|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.0, |
|
"eval_steps": 500, |
|
"global_step": 1478, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"grad_norm": 1.03125, |
|
"learning_rate": 1.3513513513513515e-06, |
|
"loss": 2.2722, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 1.140625, |
|
"learning_rate": 6.7567567567567575e-06, |
|
"loss": 2.2825, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"grad_norm": 0.953125, |
|
"learning_rate": 1.3513513513513515e-05, |
|
"loss": 2.3023, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"grad_norm": 0.73828125, |
|
"learning_rate": 2.0270270270270273e-05, |
|
"loss": 2.3044, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.439453125, |
|
"learning_rate": 2.702702702702703e-05, |
|
"loss": 2.2383, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"grad_norm": 0.326171875, |
|
"learning_rate": 3.3783783783783784e-05, |
|
"loss": 2.2166, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"grad_norm": 0.302734375, |
|
"learning_rate": 4.0540540540540545e-05, |
|
"loss": 2.1633, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 4.72972972972973e-05, |
|
"loss": 2.1206, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"grad_norm": 0.25, |
|
"learning_rate": 5.405405405405406e-05, |
|
"loss": 2.1314, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"grad_norm": 0.2412109375, |
|
"learning_rate": 6.0810810810810814e-05, |
|
"loss": 2.0912, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 6.756756756756757e-05, |
|
"loss": 2.0372, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"grad_norm": 0.283203125, |
|
"learning_rate": 7.432432432432433e-05, |
|
"loss": 1.9804, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"grad_norm": 0.6171875, |
|
"learning_rate": 8.108108108108109e-05, |
|
"loss": 1.9327, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"grad_norm": 0.2119140625, |
|
"learning_rate": 8.783783783783784e-05, |
|
"loss": 1.889, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"grad_norm": 0.2119140625, |
|
"learning_rate": 9.45945945945946e-05, |
|
"loss": 1.8393, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"grad_norm": 0.2216796875, |
|
"learning_rate": 0.00010135135135135136, |
|
"loss": 1.8477, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"grad_norm": 0.2138671875, |
|
"learning_rate": 0.00010810810810810812, |
|
"loss": 1.839, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"grad_norm": 0.197265625, |
|
"learning_rate": 0.00011486486486486487, |
|
"loss": 1.8043, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"grad_norm": 0.197265625, |
|
"learning_rate": 0.00012162162162162163, |
|
"loss": 1.8177, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"grad_norm": 0.201171875, |
|
"learning_rate": 0.0001283783783783784, |
|
"loss": 1.7922, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"grad_norm": 0.1962890625, |
|
"learning_rate": 0.00013513513513513514, |
|
"loss": 1.791, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"grad_norm": 0.2001953125, |
|
"learning_rate": 0.00014189189189189188, |
|
"loss": 1.7794, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"grad_norm": 0.203125, |
|
"learning_rate": 0.00014864864864864866, |
|
"loss": 1.7939, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"grad_norm": 0.2041015625, |
|
"learning_rate": 0.0001554054054054054, |
|
"loss": 1.7654, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"grad_norm": 0.205078125, |
|
"learning_rate": 0.00016216216216216218, |
|
"loss": 1.7436, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"grad_norm": 0.2158203125, |
|
"learning_rate": 0.00016891891891891893, |
|
"loss": 1.7431, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"grad_norm": 0.2080078125, |
|
"learning_rate": 0.00017567567567567568, |
|
"loss": 1.7665, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.00018243243243243245, |
|
"loss": 1.7625, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.0001891891891891892, |
|
"loss": 1.7485, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.00019594594594594594, |
|
"loss": 1.7564, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.00019999888409903948, |
|
"loss": 1.76, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"grad_norm": 0.224609375, |
|
"learning_rate": 0.0001999863304992469, |
|
"loss": 1.7365, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00019995983018035278, |
|
"loss": 1.7291, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.00019991938683878746, |
|
"loss": 1.7449, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"grad_norm": 0.2177734375, |
|
"learning_rate": 0.0001998650061158413, |
|
"loss": 1.7379, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"grad_norm": 0.208984375, |
|
"learning_rate": 0.0001997966955968779, |
|
"loss": 1.7308, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.00019971446481027591, |
|
"loss": 1.7395, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"grad_norm": 0.216796875, |
|
"learning_rate": 0.00019961832522610005, |
|
"loss": 1.7533, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"grad_norm": 0.2255859375, |
|
"learning_rate": 0.00019950829025450114, |
|
"loss": 1.7004, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"grad_norm": 0.216796875, |
|
"learning_rate": 0.0001993843752438457, |
|
"loss": 1.7046, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.00019924659747857484, |
|
"loss": 1.7147, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"grad_norm": 0.2314453125, |
|
"learning_rate": 0.00019909497617679348, |
|
"loss": 1.7213, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"grad_norm": 0.2265625, |
|
"learning_rate": 0.00019892953248758967, |
|
"loss": 1.7334, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"grad_norm": 0.212890625, |
|
"learning_rate": 0.00019875028948808455, |
|
"loss": 1.7004, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"grad_norm": 0.2197265625, |
|
"learning_rate": 0.0001985572721802134, |
|
"loss": 1.7027, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.00019835050748723824, |
|
"loss": 1.7132, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"grad_norm": 0.2158203125, |
|
"learning_rate": 0.0001981300242499924, |
|
"loss": 1.7003, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.0001978958532228576, |
|
"loss": 1.6947, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00019764802706947421, |
|
"loss": 1.7206, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"grad_norm": 0.2275390625, |
|
"learning_rate": 0.00019738658035818495, |
|
"loss": 1.7172, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"grad_norm": 0.216796875, |
|
"learning_rate": 0.00019711154955721335, |
|
"loss": 1.7092, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"grad_norm": 0.240234375, |
|
"learning_rate": 0.00019682297302957665, |
|
"loss": 1.6991, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"grad_norm": 0.2265625, |
|
"learning_rate": 0.00019652089102773488, |
|
"loss": 1.6896, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"grad_norm": 0.2158203125, |
|
"learning_rate": 0.00019620534568797607, |
|
"loss": 1.7012, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"grad_norm": 0.2177734375, |
|
"learning_rate": 0.000195876381024539, |
|
"loss": 1.6948, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.00019553404292347356, |
|
"loss": 1.7095, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00019517837913624048, |
|
"loss": 1.7052, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"grad_norm": 0.2236328125, |
|
"learning_rate": 0.00019480943927305058, |
|
"loss": 1.6961, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"grad_norm": 0.2177734375, |
|
"learning_rate": 0.00019442727479594484, |
|
"loss": 1.7012, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"grad_norm": 0.21484375, |
|
"learning_rate": 0.00019403193901161613, |
|
"loss": 1.6753, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"grad_norm": 0.2314453125, |
|
"learning_rate": 0.00019362348706397373, |
|
"loss": 1.7024, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.00019320197592645138, |
|
"loss": 1.6797, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"grad_norm": 0.2314453125, |
|
"learning_rate": 0.00019276746439406047, |
|
"loss": 1.6716, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"grad_norm": 0.236328125, |
|
"learning_rate": 0.00019232001307518867, |
|
"loss": 1.6757, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"grad_norm": 0.2275390625, |
|
"learning_rate": 0.00019185968438314616, |
|
"loss": 1.6809, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00019138654252745968, |
|
"loss": 1.6766, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.00019090065350491626, |
|
"loss": 1.6687, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"grad_norm": 0.2294921875, |
|
"learning_rate": 0.00019040208509035745, |
|
"loss": 1.6548, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"grad_norm": 0.2314453125, |
|
"learning_rate": 0.00018989090682722585, |
|
"loss": 1.6639, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.00018936719001786453, |
|
"loss": 1.6849, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00018883100771357157, |
|
"loss": 1.6981, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"grad_norm": 0.2236328125, |
|
"learning_rate": 0.00018828243470441028, |
|
"loss": 1.6958, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"grad_norm": 0.2177734375, |
|
"learning_rate": 0.00018772154750877696, |
|
"loss": 1.6754, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"grad_norm": 0.2275390625, |
|
"learning_rate": 0.00018714842436272773, |
|
"loss": 1.6649, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"grad_norm": 0.228515625, |
|
"learning_rate": 0.00018656314520906571, |
|
"loss": 1.6376, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 0.244140625, |
|
"learning_rate": 0.0001859657916861899, |
|
"loss": 1.6803, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 0.224609375, |
|
"learning_rate": 0.00018535644711670804, |
|
"loss": 1.6787, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.00018473519649581398, |
|
"loss": 1.6717, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"grad_norm": 0.240234375, |
|
"learning_rate": 0.00018410212647943216, |
|
"loss": 1.6846, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"grad_norm": 0.2216796875, |
|
"learning_rate": 0.00018345732537213027, |
|
"loss": 1.6464, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00018280088311480201, |
|
"loss": 1.6409, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.0001821328912721215, |
|
"loss": 1.6639, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"grad_norm": 0.2265625, |
|
"learning_rate": 0.00018145344301977127, |
|
"loss": 1.6596, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00018076263313144568, |
|
"loss": 1.6841, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"grad_norm": 0.2275390625, |
|
"learning_rate": 0.00018006055796563104, |
|
"loss": 1.6552, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00017934731545216515, |
|
"loss": 1.6723, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"grad_norm": 0.22265625, |
|
"learning_rate": 0.00017862300507857733, |
|
"loss": 1.6813, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"grad_norm": 0.220703125, |
|
"learning_rate": 0.00017788772787621126, |
|
"loss": 1.6615, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 0.0001771415864061326, |
|
"loss": 1.6444, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"grad_norm": 0.236328125, |
|
"learning_rate": 0.00017638468474482296, |
|
"loss": 1.6697, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.0001756171284696629, |
|
"loss": 1.6725, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"grad_norm": 0.236328125, |
|
"learning_rate": 0.00017483902464420506, |
|
"loss": 1.6637, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"grad_norm": 0.2392578125, |
|
"learning_rate": 0.00017405048180324043, |
|
"loss": 1.6495, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.00017325160993765932, |
|
"loss": 1.6688, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.00017244252047910892, |
|
"loss": 1.6561, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.0001716233262844502, |
|
"loss": 1.6558, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00017079414162001613, |
|
"loss": 1.6518, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.00016995508214567276, |
|
"loss": 1.6415, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"grad_norm": 0.224609375, |
|
"learning_rate": 0.00016910626489868649, |
|
"loss": 1.6574, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"grad_norm": 0.248046875, |
|
"learning_rate": 0.00016824780827739887, |
|
"loss": 1.6581, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.0001673798320247118, |
|
"loss": 1.6623, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"grad_norm": 0.232421875, |
|
"learning_rate": 0.0001665024572113848, |
|
"loss": 1.6696, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"grad_norm": 0.228515625, |
|
"learning_rate": 0.00016561580621914765, |
|
"loss": 1.6404, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00016472000272362935, |
|
"loss": 1.667, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"grad_norm": 0.240234375, |
|
"learning_rate": 0.00016381517167710755, |
|
"loss": 1.6439, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.00016290143929107912, |
|
"loss": 1.6346, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00016197893301865549, |
|
"loss": 1.6616, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"grad_norm": 0.2470703125, |
|
"learning_rate": 0.00016104778153678466, |
|
"loss": 1.651, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"grad_norm": 0.2421875, |
|
"learning_rate": 0.00016010811472830252, |
|
"loss": 1.629, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.0001591600636638161, |
|
"loss": 1.6497, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00015820376058342078, |
|
"loss": 1.6533, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"grad_norm": 0.2294921875, |
|
"learning_rate": 0.00015723933887825493, |
|
"loss": 1.6424, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"grad_norm": 0.2470703125, |
|
"learning_rate": 0.00015626693307189336, |
|
"loss": 1.6379, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.0001552866788015834, |
|
"loss": 1.6344, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00015429871279932513, |
|
"loss": 1.6492, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.0001533031728727994, |
|
"loss": 1.6245, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"grad_norm": 0.236328125, |
|
"learning_rate": 0.00015230019788614526, |
|
"loss": 1.6325, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.00015128992774059063, |
|
"loss": 1.6367, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.0001502725033549377, |
|
"loss": 1.6269, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"grad_norm": 0.236328125, |
|
"learning_rate": 0.000149248066645907, |
|
"loss": 1.6199, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"grad_norm": 0.2421875, |
|
"learning_rate": 0.00014821676050834166, |
|
"loss": 1.6358, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00014717872879527576, |
|
"loss": 1.6359, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"grad_norm": 0.2451171875, |
|
"learning_rate": 0.0001461341162978688, |
|
"loss": 1.6371, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00014508306872520912, |
|
"loss": 1.6464, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"grad_norm": 0.2431640625, |
|
"learning_rate": 0.0001440257326839897, |
|
"loss": 1.648, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.00014296225565805853, |
|
"loss": 1.625, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00014189278598784647, |
|
"loss": 1.6327, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.000140817472849676, |
|
"loss": 1.6369, |
|
"step": 635 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"grad_norm": 0.2333984375, |
|
"learning_rate": 0.00013973646623495305, |
|
"loss": 1.6194, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00013864991692924523, |
|
"loss": 1.6493, |
|
"step": 645 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"grad_norm": 0.2431640625, |
|
"learning_rate": 0.00013755797649124944, |
|
"loss": 1.6316, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"grad_norm": 0.2412109375, |
|
"learning_rate": 0.0001364607972316515, |
|
"loss": 1.618, |
|
"step": 655 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"grad_norm": 0.2421875, |
|
"learning_rate": 0.00013535853219188063, |
|
"loss": 1.6317, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"grad_norm": 0.2431640625, |
|
"learning_rate": 0.00013425133512276282, |
|
"loss": 1.6146, |
|
"step": 665 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"grad_norm": 0.244140625, |
|
"learning_rate": 0.0001331393604630741, |
|
"loss": 1.6347, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"grad_norm": 0.23828125, |
|
"learning_rate": 0.0001320227633179989, |
|
"loss": 1.6241, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"grad_norm": 0.23046875, |
|
"learning_rate": 0.00013090169943749476, |
|
"loss": 1.6324, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"grad_norm": 0.24609375, |
|
"learning_rate": 0.00012977632519456744, |
|
"loss": 1.6373, |
|
"step": 685 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"grad_norm": 0.240234375, |
|
"learning_rate": 0.00012864679756345904, |
|
"loss": 1.6453, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 0.00012751327409775228, |
|
"loss": 1.6447, |
|
"step": 695 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 0.00012637591290839376, |
|
"loss": 1.6283, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"grad_norm": 0.234375, |
|
"learning_rate": 0.00012523487264163997, |
|
"loss": 1.625, |
|
"step": 705 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00012409031245692797, |
|
"loss": 1.6049, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"grad_norm": 0.2470703125, |
|
"learning_rate": 0.00012294239200467516, |
|
"loss": 1.6418, |
|
"step": 715 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 0.00012179127140400997, |
|
"loss": 1.6229, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.0001206371112204376, |
|
"loss": 1.6021, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"grad_norm": 0.2451171875, |
|
"learning_rate": 0.00011948007244344332, |
|
"loss": 1.6184, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"grad_norm": 0.2431640625, |
|
"learning_rate": 0.00011832031646403654, |
|
"loss": 1.639, |
|
"step": 735 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 1.7224379777908325, |
|
"eval_runtime": 56.8026, |
|
"eval_samples_per_second": 7.658, |
|
"eval_steps_per_second": 0.968, |
|
"step": 739 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 0.2353515625, |
|
"learning_rate": 0.00011715800505223918, |
|
"loss": 1.5945, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"grad_norm": 0.2373046875, |
|
"learning_rate": 0.00011599330033452079, |
|
"loss": 1.5651, |
|
"step": 745 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"grad_norm": 0.2421875, |
|
"learning_rate": 0.0001148263647711842, |
|
"loss": 1.6011, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"grad_norm": 0.244140625, |
|
"learning_rate": 0.00011365736113370462, |
|
"loss": 1.5838, |
|
"step": 755 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.0001124864524820251, |
|
"loss": 1.5881, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.00011131380214181203, |
|
"loss": 1.5717, |
|
"step": 765 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.00011013957368167342, |
|
"loss": 1.5866, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.00010896393089034336, |
|
"loss": 1.5516, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"grad_norm": 0.244140625, |
|
"learning_rate": 0.00010778703775383558, |
|
"loss": 1.5718, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.00010660905843256994, |
|
"loss": 1.5438, |
|
"step": 785 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 0.00010543015723847403, |
|
"loss": 1.5686, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 0.00010425049861206411, |
|
"loss": 1.5921, |
|
"step": 795 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 0.00010307024709950774, |
|
"loss": 1.5782, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"grad_norm": 0.25, |
|
"learning_rate": 0.00010188956732967208, |
|
"loss": 1.5615, |
|
"step": 805 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"grad_norm": 0.2451171875, |
|
"learning_rate": 0.00010070862399116015, |
|
"loss": 1.5866, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 9.952758180933934e-05, |
|
"loss": 1.5762, |
|
"step": 815 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"grad_norm": 0.2490234375, |
|
"learning_rate": 9.834660552336415e-05, |
|
"loss": 1.5756, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"grad_norm": 0.244140625, |
|
"learning_rate": 9.716585986319769e-05, |
|
"loss": 1.5669, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 9.598550952663383e-05, |
|
"loss": 1.5809, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 9.480571915632421e-05, |
|
"loss": 1.5898, |
|
"step": 835 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"grad_norm": 0.25, |
|
"learning_rate": 9.362665331681294e-05, |
|
"loss": 1.5958, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"grad_norm": 0.25, |
|
"learning_rate": 9.244847647158202e-05, |
|
"loss": 1.579, |
|
"step": 845 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"grad_norm": 0.2451171875, |
|
"learning_rate": 9.127135296011101e-05, |
|
"loss": 1.5814, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"grad_norm": 0.24609375, |
|
"learning_rate": 9.009544697495374e-05, |
|
"loss": 1.5581, |
|
"step": 855 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"grad_norm": 0.2470703125, |
|
"learning_rate": 8.892092253883601e-05, |
|
"loss": 1.5927, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 8.77479434817764e-05, |
|
"loss": 1.5679, |
|
"step": 865 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 8.657667341823448e-05, |
|
"loss": 1.5774, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 8.540727572428854e-05, |
|
"loss": 1.5766, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"grad_norm": 0.25, |
|
"learning_rate": 8.423991351484716e-05, |
|
"loss": 1.5588, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 8.307474962089677e-05, |
|
"loss": 1.5613, |
|
"step": 885 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"grad_norm": 0.2490234375, |
|
"learning_rate": 8.191194656678904e-05, |
|
"loss": 1.5778, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 8.07516665475708e-05, |
|
"loss": 1.5695, |
|
"step": 895 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 7.959407140636034e-05, |
|
"loss": 1.5787, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 7.843932261177224e-05, |
|
"loss": 1.5562, |
|
"step": 905 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"grad_norm": 0.24609375, |
|
"learning_rate": 7.728758123539499e-05, |
|
"loss": 1.582, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 7.613900792932332e-05, |
|
"loss": 1.5676, |
|
"step": 915 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 7.499376290374994e-05, |
|
"loss": 1.5715, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"grad_norm": 0.271484375, |
|
"learning_rate": 7.385200590461803e-05, |
|
"loss": 1.5816, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 7.271389619133908e-05, |
|
"loss": 1.577, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 7.157959251457823e-05, |
|
"loss": 1.5808, |
|
"step": 935 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 7.044925309411093e-05, |
|
"loss": 1.5559, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 6.932303559675329e-05, |
|
"loss": 1.5794, |
|
"step": 945 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 6.820109711436988e-05, |
|
"loss": 1.5645, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 6.708359414196133e-05, |
|
"loss": 1.5781, |
|
"step": 955 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 6.59706825558357e-05, |
|
"loss": 1.5714, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"grad_norm": 0.248046875, |
|
"learning_rate": 6.486251759186572e-05, |
|
"loss": 1.5562, |
|
"step": 965 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 6.37592538238356e-05, |
|
"loss": 1.564, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 6.266104514187997e-05, |
|
"loss": 1.5725, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 6.156804473101851e-05, |
|
"loss": 1.5794, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 6.04804050497886e-05, |
|
"loss": 1.5662, |
|
"step": 985 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 5.93982778089796e-05, |
|
"loss": 1.5867, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 5.832181395047098e-05, |
|
"loss": 1.5863, |
|
"step": 995 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 5.7251163626178394e-05, |
|
"loss": 1.5694, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 5.618647617710935e-05, |
|
"loss": 1.5634, |
|
"step": 1005 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 5.5127900112532106e-05, |
|
"loss": 1.5633, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 5.407558308926083e-05, |
|
"loss": 1.5736, |
|
"step": 1015 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 5.302967189105941e-05, |
|
"loss": 1.5675, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 5.199031240816714e-05, |
|
"loss": 1.5509, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 5.095764961694922e-05, |
|
"loss": 1.5624, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 4.9931827559674115e-05, |
|
"loss": 1.5605, |
|
"step": 1035 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"grad_norm": 0.27734375, |
|
"learning_rate": 4.8912989324422164e-05, |
|
"loss": 1.559, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"grad_norm": 0.271484375, |
|
"learning_rate": 4.790127702512634e-05, |
|
"loss": 1.5778, |
|
"step": 1045 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 4.689683178174964e-05, |
|
"loss": 1.5892, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 4.589979370060037e-05, |
|
"loss": 1.5488, |
|
"step": 1055 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 4.491030185478976e-05, |
|
"loss": 1.5906, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 4.392849426483274e-05, |
|
"loss": 1.5782, |
|
"step": 1065 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 4.295450787939622e-05, |
|
"loss": 1.559, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 4.198847855619652e-05, |
|
"loss": 1.5792, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"grad_norm": 0.2734375, |
|
"learning_rate": 4.103054104304912e-05, |
|
"loss": 1.5893, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 4.0080828959073255e-05, |
|
"loss": 1.5663, |
|
"step": 1085 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 3.913947477605378e-05, |
|
"loss": 1.5665, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"grad_norm": 0.275390625, |
|
"learning_rate": 3.82066097999632e-05, |
|
"loss": 1.5795, |
|
"step": 1095 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 3.7282364152646297e-05, |
|
"loss": 1.5862, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"grad_norm": 0.26953125, |
|
"learning_rate": 3.636686675367006e-05, |
|
"loss": 1.5705, |
|
"step": 1105 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"grad_norm": 0.26953125, |
|
"learning_rate": 3.546024530234091e-05, |
|
"loss": 1.5794, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 3.45626262598926e-05, |
|
"loss": 1.5704, |
|
"step": 1115 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 3.367413483184654e-05, |
|
"loss": 1.5677, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 3.279489495054742e-05, |
|
"loss": 1.5562, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 3.19250292578762e-05, |
|
"loss": 1.5317, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 3.106465908814342e-05, |
|
"loss": 1.571, |
|
"step": 1135 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 3.021390445116462e-05, |
|
"loss": 1.5841, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 2.937288401552063e-05, |
|
"loss": 1.5712, |
|
"step": 1145 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"grad_norm": 0.25, |
|
"learning_rate": 2.8541715092005094e-05, |
|
"loss": 1.5708, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 2.7720513617260856e-05, |
|
"loss": 1.567, |
|
"step": 1155 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"grad_norm": 0.310546875, |
|
"learning_rate": 2.6909394137608868e-05, |
|
"loss": 1.563, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 2.6108469793070156e-05, |
|
"loss": 1.5789, |
|
"step": 1165 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"grad_norm": 0.275390625, |
|
"learning_rate": 2.5317852301584643e-05, |
|
"loss": 1.565, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 2.4537651943427665e-05, |
|
"loss": 1.5401, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 2.3767977545827845e-05, |
|
"loss": 1.5669, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 2.300893646778681e-05, |
|
"loss": 1.5491, |
|
"step": 1185 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 2.226063458510428e-05, |
|
"loss": 1.566, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 2.152317627560979e-05, |
|
"loss": 1.5675, |
|
"step": 1195 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 2.0796664404603416e-05, |
|
"loss": 1.5486, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 2.008120031050753e-05, |
|
"loss": 1.5786, |
|
"step": 1205 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"grad_norm": 0.271484375, |
|
"learning_rate": 1.9376883790731414e-05, |
|
"loss": 1.5747, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 1.86838130877509e-05, |
|
"loss": 1.5687, |
|
"step": 1215 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 1.8002084875404934e-05, |
|
"loss": 1.5754, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 1.7331794245410926e-05, |
|
"loss": 1.5715, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"grad_norm": 0.26953125, |
|
"learning_rate": 1.6673034694100655e-05, |
|
"loss": 1.58, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 1.6025898109378967e-05, |
|
"loss": 1.5952, |
|
"step": 1235 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 1.5390474757906446e-05, |
|
"loss": 1.5688, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"grad_norm": 0.26953125, |
|
"learning_rate": 1.4766853272508786e-05, |
|
"loss": 1.5698, |
|
"step": 1245 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 1.415512063981339e-05, |
|
"loss": 1.5674, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 1.3555362188116172e-05, |
|
"loss": 1.5391, |
|
"step": 1255 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"grad_norm": 0.271484375, |
|
"learning_rate": 1.2967661575479317e-05, |
|
"loss": 1.5762, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 1.2392100778062122e-05, |
|
"loss": 1.5743, |
|
"step": 1265 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 1.1828760078686562e-05, |
|
"loss": 1.5867, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 1.1277718055638819e-05, |
|
"loss": 1.5623, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 1.0739051571708736e-05, |
|
"loss": 1.5839, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 1.0212835763468487e-05, |
|
"loss": 1.5692, |
|
"step": 1285 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 9.699144030792162e-06, |
|
"loss": 1.5643, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 9.198048026617323e-06, |
|
"loss": 1.5515, |
|
"step": 1295 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 8.709617646950564e-06, |
|
"loss": 1.5713, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 8.233921021117863e-06, |
|
"loss": 1.5542, |
|
"step": 1305 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 7.771024502261526e-06, |
|
"loss": 1.5724, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 7.320992658084891e-06, |
|
"loss": 1.5839, |
|
"step": 1315 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 6.8838882618459165e-06, |
|
"loss": 1.5818, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 6.459772283601218e-06, |
|
"loss": 1.5659, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"grad_norm": 0.271484375, |
|
"learning_rate": 6.048703881701578e-06, |
|
"loss": 1.5646, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 5.650740394540255e-06, |
|
"loss": 1.5613, |
|
"step": 1335 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 5.265937332554849e-06, |
|
"loss": 1.566, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 4.8943483704846475e-06, |
|
"loss": 1.5748, |
|
"step": 1345 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"grad_norm": 0.26953125, |
|
"learning_rate": 4.5360253398834765e-06, |
|
"loss": 1.5706, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 4.191018221890097e-06, |
|
"loss": 1.5862, |
|
"step": 1355 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 3.859375140256371e-06, |
|
"loss": 1.5878, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 3.5411423546348075e-06, |
|
"loss": 1.562, |
|
"step": 1365 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"grad_norm": 0.2734375, |
|
"learning_rate": 3.2363642541258676e-06, |
|
"loss": 1.5348, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"grad_norm": 0.259765625, |
|
"learning_rate": 2.9450833510863307e-06, |
|
"loss": 1.5829, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"grad_norm": 0.265625, |
|
"learning_rate": 2.667340275199426e-06, |
|
"loss": 1.5797, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 2.4031737678074985e-06, |
|
"loss": 1.5903, |
|
"step": 1385 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 2.15262067650821e-06, |
|
"loss": 1.5676, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 1.9157159500146958e-06, |
|
"loss": 1.5557, |
|
"step": 1395 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"grad_norm": 0.27734375, |
|
"learning_rate": 1.6924926332807956e-06, |
|
"loss": 1.5778, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 1.4829818628916526e-06, |
|
"loss": 1.5598, |
|
"step": 1405 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 1.287212862720677e-06, |
|
"loss": 1.5683, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 1.1052129398531507e-06, |
|
"loss": 1.545, |
|
"step": 1415 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 9.370074807772966e-07, |
|
"loss": 1.5648, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 7.826199478431551e-07, |
|
"loss": 1.5441, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 6.420718759900357e-07, |
|
"loss": 1.572, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"grad_norm": 0.2578125, |
|
"learning_rate": 5.153828697425422e-07, |
|
"loss": 1.5409, |
|
"step": 1435 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 4.025706004760932e-07, |
|
"loss": 1.5713, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 3.0365080395200473e-07, |
|
"loss": 1.5608, |
|
"step": 1445 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"grad_norm": 0.25390625, |
|
"learning_rate": 2.1863727812254653e-07, |
|
"loss": 1.5541, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"grad_norm": 0.267578125, |
|
"learning_rate": 1.4754188120631452e-07, |
|
"loss": 1.5628, |
|
"step": 1455 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"grad_norm": 0.251953125, |
|
"learning_rate": 9.037453003418739e-08, |
|
"loss": 1.5841, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"grad_norm": 0.26171875, |
|
"learning_rate": 4.7143198666077524e-08, |
|
"loss": 1.5847, |
|
"step": 1465 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"grad_norm": 0.255859375, |
|
"learning_rate": 1.7853917278631838e-08, |
|
"loss": 1.5713, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 0.263671875, |
|
"learning_rate": 2.5107713241045993e-09, |
|
"loss": 1.571, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 1.7227909564971924, |
|
"eval_runtime": 56.8169, |
|
"eval_samples_per_second": 7.656, |
|
"eval_steps_per_second": 0.968, |
|
"step": 1478 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"step": 1478, |
|
"total_flos": 1.521149702706299e+17, |
|
"train_loss": 1.645228040073173, |
|
"train_runtime": 5797.8463, |
|
"train_samples_per_second": 2.039, |
|
"train_steps_per_second": 0.255 |
|
} |
|
], |
|
"logging_steps": 5, |
|
"max_steps": 1478, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 2, |
|
"save_steps": 100, |
|
"total_flos": 1.521149702706299e+17, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|