Harshkmr's picture
Model save
0f5ffb9 verified
raw
history blame
23 kB
{
"best_metric": 0.016786308959126472,
"best_model_checkpoint": "./deberta-v3-base_finetuned_bluegennx_run2.21_3e/checkpoint-48015",
"epoch": 3.0,
"eval_steps": 500,
"global_step": 48015,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.031240237425804437,
"grad_norm": 1.9453787803649902,
"learning_rate": 2.6033531188170364e-06,
"loss": 2.1742,
"step": 500
},
{
"epoch": 0.062480474851608875,
"grad_norm": 2.207657814025879,
"learning_rate": 5.206706237634073e-06,
"loss": 0.3611,
"step": 1000
},
{
"epoch": 0.09372071227741331,
"grad_norm": 3.9118669033050537,
"learning_rate": 7.81005935645111e-06,
"loss": 0.1781,
"step": 1500
},
{
"epoch": 0.12496094970321775,
"grad_norm": 1.7872343063354492,
"learning_rate": 1.0413412475268146e-05,
"loss": 0.1175,
"step": 2000
},
{
"epoch": 0.15620118712902217,
"grad_norm": 0.5929828882217407,
"learning_rate": 1.3016765594085182e-05,
"loss": 0.0815,
"step": 2500
},
{
"epoch": 0.18744142455482662,
"grad_norm": 0.8290306925773621,
"learning_rate": 1.562011871290222e-05,
"loss": 0.0766,
"step": 3000
},
{
"epoch": 0.21868166198063105,
"grad_norm": 0.4442843198776245,
"learning_rate": 1.8223471831719256e-05,
"loss": 0.0637,
"step": 3500
},
{
"epoch": 0.2499218994064355,
"grad_norm": 5.056885719299316,
"learning_rate": 2.082682495053629e-05,
"loss": 0.0588,
"step": 4000
},
{
"epoch": 0.28116213683223995,
"grad_norm": 2.825920820236206,
"learning_rate": 2.3430178069353328e-05,
"loss": 0.0525,
"step": 4500
},
{
"epoch": 0.31240237425804435,
"grad_norm": 1.8231415748596191,
"learning_rate": 2.6033531188170364e-05,
"loss": 0.0533,
"step": 5000
},
{
"epoch": 0.3436426116838488,
"grad_norm": 0.9496334195137024,
"learning_rate": 2.86368843069874e-05,
"loss": 0.06,
"step": 5500
},
{
"epoch": 0.37488284910965325,
"grad_norm": 0.6990556716918945,
"learning_rate": 3.124023742580444e-05,
"loss": 0.059,
"step": 6000
},
{
"epoch": 0.40612308653545764,
"grad_norm": 1.7713068723678589,
"learning_rate": 3.384359054462147e-05,
"loss": 0.0465,
"step": 6500
},
{
"epoch": 0.4373633239612621,
"grad_norm": 0.48251184821128845,
"learning_rate": 3.644694366343851e-05,
"loss": 0.0508,
"step": 7000
},
{
"epoch": 0.46860356138706655,
"grad_norm": 0.4298129379749298,
"learning_rate": 3.9050296782255544e-05,
"loss": 0.0531,
"step": 7500
},
{
"epoch": 0.499843798812871,
"grad_norm": 0.6740893125534058,
"learning_rate": 4.165364990107258e-05,
"loss": 0.0443,
"step": 8000
},
{
"epoch": 0.5310840362386754,
"grad_norm": 0.515224277973175,
"learning_rate": 4.425700301988962e-05,
"loss": 0.0442,
"step": 8500
},
{
"epoch": 0.5623242736644799,
"grad_norm": 0.07582231611013412,
"learning_rate": 4.6860356138706655e-05,
"loss": 0.0455,
"step": 9000
},
{
"epoch": 0.5935645110902843,
"grad_norm": 0.13481149077415466,
"learning_rate": 4.9463709257523695e-05,
"loss": 0.0502,
"step": 9500
},
{
"epoch": 0.6248047485160887,
"grad_norm": 0.06367824971675873,
"learning_rate": 4.998682293232702e-05,
"loss": 0.0497,
"step": 10000
},
{
"epoch": 0.6560449859418932,
"grad_norm": 0.107658751308918,
"learning_rate": 4.993275406892816e-05,
"loss": 0.0429,
"step": 10500
},
{
"epoch": 0.6872852233676976,
"grad_norm": 0.1800769716501236,
"learning_rate": 4.9836996765456125e-05,
"loss": 0.0425,
"step": 11000
},
{
"epoch": 0.718525460793502,
"grad_norm": 0.3818538784980774,
"learning_rate": 4.969971113148389e-05,
"loss": 0.035,
"step": 11500
},
{
"epoch": 0.7497656982193065,
"grad_norm": 0.8075230717658997,
"learning_rate": 4.952112671341091e-05,
"loss": 0.0378,
"step": 12000
},
{
"epoch": 0.7810059356451109,
"grad_norm": 1.5522449016571045,
"learning_rate": 4.9301542110653454e-05,
"loss": 0.0434,
"step": 12500
},
{
"epoch": 0.8122461730709153,
"grad_norm": 0.18798592686653137,
"learning_rate": 4.904132447637588e-05,
"loss": 0.0377,
"step": 13000
},
{
"epoch": 0.8434864104967198,
"grad_norm": 0.46252405643463135,
"learning_rate": 4.8740908903597714e-05,
"loss": 0.0386,
"step": 13500
},
{
"epoch": 0.8747266479225242,
"grad_norm": 1.5039823055267334,
"learning_rate": 4.8400797697702785e-05,
"loss": 0.036,
"step": 14000
},
{
"epoch": 0.9059668853483287,
"grad_norm": 4.865286350250244,
"learning_rate": 4.802155953656704e-05,
"loss": 0.0332,
"step": 14500
},
{
"epoch": 0.9372071227741331,
"grad_norm": 1.6047935485839844,
"learning_rate": 4.760382851970915e-05,
"loss": 0.0358,
"step": 15000
},
{
"epoch": 0.9684473601999375,
"grad_norm": 0.28227582573890686,
"learning_rate": 4.714830310805381e-05,
"loss": 0.0338,
"step": 15500
},
{
"epoch": 0.999687597625742,
"grad_norm": 1.1359831094741821,
"learning_rate": 4.6655744956080655e-05,
"loss": 0.0308,
"step": 16000
},
{
"epoch": 1.0,
"eval_AADHAR_CARD_f1": 0.935032483758121,
"eval_AGE_f1": 0.9315192743764172,
"eval_CITY_f1": 0.965787821720025,
"eval_COUNTRY_f1": 0.9705882352941178,
"eval_CREDITCARDCVV_f1": 0.9630872483221476,
"eval_CREDITCARDNUMBER_f1": 0.9281792224906655,
"eval_DATEOFBIRTH_f1": 0.8429846212265047,
"eval_DATE_f1": 0.9250582271544047,
"eval_EMAIL_f1": 0.9719218537028623,
"eval_EXPIRYDATE_f1": 0.9841930116472545,
"eval_ORGANIZATION_f1": 0.9866225478610258,
"eval_PAN_CARD_f1": 0.9696224758560141,
"eval_PERSON_f1": 0.9771855010660981,
"eval_PHONENUMBER_f1": 0.9502501389660922,
"eval_PINCODE_f1": 0.9834710743801652,
"eval_SECONDARYADDRESS_f1": 0.968014484007242,
"eval_STATE_f1": 0.9814928300036769,
"eval_TIME_f1": 0.9739364385404405,
"eval_URL_f1": 0.9844536895117144,
"eval_loss": 0.03587241843342781,
"eval_overall_accuracy": 0.9919974872098162,
"eval_overall_f1": 0.96261001579714,
"eval_overall_precision": 0.9500096174288057,
"eval_overall_recall": 0.9755491563835206,
"eval_runtime": 307.0183,
"eval_samples_per_second": 52.13,
"eval_steps_per_second": 13.035,
"step": 16005
},
{
"epoch": 1.0309278350515463,
"grad_norm": 2.9516384601593018,
"learning_rate": 4.6126977638311145e-05,
"loss": 0.0315,
"step": 16500
},
{
"epoch": 1.0621680724773508,
"grad_norm": 0.9985421299934387,
"learning_rate": 4.556288527226308e-05,
"loss": 0.0258,
"step": 17000
},
{
"epoch": 1.0934083099031553,
"grad_norm": 0.09413296729326248,
"learning_rate": 4.496441104017508e-05,
"loss": 0.0251,
"step": 17500
},
{
"epoch": 1.1246485473289598,
"grad_norm": 0.5080928802490234,
"learning_rate": 4.433255561197277e-05,
"loss": 0.0375,
"step": 18000
},
{
"epoch": 1.155888784754764,
"grad_norm": 0.6283013224601746,
"learning_rate": 4.366837547211354e-05,
"loss": 0.0238,
"step": 18500
},
{
"epoch": 1.1871290221805686,
"grad_norm": 0.09778323024511337,
"learning_rate": 4.297298115310743e-05,
"loss": 0.0258,
"step": 19000
},
{
"epoch": 1.218369259606373,
"grad_norm": 0.5504295825958252,
"learning_rate": 4.224753537866777e-05,
"loss": 0.0337,
"step": 19500
},
{
"epoch": 1.2496094970321774,
"grad_norm": 0.045782480388879776,
"learning_rate": 4.1493251119596374e-05,
"loss": 0.0299,
"step": 20000
},
{
"epoch": 1.280849734457982,
"grad_norm": 0.31934645771980286,
"learning_rate": 4.071138956565373e-05,
"loss": 0.0264,
"step": 20500
},
{
"epoch": 1.3120899718837864,
"grad_norm": 0.09766834229230881,
"learning_rate": 3.990325801680546e-05,
"loss": 0.0247,
"step": 21000
},
{
"epoch": 1.3433302093095907,
"grad_norm": 1.4730592966079712,
"learning_rate": 3.907020769737086e-05,
"loss": 0.024,
"step": 21500
},
{
"epoch": 1.3745704467353952,
"grad_norm": 0.22080545127391815,
"learning_rate": 3.8213631496728466e-05,
"loss": 0.0263,
"step": 22000
},
{
"epoch": 1.4058106841611997,
"grad_norm": 1.110324501991272,
"learning_rate": 3.733496164035603e-05,
"loss": 0.0232,
"step": 22500
},
{
"epoch": 1.437050921587004,
"grad_norm": 0.38805103302001953,
"learning_rate": 3.643566729509932e-05,
"loss": 0.0277,
"step": 23000
},
{
"epoch": 1.4682911590128085,
"grad_norm": 0.02454444207251072,
"learning_rate": 3.551725211267354e-05,
"loss": 0.0227,
"step": 23500
},
{
"epoch": 1.499531396438613,
"grad_norm": 1.0061655044555664,
"learning_rate": 3.458125171550487e-05,
"loss": 0.0258,
"step": 24000
},
{
"epoch": 1.5307716338644175,
"grad_norm": 0.09139035642147064,
"learning_rate": 3.362923112911588e-05,
"loss": 0.0238,
"step": 24500
},
{
"epoch": 1.5620118712902218,
"grad_norm": 1.6061333417892456,
"learning_rate": 3.266278216534792e-05,
"loss": 0.0228,
"step": 25000
},
{
"epoch": 1.5932521087160263,
"grad_norm": 0.014178004115819931,
"learning_rate": 3.1683520760795795e-05,
"loss": 0.0253,
"step": 25500
},
{
"epoch": 1.6244923461418308,
"grad_norm": 0.1629280000925064,
"learning_rate": 3.069308427490513e-05,
"loss": 0.0226,
"step": 26000
},
{
"epoch": 1.655732583567635,
"grad_norm": 0.6675782203674316,
"learning_rate": 2.969312875224981e-05,
"loss": 0.0209,
"step": 26500
},
{
"epoch": 1.6869728209934396,
"grad_norm": 0.20190191268920898,
"learning_rate": 2.8685326153567404e-05,
"loss": 0.0232,
"step": 27000
},
{
"epoch": 1.718213058419244,
"grad_norm": 0.22869578003883362,
"learning_rate": 2.7671361560182018e-05,
"loss": 0.0205,
"step": 27500
},
{
"epoch": 1.7494532958450484,
"grad_norm": 0.4046876132488251,
"learning_rate": 2.6652930356489275e-05,
"loss": 0.0235,
"step": 28000
},
{
"epoch": 1.780693533270853,
"grad_norm": 0.08248839527368546,
"learning_rate": 2.563173539521398e-05,
"loss": 0.0212,
"step": 28500
},
{
"epoch": 1.8119337706966574,
"grad_norm": 0.14581803977489471,
"learning_rate": 2.460948415018067e-05,
"loss": 0.022,
"step": 29000
},
{
"epoch": 1.8431740081224617,
"grad_norm": 0.07922026515007019,
"learning_rate": 2.3587885861357417e-05,
"loss": 0.0201,
"step": 29500
},
{
"epoch": 1.8744142455482662,
"grad_norm": 0.12380392849445343,
"learning_rate": 2.256864867694655e-05,
"loss": 0.0222,
"step": 30000
},
{
"epoch": 1.9056544829740707,
"grad_norm": 0.01387234590947628,
"learning_rate": 2.1553476797300967e-05,
"loss": 0.0172,
"step": 30500
},
{
"epoch": 1.936894720399875,
"grad_norm": 0.7150996923446655,
"learning_rate": 2.054406762544126e-05,
"loss": 0.0193,
"step": 31000
},
{
"epoch": 1.9681349578256795,
"grad_norm": 0.12249834090471268,
"learning_rate": 1.954210892893833e-05,
"loss": 0.0162,
"step": 31500
},
{
"epoch": 1.999375195251484,
"grad_norm": 0.20916196703910828,
"learning_rate": 1.8549276017906707e-05,
"loss": 0.0202,
"step": 32000
},
{
"epoch": 2.0,
"eval_AADHAR_CARD_f1": 0.975609756097561,
"eval_AGE_f1": 0.9585812623964532,
"eval_CITY_f1": 0.9789817842129845,
"eval_COUNTRY_f1": 0.9825646794150731,
"eval_CREDITCARDCVV_f1": 0.9866327180140038,
"eval_CREDITCARDNUMBER_f1": 0.9497256746165043,
"eval_DATEOFBIRTH_f1": 0.9060173073052928,
"eval_DATE_f1": 0.9592630501535313,
"eval_EMAIL_f1": 0.9893244984354868,
"eval_EXPIRYDATE_f1": 0.9871847871021083,
"eval_ORGANIZATION_f1": 0.9902360413309318,
"eval_PAN_CARD_f1": 0.9716031631919482,
"eval_PERSON_f1": 0.9874935533780299,
"eval_PHONENUMBER_f1": 0.9829537315570835,
"eval_PINCODE_f1": 0.9939471169162153,
"eval_SECONDARYADDRESS_f1": 0.9831296090514193,
"eval_STATE_f1": 0.9871265484576147,
"eval_TIME_f1": 0.9798612286342867,
"eval_URL_f1": 0.9927456583864586,
"eval_loss": 0.019483117386698723,
"eval_overall_accuracy": 0.9950012756613735,
"eval_overall_f1": 0.9786100681623069,
"eval_overall_precision": 0.9736953143492266,
"eval_overall_recall": 0.9835746883868889,
"eval_runtime": 307.7213,
"eval_samples_per_second": 52.011,
"eval_steps_per_second": 13.005,
"step": 32010
},
{
"epoch": 2.0306154326772883,
"grad_norm": 0.4832023084163666,
"learning_rate": 1.7567228943827265e-05,
"loss": 0.0189,
"step": 32500
},
{
"epoch": 2.0618556701030926,
"grad_norm": 0.07361295074224472,
"learning_rate": 1.6597609723882834e-05,
"loss": 0.0177,
"step": 33000
},
{
"epoch": 2.0930959075288973,
"grad_norm": 0.5370087623596191,
"learning_rate": 1.5642039595447862e-05,
"loss": 0.0108,
"step": 33500
},
{
"epoch": 2.1243361449547016,
"grad_norm": 0.2966110408306122,
"learning_rate": 1.4702116305322614e-05,
"loss": 0.0122,
"step": 34000
},
{
"epoch": 2.155576382380506,
"grad_norm": 0.25330406427383423,
"learning_rate": 1.3779411438244385e-05,
"loss": 0.0126,
"step": 34500
},
{
"epoch": 2.1868166198063106,
"grad_norm": 0.023234376683831215,
"learning_rate": 1.2875467789142694e-05,
"loss": 0.0149,
"step": 35000
},
{
"epoch": 2.218056857232115,
"grad_norm": 0.15632040798664093,
"learning_rate": 1.1991796783531975e-05,
"loss": 0.0118,
"step": 35500
},
{
"epoch": 2.2492970946579196,
"grad_norm": 2.662566900253296,
"learning_rate": 1.1129875950355095e-05,
"loss": 0.0147,
"step": 36000
},
{
"epoch": 2.280537332083724,
"grad_norm": 0.07179576903581619,
"learning_rate": 1.0291146451503033e-05,
"loss": 0.013,
"step": 36500
},
{
"epoch": 2.311777569509528,
"grad_norm": 1.6363691091537476,
"learning_rate": 9.477010672141675e-06,
"loss": 0.0101,
"step": 37000
},
{
"epoch": 2.3430178069353325,
"grad_norm": 0.0774739608168602,
"learning_rate": 8.688829875874516e-06,
"loss": 0.0175,
"step": 37500
},
{
"epoch": 2.374258044361137,
"grad_norm": 0.07443530112504959,
"learning_rate": 7.927921928662193e-06,
"loss": 0.0104,
"step": 38000
},
{
"epoch": 2.4054982817869415,
"grad_norm": 0.021464819088578224,
"learning_rate": 7.195559095304283e-06,
"loss": 0.0118,
"step": 38500
},
{
"epoch": 2.436738519212746,
"grad_norm": 0.9025238752365112,
"learning_rate": 6.492965912167923e-06,
"loss": 0.0122,
"step": 39000
},
{
"epoch": 2.4679787566385505,
"grad_norm": 0.19682565331459045,
"learning_rate": 5.821317139720001e-06,
"loss": 0.0118,
"step": 39500
},
{
"epoch": 2.4992189940643548,
"grad_norm": 0.056999098509550095,
"learning_rate": 5.1817357982864415e-06,
"loss": 0.0113,
"step": 40000
},
{
"epoch": 2.530459231490159,
"grad_norm": 0.016481826081871986,
"learning_rate": 4.575291290322764e-06,
"loss": 0.011,
"step": 40500
},
{
"epoch": 2.561699468915964,
"grad_norm": 0.011844165623188019,
"learning_rate": 4.0029976123356544e-06,
"loss": 0.0108,
"step": 41000
},
{
"epoch": 2.592939706341768,
"grad_norm": 0.02857145480811596,
"learning_rate": 3.465811659445206e-06,
"loss": 0.0134,
"step": 41500
},
{
"epoch": 2.624179943767573,
"grad_norm": 0.03046867437660694,
"learning_rate": 2.9646316254226786e-06,
"loss": 0.0126,
"step": 42000
},
{
"epoch": 2.655420181193377,
"grad_norm": 0.0065807169303298,
"learning_rate": 2.5002955008790074e-06,
"loss": 0.0089,
"step": 42500
},
{
"epoch": 2.6866604186191814,
"grad_norm": 0.16103477776050568,
"learning_rate": 2.0735796721150447e-06,
"loss": 0.0154,
"step": 43000
},
{
"epoch": 2.717900656044986,
"grad_norm": 0.08998264372348785,
"learning_rate": 1.6851976229764393e-06,
"loss": 0.0123,
"step": 43500
},
{
"epoch": 2.7491408934707904,
"grad_norm": 0.04546121135354042,
"learning_rate": 1.3357987418835955e-06,
"loss": 0.0092,
"step": 44000
},
{
"epoch": 2.7803811308965947,
"grad_norm": 0.6218581199645996,
"learning_rate": 1.0259672360314221e-06,
"loss": 0.0101,
"step": 44500
},
{
"epoch": 2.8116213683223994,
"grad_norm": 0.2742210030555725,
"learning_rate": 7.562211545744102e-07,
"loss": 0.009,
"step": 45000
},
{
"epoch": 2.8428616057482037,
"grad_norm": 0.0064840721897780895,
"learning_rate": 5.270115224302541e-07,
"loss": 0.0109,
"step": 45500
},
{
"epoch": 2.874101843174008,
"grad_norm": 0.033717311918735504,
"learning_rate": 3.3872158615035067e-07,
"loss": 0.0136,
"step": 46000
},
{
"epoch": 2.9053420805998127,
"grad_norm": 0.08778509497642517,
"learning_rate": 1.9166617311812407e-07,
"loss": 0.0122,
"step": 46500
},
{
"epoch": 2.936582318025617,
"grad_norm": 4.181912422180176,
"learning_rate": 8.609116514657622e-08,
"loss": 0.0105,
"step": 47000
},
{
"epoch": 2.9678225554514217,
"grad_norm": 0.6781991124153137,
"learning_rate": 2.2173087355270573e-08,
"loss": 0.0129,
"step": 47500
},
{
"epoch": 2.999062792877226,
"grad_norm": 0.332028865814209,
"learning_rate": 1.8813014129870354e-11,
"loss": 0.0107,
"step": 48000
},
{
"epoch": 3.0,
"eval_AADHAR_CARD_f1": 0.9865537848605577,
"eval_AGE_f1": 0.9707006369426752,
"eval_CITY_f1": 0.9867601246105918,
"eval_COUNTRY_f1": 0.9864864864864865,
"eval_CREDITCARDCVV_f1": 0.9887830687830689,
"eval_CREDITCARDNUMBER_f1": 0.9587166255328696,
"eval_DATEOFBIRTH_f1": 0.9165348101265822,
"eval_DATE_f1": 0.9642906194204092,
"eval_EMAIL_f1": 0.9894194498113901,
"eval_EXPIRYDATE_f1": 0.9921259842519685,
"eval_ORGANIZATION_f1": 0.9917026219714571,
"eval_PAN_CARD_f1": 0.9855999999999999,
"eval_PERSON_f1": 0.9883056021325078,
"eval_PHONENUMBER_f1": 0.9868345735546651,
"eval_PINCODE_f1": 0.9936386768447837,
"eval_SECONDARYADDRESS_f1": 0.9860662358642972,
"eval_STATE_f1": 0.9901351845085861,
"eval_TIME_f1": 0.9821322719959352,
"eval_URL_f1": 0.994939493949395,
"eval_loss": 0.016786308959126472,
"eval_overall_accuracy": 0.9958519111859447,
"eval_overall_f1": 0.9824886126865787,
"eval_overall_precision": 0.9772792726075866,
"eval_overall_recall": 0.9877537866580728,
"eval_runtime": 309.3329,
"eval_samples_per_second": 51.74,
"eval_steps_per_second": 12.938,
"step": 48015
},
{
"epoch": 3.0,
"step": 48015,
"total_flos": 1.8885846596837596e+16,
"train_loss": 0.0555913904110963,
"train_runtime": 9844.7347,
"train_samples_per_second": 19.509,
"train_steps_per_second": 4.877
},
{
"epoch": 3.0,
"eval_AADHAR_CARD_f1": 0.9865537848605577,
"eval_AGE_f1": 0.9707006369426752,
"eval_CITY_f1": 0.9867601246105918,
"eval_COUNTRY_f1": 0.9864864864864865,
"eval_CREDITCARDCVV_f1": 0.9887830687830689,
"eval_CREDITCARDNUMBER_f1": 0.9587166255328696,
"eval_DATEOFBIRTH_f1": 0.9165348101265822,
"eval_DATE_f1": 0.9642906194204092,
"eval_EMAIL_f1": 0.9894194498113901,
"eval_EXPIRYDATE_f1": 0.9921259842519685,
"eval_ORGANIZATION_f1": 0.9917026219714571,
"eval_PAN_CARD_f1": 0.9855999999999999,
"eval_PERSON_f1": 0.9883056021325078,
"eval_PHONENUMBER_f1": 0.9868345735546651,
"eval_PINCODE_f1": 0.9936386768447837,
"eval_SECONDARYADDRESS_f1": 0.9860662358642972,
"eval_STATE_f1": 0.9901351845085861,
"eval_TIME_f1": 0.9821322719959352,
"eval_URL_f1": 0.994939493949395,
"eval_loss": 0.016786308959126472,
"eval_overall_accuracy": 0.9958519111859447,
"eval_overall_f1": 0.9824886126865787,
"eval_overall_precision": 0.9772792726075866,
"eval_overall_recall": 0.9877537866580728,
"eval_runtime": 267.5949,
"eval_samples_per_second": 59.811,
"eval_steps_per_second": 14.955,
"step": 48015
}
],
"logging_steps": 500,
"max_steps": 48015,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"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": 1.8885846596837596e+16,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}