| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7708119218910586, | |
| "eval_steps": 500, | |
| "global_step": 1500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999979150098865e-05, | |
| "loss": 4.0418, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999989444743508e-05, | |
| "loss": 3.8361, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999974458911041e-05, | |
| "loss": 3.8737, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9999529575515425e-05, | |
| "loss": 3.806, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9999249407210515e-05, | |
| "loss": 3.4125, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9998904084925845e-05, | |
| "loss": 3.6489, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999849360956141e-05, | |
| "loss": 3.4719, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999811831976564e-05, | |
| "loss": 3.5265, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999759057166764e-05, | |
| "loss": 3.6889, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9996997673913204e-05, | |
| "loss": 3.5104, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.999633962804756e-05, | |
| "loss": 3.5778, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.999561643578575e-05, | |
| "loss": 3.3353, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.999482809901257e-05, | |
| "loss": 3.3511, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9993974619782625e-05, | |
| "loss": 3.6355, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.999305600032027e-05, | |
| "loss": 3.3915, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.999207224301965e-05, | |
| "loss": 3.4926, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.999102335044467e-05, | |
| "loss": 3.4918, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9989909325328996e-05, | |
| "loss": 3.3984, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.998873017057605e-05, | |
| "loss": 3.4989, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.998748588925897e-05, | |
| "loss": 3.6368, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9986176484620665e-05, | |
| "loss": 3.4412, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.998480196007375e-05, | |
| "loss": 3.4708, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9983362319200554e-05, | |
| "loss": 3.37, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.998185756575313e-05, | |
| "loss": 3.4524, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9980287703653225e-05, | |
| "loss": 3.3863, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.997865273699227e-05, | |
| "loss": 3.4414, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.997729789123732e-05, | |
| "loss": 3.3335, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.997554574721849e-05, | |
| "loss": 3.231, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.997372851099727e-05, | |
| "loss": 3.5962, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.997184618730983e-05, | |
| "loss": 3.2795, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.996989878106192e-05, | |
| "loss": 3.5384, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.996788629732897e-05, | |
| "loss": 3.3274, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.996580874135597e-05, | |
| "loss": 3.2292, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.996366611855753e-05, | |
| "loss": 3.4829, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.996145843451785e-05, | |
| "loss": 3.1998, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9959185694990655e-05, | |
| "loss": 3.3757, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.995684790589926e-05, | |
| "loss": 3.1001, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.99544450733365e-05, | |
| "loss": 3.2301, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.995197720356471e-05, | |
| "loss": 3.3187, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.994944430301576e-05, | |
| "loss": 3.527, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.994684637829098e-05, | |
| "loss": 3.3808, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.994418343616119e-05, | |
| "loss": 3.2406, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.994145548356664e-05, | |
| "loss": 3.3463, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.993866252761702e-05, | |
| "loss": 3.2972, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.993580457559145e-05, | |
| "loss": 3.4135, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9932881634938424e-05, | |
| "loss": 3.4025, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992989371327581e-05, | |
| "loss": 3.662, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992684081839086e-05, | |
| "loss": 3.4279, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9923722958240124e-05, | |
| "loss": 3.3075, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.992054014094951e-05, | |
| "loss": 3.2826, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.991729237481417e-05, | |
| "loss": 3.4992, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.991397966829856e-05, | |
| "loss": 3.2421, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.991060203003639e-05, | |
| "loss": 3.2615, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9907159468830585e-05, | |
| "loss": 3.3299, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.990365199365327e-05, | |
| "loss": 3.3117, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9900079613645754e-05, | |
| "loss": 3.4729, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9896442338118524e-05, | |
| "loss": 3.2521, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.989274017655117e-05, | |
| "loss": 3.261, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9888973138592414e-05, | |
| "loss": 3.4403, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.988514123406004e-05, | |
| "loss": 3.3408, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9881244472940914e-05, | |
| "loss": 3.3208, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9877282865390904e-05, | |
| "loss": 3.6889, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9873256421734915e-05, | |
| "loss": 3.1537, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.986916515246681e-05, | |
| "loss": 3.1837, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.986500906824941e-05, | |
| "loss": 3.23, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9860788179914456e-05, | |
| "loss": 3.4343, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.985650249846258e-05, | |
| "loss": 3.3317, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9852152035063294e-05, | |
| "loss": 3.155, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.984773680105493e-05, | |
| "loss": 3.3527, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.984415798673684e-05, | |
| "loss": 3.3783, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9839626194743575e-05, | |
| "loss": 3.2567, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.983502966478656e-05, | |
| "loss": 3.1278, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9830368408845425e-05, | |
| "loss": 3.2344, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.982564243906852e-05, | |
| "loss": 3.3458, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9820851767772844e-05, | |
| "loss": 3.1467, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.981599640744402e-05, | |
| "loss": 3.217, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.981107637073628e-05, | |
| "loss": 3.4011, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.98060916704724e-05, | |
| "loss": 3.4098, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9801042319643686e-05, | |
| "loss": 3.3332, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.979592833140996e-05, | |
| "loss": 3.4042, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.979074971909948e-05, | |
| "loss": 3.3127, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.978550649620894e-05, | |
| "loss": 3.1726, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.978019867640342e-05, | |
| "loss": 3.3029, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.977482627351637e-05, | |
| "loss": 3.3519, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.976938930154954e-05, | |
| "loss": 3.2507, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.976388777467297e-05, | |
| "loss": 3.1322, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9758321707224954e-05, | |
| "loss": 3.1306, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.975269111371197e-05, | |
| "loss": 3.189, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.974699600880869e-05, | |
| "loss": 3.2598, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.974123640735791e-05, | |
| "loss": 3.3483, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.973541232437052e-05, | |
| "loss": 3.4678, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.972952377502545e-05, | |
| "loss": 2.9668, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9723570774669687e-05, | |
| "loss": 3.372, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.971755333881814e-05, | |
| "loss": 3.2808, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.971147148315367e-05, | |
| "loss": 3.2235, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9705325223527057e-05, | |
| "loss": 3.0441, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.96991145759569e-05, | |
| "loss": 3.4724, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.969283955662963e-05, | |
| "loss": 3.389, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.968650018189943e-05, | |
| "loss": 3.3817, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.968009646828822e-05, | |
| "loss": 3.2597, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.967362843248559e-05, | |
| "loss": 3.253, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.966709609134878e-05, | |
| "loss": 3.1903, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9660499461902645e-05, | |
| "loss": 3.1286, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.965383856133953e-05, | |
| "loss": 3.292, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.964711340701935e-05, | |
| "loss": 3.4217, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.964032401646944e-05, | |
| "loss": 3.2587, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.963347040738457e-05, | |
| "loss": 3.3325, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.962655259762687e-05, | |
| "loss": 3.3475, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.961957060522581e-05, | |
| "loss": 3.2882, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.961252444837809e-05, | |
| "loss": 3.3631, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9605414145447696e-05, | |
| "loss": 3.2589, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.959823971496574e-05, | |
| "loss": 3.4855, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.959100117563052e-05, | |
| "loss": 3.1354, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.958369854630737e-05, | |
| "loss": 3.3037, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.957633184602867e-05, | |
| "loss": 3.1023, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.956890109399381e-05, | |
| "loss": 3.1933, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.956140630956908e-05, | |
| "loss": 3.3994, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.955384751228767e-05, | |
| "loss": 3.1178, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.954622472184961e-05, | |
| "loss": 3.2513, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.9538537958121686e-05, | |
| "loss": 3.5002, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.953078724113744e-05, | |
| "loss": 3.1867, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.9522972591097074e-05, | |
| "loss": 3.3883, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9515094028367425e-05, | |
| "loss": 3.3708, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.950715157348191e-05, | |
| "loss": 3.2425, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9499145247140434e-05, | |
| "loss": 3.1046, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.9491075070209385e-05, | |
| "loss": 3.3236, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9482941063721576e-05, | |
| "loss": 3.4957, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.947474324887614e-05, | |
| "loss": 3.2225, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9466481647038554e-05, | |
| "loss": 3.5314, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.9458156279740484e-05, | |
| "loss": 3.1028, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.944976716867984e-05, | |
| "loss": 3.2322, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.9441314335720616e-05, | |
| "loss": 3.1332, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.943279780289292e-05, | |
| "loss": 3.2028, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.9424217592392854e-05, | |
| "loss": 3.2408, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.9415573726582465e-05, | |
| "loss": 3.0121, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.940686622798974e-05, | |
| "loss": 3.2182, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.93980951193085e-05, | |
| "loss": 3.4794, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.93892604233983e-05, | |
| "loss": 3.3226, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.938036216328448e-05, | |
| "loss": 3.1917, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.9371400362158016e-05, | |
| "loss": 3.2176, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.936237504337549e-05, | |
| "loss": 2.933, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.935328623045902e-05, | |
| "loss": 3.134, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.934413394709621e-05, | |
| "loss": 2.983, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.933491821714009e-05, | |
| "loss": 3.4337, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.9325639064609045e-05, | |
| "loss": 3.4699, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.931629651368673e-05, | |
| "loss": 3.4081, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.930689058872208e-05, | |
| "loss": 3.2174, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.929742131422913e-05, | |
| "loss": 3.4953, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.92878887148871e-05, | |
| "loss": 3.217, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.927829281554017e-05, | |
| "loss": 3.2689, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.926863364119756e-05, | |
| "loss": 3.2994, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.9258911217033355e-05, | |
| "loss": 3.3868, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.924912556838651e-05, | |
| "loss": 3.3169, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.923927672076075e-05, | |
| "loss": 3.0466, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.92293646998245e-05, | |
| "loss": 3.0857, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.921938953141084e-05, | |
| "loss": 3.5001, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.920935124151744e-05, | |
| "loss": 3.1633, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.919924985630647e-05, | |
| "loss": 3.4816, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.9189085402104516e-05, | |
| "loss": 3.2032, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.917885790540258e-05, | |
| "loss": 3.164, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.916856739285593e-05, | |
| "loss": 3.3047, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.91582138912841e-05, | |
| "loss": 3.2366, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.9147797427670766e-05, | |
| "loss": 3.3105, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.91373180291637e-05, | |
| "loss": 3.4871, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.912677572307473e-05, | |
| "loss": 3.0442, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.911617053687957e-05, | |
| "loss": 3.1934, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.910550249821787e-05, | |
| "loss": 3.27, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.9094771634893075e-05, | |
| "loss": 3.3252, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.908397797487237e-05, | |
| "loss": 2.9852, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.907312154628658e-05, | |
| "loss": 3.1331, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.906220237743014e-05, | |
| "loss": 3.1444, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.905122049676101e-05, | |
| "loss": 3.3702, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.904017593290056e-05, | |
| "loss": 3.0531, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.902906871463355e-05, | |
| "loss": 3.2007, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.901789887090804e-05, | |
| "loss": 3.0837, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.900666643083528e-05, | |
| "loss": 3.2268, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.8995371423689684e-05, | |
| "loss": 2.9249, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.898401387890873e-05, | |
| "loss": 3.2058, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.897259382609285e-05, | |
| "loss": 3.0971, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.8961111295005446e-05, | |
| "loss": 3.2227, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.894956631557269e-05, | |
| "loss": 3.2907, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.893795891788357e-05, | |
| "loss": 3.1382, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.892628913218969e-05, | |
| "loss": 2.8996, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.8914556988905315e-05, | |
| "loss": 2.9693, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.890276251860717e-05, | |
| "loss": 3.0134, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.889090575203445e-05, | |
| "loss": 2.9696, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.887898672008871e-05, | |
| "loss": 3.3306, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.8867005453833774e-05, | |
| "loss": 3.2199, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.885496198449567e-05, | |
| "loss": 3.2795, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.8842856343462535e-05, | |
| "loss": 3.3482, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.883068856228454e-05, | |
| "loss": 3.322, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.881845867267382e-05, | |
| "loss": 3.0443, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.8806166706504375e-05, | |
| "loss": 3.1826, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.879381269581197e-05, | |
| "loss": 3.3732, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.878139667279411e-05, | |
| "loss": 3.3457, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.876891866980988e-05, | |
| "loss": 3.182, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.875637871937994e-05, | |
| "loss": 3.2416, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.874377685418635e-05, | |
| "loss": 3.2411, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.873111310707259e-05, | |
| "loss": 3.2836, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.8718387511043385e-05, | |
| "loss": 3.1773, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.870560009926465e-05, | |
| "loss": 3.1723, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8692750905063424e-05, | |
| "loss": 3.1535, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.867983996192776e-05, | |
| "loss": 2.977, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.866686730350663e-05, | |
| "loss": 3.1236, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.865383296360987e-05, | |
| "loss": 3.4016, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.864073697620807e-05, | |
| "loss": 3.1547, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.862757937543248e-05, | |
| "loss": 3.143, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.861436019557492e-05, | |
| "loss": 3.1989, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.860107947108772e-05, | |
| "loss": 3.0308, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.858773723658361e-05, | |
| "loss": 3.409, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.8574333526835605e-05, | |
| "loss": 3.2011, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.856086837677697e-05, | |
| "loss": 3.1954, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.854734182150107e-05, | |
| "loss": 3.1366, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.853375389626133e-05, | |
| "loss": 3.3789, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.85201046364711e-05, | |
| "loss": 3.1337, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.85063940777036e-05, | |
| "loss": 3.2278, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.849262225569181e-05, | |
| "loss": 3.1339, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.847878920632834e-05, | |
| "loss": 2.954, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.846489496566542e-05, | |
| "loss": 3.0878, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.845093956991475e-05, | |
| "loss": 3.2329, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.843692305544738e-05, | |
| "loss": 3.2245, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.84228454587937e-05, | |
| "loss": 3.2865, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.8408706816643254e-05, | |
| "loss": 3.1569, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.83945071658447e-05, | |
| "loss": 3.2279, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.838024654340571e-05, | |
| "loss": 3.0554, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.836592498649283e-05, | |
| "loss": 3.3924, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.835154253243147e-05, | |
| "loss": 3.0695, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.8337099218705695e-05, | |
| "loss": 3.2666, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.832259508295822e-05, | |
| "loss": 3.135, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.830803016299027e-05, | |
| "loss": 2.947, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.829340449676147e-05, | |
| "loss": 3.13, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.82787181223898e-05, | |
| "loss": 3.038, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.826397107815142e-05, | |
| "loss": 3.0761, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.824916340248064e-05, | |
| "loss": 3.1496, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.823429513396977e-05, | |
| "loss": 2.8758, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.821936631136906e-05, | |
| "loss": 3.218, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.820437697358656e-05, | |
| "loss": 3.0933, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.8189327159688035e-05, | |
| "loss": 3.2388, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.817421690889689e-05, | |
| "loss": 3.1914, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.815904626059401e-05, | |
| "loss": 3.2259, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.814381525431771e-05, | |
| "loss": 3.1867, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.8128523929763626e-05, | |
| "loss": 3.1134, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.811317232678456e-05, | |
| "loss": 3.3846, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8097760485390444e-05, | |
| "loss": 3.2507, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8082288445748194e-05, | |
| "loss": 3.0992, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.806675624818162e-05, | |
| "loss": 3.0715, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.8051163933171316e-05, | |
| "loss": 3.2028, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.8035511541354547e-05, | |
| "loss": 3.2714, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.801979911352519e-05, | |
| "loss": 3.1238, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.800402669063353e-05, | |
| "loss": 3.2508, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.7988194313786275e-05, | |
| "loss": 3.2359, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.797230202424633e-05, | |
| "loss": 2.8538, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.7956349863432794e-05, | |
| "loss": 3.1991, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.794033787292078e-05, | |
| "loss": 3.0467, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7924266094441346e-05, | |
| "loss": 3.1088, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7908134569881344e-05, | |
| "loss": 3.128, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.789194334128338e-05, | |
| "loss": 3.3195, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.7875692450845624e-05, | |
| "loss": 3.2049, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.785938194092176e-05, | |
| "loss": 3.1477, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7843011854020846e-05, | |
| "loss": 3.2506, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7826582232807213e-05, | |
| "loss": 3.1822, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7810093120100374e-05, | |
| "loss": 2.9576, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.779354455887485e-05, | |
| "loss": 3.2086, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.777693659226013e-05, | |
| "loss": 3.2935, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.776026926354052e-05, | |
| "loss": 2.9873, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.774354261615503e-05, | |
| "loss": 3.2671, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.772675669369728e-05, | |
| "loss": 3.0882, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.770991153991537e-05, | |
| "loss": 3.2727, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.769300719871177e-05, | |
| "loss": 2.9907, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7676043714143195e-05, | |
| "loss": 3.1303, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.765902113042053e-05, | |
| "loss": 3.1412, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7641939491908674e-05, | |
| "loss": 3.3585, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7624798843126404e-05, | |
| "loss": 3.1958, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.7607599228746355e-05, | |
| "loss": 3.1581, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.759034069359478e-05, | |
| "loss": 3.224, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.7573023282651536e-05, | |
| "loss": 3.1031, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.75556470410499e-05, | |
| "loss": 3.0914, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.753821201407648e-05, | |
| "loss": 3.2474, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.75207182471711e-05, | |
| "loss": 3.2045, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7503165785926676e-05, | |
| "loss": 2.9288, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.748555467608909e-05, | |
| "loss": 3.3084, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.7467884963557064e-05, | |
| "loss": 3.1761, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.745015669438207e-05, | |
| "loss": 3.0644, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.7432369914768185e-05, | |
| "loss": 3.0874, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.7414524671071994e-05, | |
| "loss": 3.1106, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.739662100980242e-05, | |
| "loss": 2.8364, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.737865897762066e-05, | |
| "loss": 3.0884, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.736063862134004e-05, | |
| "loss": 3.1134, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.734255998792589e-05, | |
| "loss": 3.3668, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.732442312449541e-05, | |
| "loss": 2.9823, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.730622807831758e-05, | |
| "loss": 3.0893, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.728797489681301e-05, | |
| "loss": 3.1091, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.726966362755383e-05, | |
| "loss": 2.9348, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.725129431826355e-05, | |
| "loss": 3.2506, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.723286701681695e-05, | |
| "loss": 2.9876, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.7214381771239946e-05, | |
| "loss": 2.9158, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.719583862970949e-05, | |
| "loss": 3.2035, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.7177237640553414e-05, | |
| "loss": 2.9031, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.7158578852250304e-05, | |
| "loss": 3.2779, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.7139862313429385e-05, | |
| "loss": 3.1219, | |
| "step": 1500 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 9730, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 300, | |
| "total_flos": 1.7702871849566208e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |