| { | |
| "best_metric": 3.0925960540771484, | |
| "best_model_checkpoint": "matching-buyback/checkpoint-2648", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 2648, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0377643504531722, | |
| "grad_norm": 2.148848295211792, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 7.1736, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0755287009063444, | |
| "grad_norm": 1.9532533884048462, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 7.1696, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11329305135951662, | |
| "grad_norm": 1.8605287075042725, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 7.1706, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1510574018126888, | |
| "grad_norm": 1.7690376043319702, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 7.168, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18882175226586104, | |
| "grad_norm": 1.6989686489105225, | |
| "learning_rate": 2.358490566037736e-05, | |
| "loss": 7.1623, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22658610271903323, | |
| "grad_norm": 1.9419589042663574, | |
| "learning_rate": 2.830188679245283e-05, | |
| "loss": 7.1582, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.26435045317220546, | |
| "grad_norm": 6.193149089813232, | |
| "learning_rate": 3.30188679245283e-05, | |
| "loss": 7.142, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3021148036253776, | |
| "grad_norm": 3.8507258892059326, | |
| "learning_rate": 3.7735849056603776e-05, | |
| "loss": 7.1296, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.33987915407854985, | |
| "grad_norm": 3.860013246536255, | |
| "learning_rate": 4.245283018867925e-05, | |
| "loss": 7.129, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3776435045317221, | |
| "grad_norm": 3.5848217010498047, | |
| "learning_rate": 4.716981132075472e-05, | |
| "loss": 7.104, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.41540785498489424, | |
| "grad_norm": 3.5513248443603516, | |
| "learning_rate": 4.979018044481746e-05, | |
| "loss": 7.0868, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.45317220543806647, | |
| "grad_norm": 3.685840368270874, | |
| "learning_rate": 4.9265631556861106e-05, | |
| "loss": 6.9957, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4909365558912387, | |
| "grad_norm": 3.6940855979919434, | |
| "learning_rate": 4.874108266890474e-05, | |
| "loss": 6.9211, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5287009063444109, | |
| "grad_norm": 4.479358673095703, | |
| "learning_rate": 4.8216533780948384e-05, | |
| "loss": 6.8176, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5664652567975831, | |
| "grad_norm": 4.331283092498779, | |
| "learning_rate": 4.769198489299203e-05, | |
| "loss": 6.7406, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.6042296072507553, | |
| "grad_norm": 4.426012992858887, | |
| "learning_rate": 4.716743600503567e-05, | |
| "loss": 6.6516, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6419939577039275, | |
| "grad_norm": 4.232052326202393, | |
| "learning_rate": 4.664288711707932e-05, | |
| "loss": 6.6214, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.6797583081570997, | |
| "grad_norm": 5.0197601318359375, | |
| "learning_rate": 4.6118338229122954e-05, | |
| "loss": 6.4544, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7175226586102719, | |
| "grad_norm": 5.322621822357178, | |
| "learning_rate": 4.5593789341166596e-05, | |
| "loss": 6.4331, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.7552870090634441, | |
| "grad_norm": 5.3774871826171875, | |
| "learning_rate": 4.5069240453210245e-05, | |
| "loss": 6.3182, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7930513595166163, | |
| "grad_norm": 5.22687292098999, | |
| "learning_rate": 4.454469156525388e-05, | |
| "loss": 6.2782, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.8308157099697885, | |
| "grad_norm": 4.664173603057861, | |
| "learning_rate": 4.402014267729753e-05, | |
| "loss": 6.2099, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.8685800604229608, | |
| "grad_norm": 5.771005153656006, | |
| "learning_rate": 4.349559378934117e-05, | |
| "loss": 6.1739, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.9063444108761329, | |
| "grad_norm": 5.225096225738525, | |
| "learning_rate": 4.297104490138481e-05, | |
| "loss": 6.1653, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.9441087613293051, | |
| "grad_norm": 5.635812282562256, | |
| "learning_rate": 4.244649601342846e-05, | |
| "loss": 6.0278, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.9818731117824774, | |
| "grad_norm": 5.488772392272949, | |
| "learning_rate": 4.192194712547209e-05, | |
| "loss": 6.0308, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.026465028355387523, | |
| "eval_f1_macro": 0.0018394084225980786, | |
| "eval_f1_micro": 0.026465028355387523, | |
| "eval_f1_weighted": 0.002693831801885104, | |
| "eval_loss": 5.7428083419799805, | |
| "eval_precision_macro": 0.0011073227324794681, | |
| "eval_precision_micro": 0.026465028355387523, | |
| "eval_precision_weighted": 0.0015115244261337012, | |
| "eval_recall_macro": 0.014671814671814672, | |
| "eval_recall_micro": 0.026465028355387523, | |
| "eval_recall_weighted": 0.026465028355387523, | |
| "eval_runtime": 258.8851, | |
| "eval_samples_per_second": 10.217, | |
| "eval_steps_per_second": 0.321, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.0196374622356494, | |
| "grad_norm": 5.648388385772705, | |
| "learning_rate": 4.139739823751574e-05, | |
| "loss": 5.8178, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.0574018126888218, | |
| "grad_norm": 6.245723724365234, | |
| "learning_rate": 4.0872849349559384e-05, | |
| "loss": 5.6123, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.095166163141994, | |
| "grad_norm": 6.041352272033691, | |
| "learning_rate": 4.034830046160302e-05, | |
| "loss": 5.7037, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.1329305135951662, | |
| "grad_norm": 6.065165996551514, | |
| "learning_rate": 3.982375157364667e-05, | |
| "loss": 5.6632, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.1706948640483383, | |
| "grad_norm": 5.901498794555664, | |
| "learning_rate": 3.9299202685690305e-05, | |
| "loss": 5.6154, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.2084592145015105, | |
| "grad_norm": 6.365645885467529, | |
| "learning_rate": 3.8774653797733954e-05, | |
| "loss": 5.4768, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.2462235649546827, | |
| "grad_norm": 6.6829705238342285, | |
| "learning_rate": 3.8250104909777596e-05, | |
| "loss": 5.3382, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.283987915407855, | |
| "grad_norm": 5.9976019859313965, | |
| "learning_rate": 3.772555602182123e-05, | |
| "loss": 5.4, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.3217522658610272, | |
| "grad_norm": 6.049790859222412, | |
| "learning_rate": 3.720100713386488e-05, | |
| "loss": 5.3734, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.3595166163141994, | |
| "grad_norm": 7.126038074493408, | |
| "learning_rate": 3.667645824590852e-05, | |
| "loss": 5.2598, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.3972809667673716, | |
| "grad_norm": 6.60282564163208, | |
| "learning_rate": 3.6151909357952166e-05, | |
| "loss": 5.2374, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.4350453172205437, | |
| "grad_norm": 6.773179054260254, | |
| "learning_rate": 3.562736046999581e-05, | |
| "loss": 5.172, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.4728096676737161, | |
| "grad_norm": 6.833974361419678, | |
| "learning_rate": 3.5102811582039444e-05, | |
| "loss": 5.1996, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.510574018126888, | |
| "grad_norm": 6.419187068939209, | |
| "learning_rate": 3.457826269408309e-05, | |
| "loss": 5.1801, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.5483383685800605, | |
| "grad_norm": 6.803603649139404, | |
| "learning_rate": 3.405371380612673e-05, | |
| "loss": 4.988, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.5861027190332326, | |
| "grad_norm": 6.773290157318115, | |
| "learning_rate": 3.352916491817038e-05, | |
| "loss": 5.0186, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.6238670694864048, | |
| "grad_norm": 6.730032920837402, | |
| "learning_rate": 3.300461603021402e-05, | |
| "loss": 5.0032, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.6616314199395772, | |
| "grad_norm": 7.288454532623291, | |
| "learning_rate": 3.2480067142257656e-05, | |
| "loss": 4.908, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6993957703927491, | |
| "grad_norm": 6.888255596160889, | |
| "learning_rate": 3.1955518254301305e-05, | |
| "loss": 4.937, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.7371601208459215, | |
| "grad_norm": 6.915685176849365, | |
| "learning_rate": 3.143096936634494e-05, | |
| "loss": 4.9246, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.7749244712990937, | |
| "grad_norm": 6.667412757873535, | |
| "learning_rate": 3.090642047838859e-05, | |
| "loss": 4.8717, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.8126888217522659, | |
| "grad_norm": 6.816399097442627, | |
| "learning_rate": 3.0381871590432233e-05, | |
| "loss": 4.9049, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.850453172205438, | |
| "grad_norm": 7.150826930999756, | |
| "learning_rate": 2.985732270247587e-05, | |
| "loss": 4.705, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.8882175226586102, | |
| "grad_norm": 7.209892272949219, | |
| "learning_rate": 2.9332773814519514e-05, | |
| "loss": 4.8234, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.9259818731117826, | |
| "grad_norm": 7.008028507232666, | |
| "learning_rate": 2.8808224926563153e-05, | |
| "loss": 4.6716, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.9637462235649545, | |
| "grad_norm": 7.195932865142822, | |
| "learning_rate": 2.82836760386068e-05, | |
| "loss": 4.6394, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.21890359168241966, | |
| "eval_f1_macro": 0.07847614280788937, | |
| "eval_f1_micro": 0.21890359168241966, | |
| "eval_f1_weighted": 0.14378778601568412, | |
| "eval_loss": 4.332443714141846, | |
| "eval_precision_macro": 0.07050179882195226, | |
| "eval_precision_micro": 0.21890359168241966, | |
| "eval_precision_weighted": 0.12910697824366127, | |
| "eval_recall_macro": 0.12453024453024451, | |
| "eval_recall_micro": 0.21890359168241966, | |
| "eval_recall_weighted": 0.21890359168241966, | |
| "eval_runtime": 274.1069, | |
| "eval_samples_per_second": 9.65, | |
| "eval_steps_per_second": 0.303, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 2.001510574018127, | |
| "grad_norm": 7.194442272186279, | |
| "learning_rate": 2.7759127150650445e-05, | |
| "loss": 4.5155, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.039274924471299, | |
| "grad_norm": 7.649657726287842, | |
| "learning_rate": 2.7234578262694084e-05, | |
| "loss": 4.3742, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.0770392749244713, | |
| "grad_norm": 6.8479084968566895, | |
| "learning_rate": 2.6710029374737726e-05, | |
| "loss": 4.4017, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.1148036253776437, | |
| "grad_norm": 7.72012186050415, | |
| "learning_rate": 2.6185480486781372e-05, | |
| "loss": 4.2114, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.1525679758308156, | |
| "grad_norm": 7.429843425750732, | |
| "learning_rate": 2.566093159882501e-05, | |
| "loss": 4.3608, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.190332326283988, | |
| "grad_norm": 7.590367794036865, | |
| "learning_rate": 2.5136382710868657e-05, | |
| "loss": 4.2724, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.22809667673716, | |
| "grad_norm": 7.643247127532959, | |
| "learning_rate": 2.46118338229123e-05, | |
| "loss": 4.149, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.2658610271903323, | |
| "grad_norm": 7.686148643493652, | |
| "learning_rate": 2.4087284934955938e-05, | |
| "loss": 4.1981, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.3036253776435047, | |
| "grad_norm": 7.800241470336914, | |
| "learning_rate": 2.356273604699958e-05, | |
| "loss": 4.2175, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.3413897280966767, | |
| "grad_norm": 7.715053558349609, | |
| "learning_rate": 2.3038187159043223e-05, | |
| "loss": 4.1977, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.379154078549849, | |
| "grad_norm": 7.797581195831299, | |
| "learning_rate": 2.251363827108687e-05, | |
| "loss": 4.1727, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.416918429003021, | |
| "grad_norm": 7.720523834228516, | |
| "learning_rate": 2.198908938313051e-05, | |
| "loss": 4.0437, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.4546827794561934, | |
| "grad_norm": 7.72569465637207, | |
| "learning_rate": 2.146454049517415e-05, | |
| "loss": 4.0182, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.4924471299093653, | |
| "grad_norm": 7.541537761688232, | |
| "learning_rate": 2.0939991607217792e-05, | |
| "loss": 4.0712, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.5302114803625377, | |
| "grad_norm": 7.888173580169678, | |
| "learning_rate": 2.0415442719261435e-05, | |
| "loss": 3.9628, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.56797583081571, | |
| "grad_norm": 7.701706886291504, | |
| "learning_rate": 1.989089383130508e-05, | |
| "loss": 3.9174, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.605740181268882, | |
| "grad_norm": 7.674171447753906, | |
| "learning_rate": 1.9366344943348723e-05, | |
| "loss": 3.8891, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.6435045317220545, | |
| "grad_norm": 7.812932014465332, | |
| "learning_rate": 1.8841796055392362e-05, | |
| "loss": 3.9667, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.6812688821752264, | |
| "grad_norm": 8.138640403747559, | |
| "learning_rate": 1.8317247167436005e-05, | |
| "loss": 3.9761, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.719033232628399, | |
| "grad_norm": 7.94352912902832, | |
| "learning_rate": 1.7792698279479647e-05, | |
| "loss": 3.8518, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.756797583081571, | |
| "grad_norm": 7.7708539962768555, | |
| "learning_rate": 1.7268149391523293e-05, | |
| "loss": 3.6874, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.794561933534743, | |
| "grad_norm": 8.06053638458252, | |
| "learning_rate": 1.6743600503566935e-05, | |
| "loss": 3.8025, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.8323262839879155, | |
| "grad_norm": 7.878209590911865, | |
| "learning_rate": 1.6219051615610574e-05, | |
| "loss": 3.7442, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.8700906344410875, | |
| "grad_norm": 8.10824203491211, | |
| "learning_rate": 1.5694502727654217e-05, | |
| "loss": 3.8247, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.90785498489426, | |
| "grad_norm": 7.619449138641357, | |
| "learning_rate": 1.5169953839697862e-05, | |
| "loss": 3.6439, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.9456193353474323, | |
| "grad_norm": 8.001503944396973, | |
| "learning_rate": 1.4645404951741503e-05, | |
| "loss": 3.6267, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.983383685800604, | |
| "grad_norm": 7.938083648681641, | |
| "learning_rate": 1.4120856063785145e-05, | |
| "loss": 3.6283, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.45255198487712667, | |
| "eval_f1_macro": 0.22372025748723245, | |
| "eval_f1_micro": 0.45255198487712667, | |
| "eval_f1_weighted": 0.36915380230770156, | |
| "eval_loss": 3.410895347595215, | |
| "eval_precision_macro": 0.210819340352679, | |
| "eval_precision_micro": 0.45255198487712667, | |
| "eval_precision_weighted": 0.34998443503067106, | |
| "eval_recall_macro": 0.28555984555984554, | |
| "eval_recall_micro": 0.45255198487712667, | |
| "eval_recall_weighted": 0.45255198487712667, | |
| "eval_runtime": 240.6652, | |
| "eval_samples_per_second": 10.99, | |
| "eval_steps_per_second": 0.345, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 3.0211480362537766, | |
| "grad_norm": 8.365966796875, | |
| "learning_rate": 1.3596307175828788e-05, | |
| "loss": 3.6007, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0589123867069485, | |
| "grad_norm": 8.145760536193848, | |
| "learning_rate": 1.3071758287872429e-05, | |
| "loss": 3.5455, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.096676737160121, | |
| "grad_norm": 7.752460479736328, | |
| "learning_rate": 1.2547209399916074e-05, | |
| "loss": 3.3978, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.134441087613293, | |
| "grad_norm": 8.035235404968262, | |
| "learning_rate": 1.2022660511959715e-05, | |
| "loss": 3.5076, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.1722054380664653, | |
| "grad_norm": 8.423967361450195, | |
| "learning_rate": 1.1498111624003358e-05, | |
| "loss": 3.4326, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.2099697885196377, | |
| "grad_norm": 8.095000267028809, | |
| "learning_rate": 1.0973562736047e-05, | |
| "loss": 3.4518, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.2477341389728096, | |
| "grad_norm": 8.297784805297852, | |
| "learning_rate": 1.0449013848090642e-05, | |
| "loss": 3.4024, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.285498489425982, | |
| "grad_norm": 8.063372611999512, | |
| "learning_rate": 9.924464960134285e-06, | |
| "loss": 3.4202, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.323262839879154, | |
| "grad_norm": 7.8603515625, | |
| "learning_rate": 9.399916072177927e-06, | |
| "loss": 3.4905, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.3610271903323263, | |
| "grad_norm": 8.366976737976074, | |
| "learning_rate": 8.87536718422157e-06, | |
| "loss": 3.395, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.3987915407854983, | |
| "grad_norm": 8.408001899719238, | |
| "learning_rate": 8.350818296265214e-06, | |
| "loss": 3.4572, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.4365558912386707, | |
| "grad_norm": 7.793301582336426, | |
| "learning_rate": 7.826269408308854e-06, | |
| "loss": 3.3398, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.474320241691843, | |
| "grad_norm": 8.445042610168457, | |
| "learning_rate": 7.301720520352497e-06, | |
| "loss": 3.3126, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.512084592145015, | |
| "grad_norm": 8.048064231872559, | |
| "learning_rate": 6.77717163239614e-06, | |
| "loss": 3.3747, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.5498489425981874, | |
| "grad_norm": 8.69857120513916, | |
| "learning_rate": 6.252622744439782e-06, | |
| "loss": 3.424, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.5876132930513593, | |
| "grad_norm": 8.197694778442383, | |
| "learning_rate": 5.728073856483424e-06, | |
| "loss": 3.3231, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.6253776435045317, | |
| "grad_norm": 8.308802604675293, | |
| "learning_rate": 5.2035249685270664e-06, | |
| "loss": 3.2876, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.663141993957704, | |
| "grad_norm": 8.097749710083008, | |
| "learning_rate": 4.67897608057071e-06, | |
| "loss": 3.201, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.700906344410876, | |
| "grad_norm": 8.310770988464355, | |
| "learning_rate": 4.154427192614352e-06, | |
| "loss": 3.3542, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.7386706948640485, | |
| "grad_norm": 8.366157531738281, | |
| "learning_rate": 3.6298783046579945e-06, | |
| "loss": 3.4844, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.7764350453172204, | |
| "grad_norm": 8.279153823852539, | |
| "learning_rate": 3.1053294167016365e-06, | |
| "loss": 3.2068, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.814199395770393, | |
| "grad_norm": 8.165327072143555, | |
| "learning_rate": 2.5807805287452793e-06, | |
| "loss": 3.3177, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.851963746223565, | |
| "grad_norm": 8.91379165649414, | |
| "learning_rate": 2.0562316407889217e-06, | |
| "loss": 3.2367, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.889728096676737, | |
| "grad_norm": 8.2485990524292, | |
| "learning_rate": 1.5316827528325641e-06, | |
| "loss": 3.3136, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.9274924471299095, | |
| "grad_norm": 7.780291557312012, | |
| "learning_rate": 1.0071338648762065e-06, | |
| "loss": 3.273, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.9652567975830815, | |
| "grad_norm": 8.575440406799316, | |
| "learning_rate": 4.82584976919849e-07, | |
| "loss": 3.3768, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.5051039697542533, | |
| "eval_f1_macro": 0.2675876194485524, | |
| "eval_f1_micro": 0.5051039697542533, | |
| "eval_f1_weighted": 0.4266553262031796, | |
| "eval_loss": 3.0925960540771484, | |
| "eval_precision_macro": 0.2522810784129609, | |
| "eval_precision_micro": 0.5051039697542533, | |
| "eval_precision_weighted": 0.40593727556638787, | |
| "eval_recall_macro": 0.33, | |
| "eval_recall_micro": 0.5051039697542533, | |
| "eval_recall_weighted": 0.5051039697542533, | |
| "eval_runtime": 244.9673, | |
| "eval_samples_per_second": 10.797, | |
| "eval_steps_per_second": 0.339, | |
| "step": 2648 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 2648, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 2867373216675840.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |