| { | |
| "best_global_step": 2712, | |
| "best_metric": 0.06098712, | |
| "best_model_checkpoint": "/home/mithil/PycharmProjects/HebrewSearch/output/HIT-TMG/KaLM-embedding-multilingual-mini-instruct-v2/v0-20250927-150301/checkpoint-2712", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 3616, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0011061946902654867, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 5.999998867771367e-05, | |
| "loss": 0.5382940769195557, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0055309734513274336, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 5.999971694326911e-05, | |
| "loss": 0.10738233476877213, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011061946902654867, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 5.9998867778417846e-05, | |
| "loss": 0.11313067674636841, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.016592920353982302, | |
| "grad_norm": 1.0390625, | |
| "learning_rate": 5.9997452521470316e-05, | |
| "loss": 0.09977512359619141, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.022123893805309734, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 5.9995471199133074e-05, | |
| "loss": 0.07893570661544799, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02765486725663717, | |
| "grad_norm": 1.34375, | |
| "learning_rate": 5.999292384879455e-05, | |
| "loss": 0.08480485081672669, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.033185840707964605, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 5.9989810518524385e-05, | |
| "loss": 0.08676509857177735, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03871681415929203, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 5.9986131267072534e-05, | |
| "loss": 0.09130167365074157, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04424778761061947, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 5.998188616386811e-05, | |
| "loss": 0.1171728253364563, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.049778761061946904, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 5.997707528901811e-05, | |
| "loss": 0.07707504630088806, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05530973451327434, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 5.997169873330592e-05, | |
| "loss": 0.07726281881332397, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06084070796460177, | |
| "grad_norm": 1.3046875, | |
| "learning_rate": 5.996575659818954e-05, | |
| "loss": 0.08723951578140259, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06637168141592921, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 5.995924899579974e-05, | |
| "loss": 0.07960972785949708, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07190265486725664, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 5.995217604893789e-05, | |
| "loss": 0.07275732159614563, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07743362831858407, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 5.994453789107367e-05, | |
| "loss": 0.0901167094707489, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08296460176991151, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.9936334666342555e-05, | |
| "loss": 0.06990140080451965, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.08849557522123894, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 5.992756652954306e-05, | |
| "loss": 0.06333508491516113, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09402654867256637, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 5.991823364613388e-05, | |
| "loss": 0.06881271004676819, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.09955752212389381, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 5.9908336192230694e-05, | |
| "loss": 0.06817507743835449, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.10508849557522124, | |
| "grad_norm": 0.98046875, | |
| "learning_rate": 5.989787435460292e-05, | |
| "loss": 0.0841898500919342, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.11061946902654868, | |
| "grad_norm": 0.9921875, | |
| "learning_rate": 5.988684833067011e-05, | |
| "loss": 0.10564028024673462, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1161504424778761, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 5.9875258328498294e-05, | |
| "loss": 0.08748467564582825, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.12168141592920353, | |
| "grad_norm": 1.21875, | |
| "learning_rate": 5.9863104566796004e-05, | |
| "loss": 0.07313825488090515, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12721238938053098, | |
| "grad_norm": 1.0, | |
| "learning_rate": 5.985038727491019e-05, | |
| "loss": 0.062137752771377563, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.13274336283185842, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.9837106692821844e-05, | |
| "loss": 0.0795306146144867, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.13827433628318583, | |
| "grad_norm": 1.25, | |
| "learning_rate": 5.9823263071141524e-05, | |
| "loss": 0.08484375476837158, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.14380530973451328, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 5.980885667110457e-05, | |
| "loss": 0.08816001415252686, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14933628318584072, | |
| "grad_norm": 1.1953125, | |
| "learning_rate": 5.9793887764566214e-05, | |
| "loss": 0.08630930185317993, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.15486725663716813, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 5.9778356633996446e-05, | |
| "loss": 0.08513998985290527, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16039823008849557, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.976226357247467e-05, | |
| "loss": 0.08681740760803222, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.16592920353982302, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 5.974560888368417e-05, | |
| "loss": 0.09298715591430665, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17146017699115043, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 5.972839288190641e-05, | |
| "loss": 0.08679794669151306, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.17699115044247787, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 5.9710615892015064e-05, | |
| "loss": 0.08733226656913758, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18252212389380532, | |
| "grad_norm": 1.0390625, | |
| "learning_rate": 5.96922782494699e-05, | |
| "loss": 0.08682148456573487, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.18805309734513273, | |
| "grad_norm": 0.625, | |
| "learning_rate": 5.967338030031047e-05, | |
| "loss": 0.0575802743434906, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19358407079646017, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 5.9653922401149554e-05, | |
| "loss": 0.0705379605293274, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.19911504424778761, | |
| "grad_norm": 1.265625, | |
| "learning_rate": 5.963390491916644e-05, | |
| "loss": 0.0910179078578949, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.20464601769911506, | |
| "grad_norm": 1.234375, | |
| "learning_rate": 5.961332823209998e-05, | |
| "loss": 0.07859574556350708, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.21017699115044247, | |
| "grad_norm": 1.3984375, | |
| "learning_rate": 5.9592192728241516e-05, | |
| "loss": 0.0651554822921753, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2157079646017699, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 5.957049880642748e-05, | |
| "loss": 0.06760371923446655, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.22123893805309736, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 5.9548246876031906e-05, | |
| "loss": 0.06698645949363709, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22676991150442477, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 5.9525437356958724e-05, | |
| "loss": 0.07442606687545776, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.2323008849557522, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 5.950207067963377e-05, | |
| "loss": 0.09590929746627808, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23783185840707965, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 5.9478147284996735e-05, | |
| "loss": 0.07985970377922058, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.24336283185840707, | |
| "grad_norm": 1.0, | |
| "learning_rate": 5.9453667624492834e-05, | |
| "loss": 0.0752497673034668, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2488938053097345, | |
| "grad_norm": 1.125, | |
| "learning_rate": 5.942863216006421e-05, | |
| "loss": 0.0798652172088623, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.25442477876106195, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.940304136414135e-05, | |
| "loss": 0.07292602658271789, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.25995575221238937, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 5.937689571963402e-05, | |
| "loss": 0.08627229332923889, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.26548672566371684, | |
| "grad_norm": 1.1796875, | |
| "learning_rate": 5.935019571992228e-05, | |
| "loss": 0.0845943570137024, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27101769911504425, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 5.932294186884711e-05, | |
| "loss": 0.09052139520645142, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.27654867256637167, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 5.9295134680700905e-05, | |
| "loss": 0.06385719776153564, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28207964601769914, | |
| "grad_norm": 1.0390625, | |
| "learning_rate": 5.9266774680217786e-05, | |
| "loss": 0.06568827629089355, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.28761061946902655, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 5.923786240256368e-05, | |
| "loss": 0.06642733216285705, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29314159292035397, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 5.920839839332624e-05, | |
| "loss": 0.08733498454093933, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.29867256637168144, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.9178383208504556e-05, | |
| "loss": 0.0854425847530365, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.30420353982300885, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 5.914781741449862e-05, | |
| "loss": 0.0675132155418396, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.30973451327433627, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 5.9116701588098674e-05, | |
| "loss": 0.06759149432182313, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31526548672566373, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 5.908503631647435e-05, | |
| "loss": 0.07054391503334045, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.32079646017699115, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 5.905282219716351e-05, | |
| "loss": 0.0704015076160431, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32632743362831856, | |
| "grad_norm": 1.3515625, | |
| "learning_rate": 5.9020059838061054e-05, | |
| "loss": 0.10305986404418946, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.33185840707964603, | |
| "grad_norm": 0.96484375, | |
| "learning_rate": 5.89867498574074e-05, | |
| "loss": 0.09359662532806397, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33738938053097345, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 5.895289288377682e-05, | |
| "loss": 0.06781641840934753, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.34292035398230086, | |
| "grad_norm": 0.98828125, | |
| "learning_rate": 5.8918489556065595e-05, | |
| "loss": 0.07854382991790772, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34845132743362833, | |
| "grad_norm": 1.2421875, | |
| "learning_rate": 5.888354052347997e-05, | |
| "loss": 0.07902944087982178, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.35398230088495575, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 5.8848046445523876e-05, | |
| "loss": 0.07207826375961304, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35951327433628316, | |
| "grad_norm": 1.1875, | |
| "learning_rate": 5.881200799198647e-05, | |
| "loss": 0.06817350387573243, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.36504424778761063, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.877542584292956e-05, | |
| "loss": 0.08363889455795288, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.37057522123893805, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 5.8738300688674716e-05, | |
| "loss": 0.0724075436592102, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.37610619469026546, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 5.870063322979024e-05, | |
| "loss": 0.06965845823287964, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.38163716814159293, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 5.866242417707799e-05, | |
| "loss": 0.09595088362693786, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.38716814159292035, | |
| "grad_norm": 1.3046875, | |
| "learning_rate": 5.8623674251559944e-05, | |
| "loss": 0.08167945146560669, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3926991150442478, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 5.858438418446458e-05, | |
| "loss": 0.0965988039970398, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.39823008849557523, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.8544554717213096e-05, | |
| "loss": 0.06779979467391968, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.40376106194690264, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 5.8504186601405424e-05, | |
| "loss": 0.06773656606674194, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.4092920353982301, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 5.8463280598806e-05, | |
| "loss": 0.08192671537399292, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41482300884955753, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 5.842183748132945e-05, | |
| "loss": 0.0629011332988739, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.42035398230088494, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 5.8379858031026016e-05, | |
| "loss": 0.06778149008750915, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.4258849557522124, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 5.8337343040066744e-05, | |
| "loss": 0.0666874349117279, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.4314159292035398, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 5.82942933107286e-05, | |
| "loss": 0.06856505870819092, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43694690265486724, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 5.825070965537931e-05, | |
| "loss": 0.07286258935928344, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.4424778761061947, | |
| "grad_norm": 1.265625, | |
| "learning_rate": 5.820659289646197e-05, | |
| "loss": 0.08377938270568848, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4480088495575221, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 5.816194386647965e-05, | |
| "loss": 0.07762463092803955, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.45353982300884954, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 5.811676340797957e-05, | |
| "loss": 0.06519424319267272, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.459070796460177, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 5.807105237353725e-05, | |
| "loss": 0.053890776634216306, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.4646017699115044, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 5.802481162574043e-05, | |
| "loss": 0.05748436450958252, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.47013274336283184, | |
| "grad_norm": 1.171875, | |
| "learning_rate": 5.797804203717277e-05, | |
| "loss": 0.07670645713806153, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.4756637168141593, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 5.793074449039738e-05, | |
| "loss": 0.055212479829788205, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.4811946902654867, | |
| "grad_norm": 0.875, | |
| "learning_rate": 5.788291987794021e-05, | |
| "loss": 0.09868308901786804, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.48672566371681414, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 5.783456910227315e-05, | |
| "loss": 0.07782126665115356, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.4922566371681416, | |
| "grad_norm": 0.98828125, | |
| "learning_rate": 5.778569307579702e-05, | |
| "loss": 0.07843884229660034, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.497787610619469, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 5.773629272082439e-05, | |
| "loss": 0.0664928674697876, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5033185840707964, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 5.768636896956212e-05, | |
| "loss": 0.07253829240798951, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.5088495575221239, | |
| "grad_norm": 1.125, | |
| "learning_rate": 5.7635922764093794e-05, | |
| "loss": 0.08696393966674805, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5143805309734514, | |
| "grad_norm": 1.0, | |
| "learning_rate": 5.758495505636195e-05, | |
| "loss": 0.06448702812194824, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.5199115044247787, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 5.7533466808150104e-05, | |
| "loss": 0.07809814810752869, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.5254424778761062, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 5.74814589910646e-05, | |
| "loss": 0.054554694890975954, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.5309734513274337, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 5.7428932586516285e-05, | |
| "loss": 0.07724121809005738, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.536504424778761, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 5.7375888585701986e-05, | |
| "loss": 0.07646011114120484, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.5420353982300885, | |
| "grad_norm": 1.0390625, | |
| "learning_rate": 5.732232798958579e-05, | |
| "loss": 0.08474193215370178, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.547566371681416, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 5.72682518088802e-05, | |
| "loss": 0.09357210397720336, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.5530973451327433, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 5.7213661064026994e-05, | |
| "loss": 0.0862844467163086, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5586283185840708, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 5.7158556785178026e-05, | |
| "loss": 0.08219806551933288, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.5641592920353983, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 5.7102940012175775e-05, | |
| "loss": 0.0511793851852417, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.5696902654867256, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 5.70468117945337e-05, | |
| "loss": 0.05884026288986206, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.5752212389380531, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 5.699017319141644e-05, | |
| "loss": 0.08477156162261963, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.5807522123893806, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 5.6933025271619865e-05, | |
| "loss": 0.0648829996585846, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.5862831858407079, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.6875369113550855e-05, | |
| "loss": 0.06479321718215943, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.5918141592920354, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 5.6817205805207005e-05, | |
| "loss": 0.05976671576499939, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.5973451327433629, | |
| "grad_norm": 0.99609375, | |
| "learning_rate": 5.675853644415602e-05, | |
| "loss": 0.06818843483924866, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.6028761061946902, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 5.669936213751508e-05, | |
| "loss": 0.07284531593322754, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.6084070796460177, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 5.663968400192991e-05, | |
| "loss": 0.08285737037658691, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6139380530973452, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 5.65795031635537e-05, | |
| "loss": 0.07132169008255004, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.6194690265486725, | |
| "grad_norm": 1.2890625, | |
| "learning_rate": 5.651882075802588e-05, | |
| "loss": 0.06471747159957886, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 1.234375, | |
| "learning_rate": 5.645763793045065e-05, | |
| "loss": 0.04916921257972717, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.6305309734513275, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 5.639595583537545e-05, | |
| "loss": 0.06438320279121398, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.6360619469026548, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 5.6333775636769084e-05, | |
| "loss": 0.06726735234260559, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.6415929203539823, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 5.6271098507999796e-05, | |
| "loss": 0.05601220726966858, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.6471238938053098, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 5.6207925631813133e-05, | |
| "loss": 0.050103670358657836, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.6526548672566371, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 5.6144258200309627e-05, | |
| "loss": 0.05740664005279541, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.6581858407079646, | |
| "grad_norm": 1.3125, | |
| "learning_rate": 5.608009741492226e-05, | |
| "loss": 0.08702932000160217, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.6637168141592921, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.601544448639384e-05, | |
| "loss": 0.07099788784980773, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6692477876106194, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 5.595030063475417e-05, | |
| "loss": 0.07439646124839783, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.6747787610619469, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 5.588466708929694e-05, | |
| "loss": 0.0887711524963379, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.6803097345132744, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 5.581854508855661e-05, | |
| "loss": 0.0802302360534668, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.6858407079646017, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 5.5751935880285e-05, | |
| "loss": 0.06470907926559448, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.6913716814159292, | |
| "grad_norm": 1.2421875, | |
| "learning_rate": 5.568484072142776e-05, | |
| "loss": 0.07490586042404175, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.6969026548672567, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 5.561726087810065e-05, | |
| "loss": 0.07720963358879089, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.702433628318584, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 5.554919762556564e-05, | |
| "loss": 0.05913523435592651, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.7079646017699115, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 5.548065224820683e-05, | |
| "loss": 0.0728110432624817, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.713495575221239, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 5.541162603950627e-05, | |
| "loss": 0.06673948168754577, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.7190265486725663, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.5342120302019485e-05, | |
| "loss": 0.06224437952041626, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7245575221238938, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 5.527213634735093e-05, | |
| "loss": 0.07510324120521546, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.7300884955752213, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 5.520167549612923e-05, | |
| "loss": 0.08394646644592285, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.7356194690265486, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 5.513073907798227e-05, | |
| "loss": 0.08188437223434449, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.7411504424778761, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 5.5059328431512085e-05, | |
| "loss": 0.0681014358997345, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.7466814159292036, | |
| "grad_norm": 1.390625, | |
| "learning_rate": 5.498744490426963e-05, | |
| "loss": 0.10296804904937744, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.7522123893805309, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 5.49150898527293e-05, | |
| "loss": 0.06941713690757752, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.7577433628318584, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 5.4842264642263406e-05, | |
| "loss": 0.08192881941795349, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.7632743362831859, | |
| "grad_norm": 1.6484375, | |
| "learning_rate": 5.476897064711633e-05, | |
| "loss": 0.07000873684883117, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.7688053097345132, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 5.469520925037867e-05, | |
| "loss": 0.08420374393463134, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.7743362831858407, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 5.462098184396106e-05, | |
| "loss": 0.04759725034236908, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7798672566371682, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 5.454628982856798e-05, | |
| "loss": 0.07408639192581176, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.7853982300884956, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 5.447113461367127e-05, | |
| "loss": 0.06126109957695007, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.790929203539823, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 5.439551761748357e-05, | |
| "loss": 0.06557047367095947, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.7964601769911505, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 5.431944026693152e-05, | |
| "loss": 0.06162294149398804, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.8019911504424779, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 5.424290399762886e-05, | |
| "loss": 0.07381423711776733, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.8075221238938053, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 5.416591025384935e-05, | |
| "loss": 0.080774986743927, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.8130530973451328, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 5.4088460488499465e-05, | |
| "loss": 0.07907372713088989, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.8185840707964602, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 5.4010556163091045e-05, | |
| "loss": 0.07035250663757324, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8241150442477876, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 5.393219874771366e-05, | |
| "loss": 0.08394900560379029, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.8296460176991151, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 5.38533897210069e-05, | |
| "loss": 0.05907307863235474, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8351769911504425, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 5.377413057013247e-05, | |
| "loss": 0.07112559676170349, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.8407079646017699, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 5.369442279074611e-05, | |
| "loss": 0.06508824825286866, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.8462389380530974, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 5.3614267886969374e-05, | |
| "loss": 0.07256530523300171, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.8517699115044248, | |
| "grad_norm": 0.58203125, | |
| "learning_rate": 5.3533667371361276e-05, | |
| "loss": 0.05574750900268555, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.8573008849557522, | |
| "grad_norm": 1.0, | |
| "learning_rate": 5.345262276488971e-05, | |
| "loss": 0.05882116556167603, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.8628318584070797, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 5.3371135596902766e-05, | |
| "loss": 0.059357255697250366, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.8683628318584071, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 5.328920740509987e-05, | |
| "loss": 0.08527318239212037, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.8738938053097345, | |
| "grad_norm": 0.95703125, | |
| "learning_rate": 5.320683973550276e-05, | |
| "loss": 0.0792492926120758, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.879424778761062, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 5.312403414242633e-05, | |
| "loss": 0.07450472116470337, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.8849557522123894, | |
| "grad_norm": 0.75, | |
| "learning_rate": 5.304079218844926e-05, | |
| "loss": 0.07032421827316285, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8904867256637168, | |
| "grad_norm": 1.125, | |
| "learning_rate": 5.295711544438461e-05, | |
| "loss": 0.06430639028549194, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.8960176991150443, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 5.2873005489250046e-05, | |
| "loss": 0.07071130275726319, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.9015486725663717, | |
| "grad_norm": 1.6953125, | |
| "learning_rate": 5.2788463910238204e-05, | |
| "loss": 0.07589119672775269, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.9070796460176991, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 5.2703492302686586e-05, | |
| "loss": 0.08046534061431884, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.9126106194690266, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 5.261809227004756e-05, | |
| "loss": 0.07982348203659058, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.918141592920354, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 5.253226542385808e-05, | |
| "loss": 0.049767687916755676, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9236725663716814, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 5.244601338370923e-05, | |
| "loss": 0.057305270433425905, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.9292035398230089, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 5.235933777721573e-05, | |
| "loss": 0.09568209648132324, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.9347345132743363, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 5.227224023998513e-05, | |
| "loss": 0.07460667490959168, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.9402654867256637, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 5.218472241558709e-05, | |
| "loss": 0.08460034728050232, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.9457964601769911, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 5.209678595552219e-05, | |
| "loss": 0.06128603219985962, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.9513274336283186, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 5.200843251919091e-05, | |
| "loss": 0.06983382701873779, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.956858407079646, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 5.1919663773862225e-05, | |
| "loss": 0.05572186708450318, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.9623893805309734, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 5.1830481394642204e-05, | |
| "loss": 0.06615583896636963, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.9679203539823009, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 5.174088706444236e-05, | |
| "loss": 0.0761763870716095, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.9734513274336283, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 5.16508824739479e-05, | |
| "loss": 0.06981696486473084, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.9789823008849557, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 5.156046932158586e-05, | |
| "loss": 0.07374089360237121, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.9845132743362832, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 5.146964931349295e-05, | |
| "loss": 0.06346363425254822, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.9900442477876106, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 5.137842416348352e-05, | |
| "loss": 0.0630080223083496, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.995575221238938, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 5.128679559301706e-05, | |
| "loss": 0.08187967538833618, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.07009916007518768, | |
| "eval_pearson_cosine": 0.46947768330574036, | |
| "eval_pearson_dot_product": 0.4694776237010956, | |
| "eval_pearson_euclidean": 0.4703822433948517, | |
| "eval_pearson_manhattan": 0.46604230999946594, | |
| "eval_runtime": 111.7809, | |
| "eval_samples_per_second": 72.785, | |
| "eval_spearman_cosine": 0.47130851137552526, | |
| "eval_spearman_dot_product": 0.47130876501958957, | |
| "eval_spearman_euclidean": 0.4713084809389775, | |
| "eval_spearman_manhattan": 0.46547332576597833, | |
| "eval_steps_per_second": 12.131, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.0011061946902655, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 5.119476533116581e-05, | |
| "loss": 0.06849802136421204, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.0066371681415929, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 5.1102335114582134e-05, | |
| "loss": 0.049222135543823244, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.0121681415929205, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 5.1009506687465665e-05, | |
| "loss": 0.039334213733673094, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.0176991150442478, | |
| "grad_norm": 0.96875, | |
| "learning_rate": 5.09162818015305e-05, | |
| "loss": 0.04568702876567841, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.0232300884955752, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 5.082266221597204e-05, | |
| "loss": 0.04934586882591248, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.0287610619469028, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 5.07286496974339e-05, | |
| "loss": 0.060620611906051634, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0342920353982301, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 5.063424601997449e-05, | |
| "loss": 0.05456501245498657, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.0398230088495575, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 5.053945296503355e-05, | |
| "loss": 0.04340373575687408, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.045353982300885, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 5.0444272321398565e-05, | |
| "loss": 0.03940421938896179, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.0508849557522124, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 5.034870588517101e-05, | |
| "loss": 0.05013185739517212, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0564159292035398, | |
| "grad_norm": 0.625, | |
| "learning_rate": 5.02527554597324e-05, | |
| "loss": 0.06063980460166931, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.0619469026548674, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 5.015642285571032e-05, | |
| "loss": 0.049744722247123715, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0674778761061947, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 5.005970989094424e-05, | |
| "loss": 0.05051639676094055, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.073008849557522, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 4.9962618390451205e-05, | |
| "loss": 0.043956026434898376, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.0785398230088497, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 4.98651501863914e-05, | |
| "loss": 0.04485152959823609, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.084070796460177, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 4.976730711803355e-05, | |
| "loss": 0.0462716281414032, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.0896017699115044, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 4.966909103172029e-05, | |
| "loss": 0.044433417916297915, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.095132743362832, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 4.957050378083321e-05, | |
| "loss": 0.052460002899169925, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.1006637168141593, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 4.9471547225758e-05, | |
| "loss": 0.05548023581504822, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.1061946902654867, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 4.9372223233849253e-05, | |
| "loss": 0.048399418592453, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.1117256637168142, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.927253367939525e-05, | |
| "loss": 0.0483365923166275, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.1172566371681416, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 4.917248044358263e-05, | |
| "loss": 0.05342907905578613, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.122787610619469, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 4.9072065414460856e-05, | |
| "loss": 0.053766649961471555, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.1283185840707965, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 4.8971290486906565e-05, | |
| "loss": 0.04207713305950165, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.133849557522124, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 4.887015756258787e-05, | |
| "loss": 0.04382403790950775, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.1393805309734513, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 4.876866854992845e-05, | |
| "loss": 0.048118185997009275, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.1449115044247788, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 4.8666825364071486e-05, | |
| "loss": 0.03880670666694641, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.1504424778761062, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 4.856462992684362e-05, | |
| "loss": 0.04380129873752594, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.1559734513274336, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 4.84620841667186e-05, | |
| "loss": 0.05868492722511291, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.1615044247787611, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 4.8359190018780925e-05, | |
| "loss": 0.04839249551296234, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.1670353982300885, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 4.825594942468936e-05, | |
| "loss": 0.04033098518848419, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.1725663716814159, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 4.815236433264022e-05, | |
| "loss": 0.057745051383972165, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.1780973451327434, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 4.804843669733068e-05, | |
| "loss": 0.03872983753681183, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.1836283185840708, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 4.794416847992186e-05, | |
| "loss": 0.046683937311172485, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.1891592920353982, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 4.7839561648001816e-05, | |
| "loss": 0.04324590265750885, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.1946902654867257, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 4.773461817554838e-05, | |
| "loss": 0.04969582259654999, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.200221238938053, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 4.7629340042892005e-05, | |
| "loss": 0.0490993469953537, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.2057522123893805, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 4.7523729236678264e-05, | |
| "loss": 0.04340413212776184, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.211283185840708, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 4.7417787749830465e-05, | |
| "loss": 0.03859983086585998, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.2168141592920354, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 4.7311517581512014e-05, | |
| "loss": 0.04902164340019226, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.2223451327433628, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 4.7204920737088674e-05, | |
| "loss": 0.057264649868011476, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.2278761061946903, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 4.7097999228090725e-05, | |
| "loss": 0.04205987453460693, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.2334070796460177, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 4.699075507217503e-05, | |
| "loss": 0.05334398150444031, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.238938053097345, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 4.6883190293086916e-05, | |
| "loss": 0.06391674876213074, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.2444690265486726, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 4.677530692062205e-05, | |
| "loss": 0.058903437852859494, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 4.6667106990588066e-05, | |
| "loss": 0.03916058540344238, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.2555309734513274, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 4.6558592544766214e-05, | |
| "loss": 0.04341459572315216, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.261061946902655, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 4.6449765630872756e-05, | |
| "loss": 0.04388400912284851, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.2665929203539823, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 4.634062830252041e-05, | |
| "loss": 0.05491511225700378, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.2721238938053097, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 4.623118261917953e-05, | |
| "loss": 0.04583500921726227, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.2776548672566372, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 4.6121430646139274e-05, | |
| "loss": 0.05844584107398987, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.2831858407079646, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 4.601137445446861e-05, | |
| "loss": 0.051510608196258544, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.288716814159292, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 4.5901016120977274e-05, | |
| "loss": 0.05934204459190369, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.2942477876106195, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 4.579035772817653e-05, | |
| "loss": 0.03599594533443451, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.299778761061947, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 4.567940136423991e-05, | |
| "loss": 0.035601720213890076, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.3053097345132743, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 4.556814912296379e-05, | |
| "loss": 0.04793946444988251, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.3108407079646018, | |
| "grad_norm": 1.21875, | |
| "learning_rate": 4.545660310372788e-05, | |
| "loss": 0.06304795742034912, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.3163716814159292, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 4.534476541145563e-05, | |
| "loss": 0.03024916648864746, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.3219026548672566, | |
| "grad_norm": 0.60546875, | |
| "learning_rate": 4.523263815657447e-05, | |
| "loss": 0.06463109254837036, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.3274336283185841, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 4.5120223454976e-05, | |
| "loss": 0.046007287502288816, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.3329646017699115, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.50075234279761e-05, | |
| "loss": 0.039402684569358824, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 1.338495575221239, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.489454020227485e-05, | |
| "loss": 0.056322628259658815, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.3440265486725664, | |
| "grad_norm": 0.75, | |
| "learning_rate": 4.47812759099164e-05, | |
| "loss": 0.06399442553520203, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 1.3495575221238938, | |
| "grad_norm": 1.21875, | |
| "learning_rate": 4.46677326882488e-05, | |
| "loss": 0.04439619183540344, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.3550884955752212, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 4.455391267988358e-05, | |
| "loss": 0.04778734147548676, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.3606194690265487, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 4.443981803265536e-05, | |
| "loss": 0.0571810781955719, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.366150442477876, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 4.432545089958135e-05, | |
| "loss": 0.04388399422168732, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.3716814159292037, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 4.421081343882065e-05, | |
| "loss": 0.055164027214050296, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.377212389380531, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 4.40959078136336e-05, | |
| "loss": 0.038369673490524295, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.3827433628318584, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 4.39807361923409e-05, | |
| "loss": 0.036572834849357604, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.3882743362831858, | |
| "grad_norm": 0.875, | |
| "learning_rate": 4.3865300748282744e-05, | |
| "loss": 0.052336317300796506, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.3938053097345133, | |
| "grad_norm": 0.99609375, | |
| "learning_rate": 4.374960365977772e-05, | |
| "loss": 0.05086069107055664, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.3993362831858407, | |
| "grad_norm": 0.48046875, | |
| "learning_rate": 4.363364711008185e-05, | |
| "loss": 0.05480511784553528, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.4048672566371683, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 4.3517433287347235e-05, | |
| "loss": 0.043598875403404236, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.4103982300884956, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 4.340096438458086e-05, | |
| "loss": 0.048217812180519105, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.415929203539823, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 4.328424259960319e-05, | |
| "loss": 0.0585234522819519, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.4214601769911503, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 4.316727013500667e-05, | |
| "loss": 0.04403064548969269, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.426991150442478, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 4.305004919811421e-05, | |
| "loss": 0.06126787066459656, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.4325221238938053, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.2932582000937476e-05, | |
| "loss": 0.03779804706573486, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.4380530973451329, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 4.2814870760135214e-05, | |
| "loss": 0.05430150032043457, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.4435840707964602, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 4.2696917696971324e-05, | |
| "loss": 0.044951480627059934, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.4491150442477876, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.257872503727306e-05, | |
| "loss": 0.03592209219932556, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.454646017699115, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 4.246029501138894e-05, | |
| "loss": 0.06765668988227844, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.4601769911504425, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 4.234162985414668e-05, | |
| "loss": 0.042176234722137454, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.4657079646017699, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 4.222273180481109e-05, | |
| "loss": 0.059264647960662845, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.4712389380530975, | |
| "grad_norm": 0.48046875, | |
| "learning_rate": 4.2103603107041664e-05, | |
| "loss": 0.05022870302200318, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.4767699115044248, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 4.198424600885041e-05, | |
| "loss": 0.04392969608306885, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.4823008849557522, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 4.186466276255932e-05, | |
| "loss": 0.044096311926841734, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.4878318584070795, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 4.174485562475793e-05, | |
| "loss": 0.05090985298156738, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.4933628318584071, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 4.162482685626066e-05, | |
| "loss": 0.06884741187095642, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.4988938053097345, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 4.1504578722064256e-05, | |
| "loss": 0.048919641971588136, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.504424778761062, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 4.1384113491304966e-05, | |
| "loss": 0.04950947761535644, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.5099557522123894, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 4.126343343721574e-05, | |
| "loss": 0.04636361598968506, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.5154867256637168, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 4.114254083708336e-05, | |
| "loss": 0.07762627005577087, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.5210176991150441, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 4.102143797220543e-05, | |
| "loss": 0.04796005189418793, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.5265486725663717, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 4.0900127127847364e-05, | |
| "loss": 0.05315868258476257, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.532079646017699, | |
| "grad_norm": 0.59765625, | |
| "learning_rate": 4.077861059319921e-05, | |
| "loss": 0.0509723424911499, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 1.5376106194690267, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 4.065689066133254e-05, | |
| "loss": 0.043031933903694156, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.543141592920354, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 4.0534969629157045e-05, | |
| "loss": 0.042351946234703064, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 1.5486725663716814, | |
| "grad_norm": 0.625, | |
| "learning_rate": 4.041284979737735e-05, | |
| "loss": 0.0328233003616333, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.5542035398230087, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 4.0290533470449445e-05, | |
| "loss": 0.04973614513874054, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 1.5597345132743363, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 4.0168022956537325e-05, | |
| "loss": 0.04321259260177612, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.5652654867256637, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 4.0045320567469366e-05, | |
| "loss": 0.056918442249298096, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 1.5707964601769913, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 3.992242861869468e-05, | |
| "loss": 0.054710114002227785, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.5763274336283186, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 3.9799349429239534e-05, | |
| "loss": 0.05433679223060608, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.581858407079646, | |
| "grad_norm": 1.1484375, | |
| "learning_rate": 3.9676085321663415e-05, | |
| "loss": 0.06333144903182983, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.5873893805309733, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 3.9552638622015374e-05, | |
| "loss": 0.05038842558860779, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 1.592920353982301, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 3.942901165979001e-05, | |
| "loss": 0.05301803946495056, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.5984513274336283, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 3.930520676788359e-05, | |
| "loss": 0.04807696044445038, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 1.6039823008849559, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 3.918122628254996e-05, | |
| "loss": 0.06260589361190796, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.6095132743362832, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 3.905707254335653e-05, | |
| "loss": 0.05578850507736206, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.6150442477876106, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 3.893274789314006e-05, | |
| "loss": 0.04979705512523651, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.620575221238938, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 3.880825467796248e-05, | |
| "loss": 0.04659028947353363, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 1.6261061946902655, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 3.8683595247066646e-05, | |
| "loss": 0.0442796915769577, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.6316371681415929, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 3.8558771952831945e-05, | |
| "loss": 0.03888770043849945, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.6371681415929205, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.8433787150729935e-05, | |
| "loss": 0.05296626091003418, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.6426991150442478, | |
| "grad_norm": 0.625, | |
| "learning_rate": 3.8308643199279936e-05, | |
| "loss": 0.05492118000984192, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 1.6482300884955752, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 3.818334246000447e-05, | |
| "loss": 0.03605894148349762, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.6537610619469025, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 3.805788729738469e-05, | |
| "loss": 0.05306194424629211, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.6592920353982301, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 3.7932280078815834e-05, | |
| "loss": 0.03526887595653534, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.6648230088495575, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 3.7806523174562466e-05, | |
| "loss": 0.03773034811019897, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 1.670353982300885, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 3.76806189577138e-05, | |
| "loss": 0.03187394142150879, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.6758849557522124, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 3.755456980413892e-05, | |
| "loss": 0.055419129133224485, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 1.6814159292035398, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 3.742837809244189e-05, | |
| "loss": 0.03877014815807343, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.6869469026548671, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 3.730204620391696e-05, | |
| "loss": 0.03875821828842163, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.6924778761061947, | |
| "grad_norm": 0.61328125, | |
| "learning_rate": 3.717557652250353e-05, | |
| "loss": 0.03490549921989441, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.698008849557522, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 3.704897143474126e-05, | |
| "loss": 0.05157696008682251, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 1.7035398230088497, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 3.692223332972494e-05, | |
| "loss": 0.04353027641773224, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.709070796460177, | |
| "grad_norm": 1.2109375, | |
| "learning_rate": 3.679536459905953e-05, | |
| "loss": 0.03230733871459961, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 1.7146017699115044, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 3.6668367636814866e-05, | |
| "loss": 0.04994140565395355, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.7201327433628317, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 3.6541244839480614e-05, | |
| "loss": 0.04984952807426453, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 1.7256637168141593, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 3.641399860592103e-05, | |
| "loss": 0.03698314130306244, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.731194690265487, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 3.6286631337329615e-05, | |
| "loss": 0.0457876056432724, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 1.7367256637168142, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 3.615914543718389e-05, | |
| "loss": 0.06582319736480713, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.7422566371681416, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 3.603154331119999e-05, | |
| "loss": 0.0547947883605957, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.747787610619469, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 3.590382736728731e-05, | |
| "loss": 0.053344517946243286, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.7533185840707963, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 3.5776000015503e-05, | |
| "loss": 0.04027817845344543, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 1.758849557522124, | |
| "grad_norm": 0.546875, | |
| "learning_rate": 3.564806366800655e-05, | |
| "loss": 0.05831453800201416, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.7643805309734515, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 3.552002073901426e-05, | |
| "loss": 0.049893522262573244, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 1.7699115044247788, | |
| "grad_norm": 0.96875, | |
| "learning_rate": 3.539187364475364e-05, | |
| "loss": 0.04103112518787384, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.7754424778761062, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 3.526362480341788e-05, | |
| "loss": 0.04785359501838684, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 1.7809734513274336, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 3.513527663512015e-05, | |
| "loss": 0.05216856002807617, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.786504424778761, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 3.500683156184797e-05, | |
| "loss": 0.04190321564674378, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 1.7920353982300885, | |
| "grad_norm": 0.55859375, | |
| "learning_rate": 3.487829200741753e-05, | |
| "loss": 0.040077108144760135, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.797566371681416, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 3.47496603974279e-05, | |
| "loss": 0.046674606204032895, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.8030973451327434, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 3.462093915921526e-05, | |
| "loss": 0.046833425760269165, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.8086283185840708, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 3.449213072180716e-05, | |
| "loss": 0.04743701219558716, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 1.8141592920353982, | |
| "grad_norm": 0.984375, | |
| "learning_rate": 3.43632375158766e-05, | |
| "loss": 0.05728857517242432, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.8196902654867255, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 3.423426197369621e-05, | |
| "loss": 0.046563762426376346, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 1.825221238938053, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 3.4105206529092364e-05, | |
| "loss": 0.0623621940612793, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.8307522123893807, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 3.397607361739919e-05, | |
| "loss": 0.059157317876815795, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 1.836283185840708, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 3.3846865675412695e-05, | |
| "loss": 0.04876360595226288, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.8418141592920354, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 3.371758514134471e-05, | |
| "loss": 0.04870246946811676, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 1.8473451327433628, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.358823445477694e-05, | |
| "loss": 0.040651172399520874, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.8528761061946901, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 3.3458816056614865e-05, | |
| "loss": 0.0449719101190567, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.8584070796460177, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 3.3329332389041745e-05, | |
| "loss": 0.04288926720619202, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.8639380530973453, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 3.3199785895472484e-05, | |
| "loss": 0.034792694449424746, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 1.8694690265486726, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 3.307017902050754e-05, | |
| "loss": 0.05094568133354187, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 3.294051420988683e-05, | |
| "loss": 0.0596714973449707, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 1.8805309734513274, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 3.281079391044347e-05, | |
| "loss": 0.04051285088062286, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.8860619469026547, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 3.268102057005776e-05, | |
| "loss": 0.04296950399875641, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 1.8915929203539823, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 3.255119663761083e-05, | |
| "loss": 0.05117958188056946, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.8971238938053099, | |
| "grad_norm": 0.52734375, | |
| "learning_rate": 3.2421324562938584e-05, | |
| "loss": 0.044080698490142824, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 1.9026548672566372, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 3.22914067967853e-05, | |
| "loss": 0.05295083522796631, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.9081858407079646, | |
| "grad_norm": 0.94140625, | |
| "learning_rate": 3.216144579075757e-05, | |
| "loss": 0.03862910270690918, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.913716814159292, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 3.203144399727785e-05, | |
| "loss": 0.03501943051815033, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.9192477876106193, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.1901403869538356e-05, | |
| "loss": 0.0491629958152771, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.924778761061947, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 3.1771327861454624e-05, | |
| "loss": 0.0392573356628418, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.9303097345132745, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 3.164121842761931e-05, | |
| "loss": 0.06640264987945557, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 1.9358407079646018, | |
| "grad_norm": 0.412109375, | |
| "learning_rate": 3.151107802325581e-05, | |
| "loss": 0.04343979060649872, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.9413716814159292, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 3.138090910417195e-05, | |
| "loss": 0.048336106538772586, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 1.9469026548672566, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 3.125071412671365e-05, | |
| "loss": 0.06405719518661498, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.952433628318584, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 3.112049554771853e-05, | |
| "loss": 0.03451224863529205, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 1.9579646017699115, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 3.0990255824469654e-05, | |
| "loss": 0.030917057394981386, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.963495575221239, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 3.0859997414649e-05, | |
| "loss": 0.03901076912879944, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.9690265486725664, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 3.0729722776291234e-05, | |
| "loss": 0.04505694508552551, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.9745575221238938, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 3.059943436773724e-05, | |
| "loss": 0.04660122394561768, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 1.9800884955752212, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 3.0469134647587736e-05, | |
| "loss": 0.05220373272895813, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.9856194690265485, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 3.0338826074656914e-05, | |
| "loss": 0.04760865271091461, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 1.991150442477876, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 3.0208511107926032e-05, | |
| "loss": 0.03707890212535858, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.9966814159292037, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 3.0078192206496984e-05, | |
| "loss": 0.043943649530410765, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.06138954311609268, | |
| "eval_pearson_cosine": 0.5632613301277161, | |
| "eval_pearson_dot_product": 0.5632612705230713, | |
| "eval_pearson_euclidean": 0.5640513896942139, | |
| "eval_pearson_manhattan": 0.5613865852355957, | |
| "eval_runtime": 110.6327, | |
| "eval_samples_per_second": 73.541, | |
| "eval_spearman_cosine": 0.5579265308325535, | |
| "eval_spearman_dot_product": 0.5579263425192048, | |
| "eval_spearman_euclidean": 0.557926437983032, | |
| "eval_spearman_manhattan": 0.5538608641039989, | |
| "eval_steps_per_second": 12.257, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 2.002212389380531, | |
| "grad_norm": 0.359375, | |
| "learning_rate": 2.9947871829545897e-05, | |
| "loss": 0.04351601302623749, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.0077433628318584, | |
| "grad_norm": 0.408203125, | |
| "learning_rate": 2.981755243627679e-05, | |
| "loss": 0.01824622005224228, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 2.0132743362831858, | |
| "grad_norm": 0.546875, | |
| "learning_rate": 2.9687236485875078e-05, | |
| "loss": 0.02456129193305969, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.018805309734513, | |
| "grad_norm": 0.455078125, | |
| "learning_rate": 2.9556926437461227e-05, | |
| "loss": 0.020082639157772066, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.024336283185841, | |
| "grad_norm": 0.478515625, | |
| "learning_rate": 2.9426624750044306e-05, | |
| "loss": 0.021497070789337158, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.0298672566371683, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 2.9296333882475645e-05, | |
| "loss": 0.02289869785308838, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 2.0353982300884956, | |
| "grad_norm": 0.37890625, | |
| "learning_rate": 2.9166056293402382e-05, | |
| "loss": 0.016062986850738526, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.040929203539823, | |
| "grad_norm": 0.51953125, | |
| "learning_rate": 2.903579444122107e-05, | |
| "loss": 0.016185095906257628, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 2.0464601769911503, | |
| "grad_norm": 0.37890625, | |
| "learning_rate": 2.890555078403132e-05, | |
| "loss": 0.021845296025276184, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.0519911504424777, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 2.877532777958938e-05, | |
| "loss": 0.022076010704040527, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 2.0575221238938055, | |
| "grad_norm": 0.341796875, | |
| "learning_rate": 2.864512788526178e-05, | |
| "loss": 0.022153703868389128, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.063053097345133, | |
| "grad_norm": 0.498046875, | |
| "learning_rate": 2.8514953557978948e-05, | |
| "loss": 0.01988862007856369, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 2.0685840707964602, | |
| "grad_norm": 0.3125, | |
| "learning_rate": 2.8384807254188867e-05, | |
| "loss": 0.020245073735713957, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.0741150442477876, | |
| "grad_norm": 0.546875, | |
| "learning_rate": 2.825469142981067e-05, | |
| "loss": 0.02106291651725769, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.079646017699115, | |
| "grad_norm": 0.5, | |
| "learning_rate": 2.8124608540188368e-05, | |
| "loss": 0.025235247611999512, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.0851769911504423, | |
| "grad_norm": 0.48046875, | |
| "learning_rate": 2.799456104004444e-05, | |
| "loss": 0.026167386770248414, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 2.09070796460177, | |
| "grad_norm": 0.59765625, | |
| "learning_rate": 2.7864551383433577e-05, | |
| "loss": 0.03409063220024109, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.0962389380530975, | |
| "grad_norm": 0.39453125, | |
| "learning_rate": 2.7734582023696348e-05, | |
| "loss": 0.024271368980407715, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 2.101769911504425, | |
| "grad_norm": 0.390625, | |
| "learning_rate": 2.7604655413412882e-05, | |
| "loss": 0.02285452336072922, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.107300884955752, | |
| "grad_norm": 0.61328125, | |
| "learning_rate": 2.7474774004356625e-05, | |
| "loss": 0.019000703096389772, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 2.1128318584070795, | |
| "grad_norm": 0.59765625, | |
| "learning_rate": 2.734494024744803e-05, | |
| "loss": 0.023629143834114075, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.118362831858407, | |
| "grad_norm": 0.61328125, | |
| "learning_rate": 2.721515659270837e-05, | |
| "loss": 0.016240663826465607, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 2.1238938053097347, | |
| "grad_norm": 0.494140625, | |
| "learning_rate": 2.7085425489213425e-05, | |
| "loss": 0.023873592913150787, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.129424778761062, | |
| "grad_norm": 0.50390625, | |
| "learning_rate": 2.6955749385047348e-05, | |
| "loss": 0.02274503409862518, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.1349557522123894, | |
| "grad_norm": 0.47265625, | |
| "learning_rate": 2.682613072725642e-05, | |
| "loss": 0.02265378534793854, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.140486725663717, | |
| "grad_norm": 0.40234375, | |
| "learning_rate": 2.6696571961802857e-05, | |
| "loss": 0.021552833914756774, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 2.146017699115044, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 2.656707553351871e-05, | |
| "loss": 0.0170512780547142, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.1515486725663715, | |
| "grad_norm": 0.458984375, | |
| "learning_rate": 2.64376438860597e-05, | |
| "loss": 0.03171337842941284, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 2.1570796460176993, | |
| "grad_norm": 0.396484375, | |
| "learning_rate": 2.6308279461859066e-05, | |
| "loss": 0.01804209202528, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.1626106194690267, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 2.6178984702081562e-05, | |
| "loss": 0.01640341579914093, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 2.168141592920354, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 2.6049762046577327e-05, | |
| "loss": 0.026275703310966493, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.1736725663716814, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 2.5920613933835843e-05, | |
| "loss": 0.01784980744123459, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 2.1792035398230087, | |
| "grad_norm": 0.384765625, | |
| "learning_rate": 2.579154280093996e-05, | |
| "loss": 0.014947080612182617, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.184734513274336, | |
| "grad_norm": 0.400390625, | |
| "learning_rate": 2.5662551083519864e-05, | |
| "loss": 0.017587074637413026, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.190265486725664, | |
| "grad_norm": 0.4921875, | |
| "learning_rate": 2.5533641215707136e-05, | |
| "loss": 0.02009725272655487, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.1957964601769913, | |
| "grad_norm": 0.546875, | |
| "learning_rate": 2.540481563008886e-05, | |
| "loss": 0.028231135010719298, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 2.2013274336283186, | |
| "grad_norm": 0.455078125, | |
| "learning_rate": 2.527607675766161e-05, | |
| "loss": 0.020630866289138794, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.206858407079646, | |
| "grad_norm": 0.52734375, | |
| "learning_rate": 2.5147427027785703e-05, | |
| "loss": 0.021563749015331268, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 2.2123893805309733, | |
| "grad_norm": 0.4453125, | |
| "learning_rate": 2.501886886813925e-05, | |
| "loss": 0.014575961232185363, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.2179203539823007, | |
| "grad_norm": 0.46484375, | |
| "learning_rate": 2.4890404704672424e-05, | |
| "loss": 0.018347996473312377, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 2.2234513274336285, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 2.4762036961561617e-05, | |
| "loss": 0.024746355414390565, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.228982300884956, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.463376806116376e-05, | |
| "loss": 0.0214070588350296, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 2.234513274336283, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 2.4505600423970563e-05, | |
| "loss": 0.0181646853685379, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.2400442477876106, | |
| "grad_norm": 0.546875, | |
| "learning_rate": 2.437753646856285e-05, | |
| "loss": 0.024569767713546752, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.245575221238938, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 2.4249578611564915e-05, | |
| "loss": 0.017617473006248476, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.2511061946902653, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 2.412172926759895e-05, | |
| "loss": 0.0201715812087059, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 2.256637168141593, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 2.399399084923943e-05, | |
| "loss": 0.017509616911411285, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.2621681415929205, | |
| "grad_norm": 0.40625, | |
| "learning_rate": 2.386636576696764e-05, | |
| "loss": 0.019830062985420227, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 2.267699115044248, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 2.3738856429126156e-05, | |
| "loss": 0.026457592844963074, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.273230088495575, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 2.3611465241873392e-05, | |
| "loss": 0.027082833647727966, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 2.2787610619469025, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 2.3484194609138216e-05, | |
| "loss": 0.01795403063297272, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.28429203539823, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 2.3357046932574577e-05, | |
| "loss": 0.023323030769824983, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 2.2898230088495577, | |
| "grad_norm": 0.376953125, | |
| "learning_rate": 2.323002461151618e-05, | |
| "loss": 0.020825691521167755, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.295353982300885, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 2.3103130042931236e-05, | |
| "loss": 0.02453443259000778, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.3008849557522124, | |
| "grad_norm": 0.44140625, | |
| "learning_rate": 2.2976365621377187e-05, | |
| "loss": 0.027629950642585756, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.3064159292035398, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 2.2849733738955554e-05, | |
| "loss": 0.019730879366397856, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 2.311946902654867, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.272323678526677e-05, | |
| "loss": 0.023601703345775604, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.317477876106195, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 2.2596877147365133e-05, | |
| "loss": 0.024773350358009337, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 2.3230088495575223, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 2.2470657209713676e-05, | |
| "loss": 0.023515698313713074, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.3285398230088497, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 2.2344579354139307e-05, | |
| "loss": 0.018206067383289337, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 2.334070796460177, | |
| "grad_norm": 0.474609375, | |
| "learning_rate": 2.2218645959787704e-05, | |
| "loss": 0.021591377258300782, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.3396017699115044, | |
| "grad_norm": 0.458984375, | |
| "learning_rate": 2.2092859403078545e-05, | |
| "loss": 0.025655537843704224, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 2.3451327433628317, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 2.1967222057660584e-05, | |
| "loss": 0.016385671496391297, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.350663716814159, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 2.1841736294366924e-05, | |
| "loss": 0.01837053894996643, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.356194690265487, | |
| "grad_norm": 0.58203125, | |
| "learning_rate": 2.1716404481170193e-05, | |
| "loss": 0.022889074683189393, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.3617256637168142, | |
| "grad_norm": 0.48046875, | |
| "learning_rate": 2.159122898313798e-05, | |
| "loss": 0.023278559744358062, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 2.3672566371681416, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 2.1466212162388096e-05, | |
| "loss": 0.02303392142057419, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.372787610619469, | |
| "grad_norm": 0.380859375, | |
| "learning_rate": 2.1341356378044037e-05, | |
| "loss": 0.019359081983566284, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 2.3783185840707963, | |
| "grad_norm": 0.470703125, | |
| "learning_rate": 2.1216663986190484e-05, | |
| "loss": 0.019042085111141204, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.383849557522124, | |
| "grad_norm": 0.34765625, | |
| "learning_rate": 2.1092137339828824e-05, | |
| "loss": 0.015889045596122742, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 2.3893805309734515, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 2.0967778788832747e-05, | |
| "loss": 0.0194756418466568, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.394911504424779, | |
| "grad_norm": 0.44921875, | |
| "learning_rate": 2.0843590679903925e-05, | |
| "loss": 0.017387692630290986, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 2.400442477876106, | |
| "grad_norm": 0.484375, | |
| "learning_rate": 2.071957535652769e-05, | |
| "loss": 0.017639362812042238, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.4059734513274336, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 2.059573515892885e-05, | |
| "loss": 0.023314842581748964, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.411504424778761, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 2.04720724240275e-05, | |
| "loss": 0.018998470902442933, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.4170353982300883, | |
| "grad_norm": 0.52734375, | |
| "learning_rate": 2.034858948539493e-05, | |
| "loss": 0.02109953910112381, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 2.422566371681416, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 2.0225288673209614e-05, | |
| "loss": 0.022196565568447114, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.4280973451327434, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 2.010217231421321e-05, | |
| "loss": 0.018953518569469453, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 2.433628318584071, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 1.9979242731666645e-05, | |
| "loss": 0.023169419169425963, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.439159292035398, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 1.9856502245306316e-05, | |
| "loss": 0.02507864236831665, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 2.4446902654867255, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 1.9733953171300263e-05, | |
| "loss": 0.02877911627292633, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 2.4502212389380533, | |
| "grad_norm": 0.625, | |
| "learning_rate": 1.961159782220452e-05, | |
| "loss": 0.01942291557788849, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 2.4557522123893807, | |
| "grad_norm": 0.404296875, | |
| "learning_rate": 1.9489438506919408e-05, | |
| "loss": 0.015319710969924927, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.461283185840708, | |
| "grad_norm": 0.39453125, | |
| "learning_rate": 1.9367477530646034e-05, | |
| "loss": 0.018948142230510712, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.4668141592920354, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 1.924571719484276e-05, | |
| "loss": 0.02216297686100006, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.4723451327433628, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.9124159797181738e-05, | |
| "loss": 0.021475856006145478, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 2.47787610619469, | |
| "grad_norm": 0.373046875, | |
| "learning_rate": 1.9002807631505627e-05, | |
| "loss": 0.016713853180408477, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.4834070796460175, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 1.8881662987784224e-05, | |
| "loss": 0.026210951805114745, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 2.4889380530973453, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 1.8760728152071326e-05, | |
| "loss": 0.017660804092884064, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.4944690265486726, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.864000540646157e-05, | |
| "loss": 0.019666454195976256, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.4453125, | |
| "learning_rate": 1.8519497029047307e-05, | |
| "loss": 0.013202360272407532, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.5055309734513274, | |
| "grad_norm": 0.443359375, | |
| "learning_rate": 1.8399205293875715e-05, | |
| "loss": 0.014842431247234344, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 2.5110619469026547, | |
| "grad_norm": 0.408203125, | |
| "learning_rate": 1.827913247090581e-05, | |
| "loss": 0.022717979550361634, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.5165929203539825, | |
| "grad_norm": 0.4609375, | |
| "learning_rate": 1.8159280825965628e-05, | |
| "loss": 0.019391798973083497, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.52212389380531, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 1.8039652620709506e-05, | |
| "loss": 0.02129274606704712, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.5276548672566372, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 1.7920250112575356e-05, | |
| "loss": 0.02325495332479477, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 2.5331858407079646, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 1.7801075554742073e-05, | |
| "loss": 0.020874102413654328, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.538716814159292, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 1.768213119608706e-05, | |
| "loss": 0.02344939112663269, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 2.5442477876106193, | |
| "grad_norm": 0.42578125, | |
| "learning_rate": 1.7563419281143717e-05, | |
| "loss": 0.01531343162059784, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.5497787610619467, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 1.7444942050059157e-05, | |
| "loss": 0.01874540001153946, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 2.5553097345132745, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 1.7326701738551897e-05, | |
| "loss": 0.026428407430648802, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.560840707964602, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 1.7208700577869668e-05, | |
| "loss": 0.02873435914516449, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 2.566371681415929, | |
| "grad_norm": 0.470703125, | |
| "learning_rate": 1.709094079474733e-05, | |
| "loss": 0.025739896297454833, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.5719026548672566, | |
| "grad_norm": 0.484375, | |
| "learning_rate": 1.6973424611364832e-05, | |
| "loss": 0.028698429465293884, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.577433628318584, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.685615424530528e-05, | |
| "loss": 0.023217637836933137, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.5829646017699117, | |
| "grad_norm": 0.400390625, | |
| "learning_rate": 1.6739131909513084e-05, | |
| "loss": 0.02129518836736679, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 2.588495575221239, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 1.662235981225227e-05, | |
| "loss": 0.018684764206409455, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.5940265486725664, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 1.6505840157064706e-05, | |
| "loss": 0.01910308301448822, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 2.599557522123894, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 1.6389575142728553e-05, | |
| "loss": 0.01732843816280365, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.605088495575221, | |
| "grad_norm": 0.404296875, | |
| "learning_rate": 1.6273566963216807e-05, | |
| "loss": 0.018715257942676543, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 2.6106194690265485, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 1.6157817807655876e-05, | |
| "loss": 0.018781982362270355, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.616150442477876, | |
| "grad_norm": 0.5, | |
| "learning_rate": 1.604232986028426e-05, | |
| "loss": 0.01506403237581253, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 2.6216814159292037, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 1.5927105300411357e-05, | |
| "loss": 0.015091025829315185, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.627212389380531, | |
| "grad_norm": 0.357421875, | |
| "learning_rate": 1.5812146302376305e-05, | |
| "loss": 0.016748161613941194, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.6327433628318584, | |
| "grad_norm": 0.41796875, | |
| "learning_rate": 1.5697455035507e-05, | |
| "loss": 0.018306872248649596, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.6382743362831858, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 1.5583033664079092e-05, | |
| "loss": 0.01826050728559494, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 2.643805309734513, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.546888434727521e-05, | |
| "loss": 0.022905944287776946, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.649336283185841, | |
| "grad_norm": 0.431640625, | |
| "learning_rate": 1.535500923914419e-05, | |
| "loss": 0.0194864958524704, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 2.6548672566371683, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 1.5241410488560417e-05, | |
| "loss": 0.01883869767189026, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.6603982300884956, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 1.5128090239183286e-05, | |
| "loss": 0.023661671578884123, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 2.665929203539823, | |
| "grad_norm": 0.58203125, | |
| "learning_rate": 1.5015050629416757e-05, | |
| "loss": 0.017618557810783385, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.6714601769911503, | |
| "grad_norm": 0.5625, | |
| "learning_rate": 1.4902293792369004e-05, | |
| "loss": 0.021889759600162505, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 2.676991150442478, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 1.4789821855812113e-05, | |
| "loss": 0.02251184284687042, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.682522123893805, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 1.4677636942141993e-05, | |
| "loss": 0.020465926826000215, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.688053097345133, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 1.456574116833833e-05, | |
| "loss": 0.016478195786476135, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.6935840707964602, | |
| "grad_norm": 0.48828125, | |
| "learning_rate": 1.4454136645924589e-05, | |
| "loss": 0.013935254514217376, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 2.6991150442477876, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 1.4342825480928163e-05, | |
| "loss": 0.02402741014957428, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.704646017699115, | |
| "grad_norm": 0.46875, | |
| "learning_rate": 1.4231809773840694e-05, | |
| "loss": 0.021904829144477844, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 2.7101769911504423, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 1.4121091619578395e-05, | |
| "loss": 0.01800040900707245, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.71570796460177, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.4010673107442513e-05, | |
| "loss": 0.017484547197818757, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 2.7212389380530975, | |
| "grad_norm": 0.474609375, | |
| "learning_rate": 1.390055632107992e-05, | |
| "loss": 0.0166101336479187, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.726769911504425, | |
| "grad_norm": 0.59375, | |
| "learning_rate": 1.3790743338443796e-05, | |
| "loss": 0.021221336722373963, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 2.732300884955752, | |
| "grad_norm": 0.404296875, | |
| "learning_rate": 1.3681236231754397e-05, | |
| "loss": 0.02030891180038452, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.7378318584070795, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.3572037067459953e-05, | |
| "loss": 0.018915680050849915, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.7433628318584073, | |
| "grad_norm": 0.375, | |
| "learning_rate": 1.3463147906197702e-05, | |
| "loss": 0.01837272197008133, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.7488938053097343, | |
| "grad_norm": 0.412109375, | |
| "learning_rate": 1.3354570802754981e-05, | |
| "loss": 0.022306661307811736, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 2.754424778761062, | |
| "grad_norm": 0.53515625, | |
| "learning_rate": 1.3246307806030447e-05, | |
| "loss": 0.020701563358306883, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.7599557522123894, | |
| "grad_norm": 0.486328125, | |
| "learning_rate": 1.3138360958995436e-05, | |
| "loss": 0.01721554100513458, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 2.765486725663717, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 1.3030732298655382e-05, | |
| "loss": 0.017794930934906007, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.771017699115044, | |
| "grad_norm": 0.58984375, | |
| "learning_rate": 1.2923423856011412e-05, | |
| "loss": 0.01889959126710892, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 2.7765486725663715, | |
| "grad_norm": 0.486328125, | |
| "learning_rate": 1.2816437656021994e-05, | |
| "loss": 0.015588635206222534, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 2.7820796460176993, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 1.2709775717564696e-05, | |
| "loss": 0.01964143365621567, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 2.7876106194690267, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 1.2603440053398195e-05, | |
| "loss": 0.0225391149520874, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 2.793141592920354, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.2497432670124191e-05, | |
| "loss": 0.02650713324546814, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.7986725663716814, | |
| "grad_norm": 0.404296875, | |
| "learning_rate": 1.2391755568149547e-05, | |
| "loss": 0.017137607932090758, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 2.8042035398230087, | |
| "grad_norm": 0.46875, | |
| "learning_rate": 1.2286410741648615e-05, | |
| "loss": 0.018618327379226685, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 2.8097345132743365, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 1.2181400178525538e-05, | |
| "loss": 0.021288736164569853, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 2.8152654867256635, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 1.2076725860376761e-05, | |
| "loss": 0.024186788499355315, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 2.8207964601769913, | |
| "grad_norm": 0.58203125, | |
| "learning_rate": 1.1972389762453642e-05, | |
| "loss": 0.022785373032093048, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.8263274336283186, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 1.1868393853625166e-05, | |
| "loss": 0.024043235182762145, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 2.831858407079646, | |
| "grad_norm": 0.462890625, | |
| "learning_rate": 1.1764740096340806e-05, | |
| "loss": 0.02035459876060486, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 2.8373893805309733, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 1.1661430446593455e-05, | |
| "loss": 0.015686513483524324, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 2.8429203539823007, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 1.1558466853882572e-05, | |
| "loss": 0.016961218416690828, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 2.8484513274336285, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 1.1455851261177355e-05, | |
| "loss": 0.024346837401390077, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.853982300884956, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 1.1353585604880087e-05, | |
| "loss": 0.022164295613765716, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 2.859513274336283, | |
| "grad_norm": 0.34375, | |
| "learning_rate": 1.1251671814789586e-05, | |
| "loss": 0.018967586755752563, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 2.8650442477876106, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.1150111814064802e-05, | |
| "loss": 0.019415172934532165, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 2.870575221238938, | |
| "grad_norm": 0.5390625, | |
| "learning_rate": 1.1048907519188528e-05, | |
| "loss": 0.018364152312278746, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 2.8761061946902657, | |
| "grad_norm": 0.6015625, | |
| "learning_rate": 1.0948060839931227e-05, | |
| "loss": 0.019008655846118928, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.8816371681415927, | |
| "grad_norm": 0.55078125, | |
| "learning_rate": 1.0847573679314957e-05, | |
| "loss": 0.024682247638702394, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 2.8871681415929205, | |
| "grad_norm": 0.5859375, | |
| "learning_rate": 1.0747447933577563e-05, | |
| "loss": 0.022752675414085387, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 2.892699115044248, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 1.0647685492136797e-05, | |
| "loss": 0.02054031044244766, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 2.898230088495575, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.0548288237554672e-05, | |
| "loss": 0.018094034492969514, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 2.9037610619469025, | |
| "grad_norm": 0.53515625, | |
| "learning_rate": 1.0449258045501999e-05, | |
| "loss": 0.020281152427196504, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.90929203539823, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 1.0350596784722931e-05, | |
| "loss": 0.023686909675598146, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 2.9148230088495577, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 1.0252306316999743e-05, | |
| "loss": 0.017267668247222902, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 2.920353982300885, | |
| "grad_norm": 0.412109375, | |
| "learning_rate": 1.0154388497117654e-05, | |
| "loss": 0.017788177728652953, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 2.9258849557522124, | |
| "grad_norm": 0.5, | |
| "learning_rate": 1.0056845172829871e-05, | |
| "loss": 0.029543858766555787, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 2.9314159292035398, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 9.9596781848227e-06, | |
| "loss": 0.02440253645181656, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.936946902654867, | |
| "grad_norm": 0.4765625, | |
| "learning_rate": 9.862889366680788e-06, | |
| "loss": 0.014462122321128845, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 2.942477876106195, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 9.766480544852574e-06, | |
| "loss": 0.014246387779712677, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 2.948008849557522, | |
| "grad_norm": 0.5, | |
| "learning_rate": 9.670453538615781e-06, | |
| "loss": 0.023418280482292175, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 2.9535398230088497, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 9.57481016004313e-06, | |
| "loss": 0.018582333624362946, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 2.959070796460177, | |
| "grad_norm": 0.52734375, | |
| "learning_rate": 9.479552213968065e-06, | |
| "loss": 0.017685665190219878, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.9646017699115044, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 9.384681497950785e-06, | |
| "loss": 0.025341919064521788, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 2.9701327433628317, | |
| "grad_norm": 0.435546875, | |
| "learning_rate": 9.290199802244273e-06, | |
| "loss": 0.01513555645942688, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 2.975663716814159, | |
| "grad_norm": 0.48828125, | |
| "learning_rate": 9.196108909760526e-06, | |
| "loss": 0.01928664892911911, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 2.981194690265487, | |
| "grad_norm": 0.53125, | |
| "learning_rate": 9.102410596036883e-06, | |
| "loss": 0.0181598424911499, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 2.9867256637168142, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 9.009106629202602e-06, | |
| "loss": 0.019247274100780486, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.9922566371681416, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 8.916198769945394e-06, | |
| "loss": 0.0226952463388443, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 2.997787610619469, | |
| "grad_norm": 0.49609375, | |
| "learning_rate": 8.823688771478244e-06, | |
| "loss": 0.018601973354816437, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.06098712235689163, | |
| "eval_pearson_cosine": 0.5689224600791931, | |
| "eval_pearson_dot_product": 0.5689224004745483, | |
| "eval_pearson_euclidean": 0.5670096278190613, | |
| "eval_pearson_manhattan": 0.5644800662994385, | |
| "eval_runtime": 110.4106, | |
| "eval_samples_per_second": 73.689, | |
| "eval_spearman_cosine": 0.5631030730927976, | |
| "eval_spearman_dot_product": 0.5631029475155491, | |
| "eval_spearman_euclidean": 0.5631028998526922, | |
| "eval_spearman_manhattan": 0.5578877185596505, | |
| "eval_steps_per_second": 12.281, | |
| "step": 2712 | |
| }, | |
| { | |
| "epoch": 3.0033185840707963, | |
| "grad_norm": 0.4375, | |
| "learning_rate": 8.73157837950634e-06, | |
| "loss": 0.01810520589351654, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 3.0088495575221237, | |
| "grad_norm": 0.408203125, | |
| "learning_rate": 8.63986933219411e-06, | |
| "loss": 0.015537238121032715, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 3.0143805309734515, | |
| "grad_norm": 0.318359375, | |
| "learning_rate": 8.548563360132433e-06, | |
| "loss": 0.010170294344425202, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.019911504424779, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 8.457662186305967e-06, | |
| "loss": 0.018869960308074953, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 3.025442477876106, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 8.367167526060655e-06, | |
| "loss": 0.010791405290365218, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 3.0309734513274336, | |
| "grad_norm": 0.400390625, | |
| "learning_rate": 8.277081087071355e-06, | |
| "loss": 0.008226613700389861, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 3.036504424778761, | |
| "grad_norm": 0.40234375, | |
| "learning_rate": 8.187404569309575e-06, | |
| "loss": 0.013960433006286622, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 3.0420353982300883, | |
| "grad_norm": 0.470703125, | |
| "learning_rate": 8.09813966501145e-06, | |
| "loss": 0.01127820611000061, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.047566371681416, | |
| "grad_norm": 0.328125, | |
| "learning_rate": 8.009288058645778e-06, | |
| "loss": 0.014376454055309296, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 3.0530973451327434, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 7.920851426882262e-06, | |
| "loss": 0.011377556622028351, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.058628318584071, | |
| "grad_norm": 0.31640625, | |
| "learning_rate": 7.832831438559805e-06, | |
| "loss": 0.01148907095193863, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 3.064159292035398, | |
| "grad_norm": 0.3671875, | |
| "learning_rate": 7.745229754655091e-06, | |
| "loss": 0.010777175426483154, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 3.0696902654867255, | |
| "grad_norm": 0.4296875, | |
| "learning_rate": 7.658048028251198e-06, | |
| "loss": 0.011512097716331483, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.0752212389380533, | |
| "grad_norm": 0.51953125, | |
| "learning_rate": 7.571287904506444e-06, | |
| "loss": 0.01386881023645401, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 3.0807522123893807, | |
| "grad_norm": 0.2314453125, | |
| "learning_rate": 7.484951020623264e-06, | |
| "loss": 0.009709823131561279, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 3.086283185840708, | |
| "grad_norm": 0.474609375, | |
| "learning_rate": 7.39903900581742e-06, | |
| "loss": 0.014929243922233581, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 3.0918141592920354, | |
| "grad_norm": 0.310546875, | |
| "learning_rate": 7.31355348128719e-06, | |
| "loss": 0.012847021222114563, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 3.0973451327433628, | |
| "grad_norm": 0.453125, | |
| "learning_rate": 7.228496060182761e-06, | |
| "loss": 0.012225713580846786, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.10287610619469, | |
| "grad_norm": 0.322265625, | |
| "learning_rate": 7.1438683475758435e-06, | |
| "loss": 0.01430513709783554, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 3.1084070796460175, | |
| "grad_norm": 0.3359375, | |
| "learning_rate": 7.059671940429353e-06, | |
| "loss": 0.011279582977294922, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 3.1139380530973453, | |
| "grad_norm": 0.333984375, | |
| "learning_rate": 6.975908427567268e-06, | |
| "loss": 0.008500397950410844, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 3.1194690265486726, | |
| "grad_norm": 0.3203125, | |
| "learning_rate": 6.892579389644669e-06, | |
| "loss": 0.006469354778528213, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 0.193359375, | |
| "learning_rate": 6.8096863991178906e-06, | |
| "loss": 0.011107449978590011, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.1305309734513274, | |
| "grad_norm": 0.400390625, | |
| "learning_rate": 6.72723102021486e-06, | |
| "loss": 0.01164601743221283, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 3.1360619469026547, | |
| "grad_norm": 0.388671875, | |
| "learning_rate": 6.645214808905587e-06, | |
| "loss": 0.008411216735839843, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 3.1415929203539825, | |
| "grad_norm": 0.451171875, | |
| "learning_rate": 6.563639312872754e-06, | |
| "loss": 0.008828174322843552, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 3.14712389380531, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 6.482506071482605e-06, | |
| "loss": 0.011891142278909684, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 3.1526548672566372, | |
| "grad_norm": 0.38671875, | |
| "learning_rate": 6.401816615755808e-06, | |
| "loss": 0.008382811397314071, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.1581858407079646, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 6.3215724683385864e-06, | |
| "loss": 0.013552053272724152, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 3.163716814159292, | |
| "grad_norm": 0.349609375, | |
| "learning_rate": 6.241775143474017e-06, | |
| "loss": 0.009054893255233764, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 3.1692477876106193, | |
| "grad_norm": 0.4375, | |
| "learning_rate": 6.162426146973423e-06, | |
| "loss": 0.013751232624053955, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 3.1747787610619467, | |
| "grad_norm": 0.322265625, | |
| "learning_rate": 6.083526976187979e-06, | |
| "loss": 0.01023416519165039, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 3.1803097345132745, | |
| "grad_norm": 0.2890625, | |
| "learning_rate": 6.005079119980445e-06, | |
| "loss": 0.011419428139925003, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.185840707964602, | |
| "grad_norm": 0.271484375, | |
| "learning_rate": 5.92708405869706e-06, | |
| "loss": 0.010896737873554229, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 3.191371681415929, | |
| "grad_norm": 0.29296875, | |
| "learning_rate": 5.849543264139646e-06, | |
| "loss": 0.009924667328596115, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 3.1969026548672566, | |
| "grad_norm": 0.34765625, | |
| "learning_rate": 5.772458199537769e-06, | |
| "loss": 0.015007269382476807, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 3.202433628318584, | |
| "grad_norm": 0.447265625, | |
| "learning_rate": 5.695830319521198e-06, | |
| "loss": 0.01667655408382416, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 3.2079646017699117, | |
| "grad_norm": 0.369140625, | |
| "learning_rate": 5.6196610700924115e-06, | |
| "loss": 0.017554591596126556, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.213495575221239, | |
| "grad_norm": 0.353515625, | |
| "learning_rate": 5.543951888599329e-06, | |
| "loss": 0.01232467070221901, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 3.2190265486725664, | |
| "grad_norm": 0.34765625, | |
| "learning_rate": 5.468704203708177e-06, | |
| "loss": 0.01566605567932129, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 3.224557522123894, | |
| "grad_norm": 0.390625, | |
| "learning_rate": 5.393919435376536e-06, | |
| "loss": 0.009415168315172195, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 3.230088495575221, | |
| "grad_norm": 0.427734375, | |
| "learning_rate": 5.31959899482654e-06, | |
| "loss": 0.013145282864570618, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 3.2356194690265485, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 5.245744284518259e-06, | |
| "loss": 0.013940691947937012, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 3.241150442477876, | |
| "grad_norm": 0.1708984375, | |
| "learning_rate": 5.172356698123195e-06, | |
| "loss": 0.009199875593185424, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 3.2466814159292037, | |
| "grad_norm": 0.421875, | |
| "learning_rate": 5.099437620498061e-06, | |
| "loss": 0.011912959069013596, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 3.252212389380531, | |
| "grad_norm": 0.33203125, | |
| "learning_rate": 5.026988427658565e-06, | |
| "loss": 0.010796493291854859, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 3.2577433628318584, | |
| "grad_norm": 0.384765625, | |
| "learning_rate": 4.955010486753477e-06, | |
| "loss": 0.011171175539493561, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 3.2632743362831858, | |
| "grad_norm": 0.435546875, | |
| "learning_rate": 4.883505156038841e-06, | |
| "loss": 0.010442172735929489, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.268805309734513, | |
| "grad_norm": 0.3359375, | |
| "learning_rate": 4.812473784852344e-06, | |
| "loss": 0.012105541676282883, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 3.274336283185841, | |
| "grad_norm": 0.267578125, | |
| "learning_rate": 4.7419177135878236e-06, | |
| "loss": 0.007604041695594787, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 3.2798672566371683, | |
| "grad_norm": 0.5703125, | |
| "learning_rate": 4.671838273670011e-06, | |
| "loss": 0.01639808863401413, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 3.2853982300884956, | |
| "grad_norm": 0.435546875, | |
| "learning_rate": 4.602236787529383e-06, | |
| "loss": 0.014667351543903352, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 3.290929203539823, | |
| "grad_norm": 0.2041015625, | |
| "learning_rate": 4.533114568577223e-06, | |
| "loss": 0.011389161646366119, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.2964601769911503, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 4.464472921180801e-06, | |
| "loss": 0.01777474582195282, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 3.3019911504424777, | |
| "grad_norm": 0.43359375, | |
| "learning_rate": 4.396313140638815e-06, | |
| "loss": 0.012283781915903092, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 3.307522123893805, | |
| "grad_norm": 0.3125, | |
| "learning_rate": 4.328636513156902e-06, | |
| "loss": 0.013930252194404602, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 3.313053097345133, | |
| "grad_norm": 0.361328125, | |
| "learning_rate": 4.26144431582341e-06, | |
| "loss": 0.012156898528337479, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 3.3185840707964602, | |
| "grad_norm": 0.2578125, | |
| "learning_rate": 4.194737816585236e-06, | |
| "loss": 0.0139230415225029, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.3241150442477876, | |
| "grad_norm": 0.40625, | |
| "learning_rate": 4.128518274223959e-06, | |
| "loss": 0.010459666699171066, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 3.329646017699115, | |
| "grad_norm": 0.369140625, | |
| "learning_rate": 4.062786938332057e-06, | |
| "loss": 0.00808744803071022, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 3.3351769911504423, | |
| "grad_norm": 0.41796875, | |
| "learning_rate": 3.9975450492893385e-06, | |
| "loss": 0.016974259912967683, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 3.34070796460177, | |
| "grad_norm": 0.294921875, | |
| "learning_rate": 3.93279383823951e-06, | |
| "loss": 0.006088964641094208, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 3.3462389380530975, | |
| "grad_norm": 0.388671875, | |
| "learning_rate": 3.868534527066996e-06, | |
| "loss": 0.008859299868345261, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.351769911504425, | |
| "grad_norm": 0.5, | |
| "learning_rate": 3.8047683283738345e-06, | |
| "loss": 0.011756136268377303, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 3.357300884955752, | |
| "grad_norm": 0.345703125, | |
| "learning_rate": 3.7414964454567945e-06, | |
| "loss": 0.010117633640766144, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 3.3628318584070795, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 3.678720072284707e-06, | |
| "loss": 0.013096696138381958, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 3.368362831858407, | |
| "grad_norm": 0.451171875, | |
| "learning_rate": 3.6164403934759018e-06, | |
| "loss": 0.014904367923736572, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 3.3738938053097347, | |
| "grad_norm": 0.40234375, | |
| "learning_rate": 3.5546585842758627e-06, | |
| "loss": 0.013438491523265839, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.379424778761062, | |
| "grad_norm": 0.455078125, | |
| "learning_rate": 3.4933758105350567e-06, | |
| "loss": 0.013479460775852204, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 3.3849557522123894, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 3.4325932286869187e-06, | |
| "loss": 0.011800102144479751, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 3.390486725663717, | |
| "grad_norm": 0.384765625, | |
| "learning_rate": 3.372311985726052e-06, | |
| "loss": 0.01110667958855629, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 3.396017699115044, | |
| "grad_norm": 0.349609375, | |
| "learning_rate": 3.3125332191865467e-06, | |
| "loss": 0.010498964041471482, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 3.4015486725663715, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 3.2532580571205604e-06, | |
| "loss": 0.013484889268875122, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.4070796460176993, | |
| "grad_norm": 0.349609375, | |
| "learning_rate": 3.1944876180769943e-06, | |
| "loss": 0.011495392024517059, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 3.4126106194690267, | |
| "grad_norm": 0.349609375, | |
| "learning_rate": 3.1362230110804168e-06, | |
| "loss": 0.013613630831241608, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 3.418141592920354, | |
| "grad_norm": 0.251953125, | |
| "learning_rate": 3.0784653356100958e-06, | |
| "loss": 0.011250624060630798, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 3.4236725663716814, | |
| "grad_norm": 0.56640625, | |
| "learning_rate": 3.0212156815792867e-06, | |
| "loss": 0.011816568672657013, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 3.4292035398230087, | |
| "grad_norm": 0.3984375, | |
| "learning_rate": 2.964475129314652e-06, | |
| "loss": 0.010769976675510407, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.434734513274336, | |
| "grad_norm": 0.33203125, | |
| "learning_rate": 2.908244749535881e-06, | |
| "loss": 0.010063180327415466, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 3.440265486725664, | |
| "grad_norm": 0.25390625, | |
| "learning_rate": 2.8525256033354543e-06, | |
| "loss": 0.0068759948015213014, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 3.4457964601769913, | |
| "grad_norm": 0.341796875, | |
| "learning_rate": 2.7973187421586754e-06, | |
| "loss": 0.017900320887565612, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 3.4513274336283186, | |
| "grad_norm": 0.39453125, | |
| "learning_rate": 2.7426252077837954e-06, | |
| "loss": 0.013324187695980072, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 3.456858407079646, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 2.688446032302333e-06, | |
| "loss": 0.008624576032161713, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.4623893805309733, | |
| "grad_norm": 0.375, | |
| "learning_rate": 2.634782238099651e-06, | |
| "loss": 0.0121292382478714, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 3.4679203539823007, | |
| "grad_norm": 0.2734375, | |
| "learning_rate": 2.5816348378356226e-06, | |
| "loss": 0.008544319868087768, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 3.4734513274336285, | |
| "grad_norm": 0.34375, | |
| "learning_rate": 2.529004834425541e-06, | |
| "loss": 0.012461203336715698, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 3.478982300884956, | |
| "grad_norm": 0.4375, | |
| "learning_rate": 2.476893221021188e-06, | |
| "loss": 0.013459134101867675, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 3.484513274336283, | |
| "grad_norm": 0.51953125, | |
| "learning_rate": 2.4253009809920925e-06, | |
| "loss": 0.010717058926820755, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.4900442477876106, | |
| "grad_norm": 0.251953125, | |
| "learning_rate": 2.3742290879069726e-06, | |
| "loss": 0.010836039483547211, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 3.495575221238938, | |
| "grad_norm": 0.267578125, | |
| "learning_rate": 2.323678505515373e-06, | |
| "loss": 0.011072257906198502, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 3.5011061946902657, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 2.2736501877294624e-06, | |
| "loss": 0.013174070417881012, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 3.5066371681415927, | |
| "grad_norm": 0.375, | |
| "learning_rate": 2.2241450786060367e-06, | |
| "loss": 0.012093041092157364, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 3.5121681415929205, | |
| "grad_norm": 0.38671875, | |
| "learning_rate": 2.1751641123287434e-06, | |
| "loss": 0.011323301494121552, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.517699115044248, | |
| "grad_norm": 0.306640625, | |
| "learning_rate": 2.126708213190375e-06, | |
| "loss": 0.013007217645645141, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 3.523230088495575, | |
| "grad_norm": 0.29296875, | |
| "learning_rate": 2.078778295575499e-06, | |
| "loss": 0.010674595087766647, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 3.5287610619469025, | |
| "grad_norm": 0.4453125, | |
| "learning_rate": 2.0313752639431694e-06, | |
| "loss": 0.010046075284481048, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 3.53429203539823, | |
| "grad_norm": 0.251953125, | |
| "learning_rate": 1.9845000128098646e-06, | |
| "loss": 0.009277325868606568, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 3.5398230088495577, | |
| "grad_norm": 0.38671875, | |
| "learning_rate": 1.938153426732595e-06, | |
| "loss": 0.014334882795810699, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.545353982300885, | |
| "grad_norm": 0.267578125, | |
| "learning_rate": 1.8923363802922521e-06, | |
| "loss": 0.008805647492408752, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 3.5508849557522124, | |
| "grad_norm": 0.447265625, | |
| "learning_rate": 1.8470497380770602e-06, | |
| "loss": 0.013123226165771485, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 3.5564159292035398, | |
| "grad_norm": 0.33984375, | |
| "learning_rate": 1.8022943546662695e-06, | |
| "loss": 0.011118625849485397, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 3.561946902654867, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 1.7580710746140572e-06, | |
| "loss": 0.010188119113445282, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 3.567477876106195, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 1.714380732433558e-06, | |
| "loss": 0.010415033996105194, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.573008849557522, | |
| "grad_norm": 0.388671875, | |
| "learning_rate": 1.6712241525811323e-06, | |
| "loss": 0.01137053668498993, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 3.5785398230088497, | |
| "grad_norm": 0.373046875, | |
| "learning_rate": 1.62860214944081e-06, | |
| "loss": 0.010083390027284622, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 3.584070796460177, | |
| "grad_norm": 0.361328125, | |
| "learning_rate": 1.5865155273089172e-06, | |
| "loss": 0.011858479678630829, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 3.5896017699115044, | |
| "grad_norm": 0.251953125, | |
| "learning_rate": 1.544965080378895e-06, | |
| "loss": 0.012606756389141082, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 3.5951327433628317, | |
| "grad_norm": 0.44921875, | |
| "learning_rate": 1.503951592726328e-06, | |
| "loss": 0.010169728845357894, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.600663716814159, | |
| "grad_norm": 0.404296875, | |
| "learning_rate": 1.463475838294126e-06, | |
| "loss": 0.014370378851890565, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 3.606194690265487, | |
| "grad_norm": 0.35546875, | |
| "learning_rate": 1.4235385808779323e-06, | |
| "loss": 0.011973607540130615, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 3.6117256637168142, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 1.384140574111732e-06, | |
| "loss": 0.013000024855136872, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 3.6172566371681416, | |
| "grad_norm": 0.515625, | |
| "learning_rate": 1.3452825614535802e-06, | |
| "loss": 0.012618395686149596, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 3.622787610619469, | |
| "grad_norm": 0.439453125, | |
| "learning_rate": 1.3069652761716133e-06, | |
| "loss": 0.013663342595100403, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.6283185840707963, | |
| "grad_norm": 0.466796875, | |
| "learning_rate": 1.2691894413301963e-06, | |
| "loss": 0.011588869988918305, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 3.633849557522124, | |
| "grad_norm": 0.2470703125, | |
| "learning_rate": 1.231955769776294e-06, | |
| "loss": 0.012852667272090912, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 3.6393805309734515, | |
| "grad_norm": 0.2197265625, | |
| "learning_rate": 1.1952649641259817e-06, | |
| "loss": 0.009375353902578354, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 3.644911504424779, | |
| "grad_norm": 0.33984375, | |
| "learning_rate": 1.1591177167512356e-06, | |
| "loss": 0.008105507493019104, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 3.650442477876106, | |
| "grad_norm": 0.5546875, | |
| "learning_rate": 1.1235147097668374e-06, | |
| "loss": 0.012569212913513183, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.6559734513274336, | |
| "grad_norm": 0.34765625, | |
| "learning_rate": 1.0884566150174968e-06, | |
| "loss": 0.010477407276630402, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 3.661504424778761, | |
| "grad_norm": 0.2412109375, | |
| "learning_rate": 1.0539440940651924e-06, | |
| "loss": 0.009510351717472077, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 3.6670353982300883, | |
| "grad_norm": 0.451171875, | |
| "learning_rate": 1.0199777981766878e-06, | |
| "loss": 0.008886636048555375, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 3.672566371681416, | |
| "grad_norm": 0.3359375, | |
| "learning_rate": 9.865583683112255e-07, | |
| "loss": 0.006694913655519485, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 3.6780973451327434, | |
| "grad_norm": 0.361328125, | |
| "learning_rate": 9.536864351084396e-07, | |
| "loss": 0.009480822086334228, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.683628318584071, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 9.213626188764646e-07, | |
| "loss": 0.017233464121818542, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 3.689159292035398, | |
| "grad_norm": 0.3046875, | |
| "learning_rate": 8.895875295802159e-07, | |
| "loss": 0.006251173466444016, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 3.6946902654867255, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 8.583617668298916e-07, | |
| "loss": 0.014849424362182617, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 3.7002212389380533, | |
| "grad_norm": 0.310546875, | |
| "learning_rate": 8.276859198696384e-07, | |
| "loss": 0.011075834929943084, | |
| "step": 3345 | |
| }, | |
| { | |
| "epoch": 3.7057522123893807, | |
| "grad_norm": 0.279296875, | |
| "learning_rate": 7.975605675664511e-07, | |
| "loss": 0.013128185272216797, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.711283185840708, | |
| "grad_norm": 0.220703125, | |
| "learning_rate": 7.679862783992598e-07, | |
| "loss": 0.010468773543834686, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 3.7168141592920354, | |
| "grad_norm": 0.3203125, | |
| "learning_rate": 7.389636104481601e-07, | |
| "loss": 0.017043809592723846, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 3.7223451327433628, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 7.104931113839242e-07, | |
| "loss": 0.014255988597869872, | |
| "step": 3365 | |
| }, | |
| { | |
| "epoch": 3.72787610619469, | |
| "grad_norm": 0.333984375, | |
| "learning_rate": 6.825753184576422e-07, | |
| "loss": 0.014779409766197205, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 3.7334070796460175, | |
| "grad_norm": 0.439453125, | |
| "learning_rate": 6.552107584905975e-07, | |
| "loss": 0.009380747377872468, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.7389380530973453, | |
| "grad_norm": 0.45703125, | |
| "learning_rate": 6.283999478643109e-07, | |
| "loss": 0.009793084114789963, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 3.7444690265486726, | |
| "grad_norm": 0.466796875, | |
| "learning_rate": 6.021433925108122e-07, | |
| "loss": 0.012913639843463897, | |
| "step": 3385 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 0.41796875, | |
| "learning_rate": 5.764415879030871e-07, | |
| "loss": 0.008790192008018494, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 3.7555309734513274, | |
| "grad_norm": 0.416015625, | |
| "learning_rate": 5.512950190457189e-07, | |
| "loss": 0.009441033750772477, | |
| "step": 3395 | |
| }, | |
| { | |
| "epoch": 3.7610619469026547, | |
| "grad_norm": 0.302734375, | |
| "learning_rate": 5.267041604657418e-07, | |
| "loss": 0.010161211341619491, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.7665929203539825, | |
| "grad_norm": 0.41796875, | |
| "learning_rate": 5.02669476203692e-07, | |
| "loss": 0.011450953781604767, | |
| "step": 3405 | |
| }, | |
| { | |
| "epoch": 3.77212389380531, | |
| "grad_norm": 0.361328125, | |
| "learning_rate": 4.791914198048541e-07, | |
| "loss": 0.008028075844049454, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 3.7776548672566372, | |
| "grad_norm": 0.3125, | |
| "learning_rate": 4.562704343106816e-07, | |
| "loss": 0.011647894233465194, | |
| "step": 3415 | |
| }, | |
| { | |
| "epoch": 3.7831858407079646, | |
| "grad_norm": 0.44140625, | |
| "learning_rate": 4.3390695225045685e-07, | |
| "loss": 0.013816773891448975, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 3.788716814159292, | |
| "grad_norm": 0.38671875, | |
| "learning_rate": 4.12101395633121e-07, | |
| "loss": 0.01271996796131134, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.7942477876106193, | |
| "grad_norm": 0.41015625, | |
| "learning_rate": 3.9085417593931363e-07, | |
| "loss": 0.014049507677555084, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 3.7997787610619467, | |
| "grad_norm": 0.322265625, | |
| "learning_rate": 3.7016569411360224e-07, | |
| "loss": 0.011587772518396378, | |
| "step": 3435 | |
| }, | |
| { | |
| "epoch": 3.8053097345132745, | |
| "grad_norm": 0.375, | |
| "learning_rate": 3.500363405569251e-07, | |
| "loss": 0.010679934173822403, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 3.810840707964602, | |
| "grad_norm": 0.380859375, | |
| "learning_rate": 3.304664951192171e-07, | |
| "loss": 0.009259649366140366, | |
| "step": 3445 | |
| }, | |
| { | |
| "epoch": 3.816371681415929, | |
| "grad_norm": 0.4296875, | |
| "learning_rate": 3.114565270922387e-07, | |
| "loss": 0.011192093044519425, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.8219026548672566, | |
| "grad_norm": 0.416015625, | |
| "learning_rate": 2.9300679520262163e-07, | |
| "loss": 0.014347481727600097, | |
| "step": 3455 | |
| }, | |
| { | |
| "epoch": 3.827433628318584, | |
| "grad_norm": 0.46875, | |
| "learning_rate": 2.75117647605081e-07, | |
| "loss": 0.015431442856788635, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 3.8329646017699117, | |
| "grad_norm": 0.498046875, | |
| "learning_rate": 2.577894218758636e-07, | |
| "loss": 0.009501569718122483, | |
| "step": 3465 | |
| }, | |
| { | |
| "epoch": 3.838495575221239, | |
| "grad_norm": 0.3203125, | |
| "learning_rate": 2.4102244500636026e-07, | |
| "loss": 0.010970748215913772, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 3.8440265486725664, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 2.2481703339695348e-07, | |
| "loss": 0.015523919463157653, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.849557522123894, | |
| "grad_norm": 0.2353515625, | |
| "learning_rate": 2.0917349285102584e-07, | |
| "loss": 0.00985901951789856, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 3.855088495575221, | |
| "grad_norm": 0.390625, | |
| "learning_rate": 1.940921185692146e-07, | |
| "loss": 0.013239237666130065, | |
| "step": 3485 | |
| }, | |
| { | |
| "epoch": 3.8606194690265485, | |
| "grad_norm": 0.490234375, | |
| "learning_rate": 1.7957319514381286e-07, | |
| "loss": 0.015627865493297578, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 3.866150442477876, | |
| "grad_norm": 0.28125, | |
| "learning_rate": 1.6561699655342378e-07, | |
| "loss": 0.014176610112190246, | |
| "step": 3495 | |
| }, | |
| { | |
| "epoch": 3.8716814159292037, | |
| "grad_norm": 0.30078125, | |
| "learning_rate": 1.5222378615777798e-07, | |
| "loss": 0.006326702237129211, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.877212389380531, | |
| "grad_norm": 0.39453125, | |
| "learning_rate": 1.393938166927644e-07, | |
| "loss": 0.010394010692834854, | |
| "step": 3505 | |
| }, | |
| { | |
| "epoch": 3.8827433628318584, | |
| "grad_norm": 0.447265625, | |
| "learning_rate": 1.2712733026566392e-07, | |
| "loss": 0.014642159640789031, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 3.8882743362831858, | |
| "grad_norm": 0.4296875, | |
| "learning_rate": 1.1542455835057975e-07, | |
| "loss": 0.01353752166032791, | |
| "step": 3515 | |
| }, | |
| { | |
| "epoch": 3.893805309734513, | |
| "grad_norm": 0.302734375, | |
| "learning_rate": 1.0428572178407092e-07, | |
| "loss": 0.010867173224687577, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 3.899336283185841, | |
| "grad_norm": 0.494140625, | |
| "learning_rate": 9.371103076097898e-08, | |
| "loss": 0.014151638746261597, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.9048672566371683, | |
| "grad_norm": 0.3828125, | |
| "learning_rate": 8.37006848304711e-08, | |
| "loss": 0.01013147011399269, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 3.9103982300884956, | |
| "grad_norm": 0.357421875, | |
| "learning_rate": 7.425487289226318e-08, | |
| "loss": 0.012769846618175507, | |
| "step": 3535 | |
| }, | |
| { | |
| "epoch": 3.915929203539823, | |
| "grad_norm": 0.267578125, | |
| "learning_rate": 6.537377319307258e-08, | |
| "loss": 0.00898270383477211, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 3.9214601769911503, | |
| "grad_norm": 0.310546875, | |
| "learning_rate": 5.7057553323227594e-08, | |
| "loss": 0.010919363051652909, | |
| "step": 3545 | |
| }, | |
| { | |
| "epoch": 3.926991150442478, | |
| "grad_norm": 0.466796875, | |
| "learning_rate": 4.93063702135299e-08, | |
| "loss": 0.012932133674621583, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.932522123893805, | |
| "grad_norm": 0.482421875, | |
| "learning_rate": 4.212037013228698e-08, | |
| "loss": 0.014183707535266876, | |
| "step": 3555 | |
| }, | |
| { | |
| "epoch": 3.938053097345133, | |
| "grad_norm": 0.5234375, | |
| "learning_rate": 3.549968868254427e-08, | |
| "loss": 0.013312140107154846, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 3.9435840707964602, | |
| "grad_norm": 0.55078125, | |
| "learning_rate": 2.944445079953062e-08, | |
| "loss": 0.009294530749320984, | |
| "step": 3565 | |
| }, | |
| { | |
| "epoch": 3.9491150442477876, | |
| "grad_norm": 0.322265625, | |
| "learning_rate": 2.3954770748303433e-08, | |
| "loss": 0.010873483866453171, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 3.954646017699115, | |
| "grad_norm": 0.388671875, | |
| "learning_rate": 1.9030752121583785e-08, | |
| "loss": 0.012081097066402435, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.9601769911504423, | |
| "grad_norm": 0.388671875, | |
| "learning_rate": 1.4672487837817939e-08, | |
| "loss": 0.01202463060617447, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 3.96570796460177, | |
| "grad_norm": 0.5078125, | |
| "learning_rate": 1.0880060139405456e-08, | |
| "loss": 0.01384652704000473, | |
| "step": 3585 | |
| }, | |
| { | |
| "epoch": 3.9712389380530975, | |
| "grad_norm": 0.296875, | |
| "learning_rate": 7.653540591160413e-09, | |
| "loss": 0.01071920469403267, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 3.976769911504425, | |
| "grad_norm": 0.498046875, | |
| "learning_rate": 4.992990078955817e-09, | |
| "loss": 0.01177939847111702, | |
| "step": 3595 | |
| }, | |
| { | |
| "epoch": 3.982300884955752, | |
| "grad_norm": 0.470703125, | |
| "learning_rate": 2.8984588085678722e-09, | |
| "loss": 0.010881439596414567, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.9878318584070795, | |
| "grad_norm": 0.369140625, | |
| "learning_rate": 1.3699863047500483e-09, | |
| "loss": 0.011608899384737015, | |
| "step": 3605 | |
| }, | |
| { | |
| "epoch": 3.9933628318584073, | |
| "grad_norm": 0.306640625, | |
| "learning_rate": 4.0760141045703693e-10, | |
| "loss": 0.011873122304677963, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 3.9988938053097343, | |
| "grad_norm": 0.455078125, | |
| "learning_rate": 1.1322286328496745e-11, | |
| "loss": 0.017984117567539214, | |
| "step": 3615 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.06123881787061691, | |
| "eval_pearson_cosine": 0.5666685700416565, | |
| "eval_pearson_dot_product": 0.5666685700416565, | |
| "eval_pearson_euclidean": 0.5647152066230774, | |
| "eval_pearson_manhattan": 0.562436044216156, | |
| "eval_runtime": 110.4774, | |
| "eval_samples_per_second": 73.644, | |
| "eval_spearman_cosine": 0.5600658786573235, | |
| "eval_spearman_dot_product": 0.5600658542827176, | |
| "eval_spearman_euclidean": 0.5600656976346822, | |
| "eval_spearman_manhattan": 0.5548845171480339, | |
| "eval_steps_per_second": 12.274, | |
| "step": 3616 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.06123881787061691, | |
| "eval_pearson_cosine": 0.5666685700416565, | |
| "eval_pearson_dot_product": 0.5666685700416565, | |
| "eval_pearson_euclidean": 0.5647152066230774, | |
| "eval_pearson_manhattan": 0.562436044216156, | |
| "eval_runtime": 110.6823, | |
| "eval_samples_per_second": 73.508, | |
| "eval_spearman_cosine": 0.5600658786573235, | |
| "eval_spearman_dot_product": 0.5600658542827176, | |
| "eval_spearman_euclidean": 0.5600656976346822, | |
| "eval_spearman_manhattan": 0.5548845171480339, | |
| "eval_steps_per_second": 12.251, | |
| "step": 3616 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 3616, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.784267470680883e+17, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |