SPB Git

spb/wp41_uqo

Public
8.6 KB · 38 lines
1{"event": "train", "step": 20, "loss": -5.79660177230835, "grad_norm": 1.073551893234253, "lr": 0.00012599999999999997, "ex_per_s": 237.29563612709535, "data_frac": 0.2973607573964204, "mem_mb": 62.975488, "nll": -5.79660177230835, "elapsed_s": 5.4}2{"event": "train", "step": 40, "loss": -6.048308849334717, "grad_norm": 0.7612945437431335, "lr": 0.00024599999999999996, "ex_per_s": 362.57818334165876, "data_frac": 0.22835545406328214, "mem_mb": 62.975488, "nll": -6.048308849334717, "elapsed_s": 7.1}3{"event": "train", "step": 60, "loss": -5.96046257019043, "grad_norm": 0.3893465995788574, "lr": 0.0002997798290300283, "ex_per_s": 440.0169721831753, "data_frac": 0.18569225069469597, "mem_mb": 62.975488, "nll": -5.96046257019043, "elapsed_s": 8.8}4{"event": "train", "step": 80, "loss": -5.8896894454956055, "grad_norm": 0.6535124182701111, "lr": 0.0002980227678394385, "ex_per_s": 492.24908612315767, "data_frac": 0.15879572149194213, "mem_mb": 62.975488, "nll": -5.8896894454956055, "elapsed_s": 10.4}5{"event": "train", "step": 100, "loss": -5.932180404663086, "grad_norm": 0.450558066368103, "lr": 0.00029453155143795715, "ex_per_s": 530.3066385483229, "data_frac": 0.13750631112920184, "mem_mb": 62.975488, "nll": -5.932180404663086, "elapsed_s": 12.1}6{"event": "val", "step": 100, "val_nll": -6.062143116221289, "best": -6.062143116221289, "bad": 0, "mem_mb": 62.975488, "elapsed_s": 37.7}7{"event": "train", "step": 120, "loss": -5.895845413208008, "grad_norm": 1.1955969333648682, "lr": 0.00028935169317050475, "ex_per_s": 195.39078126263567, "data_frac": 0.04244296765799833, "mem_mb": 62.975488, "nll": -5.895845413208008, "elapsed_s": 39.4}8{"event": "train", "step": 140, "loss": -5.784059524536133, "grad_norm": 0.1311793178319931, "lr": 0.0002825507204111699, "ex_per_s": 218.6930297009989, "data_frac": 0.04092078088331632, "mem_mb": 62.975488, "nll": -5.784059524536133, "elapsed_s": 41.0}9{"event": "train", "step": 160, "loss": -5.8245158195495605, "grad_norm": 0.1426067352294922, "lr": 0.00027421729424061787, "ex_per_s": 240.13111985807566, "data_frac": 0.04004858479394083, "mem_mb": 62.975488, "nll": -5.8245158195495605, "elapsed_s": 42.7}10{"event": "train", "step": 180, "loss": -5.77628755569458, "grad_norm": 0.8541438579559326, "lr": 0.00026446005361331463, "ex_per_s": 260.01114544529514, "data_frac": 0.03871767163415616, "mem_mb": 62.975488, "nll": -5.77628755569458, "elapsed_s": 44.4}11{"event": "train", "step": 200, "loss": -5.813711643218994, "grad_norm": 0.7303595542907715, "lr": 0.0002534061990826135, "ex_per_s": 278.43269177924714, "data_frac": 0.03752984905459563, "mem_mb": 62.975488, "nll": -5.813711643218994, "elapsed_s": 46.0}12{"event": "val", "step": 200, "val_nll": -6.06184728456237, "best": -6.062143116221289, "bad": 1, "mem_mb": 62.975488, "elapsed_s": 71.4}13{"event": "train", "step": 220, "loss": -5.935263156890869, "grad_norm": 0.3792751431465149, "lr": 0.00024119983454701384, "ex_per_s": 192.63223229268553, "data_frac": 0.023720295174144077, "mem_mb": 62.975488, "nll": -5.935263156890869, "elapsed_s": 73.1}14{"event": "train", "step": 240, "loss": -5.903744697570801, "grad_norm": 1.1215780973434448, "lr": 0.00022800008863546595, "ex_per_s": 205.42988839279863, "data_frac": 0.023600813470925883, "mem_mb": 62.975488, "nll": -5.903744697570801, "elapsed_s": 74.8}15{"event": "train", "step": 260, "loss": -6.040683269500732, "grad_norm": 0.6944382190704346, "lr": 0.0002139790402223365, "ex_per_s": 217.67841743542243, "data_frac": 0.023189267546029246, "mem_mb": 62.975488, "nll": -6.040683269500732, "elapsed_s": 76.5}16{"event": "train", "step": 280, "loss": -5.8309431076049805, "grad_norm": 0.33345961570739746, "lr": 0.00019931947511612076, "ex_per_s": 229.41005935662122, "data_frac": 0.022802582772367228, "mem_mb": 62.975488, "nll": -5.8309431076049805, "elapsed_s": 78.2}17{"event": "train", "step": 300, "loss": -5.931258678436279, "grad_norm": 0.2723250985145569, "lr": 0.00018421250316689352, "ex_per_s": 240.64237649156829, "data_frac": 0.022425128256172946, "mem_mb": 62.975488, "nll": -5.931258678436279, "elapsed_s": 79.8}18{"event": "val", "step": 300, "val_nll": -6.0640847682342045, "best": -6.0640847682342045, "bad": 0, "mem_mb": 62.975488, "elapsed_s": 105.3}19{"event": "train", "step": 320, "loss": -5.870183944702148, "grad_norm": 0.7430810332298279, "lr": 0.00016885506685714895, "ex_per_s": 191.5320445606552, "data_frac": 0.017023865649740753, "mem_mb": 62.975488, "nll": -5.870183944702148, "elapsed_s": 107.0}20{"event": "train", "step": 340, "loss": -5.945537567138672, "grad_norm": 0.27774932980537415, "lr": 0.00015344737385535666, "ex_per_s": 200.37367828684376, "data_frac": 0.01683891577654667, "mem_mb": 62.975488, "nll": -5.945537567138672, "elapsed_s": 108.6}21{"event": "train", "step": 360, "loss": -5.9317779541015625, "grad_norm": 0.15965408086776733, "lr": 0.0001381902870028244, "ex_per_s": 208.9521613932902, "data_frac": 0.016662224662875567, "mem_mb": 62.975488, "nll": -5.9317779541015625, "elapsed_s": 110.3}22{"event": "train", "step": 380, "loss": -5.93388557434082, "grad_norm": 0.10920978337526321, "lr": 0.00012328270575938212, "ex_per_s": 217.26589146507987, "data_frac": 0.016484160721197495, "mem_mb": 62.975488, "nll": -5.93388557434082, "elapsed_s": 112.0}23{"event": "train", "step": 400, "loss": -5.94580602645874, "grad_norm": 0.6662354469299316, "lr": 0.00010891897324474534, "ex_per_s": 225.3368983318657, "data_frac": 0.016514756961523006, "mem_mb": 62.975488, "nll": -5.94580602645874, "elapsed_s": 113.7}24{"event": "val", "step": 400, "val_nll": -6.064307921882754, "best": -6.064307921882754, "bad": 0, "mem_mb": 62.975488, "elapsed_s": 139.2}25{"event": "train", "step": 420, "loss": -5.973221302032471, "grad_norm": 0.6370285153388977, "lr": 9.528634267874014e-05, "ex_per_s": 190.89421367779767, "data_frac": 0.013383953757931362, "mem_mb": 62.975488, "nll": -5.973221302032471, "elapsed_s": 140.9}26{"event": "train", "step": 440, "loss": -5.923673152923584, "grad_norm": 0.11263495683670044, "lr": 8.256253624921584e-05, "ex_per_s": 197.63820912363855, "data_frac": 0.013288505402742307, "mem_mb": 62.975488, "nll": -5.923673152923584, "elapsed_s": 142.5}27{"event": "train", "step": 460, "loss": -5.887053966522217, "grad_norm": 0.30858540534973145, "lr": 7.09134282315354e-05, "ex_per_s": 204.21772137137194, "data_frac": 0.013187767853577042, "mem_mb": 62.975488, "nll": -5.887053966522217, "elapsed_s": 144.2}28{"event": "train", "step": 480, "loss": -5.9157633781433105, "grad_norm": 0.4017937183380127, "lr": 6.049088256372386e-05, "ex_per_s": 210.64344849109528, "data_frac": 0.013254421004791581, "mem_mb": 62.975488, "nll": -5.9157633781433105, "elapsed_s": 145.9}29{"event": "train", "step": 500, "loss": -5.957696914672852, "grad_norm": 0.886857807636261, "lr": 5.143077306779053e-05, "ex_per_s": 216.92659905679682, "data_frac": 0.01316597849813077, "mem_mb": 62.975488, "nll": -5.957696914672852, "elapsed_s": 147.6}30{"event": "val", "step": 500, "val_nll": -6.064198265895228, "best": -6.064307921882754, "bad": 1, "mem_mb": 62.975488, "elapsed_s": 173.0}31{"event": "train", "step": 520, "loss": -5.997697830200195, "grad_norm": 0.18197184801101685, "lr": 4.385121212666663e-05, "ex_per_s": 190.5486483498406, "data_frac": 0.011170884646488289, "mem_mb": 62.975488, "nll": -5.997697830200195, "elapsed_s": 174.7}32{"event": "train", "step": 540, "loss": -5.948782920837402, "grad_norm": 0.5912473797798157, "lr": 3.785101090866418e-05, "ex_per_s": 195.99118706718582, "data_frac": 0.01123519242138573, "mem_mb": 62.975488, "nll": -5.948782920837402, "elapsed_s": 176.4}33{"event": "train", "step": 560, "loss": -5.8445210456848145, "grad_norm": 0.6760861277580261, "lr": 3.3508391212784984e-05, "ex_per_s": 201.34114429074137, "data_frac": 0.011180878653684835, "mem_mb": 62.975488, "nll": -5.8445210456848145, "elapsed_s": 178.1}34{"event": "train", "step": 580, "loss": -5.819464206695557, "grad_norm": 0.2640511095523834, "lr": 3.087996572794572e-05, "ex_per_s": 206.59323148639226, "data_frac": 0.011124701048004365, "mem_mb": 62.975488, "nll": -5.819464206695557, "elapsed_s": 179.7}35{"event": "train", "step": 600, "loss": -5.882576942443848, "grad_norm": 0.6143320798873901, "lr": 2.9999999999999997e-05, "ex_per_s": 211.74905713811438, "data_frac": 0.011069131901058838, "mem_mb": 62.975488, "nll": -5.882576942443848, "elapsed_s": 181.4}36{"event": "val", "step": 600, "val_nll": -6.066347980807233, "best": -6.066347980807233, "bad": 0, "mem_mb": 62.975488, "elapsed_s": 206.9}37{"event": "done", "steps": 600, "best_val": -6.066347980807233, "wall_s": 206.8860673904419, "ex_per_s": 185.6094056229041, "peak_mem_mb": 62.975488, "n_params": 3455619, "device": "mps", "data_time_frac": 0.009702687797929493, "elapsed_s": 206.9}38
Files · main
Menu

Theme