_rate": 1.6885068677185989e-06, "epoch": 3.0442477876106193, "percentage": 61.03, "elapsed_time": "1:44:26", "remaining_time": "1:06:40", "throughput": "499.66", "total_tokens": 3130992} | |
| {"current_steps": 474, "total_steps": 775, "loss": 0.0075, "learning_rate": 1.678803009926841e-06, "epoch": 3.0506838294448912, "percentage": 61.16, "elapsed_time": "1:44:39", "remaining_time": "1:06:27", "throughput": "499.68", "total_tokens": 3137696} | |
| {"current_steps": 475, "total_steps": 775, "loss": 0.0069, "learning_rate": 1.6691130013008514e-06, "epoch": 3.057119871279163, "percentage": 61.29, "elapsed_time": "1:44:52", "remaining_time": "1:06:14", "throughput": "499.72", "total_tokens": 3144560} | |
| {"current_steps": 476, "total_steps": 775, "loss": 0.0009, "learning_rate": 1.6594370052588328e-06, "epoch": 3.063555913113435, "percentage": 61.42, "elapsed_time": "1:45:05", "remaining_time": "1:06:01", "throughput": "499.71", "total_tokens": 3151072} | |
| {"current_steps": 477, "total_steps": 775, "loss": 0.0015, "learning_rate": 1.6497751849826692e-06, "epoch": 3.069991954947707, "percentage": 61.55, "elapsed_time": "1:45:19", "remaining_time": "1:05:47", "throughput": "499.76", "total_tokens": 3158016} | |
| {"current_steps": 478, "total_steps": 775, "loss": 0.0007, "learning_rate": 1.6401277034151798e-06, "epoch": 3.076427996781979, "percentage": 61.68, "elapsed_time": "1:45:32", "remaining_time": "1:05:34", "throughput": "499.75", "total_tokens": 3164560} | |
| {"current_steps": 479, "total_steps": 775, "loss": 0.0012, "learning_rate": 1.630494723257363e-06, "epoch": 3.082864038616251, "percentage": 61.81, "elapsed_time": "1:45:45", "remaining_time": "1:05:21", "throughput": "499.74", "total_tokens": 3171088} | |
| {"current_steps": 480, "total_steps": 775, "loss": 0.0018, "learning_rate": 1.620876406965658e-06, "epoch": 3.089300080450523, "percentage": 61.94, "elapsed_time": "1:45:58", "remaining_time": "1:05:07", "throughput": "499.79", "total_tokens": 3177952} | |
| {"current_steps": 481, "total_steps": 775, "loss": 0.0003, "learning_rate": 1.611272916749205e-06, "epoch": 3.0957361222847948, "percentage": 62.06, "elapsed_time": "1:46:11", "remaining_time": "1:04:54", "throughput": "499.79", "total_tokens": 3184592} | |
| {"current_steps": 482, "total_steps": 775, "loss": 0.0044, "learning_rate": 1.6016844145671062e-06, "epoch": 3.1021721641190667, "percentage": 62.19, "elapsed_time": "1:46:25", "remaining_time": "1:04:41", "throughput": "499.75", "total_tokens": 3190896} | |
| {"current_steps": 483, "total_steps": 775, "loss": 0.0019, "learning_rate": 1.5921110621256972e-06, "epoch": 3.1086082059533386, "percentage": 62.32, "elapsed_time": "1:46:38", "remaining_time": "1:04:28", "throughput": "499.73", "total_tokens": 3197376} | |
| {"current_steps": 484, "total_steps": 775, "loss": 0.0059, "learning_rate": 1.58255302087582e-06, "epoch": 3.1150442477876106, "percentage": 62.45, "elapsed_time": "1:46:51", "remaining_time": "1:04:14", "throughput": "499.70", "total_tokens": 3203776} | |
| {"current_steps": 485, "total_steps": 775, "loss": 0.0036, "learning_rate": 1.5730104520100984e-06, "epoch": 3.1214802896218825, "percentage": 62.58, "elapsed_time": "1:47:04", "remaining_time": "1:04:01", "throughput": "499.71", "total_tokens": 3210464} | |
| {"current_steps": 486, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.56348351646022e-06, "epoch": 3.1279163314561544, "percentage": 62.71, "elapsed_time": "1:47:17", "remaining_time": "1:03:48", "throughput": "499.71", "total_tokens": 3217056} | |
| {"current_steps": 487, "total_steps": 775, "loss": 0.0019, "learning_rate": 1.5539723748942246e-06, "epoch": 3.1343523732904264, "percentage": 62.84, "elapsed_time": "1:47:31", "remaining_time": "1:03:34", "throughput": "499.74", "total_tokens": 3223840} | |
| {"current_steps": 488, "total_steps": 775, "loss": 0.0009, "learning_rate": 1.544477187713791e-06, "epoch": 3.1407884151246983, "percentage": 62.97, "elapsed_time": "1:47:44", "remaining_time": "1:03:21", "throughput": "499.76", "total_tokens": 3230592} | |
| {"current_steps": 489, "total_steps": 775, "loss": 0.0318, "learning_rate": 1.534998115051533e-06, "epoch": 3.1472244569589702, "percentage": 63.1, "elapsed_time": "1:47:57", "remaining_time": "1:03:08", "throughput": "499.77", "total_tokens": 3237216} | |
| {"current_steps": 490, "total_steps": 775, "loss": 0.0216, "learning_rate": 1.5255353167683017e-06, "epoch": 3.153660498793242, "percentage": 63.23, "elapsed_time": "1:48:10", "remaining_time": "1:02:55", "throughput": "499.79", "total_tokens": 3243920} | |
| {"current_steps": 491, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.5160889524504857e-06, "epoch": 3.160096540627514, "percentage": 63.35, "elapsed_time": "1:48:23", "remaining_time": "1:02:41", "throughput": "499.81", "total_tokens": 3250656} | |
| {"current_steps": 492, "total_steps": 775, "loss": 0.0286, "learning_rate": 1.50665918140732e-06, "epoch": 3.166532582461786, "percentage": 63.48, "elapsed_time": "1:48:37", "remaining_time": "1:02:28", "throughput": "499.81", "total_tokens": 3257312} | |
| {"current_steps": 493, "total_steps": 775, "loss": 0.0254, "learning_rate": 1.4972461626682033e-06, "epoch": 3.172968624296058, "percentage": 63.61, "elapsed_time": "1:48:50", "remaining_time": "1:02:15", "throughput": "499.85", "total_tokens": 3264112} | |
| {"current_steps": 494, "total_steps": 775, "loss": 0.0039, "learning_rate": 1.4878500549800115e-06, "epoch": 3.17940466613033, "percentage": 63.74, "elapsed_time": "1:49:03", "remaining_time": "1:02:02", "throughput": "499.82", "total_tokens": 3270528} | |
| {"current_steps": 495, "total_steps": 775, "loss": 0.0005, "learning_rate": 1.4784710168044215e-06, "epoch": 3.185840707964602, "percentage": 63.87, "elapsed_time": "1:49:16", "remaining_time": "1:01:48", "throughput": "499.80", "total_tokens": 3277008} | |
| {"current_steps": 496, "total_steps": 775, "loss": 0.0196, "learning_rate": 1.4691092063152417e-06, "epoch": 3.1922767497988738, "percentage": 64.0, "elapsed_time": "1:49:29", "remaining_time": "1:01:35", "throughput": "499.76", "total_tokens": 3283376} | |
| {"current_steps": 497, "total_steps": 775, "loss": 0.0015, "learning_rate": 1.459764781395741e-06, "epoch": 3.1987127916331457, "percentage": 64.13, "elapsed_time": "1:49:43", "remaining_time": "1:01:22", "throughput": "499.72", "total_tokens": 3289664} | |
| {"current_steps": 498, "total_steps": 775, "loss": 0.0088, "learning_rate": 1.4504378996359867e-06, "epoch": 3.2051488334674176, "percentage": 64.26, "elapsed_time": "1:49:56", "remaining_time": "1:01:08", "throughput": "499.77", "total_tokens": 3296576} | |
| {"current_steps": 499, "total_steps": 775, "loss": 0.0023, "learning_rate": 1.4411287183301902e-06, "epoch": 3.2115848753016896, "percentage": 64.39, "elapsed_time": "1:50:09", "remaining_time": "1:00:55", "throughput": "499.76", "total_tokens": 3303120} | |
| {"current_steps": 500, "total_steps": 775, "loss": 0.0071, "learning_rate": 1.4318373944740485e-06, "epoch": 3.2180209171359615, "percentage": 64.52, "elapsed_time": "1:50:22", "remaining_time": "1:00:42", "throughput": "499.86", "total_tokens": 3310384} | |
| {"current_steps": 501, "total_steps": 775, "loss": 0.0006, "learning_rate": 1.4225640847621006e-06, "epoch": 3.2244569589702334, "percentage": 64.65, "elapsed_time": "1:50:35", "remaining_time": "1:00:29", "throughput": "499.83", "total_tokens": 3316768} | |
| {"current_steps": 502, "total_steps": 775, "loss": 0.0011, "learning_rate": 1.4133089455850878e-06, "epoch": 3.2308930008045054, "percentage": 64.77, "elapsed_time": "1:50:49", "remaining_time": "1:00:15", "throughput": "499.85", "total_tokens": 3323488} | |
| {"current_steps": 503, "total_steps": 775, "loss": 0.0057, "learning_rate": 1.4040721330273063e-06, "epoch": 3.2373290426387773, "percentage": 64.9, "elapsed_time": "1:51:02", "remaining_time": "1:00:02", "throughput": "499.83", "total_tokens": 3330000} | |
| {"current_steps": 504, "total_steps": 775, "loss": 0.0006, "learning_rate": 1.3948538028639851e-06, "epoch": 3.2437650844730492, "percentage": 65.03, "elapsed_time": "1:51:15", "remaining_time": "0:59:49", "throughput": "499.83", "total_tokens": 3336592} | |
| {"current_steps": 505, "total_steps": 775, "loss": 0.0066, "learning_rate": 1.3856541105586545e-06, "epoch": 3.250201126307321, "percentage": 65.16, "elapsed_time": "1:51:28", "remaining_time": "0:59:36", "throughput": "499.82", "total_tokens": 3343136} | |
| {"current_steps": 506, "total_steps": 775, "loss": 0.0079, "learning_rate": 1.3764732112605223e-06, "epoch": 3.256637168141593, "percentage": 65.29, "elapsed_time": "1:51:41", "remaining_time": "0:59:22", "throughput": "499.82", "total_tokens": 3349680} | |
| {"current_steps": 507, "total_steps": 775, "loss": 0.0215, "learning_rate": 1.367311259801863e-06, "epoch": 3.263073209975865, "percentage": 65.42, "elapsed_time": "1:51:55", "remaining_time": "0:59:09", "throughput": "499.82", "total_tokens": 3356304} | |
| {"current_steps": 508, "total_steps": 775, "loss": 0.0031, "learning_rate": 1.3581684106953987e-06, "epoch": 3.2695092518101365, "percentage": 65.55, "elapsed_time": "1:52:08", "remaining_time": "0:58:56", "throughput": "499.84", "total_tokens": 3363008} | |
| {"current_steps": 509, "total_steps": 775, "loss": 0.0024, "learning_rate": 1.3490448181317025e-06, "epoch": 3.2759452936444085, "percentage": 65.68, "elapsed_time": "1:52:21", "remaining_time": "0:58:43", "throughput": "499.85", "total_tokens": 3369728} | |
| {"current_steps": 510, "total_steps": 775, "loss": 0.0094, "learning_rate": 1.3399406359765921e-06, "epoch": 3.2823813354786804, "percentage": 65.81, "elapsed_time": "1:52:34", "remaining_time": "0:58:29", "throughput": "499.80", "total_tokens": 3375968} | |
| {"current_steps": 511, "total_steps": 775, "loss": 0.0054, "learning_rate": 1.3308560177685334e-06, "epoch": 3.2888173773129523, "percentage": 65.94, "elapsed_time": "1:52:47", "remaining_time": "0:58:16", "throughput": "499.87", "total_tokens": 3383024} | |
| {"current_steps": 512, "total_steps": 775, "loss": 0.0008, "learning_rate": 1.3217911167160575e-06, "epoch": 3.2952534191472242, "percentage": 66.06, "elapsed_time": "1:53:01", "remaining_time": "0:58:03", "throughput": "499.85", "total_tokens": 3389488} | |
| {"current_steps": 513, "total_steps": 775, "loss": 0.0053, "learning_rate": 1.3127460856951724e-06, "epoch": 3.301689460981496, "percentage": 66.19, "elapsed_time": "1:53:14", "remaining_time": "0:57:49", "throughput": "499.79", "total_tokens": 3395712} | |
| {"current_steps": 514, "total_steps": 775, "loss": 0.006, "learning_rate": 1.303721077246784e-06, "epoch": 3.308125502815768, "percentage": 66.32, "elapsed_time": "1:53:27", "remaining_time": "0:57:36", "throughput": "499.80", "total_tokens": 3402384} | |
| {"current_steps": 515, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.2947162435741278e-06, "epoch": 3.31456154465004, "percentage": 66.45, "elapsed_time": "1:53:40", "remaining_time": "0:57:23", "throughput": "499.82", "total_tokens": 3409136} | |
| {"current_steps": 516, "total_steps": 775, "loss": 0.0135, "learning_rate": 1.2857317365401997e-06, "epoch": 3.320997586484312, "percentage": 66.58, "elapsed_time": "1:53:53", "remaining_time": "0:57:10", "throughput": "499.83", "total_tokens": 3415776} | |
| {"current_steps": 517, "total_steps": 775, "loss": 0.0083, "learning_rate": 1.2767677076651913e-06, "epoch": 3.327433628318584, "percentage": 66.71, "elapsed_time": "1:54:07", "remaining_time": "0:56:56", "throughput": "499.85", "total_tokens": 3422496} | |
| {"current_steps": 518, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.2678243081239421e-06, "epoch": 3.333869670152856, "percentage": 66.84, "elapsed_time": "1:54:20", "remaining_time": "0:56:43", "throughput": "499.88", "total_tokens": 3429312} | |
| {"current_steps": 519, "total_steps": 775, "loss": 0.0038, "learning_rate": 1.2589016887433846e-06, "epoch": 3.340305711987128, "percentage": 66.97, "elapsed_time": "1:54:33", "remaining_time": "0:56:30", "throughput": "499.87", "total_tokens": 3435840} | |
| {"current_steps": 520, "total_steps": 775, "loss": 0.0269, "learning_rate": 1.2500000000000007e-06, "epoch": 3.3467417538213997, "percentage": 67.1, "elapsed_time": "1:54:46", "remaining_time": "0:56:17", "throughput": "499.83", "total_tokens": 3442176} | |
| {"current_steps": 521, "total_steps": 775, "loss": 0.0062, "learning_rate": 1.2411193920172866e-06, "epoch": 3.3531777956556716, "percentage": 67.23, "elapsed_time": "1:54:59", "remaining_time": "0:56:03", "throughput": "499.83", "total_tokens": 3448784} | |
| {"current_steps": 522, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.2322600145632204e-06, "epoch": 3.3596138374899436, "percentage": 67.35, "elapsed_time": "1:55:13", "remaining_time": "0:55:50", "throughput": "499.80", "total_tokens": 3455184} | |
| {"current_steps": 523, "total_steps": 775, "loss": 0.0011, "learning_rate": 1.2234220170477332e-06, "epoch": 3.3660498793242155, "percentage": 67.48, "elapsed_time": "1:55:26", "remaining_time": "0:55:37", "throughput": "499.80", "total_tokens": 3461792} | |
| {"current_steps": 524, "total_steps": 775, "loss": 0.0026, "learning_rate": 1.2146055485201943e-06, "epoch": 3.3724859211584874, "percentage": 67.61, "elapsed_time": "1:55:39", "remaining_time": "0:55:24", "throughput": "499.84", "total_tokens": 3468624} | |
| {"current_steps": 525, "total_steps": 775, "loss": 0.0183, "learning_rate": 1.205810757666894e-06, "epoch": 3.3789219629927594, "percentage": 67.74, "elapsed_time": "1:55:52", "remaining_time": "0:55:10", "throughput": "499.80", "total_tokens": 3474976} | |
| {"current_steps": 526, "total_steps": 775, "loss": 0.0079, "learning_rate": 1.1970377928085372e-06, "epoch": 3.3853580048270313, "percentage": 67.87, "elapsed_time": "1:56:05", "remaining_time": "0:54:57", "throughput": "499.77", "total_tokens": 3481360} | |
| {"current_steps": 527, "total_steps": 775, "loss": 0.0123, "learning_rate": 1.188286801897743e-06, "epoch": 3.3917940466613032, "percentage": 68.0, "elapsed_time": "1:56:19", "remaining_time": "0:54:44", "throughput": "499.76", "total_tokens": 3487904} | |
| {"current_steps": 528, "total_steps": 775, "loss": 0.0018, "learning_rate": 1.1795579325165448e-06, "epoch": 3.398230088495575, "percentage": 68.13, "elapsed_time": "1:56:32", "remaining_time": "0:54:31", "throughput": "499.74", "total_tokens": 3494368} | |
| {"current_steps": 529, "total_steps": 775, "loss": 0.0014, "learning_rate": 1.1708513318739096e-06, "epoch": 3.404666130329847, "percentage": 68.26, "elapsed_time": "1:56:45", "remaining_time": "0:54:17", "throughput": "499.71", "total_tokens": 3500704} | |
| {"current_steps": 530, "total_steps": 775, "loss": 0.0009, "learning_rate": 1.1621671468032495e-06, "epoch": 3.411102172164119, "percentage": 68.39, "elapsed_time": "1:56:58", "remaining_time": "0:54:04", "throughput": "499.69", "total_tokens": 3507216} | |
| {"current_steps": 531, "total_steps": 775, "loss": 0.0078, "learning_rate": 1.153505523759944e-06, "epoch": 3.417538213998391, "percentage": 68.52, "elapsed_time": "1:57:11", "remaining_time": "0:53:51", "throughput": "499.67", "total_tokens": 3513664} | |
| {"current_steps": 532, "total_steps": 775, "loss": 0.0027, "learning_rate": 1.1448666088188766e-06, "epoch": 3.423974255832663, "percentage": 68.65, "elapsed_time": "1:57:25", "remaining_time": "0:53:37", "throughput": "499.65", "total_tokens": 3520096} | |
| {"current_steps": 533, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.1362505476719662e-06, "epoch": 3.430410297666935, "percentage": 68.77, "elapsed_time": "1:57:38", "remaining_time": "0:53:24", "throughput": "499.63", "total_tokens": 3526560} | |
| {"current_steps": 534, "total_steps": 775, "loss": 0.0064, "learning_rate": 1.1276574856257097e-06, "epoch": 3.4368463395012068, "percentage": 68.9, "elapsed_time": "1:57:51", "remaining_time": "0:53:11", "throughput": "499.68", "total_tokens": 3533536} | |
| {"current_steps": 535, "total_steps": 775, "loss": 0.007, "learning_rate": 1.1190875675987355e-06, "epoch": 3.4432823813354787, "percentage": 69.03, "elapsed_time": "1:58:04", "remaining_time": "0:52:58", "throughput": "499.71", "total_tokens": 3540288} | |
| {"current_steps": 536, "total_steps": 775, "loss": 0.0005, "learning_rate": 1.1105409381193572e-06, "epoch": 3.4497184231697506, "percentage": 69.16, "elapsed_time": "1:58:17", "remaining_time": "0:52:44", "throughput": "499.68", "total_tokens": 3546720} | |
| {"current_steps": 537, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.1020177413231334e-06, "epoch": 3.4561544650040226, "percentage": 69.29, "elapsed_time": "1:58:31", "remaining_time": "0:52:31", "throughput": "499.68", "total_tokens": 3553280} | |
| {"current_steps": 538, "total_steps": 775, "loss": 0.0294, "learning_rate": 1.0935181209504422e-06, "epoch": 3.4625905068382945, "percentage": 69.42, "elapsed_time": "1:58:44", "remaining_time": "0:52:18", "throughput": "499.66", "total_tokens": 3559776} | |
| {"current_steps": 539, "total_steps": 775, "loss": 0.0036, "learning_rate": 1.0850422203440555e-06, "epoch": 3.4690265486725664, "percentage": 69.55, "elapsed_time": "1:58:57", "remaining_time": "0:52:05", "throughput": "499.73", "total_tokens": 3566848} | |
| {"current_steps": 540, "total_steps": 775, "loss": 0.0051, "learning_rate": 1.0765901824467167e-06, "epoch": 3.4754625905068384, "percentage": 69.68, "elapsed_time": "1:59:10", "remaining_time": "0:51:51", "throughput": "499.71", "total_tokens": 3573280} | |
| {"current_steps": 541, "total_steps": 775, "loss": 0.0012, "learning_rate": 1.068162149798737e-06, "epoch": 3.4818986323411103, "percentage": 69.81, "elapsed_time": "1:59:23", "remaining_time": "0:51:38", "throughput": "499.69", "total_tokens": 3579712} | |
| {"current_steps": 542, "total_steps": 775, "loss": 0.0231, "learning_rate": 1.0597582645355891e-06, "epoch": 3.4883346741753822, "percentage": 69.94, "elapsed_time": "1:59:37", "remaining_time": "0:51:25", "throughput": "499.71", "total_tokens": 3586480} | |
| {"current_steps": 543, "total_steps": 775, "loss": 0.0041, "learning_rate": 1.0513786683855062e-06, "epoch": 3.494770716009654, "percentage": 70.06, "elapsed_time": "1:59:50", "remaining_time": "0:51:12", "throughput": "499.72", "total_tokens": 3593136} | |
| {"current_steps": 544, "total_steps": 775, "loss": 0.0124, "learning_rate": 1.0430235026670979e-06, "epoch": 3.501206757843926, "percentage": 70.19, "elapsed_time": "2:00:03", "remaining_time": "0:50:58", "throughput": "499.75", "total_tokens": 3599968} | |
| {"current_steps": 545, "total_steps": 775, "loss": 0.0171, "learning_rate": 1.034692908286964e-06, "epoch": 3.507642799678198, "percentage": 70.32, "elapsed_time": "2:00:16", "remaining_time": "0:50:45", "throughput": "499.75", "total_tokens": 3606592} | |
| {"current_steps": 546, "total_steps": 775, "loss": 0.008, "learning_rate": 1.0263870257373162e-06, "epoch": 3.51407884151247, "percentage": 70.45, "elapsed_time": "2:00:29", "remaining_time": "0:50:32", "throughput": "499.74", "total_tokens": 3613072} | |
| {"current_steps": 547, "total_steps": 775, "loss": 0.0035, "learning_rate": 1.0181059950936131e-06, "epoch": 3.520514883346742, "percentage": 70.58, "elapsed_time": "2:00:43", "remaining_time": "0:50:19", "throughput": "499.74", "total_tokens": 3619696} | |
| {"current_steps": 548, "total_steps": 775, "loss": 0.0006, "learning_rate": 1.0098499560121943e-06, "epoch": 3.526950925181014, "percentage": 70.71, "elapsed_time": "2:00:56", "remaining_time": "0:50:05", "throughput": "499.73", "total_tokens": 3626240} | |
| {"current_steps": 549, "total_steps": 775, "loss": 0.002, "learning_rate": 1.0016190477279274e-06, "epoch": 3.5333869670152858, "percentage": 70.84, "elapsed_time": "2:01:09", "remaining_time": "0:49:52", "throughput": "499.71", "total_tokens": 3632704} | |
| {"current_steps": 550, "total_steps": 775, "loss": 0.0003, "learning_rate": 9.934134090518593e-07, "epoch": 3.5398230088495577, "percentage": 70.97, "elapsed_time": "2:01:22", "remaining_time": "0:49:39", "throughput": "499.72", "total_tokens": 3639360} | |
| {"current_steps": 551, "total_steps": 775, "loss": 0.0004, "learning_rate": 9.852331783688722e-07, "epoch": 3.5462590506838296, "percentage": 71.1, "elapsed_time": "2:01:35", "remaining_time": "0:49:26", "throughput": "499.74", "total_tokens": 3646112} | |
| {"current_steps": 552, "total_steps": 775, "loss": 0.0016, "learning_rate": 9.770784936353555e-07, "epoch": 3.5526950925181016, "percentage": 71.23, "elapsed_time": "2:01:49", "remaining_time": "0:49:12", "throughput": "499.74", "total_tokens": 3652704} | |
| {"current_steps": 553, "total_steps": 775, "loss": 0.0012, "learning_rate": 9.689494923768756e-07, "epoch": 3.5591311343523735, "percentage": 71.35, "elapsed_time": "2:02:02", "remaining_time": "0:48:59", "throughput": "499.80", "total_tokens": 3659696} | |
| {"current_steps": 554, "total_steps": 775, "loss": 0.0057, "learning_rate": 9.608463116858544e-07, "epoch": 3.5655671761866454, "percentage": 71.48, "elapsed_time": "2:02:15", "remaining_time": "0:48:46", "throughput": "499.80", "total_tokens": 3666288} | |
| {"current_steps": 555, "total_steps": 775, "loss": 0.0168, "learning_rate": 9.527690882192636e-07, "epoch": 3.5720032180209174, "percentage": 71.61, "elapsed_time": "2:02:28", "remaining_time": "0:48:33", "throughput": "499.83", "total_tokens": 3673104} | |
| {"current_steps": 556, "total_steps": 775, "loss": 0.0012, "learning_rate": 9.447179581963156e-07, "epoch": 3.5784392598551893, "percentage": 71.74, "elapsed_time": "2:02:41", "remaining_time": "0:48:19", "throughput": "499.85", "total_tokens": 3679872} | |
| {"current_steps": 557, "total_steps": 775, "loss": 0.0002, "learning_rate": 9.366930573961649e-07, "epoch": 3.5848753016894612, "percentage": 71.87, "elapsed_time": "2:02:55", "remaining_time": "0:48:06", "throughput": "499.82", "total_tokens": 3686288} | |
| {"current_steps": 558, "total_steps": 775, "loss": 0.0025, "learning_rate": 9.286945211556231e-07, "epoch": 3.591311343523733, "percentage": 72.0, "elapsed_time": "2:03:08", "remaining_time": "0:47:53", "throughput": "499.84", "total_tokens": 3692976} | |
| {"current_steps": 559, "total_steps": 775, "loss": 0.0194, "learning_rate": 9.207224843668733e-07, "epoch": 3.597747385358005, "percentage": 72.13, "elapsed_time": "2:03:21", "remaining_time": "0:47:39", "throughput": "499.80", "total_tokens": 3699312} | |
| {"current_steps": 560, "total_steps": 775, "loss": 0.0055, "learning_rate": 9.127770814751933e-07, "epoch": 3.604183427192277, "percentage": 72.26, "elapsed_time": "2:03:34", "remaining_time": "0:47:26", "throughput": "499.80", "total_tokens": 3705888} | |
| {"current_steps": 561, "total_steps": 775, "loss": 0.0215, "learning_rate": 9.048584464766938e-07, "epoch": 3.6106194690265485, "percentage": 72.39, "elapsed_time": "2:03:47", "remaining_time": "0:47:13", "throughput": "499.83", "total_tokens": 3712688} | |
| {"current_steps": 562, "total_steps": 775, "loss": 0.0003, "learning_rate": 8.969667129160547e-07, "epoch": 3.6170555108608204, "percentage": 72.52, "elapsed_time": "2:04:01", "remaining_time": "0:47:00", "throughput": "499.81", "total_tokens": 3719168} | |
| {"current_steps": 563, "total_steps": 775, "loss": 0.0242, "learning_rate": 8.891020138842718e-07, "epoch": 3.6234915526950924, "percentage": 72.65, "elapsed_time": "2:04:14", "remaining_time": "0:46:46", "throughput": "499.85", "total_tokens": 3726048} | |
| {"current_steps": 564, "total_steps": 775, "loss": 0.0206, "learning_rate": 8.81264482016416e-07, "epoch": 3.6299275945293643, "percentage": 72.77, "elapsed_time": "2:04:27", "remaining_time": "0:46:33", "throughput": "499.85", "total_tokens": 3732672} | |
| {"current_steps": 565, "total_steps": 775, "loss": 0.0024, "learning_rate": 8.734542494893955e-07, "epoch": 3.6363636363636362, "percentage": 72.9, "elapsed_time": "2:04:40", "remaining_time": "0:46:20", "throughput": "499.87", "total_tokens": 3739456} | |
| {"current_steps": 566, "total_steps": 775, "loss": 0.0087, "learning_rate": 8.65671448019722e-07, "epoch": 3.642799678197908, "percentage": 73.03, "elapsed_time": "2:04:54", "remaining_time": "0:46:07", "throughput": "499.89", "total_tokens": 3746160} | |
| {"current_steps": 567, "total_steps": 775, "loss": 0.0089, "learning_rate": 8.579162088612974e-07, "epoch": 3.64923572003218, "percentage": 73.16, "elapsed_time": "2:05:07", "remaining_time": "0:45:53", "throughput": "499.86", "total_tokens": 3752560} | |
| {"current_steps": 568, "total_steps": 775, "loss": 0.0003, "learning_rate": 8.501886628031941e-07, "epoch": 3.655671761866452, "percentage": 73.29, "elapsed_time": "2:05:20", "remaining_time": "0:45:40", "throughput": "499.92", "total_tokens": 3759600} | |
| {"current_steps": 569, "total_steps": 775, "loss": 0.0002, "learning_rate": 8.424889401674505e-07, "epoch": 3.662107803700724, "percentage": 73.42, "elapsed_time": "2:05:33", "remaining_time": "0:45:27", "throughput": "499.90", "total_tokens": 3766096} | |
| {"current_steps": 570, "total_steps": 775, "loss": 0.0009, "learning_rate": 8.348171708068748e-07, "epoch": 3.668543845534996, "percentage": 73.55, "elapsed_time": "2:05:46", "remaining_time": "0:45:14", "throughput": "499.94", "total_tokens": 3772944} | |
| {"current_steps": 571, "total_steps": 775, "loss": 0.0154, "learning_rate": 8.271734841028553e-07, "epoch": 3.674979887369268, "percentage": 73.68, "elapsed_time": "2:06:00", "remaining_time": "0:45:00", "throughput": "499.95", "total_tokens": 3779664} | |
| {"current_steps": 572, "total_steps": 775, "loss": 0.0012, "learning_rate": 8.195580089631733e-07, "epoch": 3.6814159292035398, "percentage": 73.81, "elapsed_time": "2:06:13", "remaining_time": "0:44:47", "throughput": "499.93", "total_tokens": 3786080} | |
| {"current_steps": 573, "total_steps": 775, "loss": 0.0008, "learning_rate": 8.119708738198395e-07, "epoch": 3.6878519710378117, "percentage": 73.94, "elapsed_time": "2:06:26", "remaining_time": "0:44:34", "throughput": "499.94", "total_tokens": 3792768} | |
| {"current_steps": 574, "total_steps": 775, "loss": 0.0081, "learning_rate": 8.04412206626915e-07, "epoch": 3.6942880128720836, "percentage": 74.06, "elapsed_time": "2:06:39", "remaining_time": "0:44:21", "throughput": "499.95", "total_tokens": 3799472} | |
| {"current_steps": 575, "total_steps": 775, "loss": 0.008, "learning_rate": 7.968821348583644e-07, "epoch": 3.7007240547063556, "percentage": 74.19, "elapsed_time": "2:06:52", "remaining_time": "0:44:07", "throughput": "499.94", "total_tokens": 3805984} | |
| {"current_steps": 576, "total_steps": 775, "loss": 0.0073, "learning_rate": 7.89380785505901e-07, "epoch": 3.7071600965406275, "percentage": 74.32, "elapsed_time": "2:07:06", "remaining_time": "0:43:54", "throughput": "500.01", "total_tokens": 3813088} | |
| {"current_steps": 577, "total_steps": 775, "loss": 0.0062, "learning_rate": 7.819082850768433e-07, "epoch": 3.7135961383748994, "percentage": 74.45, "elapsed_time": "2:07:19", "remaining_time": "0:43:41", "throughput": "500.07", "total_tokens": 3820208} | |
| {"current_steps": 578, "total_steps": 775, "loss": 0.0015, "learning_rate": 7.744647595919869e-07, "epoch": 3.7200321802091714, "percentage": 74.58, "elapsed_time": "2:07:32", "remaining_time": "0:43:28", "throughput": "500.07", "total_tokens": 3826800} | |
| {"current_steps": 579, "total_steps": 775, "loss": 0.0268, "learning_rate": 7.670503345834757e-07, "epoch": 3.7264682220434433, "percentage": 74.71, "elapsed_time": "2:07:45", "remaining_time": "0:43:14", "throughput": "500.07", "total_tokens": 3833344} | |
| {"current_steps": 580, "total_steps": 775, "loss": 0.0007, "learning_rate": 7.596651350926837e-07, "epoch": 3.7329042638777152, "percentage": 74.84, "elapsed_time": "2:07:58", "remaining_time": "0:43:01", "throughput": "500.06", "total_tokens": 3839920} | |
| {"current_steps": 581, "total_steps": 775, "loss": 0.0016, "learning_rate": 7.523092856681099e-07, "epoch": 3.739340305711987, "percentage": 74.97, "elapsed_time": "2:08:12", "remaining_time": "0:42:48", "throughput": "500.05", "total_tokens": 3846432} | |
| {"current_steps": 582, "total_steps": 775, "loss": 0.0631, "learning_rate": 7.44982910363276e-07, "epoch": 3.745776347546259, "percentage": 75.1, "elapsed_time": "2:08:25", "remaining_time": "0:42:35", "throughput": "500.07", "total_tokens": 3853216} | |
| {"current_steps": 583, "total_steps": 775, "loss": 0.0128, "learning_rate": 7.376861327346325e-07, "epoch": 3.752212389380531, "percentage": 75.23, "elapsed_time": "2:08:38", "remaining_time": "0:42:21", "throughput": "500.05", "total_tokens": 3859664} | |
| {"current_steps": 584, "total_steps": 775, "loss": 0.0034, "learning_rate": 7.304190758394775e-07, "epoch": 3.758648431214803, "percentage": 75.35, "elapsed_time": "2:08:51", "remaining_time": "0:42:08", "throughput": "500.05", "total_tokens": 3866208} | |
| {"current_steps": 585, "total_steps": 775, "loss": 0.0633, "learning_rate": 7.231818622338824e-07, "epoch": 3.765084473049075, "percentage": 75.48, "elapsed_time": "2:09:04", "remaining_time": "0:41:55", "throughput": "500.04", "total_tokens": 3872736} | |
| {"current_steps": 586, "total_steps": 775, "loss": 0.0007, "learning_rate": 7.159746139706194e-07, "epoch": 3.771520514883347, "percentage": 75.61, "elapsed_time": "2:09:18", "remaining_time": "0:41:42", "throughput": "500.03", "total_tokens": 3879264} | |
| {"current_steps": 587, "total_steps": 775, "loss": 0.0211, "learning_rate": 7.087974525971103e-07, "epoch": 3.7779565567176188, "percentage": 75.74, "elapsed_time": "2:09:31", "remaining_time": "0:41:28", "throughput": "500.01", "total_tokens": 3885744} | |
| {"current_steps": 588, "total_steps": 775, "loss": 0.009, "learning_rate": 7.016504991533727e-07, "epoch": 3.7843925985518907, "percentage": 75.87, "elapsed_time": "2:09:44", "remaining_time": "0:41:15", "throughput": "500.01", "total_tokens": 3892304} | |
| {"current_steps": 589, "total_steps": 775, "loss": 0.0152, "learning_rate": 6.94533874169977e-07, "epoch": 3.7908286403861626, "percentage": 76.0, "elapsed_time": "2:09:57", "remaining_time": "0:41:02", "throughput": "499.99", "total_tokens": 3898768} | |
| {"current_steps": 590, "total_steps": 775, "loss": 0.0022, "learning_rate": 6.874476976660185e-07, "epoch": 3.7972646822204346, "percentage": 76.13, "elapsed_time": "2:10:10", "remaining_time": "0:40:49", "throughput": "499.94", "total_tokens": 3904976} | |
| {"current_steps": 591, "total_steps": 775, "loss": 0.014, "learning_rate": 6.803920891470905e-07, "epoch": 3.8037007240547065, "percentage": 76.26, "elapsed_time": "2:10:24", "remaining_time": "0:40:35", "throughput": "499.91", "total_tokens": 3911360} | |
| {"current_steps": 592, "total_steps": 775, "loss": 0.0085, "learning_rate": 6.733671676032674e-07, "epoch": 3.8101367658889784, "percentage": 76.39, "elapsed_time": "2:10:37", "remaining_time": "0:40:22", "throughput": "499.94", "total_tokens": 3918224} | |
| {"current_steps": 593, "total_steps": 775, "loss": 0.0161, "learning_rate": 6.663730515071019e-07, "epoch": 3.8165728077232504, "percentage": 76.52, "elapsed_time": "2:10:50", "remaining_time": "0:40:09", "throughput": "499.96", "total_tokens": 3924960} | |
| {"current_steps": 594, "total_steps": 775, "loss": 0.0234, "learning_rate": 6.594098588116243e-07, "epoch": 3.823008849557522, "percentage": 76.65, "elapsed_time": "2:11:03", "remaining_time": "0:39:56", "throughput": "499.98", "total_tokens": 3931712} | |
| {"current_steps": 595, "total_steps": 775, "loss": 0.0186, "learning_rate": 6.524777069483526e-07, "epoch": 3.829444891391794, "percentage": 76.77, "elapsed_time": "2:11:16", "remaining_time": "0:39:42", "throughput": "499.98", "total_tokens": 3938304} | |
| {"current_steps": 596, "total_steps": 775, "loss": 0.0199, "learning_rate": 6.455767128253148e-07, "epoch": 3.8358809332260657, "percentage": 76.9, "elapsed_time": "2:11:30", "remaining_time": "0:39:29", "throughput": "500.02", "total_tokens": 3945200} | |
| {"current_steps": 597, "total_steps": 775, "loss": 0.0052, "learning_rate": 6.38706992825075e-07, "epoch": 3.8423169750603376, "percentage": 77.03, "elapsed_time": "2:11:43", "remaining_time": "0:39:16", "throughput": "500.02", "total_tokens": 3951808} | |
| {"current_steps": 598, "total_steps": 775, "loss": 0.0165, "learning_rate": 6.318686628027723e-07, "epoch": 3.8487530168946096, "percentage": 77.16, "elapsed_time": "2:11:56", "remaining_time": "0:39:03", "throughput": "500.03", "total_tokens": 3958480} | |
| {"current_steps": 599, "total_steps": 775, "loss": 0.01, "learning_rate": 6.250618380841661e-07, "epoch": 3.8551890587288815, "percentage": 77.29, "elapsed_time": "2:12:09", "remaining_time": "0:38:49", "throughput": "500.03", "total_tokens": 3965072} | |
| {"current_steps": 600, "total_steps": 775, "loss": 0.0047, "learning_rate": 6.182866334636889e-07, "epoch": 3.8616251005631534, "percentage": 77.42, "elapsed_time": "2:12:22", "remaining_time": "0:38:36", "throughput": "500.04", "total_tokens": 3971808} | |
| {"current_steps": 601, "total_steps": 775, "loss": 0.0039, "learning_rate": 6.115431632025154e-07, "epoch": 3.8680611423974254, "percentage": 77.55, "elapsed_time": "2:12:36", "remaining_time": "0:38:23", "throughput": "500.05", "total_tokens": 3978480} | |
| {"current_steps": 602, "total_steps": 775, "loss": 0.0022, "learning_rate": 6.048315410266326e-07, "epoch": 3.8744971842316973, "percentage": 77.68, "elapsed_time": "2:12:49", "remaining_time": "0:38:10", "throughput": "500.07", "total_tokens": 3985216} | |
| {"current_steps": 603, "total_steps": 775, "loss": 0.0042, "learning_rate": 5.981518801249192e-07, "epoch": 3.8809332260659692, "percentage": 77.81, "elapsed_time": "2:13:02", "remaining_time": "0:37:56", "throughput": "500.07", "total_tokens": 3991792} | |
| {"current_steps": 604, "total_steps": 775, "loss": 0.0076, "learning_rate": 5.915042931472426e-07, "epoch": 3.887369267900241, "percentage": 77.94, "elapsed_time": "2:13:15", "remaining_time": "0:37:43", "throughput": "500.05", "total_tokens": 3998224} | |
| {"current_steps": 605, "total_steps": 775, "loss": 0.0019, "learning_rate": 5.848888922025553e-07, "epoch": 3.893805309734513, "percentage": 78.06, "elapsed_time": "2:13:28", "remaining_time": "0:37:30", "throughput": "500.06", "total_tokens": 4004960} | |
| {"current_steps": 606, "total_steps": 775, "loss": 0.0014, "learning_rate": 5.783057888570034e-07, "epoch": 3.900241351568785, "percentage": 78.19, "elapsed_time": "2:13:42", "remaining_time": "0:37:17", "throughput": "500.11", "total_tokens": 4011984} | |
| {"current_steps": 607, "total_steps": 775, "loss": 0.0022, "learning_rate": 5.717550941320482e-07, "epoch": 3.906677393403057, "percentage": 78.32, "elapsed_time": "2:13:55", "remaining_time": "0:37:03", "throughput": "500.16", "total_tokens": 4018912} | |
| {"current_steps": 608, "total_steps": 775, "loss": 0.0024, "learning_rate": 5.65236918502593e-07, "epoch": 3.913113435237329, "percentage": 78.45, "elapsed_time": "2:14:08", "remaining_time": "0:36:50", "throughput": "500.15", "total_tokens": 4025504} | |
| {"current_steps": 609, "total_steps": 775, "loss": 0.0013, "learning_rate": 5.587513718951165e-07, "epoch": 3.919549477071601, "percentage": 78.58, "elapsed_time": "2:14:21", "remaining_time": "0:36:37", "throughput": "500.11", "total_tokens": 4031776} | |
| {"current_steps": 610, "total_steps": 775, "loss": 0.0064, "learning_rate": 5.522985636858238e-07, "epoch": 3.9259855189058728, "percentage": 78.71, "elapsed_time": "2:14:34", "remaining_time": "0:36:24", "throughput": "500.09", "total_tokens": 4038208} | |
| {"current_steps": 611, "total_steps": 775, "loss": 0.0084, "learning_rate": 5.458786026988005e-07, "epoch": 3.9324215607401447, "percentage": 78.84, "elapsed_time": "2:14:48", "remaining_time": "0:36:10", "throughput": "500.11", "total_tokens": 4044928} | |
| {"current_steps": 612, "total_steps": 775, "loss": 0.0009, "learning_rate": 5.394915972041739e-07, "epoch": 3.9388576025744166, "percentage": 78.97, "elapsed_time": "2:15:01", "remaining_time": "0:35:57", "throughput": "500.11", "total_tokens": 4051552} | |
| {"current_steps": 613, "total_steps": 775, "loss": 0.001, "learning_rate": 5.33137654916292e-07, "epoch": 3.9452936444086886, "percentage": 79.1, "elapsed_time": "2:15:14", "remaining_time": "0:35:44", "throughput": "500.13", "total_tokens": 4058304} | |
| {"current_steps": 614, "total_steps": 775, "loss": 0.0064, "learning_rate": 5.268168829919046e-07, "epoch": 3.9517296862429605, "percentage": 79.23, "elapsed_time": "2:15:27", "remaining_time": "0:35:31", "throughput": "500.10", "total_tokens": 4064720} | |
| {"current_steps": 615, "total_steps": 775, "loss": 0.0033, "learning_rate": 5.205293880283552e-07, "epoch": 3.9581657280772324, "percentage": 79.35, "elapsed_time": "2:15:40", "remaining_time": "0:35:17", "throughput": "500.09", "total_tokens": 4071216} | |
| {"current_steps": 616, "total_steps": 775, "loss": 0.0065, "learning_rate": 5.14275276061785e-07, "epoch": 3.9646017699115044, "percentage": 79.48, "elapsed_time": "2:15:54", "remaining_time": "0:35:04", "throughput": "500.10", "total_tokens": 4077904} | |
| {"current_steps": 617, "total_steps": 775, "loss": 0.0272, "learning_rate": 5.080546525653448e-07, "epoch": 3.9710378117457763, "percentage": 79.61, "elapsed_time": "2:16:07", "remaining_time": "0:34:51", "throughput": "500.12", "total_tokens": 4084656} | |
| {"current_steps": 618, "total_steps": 775, "loss": 0.0015, "learning_rate": 5.018676224474139e-07, "epoch": 3.9774738535800482, "percentage": 79.74, "elapsed_time": "2:16:20", "remaining_time": "0:34:38", "throughput": "500.16", "total_tokens": 4091584} | |
| {"current_steps": 619, "total_steps": 775, "loss": 0.0013, "learning_rate": 4.957142900498335e-07, "epoch": 3.98390989541432, "percentage": 79.87, "elapsed_time": "2:16:33", "remaining_time": "0:34:24", "throughput": "500.23", "total_tokens": 4098768} | |
| {"current_steps": 620, "total_steps": 775, "loss": 0.0148, "learning_rate": 4.895947591461456e-07, "epoch": 3.990345937248592, "percentage": 80.0, "elapsed_time": "2:16:46", "remaining_time": "0:34:11", "throughput": "500.22", "total_tokens": 4105312} | |
| {"current_steps": 621, "total_steps": 775, "loss": 0.0063, "learning_rate": 4.835091329398436e-07, "epoch": 3.996781979082864, "percentage": 80.13, "elapsed_time": "2:17:00", "remaining_time": "0:33:58", "throughput": "500.23", "total_tokens": 4112000} | |
| {"current_steps": 622, "total_steps": 775, "loss": 0.0003, "learning_rate": 4.774575140626317e-07, "epoch": 4.003218020917136, "percentage": 80.26, "elapsed_time": "2:17:13", "remaining_time": "0:33:45", "throughput": "500.24", "total_tokens": 4118624} | |
| {"current_steps": 623, "total_steps": 775, "loss": 0.001, "learning_rate": 4.714400045726919e-07, "epoch": 4.009654062751408, "percentage": 80.39, "elapsed_time": "2:17:26", "remaining_time": "0:33:32", "throughput": "500.26", "total_tokens": 4125408} | |
| {"current_steps": 624, "total_steps": 775, "loss": 0.0001, "learning_rate": 4.6545670595296686e-07, "epoch": 4.01609010458568, "percentage": 80.52, "elapsed_time": "2:17:39", "remaining_time": "0:33:18", "throughput": "500.25", "total_tokens": 4131936} | |
| {"current_steps": 625, "total_steps": 775, "loss": 0.0193, "learning_rate": 4.5950771910944603e-07, "epoch": 4.022526146419952, "percentage": 80.65, "elapsed_time": "2:17:52", "remaining_time": "0:33:05", "throughput": "500.22", "total_tokens": 4138352} | |
| {"current_steps": 626, "total_steps": 775, "loss": 0.0049, "learning_rate": 4.5359314436946275e-07, "epoch": 4.028962188254224, "percentage": 80.77, "elapsed_time": "2:18:06", "remaining_time": "0:32:52", "throughput": "500.19", "total_tokens": 4144672} | |
| {"current_steps": 627, "total_steps": 775, "loss": 0.0015, "learning_rate": 4.4771308148000487e-07, "epoch": 4.035398230088496, "percentage": 80.9, "elapsed_time": "2:18:19", "remaining_time": "0:32:39", "throughput": "500.19", "total_tokens": 4151296} | |
| {"current_steps": 628, "total_steps": 775, "loss": 0.0019, "learning_rate": 4.418676296060323e-07, "epoch": 4.041834271922768, "percentage": 81.03, "elapsed_time": "2:18:32", "remaining_time": "0:32:25", "throughput": "500.17", "total_tokens": 4157696} | |
| {"current_steps": 629, "total_steps": 775, "loss": 0.0034, "learning_rate": 4.3605688732880097e-07, "epoch": 4.0482703137570395, "percentage": 81.16, "elapsed_time": "2:18:45", "remaining_time": "0:32:12", "throughput": "500.17", "total_tokens": 4164352} | |
| {"current_steps": 630, "total_steps": 775, "loss": 0.0003, "learning_rate": 4.302809526442053e-07, "epoch": 4.054706355591311, "percentage": 81.29, "elapsed_time": "2:18:59", "remaining_time": "0:31:59", "throughput": "500.18", "total_tokens": 4170992} | |
| {"current_steps": 631, "total_steps": 775, "loss": 0.0031, "learning_rate": 4.2453992296112384e-07, "epoch": 4.061142397425583, "percentage": 81.42, "elapsed_time": "2:19:12", "remaining_time": "0:31:46", "throughput": "500.21", "total_tokens": 4177888} | |
| {"current_steps": 632, "total_steps": 775, "loss": 0.0004, "learning_rate": 4.188338950997728e-07, "epoch": 4.067578439259855, "percentage": 81.55, "elapsed_time": "2:19:25", "remaining_time": "0:31:32", "throughput": "500.25", "total_tokens": 4184800} | |
| {"current_steps": 633, "total_steps": 775, "loss": 0.0027, "learning_rate": 4.1316296529007955e-07, "epoch": 4.074014481094127, "percentage": 81.68, "elapsed_time": "2:19:38", "remaining_time": "0:31:19", "throughput": "500.22", "total_tokens": 4191136} | |
| {"current_steps": 634, "total_steps": 775, "loss": 0.0047, "learning_rate": 4.075272291700558e-07, "epoch": 4.080450522928399, "percentage": 81.81, "elapsed_time": "2:19:51", "remaining_time": "0:31:06", "throughput": "500.25", "total_tokens": 4197984} | |
| {"current_steps": 635, "total_steps": 775, "loss": 0.0003, "learning_rate": 4.019267817841835e-07, "epoch": 4.086886564762671, "percentage": 81.94, "elapsed_time": "2:20:05", "remaining_time": "0:30:53", "throughput": "500.26", "total_tokens": 4204688} | |
| {"current_steps": 636, "total_steps": 775, "loss": 0.0204, "learning_rate": 3.9636171758181657e-07, "epoch": 4.093322606596943, "percentage": 82.06, "elapsed_time": "2:20:18", "remaining_time": "0:30:39", "throughput": "500.26", "total_tokens": 4211360} | |
| {"current_steps": 637, "total_steps": 775, "loss": 0.0043, "learning_rate": 3.908321304155846e-07, "epoch": 4.099758648431215, "percentage": 82.19, "elapsed_time": "2:20:31", "remaining_time": "0:30:26", "throughput": "500.27", "total_tokens": 4218000} | |
| {"current_steps": 638, "total_steps": 775, "loss": 0.0002, "learning_rate": 3.853381135398093e-07, "epoch": 4.106194690265487, "percentage": 82.32, "elapsed_time": "2:20:44", "remaining_time": "0:30:13", "throughput": "500.26", "total_tokens": 4224544} | |
| {"current_steps": 639, "total_steps": 775, "loss": 0.0009, "learning_rate": 3.798797596089351e-07, "epoch": 4.112630732099759, "percentage": 82.45, "elapsed_time": "2:20:57", "remaining_time": "0:30:00", "throughput": "500.24", "total_tokens": 4230992} | |
| {"current_steps": 640, "total_steps": 775, "loss": 0.0004, "learning_rate": 3.7445716067596506e-07, "epoch": 4.119066773934031, "percentage": 82.58, "elapsed_time": "2:21:11", "remaining_time": "0:29:46", "throughput": "500.27", "total_tokens": 4237808} | |
| {"current_steps": 641, "total_steps": 775, "loss": 0.0003, "learning_rate": 3.6907040819090604e-07, "epoch": 4.125502815768303, "percentage": 82.71, "elapsed_time": "2:21:24", "remaining_time": "0:29:33", "throughput": "500.22", "total_tokens": 4244032} | |
| {"current_steps": 642, "total_steps": 775, "loss": 0.0015, "learning_rate": 3.63719592999231e-07, "epoch": 4.131938857602575, "percentage": 82.84, "elapsed_time": "2:21:37", "remaining_time": "0:29:20", "throughput": "500.22", "total_tokens": 4250640} | |
| {"current_steps": 643, "total_steps": 775, "loss": 0.0012, "learning_rate": 3.5840480534034355e-07, "epoch": 4.1383748994368466, "percentage": 82.97, "elapsed_time": "2:21:50", "remaining_time": "0:29:07", "throughput": "500.25", "total_tokens": 4257440} | |
| {"current_steps": 644, "total_steps": 775, "loss": 0.0041, "learning_rate": 3.5312613484605546e-07, "epoch": 4.1448109412711185, "percentage": 83.1, "elapsed_time": "2:22:03", "remaining_time": "0:28:53", "throughput": "500.23", "total_tokens": 4263936} | |
| {"current_steps": 645, "total_steps": 775, "loss": 0.0087, "learning_rate": 3.4788367053908087e-07, "epoch": 4.15124698310539, "percentage": 83.23, "elapsed_time": "2:22:17", "remaining_time": "0:28:40", "throughput": "500.19", "total_tokens": 4270144} | |
| {"current_steps": 646, "total_steps": 775, "loss": 0.0025, "learning_rate": 3.4267750083152587e-07, "epoch": 4.157683024939662, "percentage": 83.35, "elapsed_time": "2:22:30", "remaining_time": "0:28:27", "throughput": "500.21", "total_tokens": 4276944} | |
| {"current_steps": 647, "total_steps": 775, "loss": 0.0012, "learning_rate": 3.375077135234051e-07, "epoch": 4.164119066773934, "percentage": 83.48, "elapsed_time": "2:22:43", "remaining_time": "0:28:14", "throughput": "500.20", "total_tokens": 4283488} | |
| {"current_steps": 648, "total_steps": 775, "loss": 0.0111, "learning_rate": 3.323743958011588e-07, "epoch": 4.170555108608206, "percentage": 83.61, "elapsed_time": "2:22:56", "remaining_time": "0:28:00", "throughput": "500.19", "total_tokens": 4290000} | |
| {"current_steps": 649, "total_steps": 775, "loss": 0.0026, "learning_rate": 3.2727763423617915e-07, "epoch": 4.176991150442478, "percentage": 83.74, "elapsed_time": "2:23:09", "remaining_time": "0:27:47", "throughput": "500.19", "total_tokens": 4296544} | |
| {"current_steps": 650, "total_steps": 775, "loss": 0.0004, "learning_rate": 3.222175147833556e-07, "epoch": 4.18342719227675, "percentage": 83.87, "elapsed_time": "2:23:23", "remaining_time": "0:27:34", "throughput": "500.18", "total_tokens": 4303056} | |
| {"current_steps": 651, "total_steps": 775, "loss": 0.0002, "learning_rate": 3.171941227796227e-07, "epoch": 4.189863234111022, "percentage": 84.0, "elapsed_time": "2:23:36", "remaining_time": "0:27:21", "throughput": "500.18", "total_tokens": 4309664} | |
| {"current_steps": 652, "total_steps": 775, "loss": 0.0118, "learning_rate": 3.122075429425184e-07, "epoch": 4.196299275945294, "percentage": 84.13, "elapsed_time": "2:23:49", "remaining_time": "0:27:07", "throughput": "500.16", "total_tokens": 4316112} | |
| {"current_steps": 653, "total_steps": 775, "loss": 0.0001, "learning_rate": 3.072578593687606e-07, "epoch": 4.202735317779566, "percentage": 84.26, "elapsed_time": "2:24:02", "remaining_time": "0:26:54", "throughput": "500.17", "total_tokens": 4322800} | |
| {"current_steps": 654, "total_steps": 775, "loss": 0.0151, "learning_rate": 3.0234515553282523e-07, "epoch": 4.209171359613838, "percentage": 84.39, "elapsed_time": "2:24:15", "remaining_time": "0:26:41", "throughput": "500.17", "total_tokens": 4329408} | |
| {"current_steps": 655, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.9746951428553884e-07, "epoch": 4.21560740144811, "percentage": 84.52, "elapsed_time": "2:24:29", "remaining_time": "0:26:28", "throughput": "500.13", "total_tokens": 4335648} | |
| {"current_steps": 656, "total_steps": 775, "loss": 0.0019, "learning_rate": 2.9263101785268253e-07, "epoch": 4.222043443282382, "percentage": 84.65, "elapsed_time": "2:24:42", "remaining_time": "0:26:14", "throughput": "500.13", "total_tokens": 4342256} | |
| {"current_steps": 657, "total_steps": 775, "loss": 0.0009, "learning_rate": 2.8782974783360534e-07, "epoch": 4.228479485116654, "percentage": 84.77, "elapsed_time": "2:24:55", "remaining_time": "0:26:01", "throughput": "500.18", "total_tokens": 4349280} | |
| {"current_steps": 658, "total_steps": 775, "loss": 0.0051, "learning_rate": 2.8306578519984526e-07, "epoch": 4.2349155269509255, "percentage": 84.9, "elapsed_time": "2:25:08", "remaining_time": "0:25:48", "throughput": "500.20", "total_tokens": 4356128} | |
| {"current_steps": 659, "total_steps": 775, "loss": 0.0008, "learning_rate": 2.783392102937682e-07, "epoch": 4.2413515687851975, "percentage": 85.03, "elapsed_time": "2:25:21", "remaining_time": "0:25:35", "throughput": "500.20", "total_tokens": 4362672} | |
| {"current_steps": 660, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.7365010282720954e-07, "epoch": 4.247787610619469, "percentage": 85.16, "elapsed_time": "2:25:35", "remaining_time": "0:25:22", "throughput": "500.21", "total_tokens": 4369440} | |
| {"current_steps": 661, "total_steps": 775, "loss": 0.0043, "learning_rate": 2.6899854188013054e-07, "epoch": 4.254223652453741, "percentage": 85.29, "elapsed_time": "2:25:48", "remaining_time": "0:25:08", "throughput": "500.18", "total_tokens": 4375760} | |
| {"current_steps": 662, "total_steps": 775, "loss": 0.0004, "learning_rate": 2.643846058992866e-07, "epoch": 4.260659694288013, "percentage": 85.42, "elapsed_time": "2:26:01", "remaining_time": "0:24:55", "throughput": "500.23", "total_tokens": 4382768} | |
| {"current_steps": 663, "total_steps": 775, "loss": 0.0092, "learning_rate": 2.5980837269690056e-07, "epoch": 4.267095736122285, "percentage": 85.55, "elapsed_time": "2:26:14", "remaining_time": "0:24:42", "throughput": "500.23", "total_tokens": 4389424} | |
| {"current_steps": 664, "total_steps": 775, "loss": 0.0006, "learning_rate": 2.552699194493549e-07, "epoch": 4.273531777956556, "percentage": 85.68, "elapsed_time": "2:26:27", "remaining_time": "0:24:29", "throughput": "500.22", "total_tokens": 4395904} | |
| {"current_steps": 665, "total_steps": 775, "loss": 0.0049, "learning_rate": 2.507693226958871e-07, "epoch": 4.279967819790828, "percentage": 85.81, "elapsed_time": "2:26:41", "remaining_time": "0:24:15", "throughput": "500.18", "total_tokens": 4402144} | |
| {"current_steps": 666, "total_steps": 775, "loss": 0.0058, "learning_rate": 2.463066583372989e-07, "epoch": 4.2864038616251, "percentage": 85.94, "elapsed_time": "2:26:54", "remaining_time": "0:24:02", "throughput": "500.17", "total_tokens": 4408672} | |
| {"current_steps": 667, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.418820016346779e-07, "epoch": 4.292839903459372, "percentage": 86.06, "elapsed_time": "2:27:07", "remaining_time": "0:23:49", "throughput": "500.14", "total_tokens": 4415040} | |
| {"current_steps": 668, "total_steps": 775, "loss": 0.0063, "learning_rate": 2.3749542720812757e-07, "epoch": 4.299275945293644, "percentage": 86.19, "elapsed_time": "2:27:20", "remaining_time": "0:23:36", "throughput": "500.15", "total_tokens": 4421696} | |
| {"current_steps": 669, "total_steps": 775, "loss": 0.0116, "learning_rate": 2.331470090355084e-07, "epoch": 4.305711987127916, "percentage": 86.32, "elapsed_time": "2:27:34", "remaining_time": "0:23:22", "throughput": "500.12", "total_tokens": 4428096} | |
| {"current_steps": 670, "total_steps": 775, "loss": 0.003, "learning_rate": 2.2883682045119066e-07, "epoch": 4.312148028962188, "percentage": 86.45, "elapsed_time": "2:27:47", "remaining_time": "0:23:09", "throughput": "500.13", "total_tokens": 4434752} | |
| {"current_steps": 671, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.2456493414481778e-07, "epoch": 4.31858407079646, "percentage": 86.58, "elapsed_time": "2:28:00", "remaining_time": "0:22:56", "throughput": "500.16", "total_tokens": 4441584} | |
| {"current_steps": 672, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.2033142216007913e-07, "epoch": 4.325020112630732, "percentage": 86.71, "elapsed_time": "2:28:13", "remaining_time": "0:22:43", "throughput": "500.19", "total_tokens": 4448464} | |
| {"current_steps": 673, "total_steps": 775, "loss": 0.0008, "learning_rate": 2.1613635589349756e-07, "epoch": 4.331456154465004, "percentage": 86.84, "elapsed_time": "2:28:26", "remaining_time": "0:22:29", "throughput": "500.18", "total_tokens": 4454976} | |
| {"current_steps": 674, "total_steps": 775, "loss": 0.0002, "learning_rate": 2.1197980609322406e-07, "epoch": 4.337892196299276, "percentage": 86.97, "elapsed_time": "2:28:40", "remaining_time": "0:22:16", "throughput": "500.16", "total_tokens": 4461440} | |
| {"current_steps": 675, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.07861842857843e-07, "epoch": 4.3443282381335475, "percentage": 87.1, "elapsed_time": "2:28:53", "remaining_time": "0:22:03", "throughput": "500.16", "total_tokens": 4468080} | |
| {"current_steps": 676, "total_steps": 775, "loss": 0.0105, "learning_rate": 2.0378253563519247e-07, "epoch": 4.3507642799678194, "percentage": 87.23, "elapsed_time": "2:29:06", "remaining_time": "0:21:50", "throughput": "500.19", "total_tokens": 4474944} | |
| {"current_steps": 677, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.997419532211925e-07, "epoch": 4.357200321802091, "percentage": 87.35, "elapsed_time": "2:29:19", "remaining_time": "0:21:36", "throughput": "500.18", "total_tokens": 4481456} | |
| {"current_steps": 678, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.9574016375868282e-07, "epoch": 4.363636363636363, "percentage": 87.48, "elapsed_time": "2:29:32", "remaining_time": "0:21:23", "throughput": "500.12", "total_tokens": 4487472} | |
| {"current_steps": 679, "total_steps": 775, "loss": 0.0076, "learning_rate": 1.9177723473627647e-07, "epoch": 4.370072405470635, "percentage": 87.61, "elapsed_time": "2:29:46", "remaining_time": "0:21:10", "throughput": "500.14", "total_tokens": 4494320} | |
| {"current_steps": 680, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.8785323298722098e-07, "epoch": 4.376508447304907, "percentage": 87.74, "elapsed_time": "2:29:59", "remaining_time": "0:20:57", "throughput": "500.17", "total_tokens": 4501152} | |
| {"current_steps": 681, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.839682246882682e-07, "epoch": 4.382944489139179, "percentage": 87.87, "elapsed_time": "2:30:12", "remaining_time": "0:20:44", "throughput": "500.13", "total_tokens": 4507376} | |
| {"current_steps": 682, "total_steps": 775, "loss": 0.002, "learning_rate": 1.801222753585638e-07, "epoch": 4.389380530973451, "percentage": 88.0, "elapsed_time": "2:30:25", "remaining_time": "0:20:30", "throughput": "500.12", "total_tokens": 4513904} | |
| {"current_steps": 683, "total_steps": 775, "loss": 0.0088, "learning_rate": 1.7631544985853623e-07, "epoch": 4.395816572807723, "percentage": 88.13, "elapsed_time": "2:30:38", "remaining_time": "0:20:17", "throughput": "500.13", "total_tokens": 4520608} | |
| {"current_steps": 684, "total_steps": 775, "loss": 0.0023, "learning_rate": 1.725478123888083e-07, "epoch": 4.402252614641995, "percentage": 88.26, "elapsed_time": "2:30:52", "remaining_time": "0:20:04", "throughput": "500.13", "total_tokens": 4527184} | |
| {"current_steps": 685, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.6881942648911077e-07, "epoch": 4.408688656476267, "percentage": 88.39, "elapsed_time": "2:31:05", "remaining_time": "0:19:51", "throughput": "500.12", "total_tokens": 4533696} | |
| {"current_steps": 686, "total_steps": 775, "loss": 0.0017, "learning_rate": 1.6513035503721213e-07, "epoch": 4.415124698310539, "percentage": 88.52, "elapsed_time": "2:31:18", "remaining_time": "0:19:37", "throughput": "500.15", "total_tokens": 4540624} | |
| {"current_steps": 687, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.614806602478583e-07, "epoch": 4.421560740144811, "percentage": 88.65, "elapsed_time": "2:31:31", "remaining_time": "0:19:24", "throughput": "500.13", "total_tokens": 4547056} | |
| {"current_steps": 688, "total_steps": 775, "loss": 0.0073, "learning_rate": 1.5787040367172379e-07, "epoch": 4.427996781979083, "percentage": 88.77, "elapsed_time": "2:31:44", "remaining_time": "0:19:11", "throughput": "500.14", "total_tokens": 4553712} | |
| {"current_steps": 689, "total_steps": 775, "loss": 0.0013, "learning_rate": 1.542996461943716e-07, "epoch": 4.434432823813355, "percentage": 88.9, "elapsed_time": "2:31:58", "remaining_time": "0:18:58", "throughput": "500.11", "total_tokens": 4560080} | |
| {"current_steps": 690, "total_steps": 775, "loss": 0.0003, "learning_rate": 1.507684480352292e-07, "epoch": 4.4408688656476265, "percentage": 89.03, "elapsed_time": "2:32:11", "remaining_time": "0:18:44", "throughput": "500.09", "total_tokens": 4566496} | |
| {"current_steps": 691, "total_steps": 775, "loss": 0.0149, "learning_rate": 1.4727686874657143e-07, "epoch": 4.447304907481898, "percentage": 89.16, "elapsed_time": "2:32:24", "remaining_time": "0:18:31", "throughput": "500.10", "total_tokens": 4573152} | |
| {"current_steps": 692, "total_steps": 775, "loss": 0.0026, "learning_rate": 1.4382496721251526e-07, "epoch": 4.45374094931617, "percentage": 89.29, "elapsed_time": "2:32:37", "remaining_time": "0:18:18", "throughput": "500.17", "total_tokens": 4580432} | |
| {"current_steps": 693, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.4041280164802967e-07, "epoch": 4.460176991150442, "percentage": 89.42, "elapsed_time": "2:32:50", "remaining_time": "0:18:05", "throughput": "500.17", "total_tokens": 4587024} | |
| {"current_steps": 694, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.3704042959795132e-07, "epoch": 4.466613032984714, "percentage": 89.55, "elapsed_time": "2:33:04", "remaining_time": "0:17:51", "throughput": "500.13", "total_tokens": 4593296} | |
| {"current_steps": 695, "total_steps": 775, "loss": 0.006, "learning_rate": 1.3370790793601373e-07, "epoch": 4.473049074818986, "percentage": 89.68, "elapsed_time": "2:33:17", "remaining_time": "0:17:38", "throughput": "500.15", "total_tokens": 4600000} | |
| {"current_steps": 696, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.3041529286389078e-07, "epoch": 4.479485116653258, "percentage": 89.81, "elapsed_time": "2:33:30", "remaining_time": "0:17:25", "throughput": "500.14", "total_tokens": 4606560} | |
| {"current_steps": 697, "total_steps": 775, "loss": 0.0004, "learning_rate": 1.2716263991024712e-07, "epoch": 4.48592115848753, "percentage": 89.94, "elapsed_time": "2:33:43", "remaining_time": "0:17:12", "throughput": "500.17", "total_tokens": 4613424} | |
| {"current_steps": 698, "total_steps": 775, "loss": 0.0007, "learning_rate": 1.2395000392980057e-07, "epoch": 4.492357200321802, "percentage": 90.06, "elapsed_time": "2:33:56", "remaining_time": "0:16:58", "throughput": "500.17", "total_tokens": 4620064} | |
| {"current_steps": 699, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.2077743910239998e-07, "epoch": 4.498793242156074, "percentage": 90.19, "elapsed_time": "2:34:10", "remaining_time": "0:16:45", "throughput": "500.17", "total_tokens": 4626608} | |
| {"current_steps": 700, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.1764499893210879e-07, "epoch": 4.505229283990346, "percentage": 90.32, "elapsed_time": "2:34:23", "remaining_time": "0:16:32", "throughput": "500.17", "total_tokens": 4633280} | |
| {"current_steps": 701, "total_steps": 775, "loss": 0.0033, "learning_rate": 1.145527362463042e-07, "epoch": 4.511665325824618, "percentage": 90.45, "elapsed_time": "2:34:36", "remaining_time": "0:16:19", "throughput": "500.18", "total_tokens": 4639920} | |
| {"current_steps": 702, "total_steps": 775, "loss": 0.0009, "learning_rate": 1.1150070319478679e-07, "epoch": 4.51810136765889, "percentage": 90.58, "elapsed_time": "2:34:49", "remaining_time": "0:16:06", "throughput": "500.20", "total_tokens": 4646736} | |
| {"current_steps": 703, "total_steps": 775, "loss": 0.0216, "learning_rate": 1.0848895124889819e-07, "epoch": 4.524537409493162, "percentage": 90.71, "elapsed_time": "2:35:02", "remaining_time": "0:15:52", "throughput": "500.20", "total_tokens": 4653328} | |
| {"current_steps": 704, "total_steps": 775, "loss": 0.0035, "learning_rate": 1.0551753120065621e-07, "epoch": 4.530973451327434, "percentage": 90.84, "elapsed_time": "2:35:16", "remaining_time": "0:15:39", "throughput": "500.22", "total_tokens": 4660112} | |
| {"current_steps": 705, "total_steps": 775, "loss": 0.0073, "learning_rate": 1.0258649316189722e-07, "epoch": 4.5374094931617055, "percentage": 90.97, "elapsed_time": "2:35:29", "remaining_time": "0:15:26", "throughput": "500.20", "total_tokens": 4666560} | |
| {"current_steps": 706, "total_steps": 775, "loss": 0.0003, "learning_rate": 9.969588656342982e-08, "epoch": 4.543845534995977, "percentage": 91.1, "elapsed_time": "2:35:42", "remaining_time": "0:15:13", "throughput": "500.20", "total_tokens": 4673152} | |
| {"current_steps": 707, "total_steps": 775, "loss": 0.0007, "learning_rate": 9.684576015420277e-08, "epoch": 4.550281576830249, "percentage": 91.23, "elapsed_time": "2:35:55", "remaining_time": "0:14:59", "throughput": "500.16", "total_tokens": 4679360} | |
| {"current_steps": 708, "total_steps": 775, "loss": 0.0003, "learning_rate": 9.403616200048288e-08, "epoch": 4.556717618664521, "percentage": 91.35, "elapsed_time": "2:36:08", "remaining_time": "0:14:46", "throughput": "500.15", "total_tokens": 4685904} | |
| {"current_steps": 709, "total_steps": 775, "loss": 0.0168, "learning_rate": 9.12671394850423e-08, "epoch": 4.563153660498793, "percentage": 91.48, "elapsed_time": "2:36:22", "remaining_time": "0:14:33", "throughput": "500.13", "total_tokens": 4692320} | |
| {"current_steps": 710, "total_steps": 775, "loss": 0.0015, "learning_rate": 8.85387393063622e-08, "epoch": 4.569589702333065, "percentage": 91.61, "elapsed_time": "2:36:35", "remaining_time": "0:14:20", "throughput": "500.13", "total_tokens": 4698928} | |
| {"current_steps": 711, "total_steps": 775, "loss": 0.0001, "learning_rate": 8.585100747784376e-08, "epoch": 4.576025744167337, "percentage": 91.74, "elapsed_time": "2:36:48", "remaining_time": "0:14:06", "throughput": "500.18", "total_tokens": 4706000} | |
| {"current_steps": 712, "total_steps": 775, "loss": 0.0036, "learning_rate": 8.320398932703145e-08, "epoch": 4.582461786001609, "percentage": 91.87, "elapsed_time": "2:37:01", "remaining_time": "0:13:53", "throughput": "500.16", "total_tokens": 4712352} | |
| {"current_steps": 713, "total_steps": 775, "loss": 0.0022, "learning_rate": 8.059772949485068e-08, "epoch": 4.588897827835881, "percentage": 92.0, "elapsed_time": "2:37:14", "remaining_time": "0:13:40", "throughput": "500.14", "total_tokens": 4718848} | |
| {"current_steps": 714, "total_steps": 775, "loss": 0.0085, "learning_rate": 7.803227193485336e-08, "epoch": 4.595333869670153, "percentage": 92.13, "elapsed_time": "2:37:28", "remaining_time": "0:13:27", "throughput": "500.17", "total_tokens": 4725728} | |
| {"current_steps": 715, "total_steps": 775, "loss": 0.0001, "learning_rate": 7.550765991247655e-08, "epoch": 4.601769911504425, "percentage": 92.26, "elapsed_time": "2:37:41", "remaining_time": "0:13:13", "throughput": "500.17", "total_tokens": 4732352} | |
| {"current_steps": 716, "total_steps": 775, "loss": 0.0001, "learning_rate": 7.30239360043139e-08, "epoch": 4.608205953338697, "percentage": 92.39, "elapsed_time": "2:37:54", "remaining_time": "0:13:00", "throughput": "500.18", "total_tokens": 4738992} | |
| {"current_steps": 717, "total_steps": 775, "loss": 0.007, "learning_rate": 7.058114209739675e-08, "epoch": 4.614641995172969, "percentage": 92.52, "elapsed_time": "2:38:07", "remaining_time": "0:12:47", "throughput": "500.21", "total_tokens": 4745872} | |
| {"current_steps": 718, "total_steps": 775, "loss": 0.0002, "learning_rate": 6.817931938848805e-08, "epoch": 4.621078037007241, "percentage": 92.65, "elapsed_time": "2:38:21", "remaining_time": "0:12:34", "throughput": "500.21", "total_tokens": 4752544} | |
| {"current_steps": 719, "total_steps": 775, "loss": 0.0002, "learning_rate": 6.581850838338816e-08, "epoch": 4.627514078841513, "percentage": 92.77, "elapsed_time": "2:38:34", "remaining_time": "0:12:21", "throughput": "500.24", "total_tokens": 4759360} | |
| {"current_steps": 720, "total_steps": 775, "loss": 0.0008, "learning_rate": 6.349874889624963e-08, "epoch": 4.6339501206757845, "percentage": 92.9, "elapsed_time": "2:38:47", "remaining_time": "0:12:07", "throughput": "500.24", "total_tokens": 4766016} | |
| {"current_steps": 721, "total_steps": 775, "loss": 0.0043, "learning_rate": 6.12200800489085e-08, "epoch": 4.640386162510056, "percentage": 93.03, "elapsed_time": "2:39:00", "remaining_time": "0:11:54", "throughput": "500.24", "total_tokens": 4772624} | |
| {"current_steps": 722, "total_steps": 775, "loss": 0.0004, "learning_rate": 5.898254027022293e-08, "epoch": 4.646822204344328, "percentage": 93.16, "elapsed_time": "2:39:13", "remaining_time": "0:11:41", "throughput": "500.22", "total_tokens": 4778960} | |
| {"current_steps": 723, "total_steps": 775, "loss": 0.0024, "learning_rate": 5.678616729542535e-08, "epoch": 4.6532582461786, "percentage": 93.29, "elapsed_time": "2:39:27", "remaining_time": "0:11:28", "throughput": "500.22", "total_tokens": 4785600} | |
| {"current_steps": 724, "total_steps": 775, "loss": 0.0024, "learning_rate": 5.463099816548578e-08, "epoch": 4.659694288012872, "percentage": 93.42, "elapsed_time": "2:39:40", "remaining_time": "0:11:14", "throughput": "500.22", "total_tokens": 4792208} | |
| {"current_steps": 725, "total_steps": 775, "loss": 0.0001, "learning_rate": 5.2517069226488694e-08, "epoch": 4.666130329847144, "percentage": 93.55, "elapsed_time": "2:39:53", "remaining_time": "0:11:01", "throughput": "500.22", "total_tokens": 4798816} | |
| {"current_steps": 726, "total_steps": 775, "loss": 0.0001, "learning_rate": 5.044441612901768e-08, "epoch": 4.672566371681416, "percentage": 93.68, "elapsed_time": "2:40:06", "remaining_time": "0:10:48", "throughput": "500.22", "total_tokens": 4805408} | |
| {"current_steps": 727, "total_steps": 775, "loss": 0.0002, "learning_rate": 4.841307382755567e-08, "epoch": 4.679002413515688, "percentage": 93.81, "elapsed_time": "2:40:19", "remaining_time": "0:10:35", "throughput": "500.27", "total_tokens": 4812480} | |
| {"current_steps": 728, "total_steps": 775, "loss": 0.0, "learning_rate": 4.6423076579895646e-08, "epoch": 4.68543845534996, "percentage": 93.94, "elapsed_time": "2:40:33", "remaining_time": "0:10:21", "throughput": "500.30", "total_tokens": 4819360} | |
| {"current_steps": 729, "total_steps": 775, "loss": 0.0003, "learning_rate": 4.4474457946562245e-08, "epoch": 4.691874497184232, "percentage": 94.06, "elapsed_time": "2:40:46", "remaining_time": "0:10:08", "throughput": "500.29", "total_tokens": 4825904} | |
| {"current_steps": 730, "total_steps": 775, "loss": 0.0004, "learning_rate": 4.256725079024554e-08, "epoch": 4.698310539018504, "percentage": 94.19, "elapsed_time": "2:40:59", "remaining_time": "0:09:55", "throughput": "500.29", "total_tokens": 4832544} | |
| {"current_steps": 731, "total_steps": 775, "loss": 0.001, "learning_rate": 4.070148727524814e-08, "epoch": 4.704746580852776, "percentage": 94.32, "elapsed_time": "2:41:12", "remaining_time": "0:09:42", "throughput": "500.27", "total_tokens": 4838960} | |
| {"current_steps": 732, "total_steps": 775, "loss": 0.0001, "learning_rate": 3.887719886694091e-08, "epoch": 4.711182622687048, "percentage": 94.45, "elapsed_time": "2:41:25", "remaining_time": "0:09:28", "throughput": "500.27", "total_tokens": 4845584} | |
| {"current_steps": 733, "total_steps": 775, "loss": 0.0031, "learning_rate": 3.709441633123367e-08, "epoch": 4.71761866452132, "percentage": 94.58, "elapsed_time": "2:41:39", "remaining_time": "0:09:15", "throughput": "500.30", "total_tokens": 4852416} | |
| {"current_steps": 734, "total_steps": 775, "loss": 0.0006, "learning_rate": 3.535316973405672e-08, "epoch": 4.7240547063555915, "percentage": 94.71, "elapsed_time": "2:41:52", "remaining_time": "0:09:02", "throughput": "500.28", "total_tokens": 4858864} | |
| {"current_steps": 735, "total_steps": 775, "loss": 0.0001, "learning_rate": 3.3653488440851255e-08, "epoch": 4.7304907481898635, "percentage": 94.84, "elapsed_time": "2:42:05", "remaining_time": "0:08:49", "throughput": "500.29", "total_tokens": 4865552} | |
| {"current_steps": 736, "total_steps": 775, "loss": 0.006, "learning_rate": 3.1995401116077516e-08, "epoch": 4.736926790024135, "percentage": 94.97, "elapsed_time": "2:42:18", "remaining_time": "0:08:36", "throughput": "500.27", "total_tokens": 4871984} | |
| {"current_steps": 737, "total_steps": 775, "loss": 0.0001, "learning_rate": 3.037893572272937e-08, "epoch": 4.743362831858407, "percentage": 95.1, "elapsed_time": "2:42:31", "remaining_time": "0:08:22", "throughput": "500.28", "total_tokens": 4878688} | |
| {"current_steps": 738, "total_steps": 775, "loss": 0.001, "learning_rate": 2.8804119521862183e-08, "epoch": 4.749798873692679, "percentage": 95.23, "elapsed_time": "2:42:45", "remaining_time": "0:08:09", "throughput": "500.31", "total_tokens": 4885504} | |
| {"current_steps": 739, "total_steps": 775, "loss": 0.0019, "learning_rate": 2.7270979072135106e-08, "epoch": 4.756234915526951, "percentage": 95.35, "elapsed_time": "2:42:58", "remaining_time": "0:07:56", "throughput": "500.32", "total_tokens": 4892272} | |
| {"current_steps": 740, "total_steps": 775, "loss": 0.0047, "learning_rate": 2.5779540229361744e-08, "epoch": 4.762670957361223, "percentage": 95.48, "elapsed_time": "2:43:11", "remaining_time": "0:07:43", "throughput": "500.32", "total_tokens": 4898848} | |
| {"current_steps": 741, "total_steps": 775, "loss": 0.0002, "learning_rate": 2.4329828146074096e-08, "epoch": 4.769106999195495, "percentage": 95.61, "elapsed_time": "2:43:24", "remaining_time": "0:07:29", "throughput": "500.31", "total_tokens": 4905392} | |
| {"current_steps": 742, "total_steps": 775, "loss": 0.0002, "learning_rate": 2.2921867271099296e-08, "epoch": 4.775543041029767, "percentage": 95.74, "elapsed_time": "2:43:37", "remaining_time": "0:07:16", "throughput": "500.38", "total_tokens": 4912672} | |
| {"current_steps": 743, "total_steps": 775, "loss": 0.0006, "learning_rate": 2.155568134914604e-08, "epoch": 4.781979082864039, "percentage": 95.87, "elapsed_time": "2:43:51", "remaining_time": "0:07:03", "throughput": "500.40", "total_tokens": 4919472} | |
| {"current_steps": 744, "total_steps": 775, "loss": 0.0001, "learning_rate": 2.0231293420405194e-08, "epoch": 4.788415124698311, "percentage": 96.0, "elapsed_time": "2:44:04", "remaining_time": "0:06:50", "throughput": "500.44", "total_tokens": 4926448} | |
| {"current_steps": 745, "total_steps": 775, "loss": 0.0038, "learning_rate": 1.8948725820160663e-08, "epoch": 4.794851166532583, "percentage": 96.13, "elapsed_time": "2:44:17", "remaining_time": "0:06:36", "throughput": "500.45", "total_tokens": 4933152} | |
| {"current_steps": 746, "total_steps": 775, "loss": 0.0088, "learning_rate": 1.770800017841301e-08, "epoch": 4.801287208366855, "percentage": 96.26, "elapsed_time": "2:44:30", "remaining_time": "0:06:23", "throughput": "500.43", "total_tokens": 4939568} | |
| {"current_steps": 747, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.650913741951421e-08, "epoch": 4.807723250201127, "percentage": 96.39, "elapsed_time": "2:44:43", "remaining_time": "0:06:10", "throughput": "500.43", "total_tokens": 4946224} | |
| {"current_steps": 748, "total_steps": 775, "loss": 0.002, "learning_rate": 1.5352157761815978e-08, "epoch": 4.814159292035399, "percentage": 96.52, "elapsed_time": "2:44:57", "remaining_time": "0:05:57", "throughput": "500.44", "total_tokens": 4952880} | |
| {"current_steps": 749, "total_steps": 775, "loss": 0.0012, "learning_rate": 1.4237080717326712e-08, "epoch": 4.8205953338696705, "percentage": 96.65, "elapsed_time": "2:45:10", "remaining_time": "0:05:44", "throughput": "500.52", "total_tokens": 4960288} | |
| {"current_steps": 750, "total_steps": 775, "loss": 0.0038, "learning_rate": 1.3163925091384532e-08, "epoch": 4.8270313757039425, "percentage": 96.77, "elapsed_time": "2:45:23", "remaining_time": "0:05:30", "throughput": "500.49", "total_tokens": 4966656} | |
| {"current_steps": 751, "total_steps": 775, "loss": 0.0003, "learning_rate": 1.2132708982338925e-08, "epoch": 4.833467417538214, "percentage": 96.9, "elapsed_time": "2:45:36", "remaining_time": "0:05:17", "throughput": "500.48", "total_tokens": 4973184} | |
| {"current_steps": 752, "total_steps": 775, "loss": 0.0023, "learning_rate": 1.1143449781245985e-08, "epoch": 4.839903459372486, "percentage": 97.03, "elapsed_time": "2:45:49", "remaining_time": "0:05:04", "throughput": "500.51", "total_tokens": 4980080} | |
| {"current_steps": 753, "total_steps": 775, "loss": 0.0001, "learning_rate": 1.0196164171574762e-08, "epoch": 4.846339501206758, "percentage": 97.16, "elapsed_time": "2:46:03", "remaining_time": "0:04:51", "throughput": "500.56", "total_tokens": 4987104} | |
| {"current_steps": 754, "total_steps": 775, "loss": 0.0044, "learning_rate": 9.290868128926378e-09, "epoch": 4.85277554304103, "percentage": 97.29, "elapsed_time": "2:46:16", "remaining_time": "0:04:37", "throughput": "500.56", "total_tokens": 4993728} | |
| {"current_steps": 755, "total_steps": 775, "loss": 0.004, "learning_rate": 8.427576920763957e-09, "epoch": 4.859211584875302, "percentage": 97.42, "elapsed_time": "2:46:29", "remaining_time": "0:04:24", "throughput": "500.56", "total_tokens": 5000368} | |
| {"current_steps": 756, "total_steps": 775, "loss": 0.0042, "learning_rate": 7.606305106155898e-09, "epoch": 4.865647626709574, "percentage": 97.55, "elapsed_time": "2:46:42", "remaining_time": "0:04:11", "throughput": "500.54", "total_tokens": 5006768} | |
| {"current_steps": 757, "total_steps": 775, "loss": 0.0004, "learning_rate": 6.827066535529947e-09, "epoch": 4.872083668543846, "percentage": 97.68, "elapsed_time": "2:46:55", "remaining_time": "0:03:58", "throughput": "500.53", "total_tokens": 5013264} | |
| {"current_steps": 758, "total_steps": 775, "loss": 0.0005, "learning_rate": 6.089874350439507e-09, "epoch": 4.878519710378118, "percentage": 97.81, "elapsed_time": "2:47:09", "remaining_time": "0:03:44", "throughput": "500.51", "total_tokens": 5019744} | |
| {"current_steps": 759, "total_steps": 775, "loss": 0.0, "learning_rate": 5.394740983341862e-09, "epoch": 4.88495575221239, "percentage": 97.94, "elapsed_time": "2:47:22", "remaining_time": "0:03:31", "throughput": "500.49", "total_tokens": 5026080} | |
| {"current_steps": 760, "total_steps": 775, "loss": 0.0001, "learning_rate": 4.74167815738974e-09, "epoch": 4.891391794046662, "percentage": 98.06, "elapsed_time": "2:47:35", "remaining_time": "0:03:18", "throughput": "500.47", "total_tokens": 5032544} | |
| {"current_steps": 761, "total_steps": 775, "loss": 0.0089, "learning_rate": 4.130696886231744e-09, "epoch": 4.897827835880933, "percentage": 98.19, "elapsed_time": "2:47:48", "remaining_time": "0:03:05", "throughput": "500.48", "total_tokens": 5039248} | |
| {"current_steps": 762, "total_steps": 775, "loss": 0.0118, "learning_rate": 3.561807473827783e-09, "epoch": 4.904263877715205, "percentage": 98.32, "elapsed_time": "2:48:02", "remaining_time": "0:02:52", "throughput": "500.48", "total_tokens": 5045792} | |
| {"current_steps": 763, "total_steps": 775, "loss": 0.0003, "learning_rate": 3.035019514275317e-09, "epoch": 4.910699919549477, "percentage": 98.45, "elapsed_time": "2:48:15", "remaining_time": "0:02:38", "throughput": "500.47", "total_tokens": 5052320} | |
| {"current_steps": 764, "total_steps": 775, "loss": 0.0013, "learning_rate": 2.5503418916464352e-09, "epoch": 4.917135961383749, "percentage": 98.58, "elapsed_time": "2:48:28", "remaining_time": "0:02:25", "throughput": "500.49", "total_tokens": 5059200} | |
| {"current_steps": 765, "total_steps": 775, "loss": 0.0058, "learning_rate": 2.1077827798404728e-09, "epoch": 4.923572003218021, "percentage": 98.71, "elapsed_time": "2:48:41", "remaining_time": "0:02:12", "throughput": "500.50", "total_tokens": 5065824} | |
| {"current_steps": 766, "total_steps": 775, "loss": 0.0002, "learning_rate": 1.707349642442735e-09, "epoch": 4.9300080450522925, "percentage": 98.84, "elapsed_time": "2:48:54", "remaining_time": "0:01:59", "throughput": "500.46", "total_tokens": 5072080} | |
| {"current_steps": 767, "total_steps": 775, "loss": 0.0059, "learning_rate": 1.349049232601818e-09, "epoch": 4.936444086886564, "percentage": 98.97, "elapsed_time": "2:49:08", "remaining_time": "0:01:45", "throughput": "500.43", "total_tokens": 5078336} | |
| {"current_steps": 768, "total_steps": 775, "loss": 0.0068, "learning_rate": 1.0328875929138671e-09, "epoch": 4.942880128720836, "percentage": 99.1, "elapsed_time": "2:49:21", "remaining_time": "0:01:32", "throughput": "500.45", "total_tokens": 5085152} | |
| {"current_steps": 769, "total_steps": 775, "loss": 0.0003, "learning_rate": 7.588700553209926e-10, "epoch": 4.949316170555108, "percentage": 99.23, "elapsed_time": "2:49:34", "remaining_time": "0:01:19", "throughput": "500.48", "total_tokens": 5092080} | |
| {"current_steps": 770, "total_steps": 775, "loss": 0.0251, "learning_rate": 5.270012410216185e-10, "epoch": 4.95575221238938, "percentage": 99.35, "elapsed_time": "2:49:47", "remaining_time": "0:01:06", "throughput": "500.51", "total_tokens": 5098960} | |
| {"current_steps": 771, "total_steps": 775, "loss": 0.0054, "learning_rate": 3.3728506039276686e-10, "epoch": 4.962188254223652, "percentage": 99.48, "elapsed_time": "2:50:00", "remaining_time": "0:00:52", "throughput": "500.51", "total_tokens": 5105648} | |
| {"current_steps": 772, "total_steps": 775, "loss": 0.0007, "learning_rate": 1.8972471292344474e-10, "epoch": 4.968624296057924, "percentage": 99.61, "elapsed_time": "2:50:14", "remaining_time": "0:00:39", "throughput": "500.50", "total_tokens": 5112096} | |
| {"current_steps": 773, "total_steps": 775, "loss": 0.0002, "learning_rate": 8.432268716135338e-11, "epoch": 4.975060337892196, "percentage": 99.74, "elapsed_time": "2:50:27", "remaining_time": "0:00:26", "throughput": "500.52", "total_tokens": 5118960} | |
| {"current_steps": 774, "total_steps": 775, "loss": 0.0032, "learning_rate": 2.108076067014464e-11, "epoch": 4.981496379726468, "percentage": 99.87, "elapsed_time": "2:50:40", "remaining_time": "0:00:13", "throughput": "500.55", "total_tokens": 5125872} | |
| {"current_steps": 775, "total_steps": 775, "loss": 0.0107, "learning_rate": 0.0, "epoch": 4.98793242156074, "percentage": 100.0, "elapsed_time": "2:50:53", "remaining_time": "0:00:00", "throughput": "500.53", "total_tokens": 5132288} | |
| {"current_steps": 775, "total_steps": 775, "epoch": 4.98793242156074, "percentage": 100.0, "elapsed_time": "2:51:57", "remaining_time": "0:00:00", "throughput": "497.41", "total_tokens": 5132288} | |