| { | |
| "best_metric": 0.38531041145324707, | |
| "best_model_checkpoint": "bert-base-uncased-dnd/checkpoint-3341", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 3341, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0074827895839568994, | |
| "grad_norm": 7.956452369689941, | |
| "learning_rate": 3.7313432835820893e-06, | |
| "loss": 0.7158, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014965579167913799, | |
| "grad_norm": 4.472965717315674, | |
| "learning_rate": 7.4626865671641785e-06, | |
| "loss": 0.6669, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022448368751870697, | |
| "grad_norm": 6.700029373168945, | |
| "learning_rate": 1.119402985074627e-05, | |
| "loss": 0.5986, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.029931158335827598, | |
| "grad_norm": 6.568297863006592, | |
| "learning_rate": 1.4925373134328357e-05, | |
| "loss": 0.6107, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.037413947919784495, | |
| "grad_norm": 6.28012228012085, | |
| "learning_rate": 1.865671641791045e-05, | |
| "loss": 0.5449, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.044896737503741395, | |
| "grad_norm": 11.522820472717285, | |
| "learning_rate": 2.238805970149254e-05, | |
| "loss": 0.5142, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.052379527087698295, | |
| "grad_norm": 10.602967262268066, | |
| "learning_rate": 2.6119402985074626e-05, | |
| "loss": 0.5016, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.059862316671655195, | |
| "grad_norm": 12.371557235717773, | |
| "learning_rate": 2.9850746268656714e-05, | |
| "loss": 0.4982, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06734510625561209, | |
| "grad_norm": 8.330849647521973, | |
| "learning_rate": 3.358208955223881e-05, | |
| "loss": 0.5868, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.07482789583956899, | |
| "grad_norm": 7.456040382385254, | |
| "learning_rate": 3.73134328358209e-05, | |
| "loss": 0.5862, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08231068542352589, | |
| "grad_norm": 6.040958881378174, | |
| "learning_rate": 4.104477611940299e-05, | |
| "loss": 0.5032, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.08979347500748279, | |
| "grad_norm": 9.931991577148438, | |
| "learning_rate": 4.477611940298508e-05, | |
| "loss": 0.4879, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09727626459143969, | |
| "grad_norm": 9.538640975952148, | |
| "learning_rate": 4.850746268656717e-05, | |
| "loss": 0.5023, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.10475905417539659, | |
| "grad_norm": 17.958545684814453, | |
| "learning_rate": 4.975049900199601e-05, | |
| "loss": 0.4918, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11224184375935349, | |
| "grad_norm": 9.302469253540039, | |
| "learning_rate": 4.933466400532269e-05, | |
| "loss": 0.4871, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.11972463334331039, | |
| "grad_norm": 6.854007244110107, | |
| "learning_rate": 4.891882900864937e-05, | |
| "loss": 0.5076, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12720742292726728, | |
| "grad_norm": 4.675145626068115, | |
| "learning_rate": 4.8502994011976046e-05, | |
| "loss": 0.4458, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.13469021251122418, | |
| "grad_norm": 3.543985605239868, | |
| "learning_rate": 4.808715901530273e-05, | |
| "loss": 0.501, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14217300209518108, | |
| "grad_norm": 6.710892200469971, | |
| "learning_rate": 4.767132401862941e-05, | |
| "loss": 0.532, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.14965579167913798, | |
| "grad_norm": 7.534671783447266, | |
| "learning_rate": 4.725548902195609e-05, | |
| "loss": 0.4354, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.15713858126309488, | |
| "grad_norm": 5.7830729484558105, | |
| "learning_rate": 4.683965402528277e-05, | |
| "loss": 0.5325, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.16462137084705178, | |
| "grad_norm": 12.206136703491211, | |
| "learning_rate": 4.642381902860945e-05, | |
| "loss": 0.4433, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17210416043100868, | |
| "grad_norm": 14.097105979919434, | |
| "learning_rate": 4.600798403193613e-05, | |
| "loss": 0.4804, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.17958695001496558, | |
| "grad_norm": 4.995736122131348, | |
| "learning_rate": 4.559214903526281e-05, | |
| "loss": 0.4982, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18706973959892248, | |
| "grad_norm": 12.077266693115234, | |
| "learning_rate": 4.517631403858949e-05, | |
| "loss": 0.4617, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.19455252918287938, | |
| "grad_norm": 5.870637893676758, | |
| "learning_rate": 4.476047904191617e-05, | |
| "loss": 0.4668, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.20203531876683628, | |
| "grad_norm": 5.639552593231201, | |
| "learning_rate": 4.434464404524285e-05, | |
| "loss": 0.4459, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.20951810835079318, | |
| "grad_norm": 10.198214530944824, | |
| "learning_rate": 4.392880904856953e-05, | |
| "loss": 0.3699, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21700089793475008, | |
| "grad_norm": 7.611853122711182, | |
| "learning_rate": 4.351297405189621e-05, | |
| "loss": 0.5426, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.22448368751870698, | |
| "grad_norm": 8.950506210327148, | |
| "learning_rate": 4.3097139055222893e-05, | |
| "loss": 0.4314, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23196647710266388, | |
| "grad_norm": 14.771285057067871, | |
| "learning_rate": 4.268130405854957e-05, | |
| "loss": 0.4637, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.23944926668662078, | |
| "grad_norm": 9.739069938659668, | |
| "learning_rate": 4.226546906187625e-05, | |
| "loss": 0.4175, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.24693205627057768, | |
| "grad_norm": 11.961554527282715, | |
| "learning_rate": 4.184963406520293e-05, | |
| "loss": 0.4471, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.25441484585453455, | |
| "grad_norm": 4.314615249633789, | |
| "learning_rate": 4.143379906852961e-05, | |
| "loss": 0.4441, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.2618976354384915, | |
| "grad_norm": 5.674382209777832, | |
| "learning_rate": 4.101796407185629e-05, | |
| "loss": 0.5078, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.26938042502244836, | |
| "grad_norm": 61.70943832397461, | |
| "learning_rate": 4.060212907518297e-05, | |
| "loss": 0.4771, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.2768632146064053, | |
| "grad_norm": 17.10365867614746, | |
| "learning_rate": 4.018629407850965e-05, | |
| "loss": 0.3983, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.28434600419036216, | |
| "grad_norm": 30.42233657836914, | |
| "learning_rate": 3.977045908183633e-05, | |
| "loss": 0.4769, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2918287937743191, | |
| "grad_norm": 3.018568992614746, | |
| "learning_rate": 3.935462408516301e-05, | |
| "loss": 0.4831, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.29931158335827596, | |
| "grad_norm": 12.429970741271973, | |
| "learning_rate": 3.893878908848969e-05, | |
| "loss": 0.4397, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3067943729422329, | |
| "grad_norm": 12.596036911010742, | |
| "learning_rate": 3.852295409181637e-05, | |
| "loss": 0.4677, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.31427716252618976, | |
| "grad_norm": 7.590559005737305, | |
| "learning_rate": 3.810711909514305e-05, | |
| "loss": 0.4027, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3217599521101467, | |
| "grad_norm": 8.291315078735352, | |
| "learning_rate": 3.7691284098469734e-05, | |
| "loss": 0.5289, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.32924274169410356, | |
| "grad_norm": 8.881794929504395, | |
| "learning_rate": 3.727544910179641e-05, | |
| "loss": 0.4684, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.3367255312780605, | |
| "grad_norm": 5.292083740234375, | |
| "learning_rate": 3.685961410512309e-05, | |
| "loss": 0.4001, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.34420832086201736, | |
| "grad_norm": 8.29925537109375, | |
| "learning_rate": 3.644377910844977e-05, | |
| "loss": 0.4543, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.3516911104459743, | |
| "grad_norm": 3.3631722927093506, | |
| "learning_rate": 3.6027944111776455e-05, | |
| "loss": 0.4456, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.35917390002993116, | |
| "grad_norm": 6.089787483215332, | |
| "learning_rate": 3.5612109115103126e-05, | |
| "loss": 0.4167, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.36665668961388803, | |
| "grad_norm": 11.426901817321777, | |
| "learning_rate": 3.5196274118429805e-05, | |
| "loss": 0.4274, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.37413947919784496, | |
| "grad_norm": 13.190223693847656, | |
| "learning_rate": 3.478043912175648e-05, | |
| "loss": 0.5219, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.38162226878180183, | |
| "grad_norm": 6.92958927154541, | |
| "learning_rate": 3.436460412508317e-05, | |
| "loss": 0.3534, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.38910505836575876, | |
| "grad_norm": 9.719358444213867, | |
| "learning_rate": 3.394876912840985e-05, | |
| "loss": 0.5147, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.39658784794971563, | |
| "grad_norm": 9.629469871520996, | |
| "learning_rate": 3.3532934131736525e-05, | |
| "loss": 0.4547, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.40407063753367256, | |
| "grad_norm": 9.82555103302002, | |
| "learning_rate": 3.3117099135063204e-05, | |
| "loss": 0.4514, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.41155342711762943, | |
| "grad_norm": 2.007561206817627, | |
| "learning_rate": 3.270126413838989e-05, | |
| "loss": 0.4945, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.41903621670158636, | |
| "grad_norm": 7.009624481201172, | |
| "learning_rate": 3.228542914171657e-05, | |
| "loss": 0.3933, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.42651900628554323, | |
| "grad_norm": 18.308015823364258, | |
| "learning_rate": 3.1869594145043246e-05, | |
| "loss": 0.3952, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.43400179586950016, | |
| "grad_norm": 3.3795619010925293, | |
| "learning_rate": 3.1453759148369925e-05, | |
| "loss": 0.5187, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.44148458545345703, | |
| "grad_norm": 6.379357814788818, | |
| "learning_rate": 3.103792415169661e-05, | |
| "loss": 0.4747, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.44896737503741396, | |
| "grad_norm": 5.316616058349609, | |
| "learning_rate": 3.062208915502329e-05, | |
| "loss": 0.4547, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.45645016462137084, | |
| "grad_norm": 9.591498374938965, | |
| "learning_rate": 3.0206254158349967e-05, | |
| "loss": 0.4283, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.46393295420532776, | |
| "grad_norm": 4.518273830413818, | |
| "learning_rate": 2.979041916167665e-05, | |
| "loss": 0.4293, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.47141574378928464, | |
| "grad_norm": 4.459203243255615, | |
| "learning_rate": 2.9374584165003327e-05, | |
| "loss": 0.4324, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.47889853337324156, | |
| "grad_norm": 8.675421714782715, | |
| "learning_rate": 2.8958749168330006e-05, | |
| "loss": 0.4131, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.48638132295719844, | |
| "grad_norm": 9.163846969604492, | |
| "learning_rate": 2.8542914171656687e-05, | |
| "loss": 0.4151, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.49386411254115536, | |
| "grad_norm": 11.9418306350708, | |
| "learning_rate": 2.8127079174983366e-05, | |
| "loss": 0.3609, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5013469021251122, | |
| "grad_norm": 2.1281261444091797, | |
| "learning_rate": 2.7711244178310048e-05, | |
| "loss": 0.383, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.5088296917090691, | |
| "grad_norm": 13.822118759155273, | |
| "learning_rate": 2.7295409181636726e-05, | |
| "loss": 0.5024, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5163124812930261, | |
| "grad_norm": 14.914559364318848, | |
| "learning_rate": 2.6879574184963408e-05, | |
| "loss": 0.4355, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.523795270876983, | |
| "grad_norm": 5.5026984214782715, | |
| "learning_rate": 2.6463739188290087e-05, | |
| "loss": 0.2979, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5312780604609398, | |
| "grad_norm": 4.769206523895264, | |
| "learning_rate": 2.604790419161677e-05, | |
| "loss": 0.5051, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.5387608500448967, | |
| "grad_norm": 4.844708442687988, | |
| "learning_rate": 2.5632069194943447e-05, | |
| "loss": 0.3833, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5462436396288536, | |
| "grad_norm": 13.217074394226074, | |
| "learning_rate": 2.521623419827013e-05, | |
| "loss": 0.4269, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.5537264292128106, | |
| "grad_norm": 12.129449844360352, | |
| "learning_rate": 2.4800399201596807e-05, | |
| "loss": 0.3442, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.5612092187967674, | |
| "grad_norm": 6.100238800048828, | |
| "learning_rate": 2.438456420492349e-05, | |
| "loss": 0.4414, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.5686920083807243, | |
| "grad_norm": 5.530787944793701, | |
| "learning_rate": 2.3968729208250168e-05, | |
| "loss": 0.4188, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5761747979646812, | |
| "grad_norm": 3.8855533599853516, | |
| "learning_rate": 2.355289421157685e-05, | |
| "loss": 0.497, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.5836575875486382, | |
| "grad_norm": 6.773860454559326, | |
| "learning_rate": 2.3137059214903528e-05, | |
| "loss": 0.4468, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.591140377132595, | |
| "grad_norm": 6.173677921295166, | |
| "learning_rate": 2.272122421823021e-05, | |
| "loss": 0.3887, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.5986231667165519, | |
| "grad_norm": 10.444341659545898, | |
| "learning_rate": 2.2305389221556888e-05, | |
| "loss": 0.4072, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6061059563005088, | |
| "grad_norm": 6.459510803222656, | |
| "learning_rate": 2.1889554224883567e-05, | |
| "loss": 0.4517, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.6135887458844658, | |
| "grad_norm": 6.071698188781738, | |
| "learning_rate": 2.147371922821025e-05, | |
| "loss": 0.3887, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.6210715354684226, | |
| "grad_norm": 3.3191375732421875, | |
| "learning_rate": 2.1057884231536927e-05, | |
| "loss": 0.3744, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.6285543250523795, | |
| "grad_norm": 8.275821685791016, | |
| "learning_rate": 2.0642049234863606e-05, | |
| "loss": 0.3914, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.6360371146363364, | |
| "grad_norm": 3.636678457260132, | |
| "learning_rate": 2.0226214238190287e-05, | |
| "loss": 0.4828, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.6435199042202934, | |
| "grad_norm": 2.3540258407592773, | |
| "learning_rate": 1.9810379241516966e-05, | |
| "loss": 0.4719, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.6510026938042502, | |
| "grad_norm": 3.962205171585083, | |
| "learning_rate": 1.9394544244843644e-05, | |
| "loss": 0.3873, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.6584854833882071, | |
| "grad_norm": 4.5042338371276855, | |
| "learning_rate": 1.8978709248170326e-05, | |
| "loss": 0.3418, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.665968272972164, | |
| "grad_norm": 10.76093578338623, | |
| "learning_rate": 1.8562874251497005e-05, | |
| "loss": 0.3311, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.673451062556121, | |
| "grad_norm": 12.721053123474121, | |
| "learning_rate": 1.8147039254823687e-05, | |
| "loss": 0.5621, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.6809338521400778, | |
| "grad_norm": 4.653266906738281, | |
| "learning_rate": 1.7731204258150365e-05, | |
| "loss": 0.41, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.6884166417240347, | |
| "grad_norm": 15.339615821838379, | |
| "learning_rate": 1.7315369261477047e-05, | |
| "loss": 0.4568, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.6958994313079916, | |
| "grad_norm": 6.669954776763916, | |
| "learning_rate": 1.6899534264803725e-05, | |
| "loss": 0.4849, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.7033822208919486, | |
| "grad_norm": 7.463631629943848, | |
| "learning_rate": 1.6483699268130407e-05, | |
| "loss": 0.3792, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7108650104759054, | |
| "grad_norm": 9.787162780761719, | |
| "learning_rate": 1.6067864271457086e-05, | |
| "loss": 0.3469, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.7183478000598623, | |
| "grad_norm": 4.896087646484375, | |
| "learning_rate": 1.5652029274783768e-05, | |
| "loss": 0.3956, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7258305896438192, | |
| "grad_norm": 13.51064395904541, | |
| "learning_rate": 1.5236194278110446e-05, | |
| "loss": 0.4875, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.7333133792277761, | |
| "grad_norm": 2.249631404876709, | |
| "learning_rate": 1.4820359281437126e-05, | |
| "loss": 0.4543, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.740796168811733, | |
| "grad_norm": 3.9572858810424805, | |
| "learning_rate": 1.4404524284763806e-05, | |
| "loss": 0.378, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.7482789583956899, | |
| "grad_norm": 6.323542594909668, | |
| "learning_rate": 1.3988689288090487e-05, | |
| "loss": 0.4459, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.7557617479796468, | |
| "grad_norm": 2.4348385334014893, | |
| "learning_rate": 1.3572854291417167e-05, | |
| "loss": 0.3472, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.7632445375636037, | |
| "grad_norm": 11.666696548461914, | |
| "learning_rate": 1.3157019294743847e-05, | |
| "loss": 0.4992, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.7707273271475606, | |
| "grad_norm": 5.232034683227539, | |
| "learning_rate": 1.2741184298070527e-05, | |
| "loss": 0.4098, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.7782101167315175, | |
| "grad_norm": 8.55593490600586, | |
| "learning_rate": 1.2325349301397205e-05, | |
| "loss": 0.3291, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.7856929063154744, | |
| "grad_norm": 9.049400329589844, | |
| "learning_rate": 1.1909514304723886e-05, | |
| "loss": 0.4242, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.7931756958994313, | |
| "grad_norm": 5.562539100646973, | |
| "learning_rate": 1.1493679308050566e-05, | |
| "loss": 0.4245, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8006584854833882, | |
| "grad_norm": 3.8648688793182373, | |
| "learning_rate": 1.1077844311377246e-05, | |
| "loss": 0.3807, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.8081412750673451, | |
| "grad_norm": 5.182237148284912, | |
| "learning_rate": 1.0662009314703926e-05, | |
| "loss": 0.3454, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.815624064651302, | |
| "grad_norm": 11.694429397583008, | |
| "learning_rate": 1.0246174318030606e-05, | |
| "loss": 0.3823, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.8231068542352589, | |
| "grad_norm": 4.656701564788818, | |
| "learning_rate": 9.830339321357286e-06, | |
| "loss": 0.3904, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.8305896438192159, | |
| "grad_norm": 14.795564651489258, | |
| "learning_rate": 9.414504324683967e-06, | |
| "loss": 0.435, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.8380724334031727, | |
| "grad_norm": 7.387109756469727, | |
| "learning_rate": 8.998669328010647e-06, | |
| "loss": 0.4077, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.8455552229871296, | |
| "grad_norm": 3.8318941593170166, | |
| "learning_rate": 8.582834331337327e-06, | |
| "loss": 0.4102, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.8530380125710865, | |
| "grad_norm": 3.9450416564941406, | |
| "learning_rate": 8.166999334664007e-06, | |
| "loss": 0.3393, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.8605208021550435, | |
| "grad_norm": 4.179422855377197, | |
| "learning_rate": 7.751164337990686e-06, | |
| "loss": 0.3033, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.8680035917390003, | |
| "grad_norm": 17.64090919494629, | |
| "learning_rate": 7.335329341317365e-06, | |
| "loss": 0.3748, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.8754863813229572, | |
| "grad_norm": 5.624340057373047, | |
| "learning_rate": 6.919494344644045e-06, | |
| "loss": 0.3161, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.8829691709069141, | |
| "grad_norm": 7.311130523681641, | |
| "learning_rate": 6.503659347970725e-06, | |
| "loss": 0.4454, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.8904519604908709, | |
| "grad_norm": 14.026363372802734, | |
| "learning_rate": 6.0878243512974054e-06, | |
| "loss": 0.3298, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.8979347500748279, | |
| "grad_norm": 10.132341384887695, | |
| "learning_rate": 5.671989354624086e-06, | |
| "loss": 0.3178, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9054175396587848, | |
| "grad_norm": 0.5942428708076477, | |
| "learning_rate": 5.256154357950766e-06, | |
| "loss": 0.4019, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.9129003292427417, | |
| "grad_norm": 11.00678539276123, | |
| "learning_rate": 4.840319361277446e-06, | |
| "loss": 0.3388, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.9203831188266985, | |
| "grad_norm": 3.2819418907165527, | |
| "learning_rate": 4.424484364604125e-06, | |
| "loss": 0.445, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.9278659084106555, | |
| "grad_norm": 11.440841674804688, | |
| "learning_rate": 4.008649367930805e-06, | |
| "loss": 0.3301, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.9353486979946124, | |
| "grad_norm": 10.728005409240723, | |
| "learning_rate": 3.592814371257485e-06, | |
| "loss": 0.3883, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.9428314875785693, | |
| "grad_norm": 4.131738185882568, | |
| "learning_rate": 3.1769793745841653e-06, | |
| "loss": 0.3623, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.9503142771625261, | |
| "grad_norm": 14.78467082977295, | |
| "learning_rate": 2.761144377910845e-06, | |
| "loss": 0.3265, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.9577970667464831, | |
| "grad_norm": 11.148056983947754, | |
| "learning_rate": 2.345309381237525e-06, | |
| "loss": 0.4538, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.96527985633044, | |
| "grad_norm": 8.455355644226074, | |
| "learning_rate": 1.929474384564205e-06, | |
| "loss": 0.3096, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.9727626459143969, | |
| "grad_norm": 5.342031002044678, | |
| "learning_rate": 1.513639387890885e-06, | |
| "loss": 0.4483, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.9802454354983537, | |
| "grad_norm": 5.38719367980957, | |
| "learning_rate": 1.097804391217565e-06, | |
| "loss": 0.3666, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.9877282250823107, | |
| "grad_norm": 4.644260883331299, | |
| "learning_rate": 6.819693945442449e-07, | |
| "loss": 0.4146, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.9952110146662676, | |
| "grad_norm": 10.537574768066406, | |
| "learning_rate": 2.661343978709248e-07, | |
| "loss": 0.3392, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8223319862296063, | |
| "eval_auc": 0.910394893673715, | |
| "eval_f1": 0.8549786194257789, | |
| "eval_loss": 0.38531041145324707, | |
| "eval_precision": 0.8500971817298348, | |
| "eval_recall": 0.8599164413860899, | |
| "eval_runtime": 931.428, | |
| "eval_samples_per_second": 7.173, | |
| "eval_steps_per_second": 0.449, | |
| "step": 3341 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 3341, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.01 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1757844960860160.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |