sentiment-kcbert / trainer_state.json
lkw2025's picture
Upload 10 files
b74313b verified
{
"best_metric": 0.9087747458533976,
"best_model_checkpoint": "./checkpoints/beomi_kcbert-base\\checkpoint-5607",
"epoch": 3.0,
"eval_steps": 500,
"global_step": 5607,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02675227394328518,
"grad_norm": 5.228867053985596,
"learning_rate": 1.9821651507044767e-05,
"loss": 0.8035,
"step": 50
},
{
"epoch": 0.05350454788657036,
"grad_norm": 20.66253662109375,
"learning_rate": 1.9643303014089533e-05,
"loss": 0.5287,
"step": 100
},
{
"epoch": 0.08025682182985554,
"grad_norm": 9.66049575805664,
"learning_rate": 1.94649545211343e-05,
"loss": 0.5059,
"step": 150
},
{
"epoch": 0.10700909577314072,
"grad_norm": 12.687285423278809,
"learning_rate": 1.9286606028179064e-05,
"loss": 0.5266,
"step": 200
},
{
"epoch": 0.1337613697164259,
"grad_norm": 13.331902503967285,
"learning_rate": 1.910825753522383e-05,
"loss": 0.4482,
"step": 250
},
{
"epoch": 0.16051364365971107,
"grad_norm": 6.606617450714111,
"learning_rate": 1.8929909042268595e-05,
"loss": 0.4771,
"step": 300
},
{
"epoch": 0.18726591760299627,
"grad_norm": 3.6321372985839844,
"learning_rate": 1.8751560549313357e-05,
"loss": 0.5196,
"step": 350
},
{
"epoch": 0.21401819154628143,
"grad_norm": 5.700988292694092,
"learning_rate": 1.8573212056358126e-05,
"loss": 0.4296,
"step": 400
},
{
"epoch": 0.24077046548956663,
"grad_norm": 14.550703048706055,
"learning_rate": 1.8394863563402892e-05,
"loss": 0.4121,
"step": 450
},
{
"epoch": 0.2675227394328518,
"grad_norm": 11.130396842956543,
"learning_rate": 1.8216515070447654e-05,
"loss": 0.4423,
"step": 500
},
{
"epoch": 0.294275013376137,
"grad_norm": 13.779672622680664,
"learning_rate": 1.8038166577492423e-05,
"loss": 0.3953,
"step": 550
},
{
"epoch": 0.32102728731942215,
"grad_norm": 7.454031944274902,
"learning_rate": 1.7859818084537185e-05,
"loss": 0.3708,
"step": 600
},
{
"epoch": 0.3477795612627073,
"grad_norm": 5.720978260040283,
"learning_rate": 1.768146959158195e-05,
"loss": 0.421,
"step": 650
},
{
"epoch": 0.37453183520599254,
"grad_norm": 14.370809555053711,
"learning_rate": 1.750312109862672e-05,
"loss": 0.4154,
"step": 700
},
{
"epoch": 0.4012841091492777,
"grad_norm": 3.7566840648651123,
"learning_rate": 1.7324772605671482e-05,
"loss": 0.3827,
"step": 750
},
{
"epoch": 0.42803638309256287,
"grad_norm": 13.605573654174805,
"learning_rate": 1.7146424112716248e-05,
"loss": 0.3827,
"step": 800
},
{
"epoch": 0.45478865703584803,
"grad_norm": 3.2871532440185547,
"learning_rate": 1.6968075619761017e-05,
"loss": 0.4183,
"step": 850
},
{
"epoch": 0.48154093097913325,
"grad_norm": 5.473564624786377,
"learning_rate": 1.678972712680578e-05,
"loss": 0.3829,
"step": 900
},
{
"epoch": 0.5082932049224184,
"grad_norm": 20.812192916870117,
"learning_rate": 1.6611378633850545e-05,
"loss": 0.3746,
"step": 950
},
{
"epoch": 0.5350454788657036,
"grad_norm": 9.71611499786377,
"learning_rate": 1.643303014089531e-05,
"loss": 0.3721,
"step": 1000
},
{
"epoch": 0.5617977528089888,
"grad_norm": 9.370783805847168,
"learning_rate": 1.6254681647940076e-05,
"loss": 0.3902,
"step": 1050
},
{
"epoch": 0.588550026752274,
"grad_norm": 5.628559112548828,
"learning_rate": 1.607633315498484e-05,
"loss": 0.3221,
"step": 1100
},
{
"epoch": 0.6153023006955591,
"grad_norm": 9.518179893493652,
"learning_rate": 1.5897984662029607e-05,
"loss": 0.3722,
"step": 1150
},
{
"epoch": 0.6420545746388443,
"grad_norm": 12.84173583984375,
"learning_rate": 1.5719636169074372e-05,
"loss": 0.3963,
"step": 1200
},
{
"epoch": 0.6688068485821295,
"grad_norm": 2.825850486755371,
"learning_rate": 1.5541287676119138e-05,
"loss": 0.3787,
"step": 1250
},
{
"epoch": 0.6955591225254146,
"grad_norm": 18.818218231201172,
"learning_rate": 1.5362939183163904e-05,
"loss": 0.3443,
"step": 1300
},
{
"epoch": 0.7223113964686998,
"grad_norm": 8.157448768615723,
"learning_rate": 1.518459069020867e-05,
"loss": 0.3857,
"step": 1350
},
{
"epoch": 0.7490636704119851,
"grad_norm": 6.142508029937744,
"learning_rate": 1.5006242197253433e-05,
"loss": 0.3516,
"step": 1400
},
{
"epoch": 0.7758159443552702,
"grad_norm": 13.665266990661621,
"learning_rate": 1.48278937042982e-05,
"loss": 0.3564,
"step": 1450
},
{
"epoch": 0.8025682182985554,
"grad_norm": 6.96636962890625,
"learning_rate": 1.4649545211342966e-05,
"loss": 0.3241,
"step": 1500
},
{
"epoch": 0.8293204922418406,
"grad_norm": 4.158944606781006,
"learning_rate": 1.447119671838773e-05,
"loss": 0.3507,
"step": 1550
},
{
"epoch": 0.8560727661851257,
"grad_norm": 12.31186294555664,
"learning_rate": 1.4292848225432496e-05,
"loss": 0.3741,
"step": 1600
},
{
"epoch": 0.8828250401284109,
"grad_norm": 8.98780632019043,
"learning_rate": 1.4114499732477263e-05,
"loss": 0.3327,
"step": 1650
},
{
"epoch": 0.9095773140716961,
"grad_norm": 4.7375335693359375,
"learning_rate": 1.3936151239522027e-05,
"loss": 0.3146,
"step": 1700
},
{
"epoch": 0.9363295880149812,
"grad_norm": 7.906491756439209,
"learning_rate": 1.3757802746566792e-05,
"loss": 0.3437,
"step": 1750
},
{
"epoch": 0.9630818619582665,
"grad_norm": 3.6906189918518066,
"learning_rate": 1.3579454253611556e-05,
"loss": 0.3023,
"step": 1800
},
{
"epoch": 0.9898341359015517,
"grad_norm": 9.612889289855957,
"learning_rate": 1.3401105760656324e-05,
"loss": 0.2908,
"step": 1850
},
{
"epoch": 1.0,
"eval_accuracy": 0.8865703584804708,
"eval_f1": 0.8776064533107432,
"eval_loss": 0.2936590015888214,
"eval_precision": 0.879542702484244,
"eval_recall": 0.8757295845022814,
"eval_runtime": 2.8948,
"eval_samples_per_second": 1291.298,
"eval_steps_per_second": 80.836,
"step": 1869
},
{
"epoch": 1.0165864098448367,
"grad_norm": 5.979121685028076,
"learning_rate": 1.3222757267701089e-05,
"loss": 0.2547,
"step": 1900
},
{
"epoch": 1.043338683788122,
"grad_norm": 3.420891284942627,
"learning_rate": 1.3044408774745853e-05,
"loss": 0.2483,
"step": 1950
},
{
"epoch": 1.070090957731407,
"grad_norm": 6.0562520027160645,
"learning_rate": 1.286606028179062e-05,
"loss": 0.2334,
"step": 2000
},
{
"epoch": 1.0968432316746923,
"grad_norm": 5.684343338012695,
"learning_rate": 1.2687711788835386e-05,
"loss": 0.2165,
"step": 2050
},
{
"epoch": 1.1235955056179776,
"grad_norm": 12.342307090759277,
"learning_rate": 1.250936329588015e-05,
"loss": 0.2185,
"step": 2100
},
{
"epoch": 1.1503477795612627,
"grad_norm": 10.415166854858398,
"learning_rate": 1.2331014802924917e-05,
"loss": 0.2436,
"step": 2150
},
{
"epoch": 1.177100053504548,
"grad_norm": 24.330848693847656,
"learning_rate": 1.2152666309969681e-05,
"loss": 0.2218,
"step": 2200
},
{
"epoch": 1.203852327447833,
"grad_norm": 19.10972785949707,
"learning_rate": 1.1974317817014447e-05,
"loss": 0.2268,
"step": 2250
},
{
"epoch": 1.2306046013911183,
"grad_norm": 5.752458572387695,
"learning_rate": 1.1795969324059214e-05,
"loss": 0.1923,
"step": 2300
},
{
"epoch": 1.2573568753344033,
"grad_norm": 7.389070510864258,
"learning_rate": 1.1617620831103978e-05,
"loss": 0.2207,
"step": 2350
},
{
"epoch": 1.2841091492776886,
"grad_norm": 8.551554679870605,
"learning_rate": 1.1439272338148743e-05,
"loss": 0.2634,
"step": 2400
},
{
"epoch": 1.3108614232209739,
"grad_norm": 8.443785667419434,
"learning_rate": 1.126092384519351e-05,
"loss": 0.2116,
"step": 2450
},
{
"epoch": 1.337613697164259,
"grad_norm": 0.6092638969421387,
"learning_rate": 1.1082575352238275e-05,
"loss": 0.2033,
"step": 2500
},
{
"epoch": 1.3643659711075442,
"grad_norm": 2.732888698577881,
"learning_rate": 1.090422685928304e-05,
"loss": 0.2469,
"step": 2550
},
{
"epoch": 1.3911182450508293,
"grad_norm": 5.319647312164307,
"learning_rate": 1.0725878366327804e-05,
"loss": 0.2212,
"step": 2600
},
{
"epoch": 1.4178705189941145,
"grad_norm": 7.216152191162109,
"learning_rate": 1.0547529873372571e-05,
"loss": 0.2124,
"step": 2650
},
{
"epoch": 1.4446227929373996,
"grad_norm": 9.755507469177246,
"learning_rate": 1.0369181380417337e-05,
"loss": 0.2129,
"step": 2700
},
{
"epoch": 1.4713750668806849,
"grad_norm": 14.041426658630371,
"learning_rate": 1.01908328874621e-05,
"loss": 0.1971,
"step": 2750
},
{
"epoch": 1.4981273408239701,
"grad_norm": 14.255478858947754,
"learning_rate": 1.0012484394506868e-05,
"loss": 0.1675,
"step": 2800
},
{
"epoch": 1.5248796147672552,
"grad_norm": 2.1935946941375732,
"learning_rate": 9.834135901551632e-06,
"loss": 0.2042,
"step": 2850
},
{
"epoch": 1.5516318887105403,
"grad_norm": 8.990334510803223,
"learning_rate": 9.655787408596398e-06,
"loss": 0.226,
"step": 2900
},
{
"epoch": 1.5783841626538255,
"grad_norm": 2.0768239498138428,
"learning_rate": 9.477438915641163e-06,
"loss": 0.1659,
"step": 2950
},
{
"epoch": 1.6051364365971108,
"grad_norm": 9.156929016113281,
"learning_rate": 9.299090422685929e-06,
"loss": 0.2023,
"step": 3000
},
{
"epoch": 1.6318887105403959,
"grad_norm": 9.479352951049805,
"learning_rate": 9.120741929730694e-06,
"loss": 0.2452,
"step": 3050
},
{
"epoch": 1.6586409844836811,
"grad_norm": 10.488241195678711,
"learning_rate": 8.94239343677546e-06,
"loss": 0.1783,
"step": 3100
},
{
"epoch": 1.6853932584269664,
"grad_norm": 24.130321502685547,
"learning_rate": 8.764044943820226e-06,
"loss": 0.261,
"step": 3150
},
{
"epoch": 1.7121455323702515,
"grad_norm": 14.768345832824707,
"learning_rate": 8.585696450864991e-06,
"loss": 0.219,
"step": 3200
},
{
"epoch": 1.7388978063135365,
"grad_norm": 15.370089530944824,
"learning_rate": 8.407347957909757e-06,
"loss": 0.1993,
"step": 3250
},
{
"epoch": 1.7656500802568218,
"grad_norm": 3.223907232284546,
"learning_rate": 8.22899946495452e-06,
"loss": 0.1752,
"step": 3300
},
{
"epoch": 1.792402354200107,
"grad_norm": 11.258790016174316,
"learning_rate": 8.050650971999288e-06,
"loss": 0.1844,
"step": 3350
},
{
"epoch": 1.8191546281433921,
"grad_norm": 2.2796430587768555,
"learning_rate": 7.872302479044054e-06,
"loss": 0.1728,
"step": 3400
},
{
"epoch": 1.8459069020866774,
"grad_norm": 7.105353355407715,
"learning_rate": 7.693953986088817e-06,
"loss": 0.2508,
"step": 3450
},
{
"epoch": 1.8726591760299627,
"grad_norm": 3.3236451148986816,
"learning_rate": 7.515605493133583e-06,
"loss": 0.1902,
"step": 3500
},
{
"epoch": 1.8994114499732477,
"grad_norm": 17.00351905822754,
"learning_rate": 7.3372570001783495e-06,
"loss": 0.1954,
"step": 3550
},
{
"epoch": 1.9261637239165328,
"grad_norm": 5.1884331703186035,
"learning_rate": 7.158908507223114e-06,
"loss": 0.1975,
"step": 3600
},
{
"epoch": 1.952915997859818,
"grad_norm": 7.371060848236084,
"learning_rate": 6.98056001426788e-06,
"loss": 0.1979,
"step": 3650
},
{
"epoch": 1.9796682718031033,
"grad_norm": 14.774871826171875,
"learning_rate": 6.8022115213126446e-06,
"loss": 0.1868,
"step": 3700
},
{
"epoch": 2.0,
"eval_accuracy": 0.9018191546281434,
"eval_f1": 0.8926409675168582,
"eval_loss": 0.2982310950756073,
"eval_precision": 0.9016177612582297,
"eval_recall": 0.8848553089710288,
"eval_runtime": 3.0671,
"eval_samples_per_second": 1218.757,
"eval_steps_per_second": 76.295,
"step": 3738
},
{
"epoch": 2.0064205457463884,
"grad_norm": 2.9711053371429443,
"learning_rate": 6.623863028357411e-06,
"loss": 0.1811,
"step": 3750
},
{
"epoch": 2.0331728196896734,
"grad_norm": 11.470788955688477,
"learning_rate": 6.445514535402177e-06,
"loss": 0.1382,
"step": 3800
},
{
"epoch": 2.059925093632959,
"grad_norm": 3.6617465019226074,
"learning_rate": 6.267166042446941e-06,
"loss": 0.1374,
"step": 3850
},
{
"epoch": 2.086677367576244,
"grad_norm": 14.748554229736328,
"learning_rate": 6.088817549491707e-06,
"loss": 0.0932,
"step": 3900
},
{
"epoch": 2.113429641519529,
"grad_norm": 2.309537172317505,
"learning_rate": 5.910469056536473e-06,
"loss": 0.1017,
"step": 3950
},
{
"epoch": 2.140181915462814,
"grad_norm": 5.085292339324951,
"learning_rate": 5.732120563581238e-06,
"loss": 0.1598,
"step": 4000
},
{
"epoch": 2.1669341894060996,
"grad_norm": 15.087725639343262,
"learning_rate": 5.553772070626004e-06,
"loss": 0.1095,
"step": 4050
},
{
"epoch": 2.1936864633493847,
"grad_norm": 20.651540756225586,
"learning_rate": 5.3754235776707685e-06,
"loss": 0.1128,
"step": 4100
},
{
"epoch": 2.2204387372926697,
"grad_norm": 23.393356323242188,
"learning_rate": 5.197075084715535e-06,
"loss": 0.099,
"step": 4150
},
{
"epoch": 2.247191011235955,
"grad_norm": 30.76633644104004,
"learning_rate": 5.0187265917603005e-06,
"loss": 0.1015,
"step": 4200
},
{
"epoch": 2.2739432851792403,
"grad_norm": 10.410665512084961,
"learning_rate": 4.840378098805065e-06,
"loss": 0.1583,
"step": 4250
},
{
"epoch": 2.3006955591225253,
"grad_norm": 0.0475541315972805,
"learning_rate": 4.662029605849831e-06,
"loss": 0.118,
"step": 4300
},
{
"epoch": 2.3274478330658104,
"grad_norm": 0.0829029381275177,
"learning_rate": 4.4836811128945964e-06,
"loss": 0.1169,
"step": 4350
},
{
"epoch": 2.354200107009096,
"grad_norm": 25.78684425354004,
"learning_rate": 4.305332619939362e-06,
"loss": 0.0962,
"step": 4400
},
{
"epoch": 2.380952380952381,
"grad_norm": 1.0623866319656372,
"learning_rate": 4.126984126984127e-06,
"loss": 0.179,
"step": 4450
},
{
"epoch": 2.407704654895666,
"grad_norm": 8.90683650970459,
"learning_rate": 3.948635634028893e-06,
"loss": 0.1481,
"step": 4500
},
{
"epoch": 2.4344569288389515,
"grad_norm": 3.296332836151123,
"learning_rate": 3.770287141073658e-06,
"loss": 0.1109,
"step": 4550
},
{
"epoch": 2.4612092027822365,
"grad_norm": 11.87473201751709,
"learning_rate": 3.5919386481184236e-06,
"loss": 0.0927,
"step": 4600
},
{
"epoch": 2.4879614767255216,
"grad_norm": 13.711977005004883,
"learning_rate": 3.4135901551631887e-06,
"loss": 0.1389,
"step": 4650
},
{
"epoch": 2.5147137506688066,
"grad_norm": 48.17192840576172,
"learning_rate": 3.2352416622079547e-06,
"loss": 0.0996,
"step": 4700
},
{
"epoch": 2.541466024612092,
"grad_norm": 0.33165135979652405,
"learning_rate": 3.05689316925272e-06,
"loss": 0.1331,
"step": 4750
},
{
"epoch": 2.568218298555377,
"grad_norm": 0.20608340203762054,
"learning_rate": 2.8785446762974855e-06,
"loss": 0.1066,
"step": 4800
},
{
"epoch": 2.5949705724986623,
"grad_norm": 0.3617757260799408,
"learning_rate": 2.7001961833422507e-06,
"loss": 0.1585,
"step": 4850
},
{
"epoch": 2.6217228464419478,
"grad_norm": 5.1519670486450195,
"learning_rate": 2.5218476903870167e-06,
"loss": 0.1266,
"step": 4900
},
{
"epoch": 2.648475120385233,
"grad_norm": 35.215240478515625,
"learning_rate": 2.343499197431782e-06,
"loss": 0.1083,
"step": 4950
},
{
"epoch": 2.675227394328518,
"grad_norm": 10.100606918334961,
"learning_rate": 2.1651507044765475e-06,
"loss": 0.1342,
"step": 5000
},
{
"epoch": 2.701979668271803,
"grad_norm": 15.540501594543457,
"learning_rate": 1.986802211521313e-06,
"loss": 0.17,
"step": 5050
},
{
"epoch": 2.7287319422150884,
"grad_norm": 0.2651878297328949,
"learning_rate": 1.8084537185660782e-06,
"loss": 0.0879,
"step": 5100
},
{
"epoch": 2.7554842161583735,
"grad_norm": 0.8803261518478394,
"learning_rate": 1.6301052256108438e-06,
"loss": 0.0985,
"step": 5150
},
{
"epoch": 2.7822364901016585,
"grad_norm": 0.043911464512348175,
"learning_rate": 1.4517567326556092e-06,
"loss": 0.1241,
"step": 5200
},
{
"epoch": 2.808988764044944,
"grad_norm": 0.012229884043335915,
"learning_rate": 1.2734082397003748e-06,
"loss": 0.1167,
"step": 5250
},
{
"epoch": 2.835741037988229,
"grad_norm": 5.769114017486572,
"learning_rate": 1.0950597467451402e-06,
"loss": 0.1411,
"step": 5300
},
{
"epoch": 2.862493311931514,
"grad_norm": 14.51673698425293,
"learning_rate": 9.167112537899057e-07,
"loss": 0.085,
"step": 5350
},
{
"epoch": 2.889245585874799,
"grad_norm": 7.242597579956055,
"learning_rate": 7.38362760834671e-07,
"loss": 0.1215,
"step": 5400
},
{
"epoch": 2.9159978598180847,
"grad_norm": 3.783128023147583,
"learning_rate": 5.600142678794364e-07,
"loss": 0.0713,
"step": 5450
},
{
"epoch": 2.9427501337613697,
"grad_norm": 0.4435977637767792,
"learning_rate": 3.816657749242019e-07,
"loss": 0.1135,
"step": 5500
},
{
"epoch": 2.969502407704655,
"grad_norm": 22.0134220123291,
"learning_rate": 2.0331728196896737e-07,
"loss": 0.0976,
"step": 5550
},
{
"epoch": 2.9962546816479403,
"grad_norm": 0.05080915242433548,
"learning_rate": 2.4968789013732835e-08,
"loss": 0.1141,
"step": 5600
},
{
"epoch": 3.0,
"eval_accuracy": 0.9087747458533976,
"eval_f1": 0.901188417064033,
"eval_loss": 0.41310834884643555,
"eval_precision": 0.9053103806790227,
"eval_recall": 0.8972911214880508,
"eval_runtime": 3.2883,
"eval_samples_per_second": 1136.766,
"eval_steps_per_second": 71.162,
"step": 5607
}
],
"logging_steps": 50,
"max_steps": 5607,
"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": 922607713156626.0,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}