egotools-8b-v2 / checkpoint-500 /trainer_state.json
shulin16's picture
Upload checkpoint-500 metadata
47f3f5f verified
Raw
History Blame Contribute Delete
23.4 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.5329780146568954,
"eval_steps": 500.0,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0010659560293137908,
"grad_norm": 65.16799013627562,
"learning_rate": 1.7241379310344828e-07,
"loss": 1.7501227855682373,
"step": 1,
"token_acc": 0.5974304068522484
},
{
"epoch": 0.005329780146568954,
"grad_norm": 49.60785562728193,
"learning_rate": 8.620689655172415e-07,
"loss": 1.9290522336959839,
"step": 5,
"token_acc": 0.6
},
{
"epoch": 0.010659560293137908,
"grad_norm": 19.615574554346505,
"learning_rate": 1.724137931034483e-06,
"loss": 1.787750244140625,
"step": 10,
"token_acc": 0.6390234811777861
},
{
"epoch": 0.015989340439706862,
"grad_norm": 11.065195164896805,
"learning_rate": 2.5862068965517246e-06,
"loss": 1.4807558059692383,
"step": 15,
"token_acc": 0.6601166054847765
},
{
"epoch": 0.021319120586275817,
"grad_norm": 6.817900936636832,
"learning_rate": 3.448275862068966e-06,
"loss": 1.398179054260254,
"step": 20,
"token_acc": 0.6557256131292643
},
{
"epoch": 0.02664890073284477,
"grad_norm": 6.32783528737025,
"learning_rate": 4.310344827586207e-06,
"loss": 1.3023101806640625,
"step": 25,
"token_acc": 0.6763397175469142
},
{
"epoch": 0.031978680879413725,
"grad_norm": 6.367128597932351,
"learning_rate": 4.9999851020489695e-06,
"loss": 1.2105424880981446,
"step": 30,
"token_acc": 0.6806692125841062
},
{
"epoch": 0.037308461025982675,
"grad_norm": 5.745679317649865,
"learning_rate": 4.99946369240638e-06,
"loss": 1.0424203872680664,
"step": 35,
"token_acc": 0.720532319391635
},
{
"epoch": 0.04263824117255163,
"grad_norm": 5.84995920075286,
"learning_rate": 4.998197562761914e-06,
"loss": 1.0592132568359376,
"step": 40,
"token_acc": 0.7028319235012872
},
{
"epoch": 0.047968021319120584,
"grad_norm": 6.027890463744797,
"learning_rate": 4.9961870903613266e-06,
"loss": 0.9895943641662598,
"step": 45,
"token_acc": 0.7256235827664399
},
{
"epoch": 0.05329780146568954,
"grad_norm": 5.138993659750872,
"learning_rate": 4.9934328742287285e-06,
"loss": 1.0118780136108398,
"step": 50,
"token_acc": 0.7169505271695052
},
{
"epoch": 0.05862758161225849,
"grad_norm": 5.517701738242315,
"learning_rate": 4.989935734988098e-06,
"loss": 1.1720707893371582,
"step": 55,
"token_acc": 0.6896987191051777
},
{
"epoch": 0.06395736175882745,
"grad_norm": 6.246846876317293,
"learning_rate": 4.985696714618781e-06,
"loss": 1.0461352348327637,
"step": 60,
"token_acc": 0.7217303822937626
},
{
"epoch": 0.06928714190539641,
"grad_norm": 5.1838327118545795,
"learning_rate": 4.98071707614503e-06,
"loss": 1.0784454345703125,
"step": 65,
"token_acc": 0.7065761258041459
},
{
"epoch": 0.07461692205196535,
"grad_norm": 5.207457333274206,
"learning_rate": 4.9749983032596775e-06,
"loss": 1.0711910247802734,
"step": 70,
"token_acc": 0.7140947288006112
},
{
"epoch": 0.07994670219853431,
"grad_norm": 5.076819409823014,
"learning_rate": 4.968542099882077e-06,
"loss": 0.9525423049926758,
"step": 75,
"token_acc": 0.7458682120626744
},
{
"epoch": 0.08527648234510327,
"grad_norm": 5.145333113069056,
"learning_rate": 4.961350389650409e-06,
"loss": 1.0395957946777343,
"step": 80,
"token_acc": 0.7181391712275215
},
{
"epoch": 0.09060626249167222,
"grad_norm": 5.708204831944594,
"learning_rate": 4.953425315348534e-06,
"loss": 1.015221405029297,
"step": 85,
"token_acc": 0.7219031993437244
},
{
"epoch": 0.09593604263824117,
"grad_norm": 5.356494906382105,
"learning_rate": 4.944769238267542e-06,
"loss": 1.065409278869629,
"step": 90,
"token_acc": 0.7133068878470971
},
{
"epoch": 0.10126582278481013,
"grad_norm": 5.561086978061129,
"learning_rate": 4.935384737502204e-06,
"loss": 0.8904417037963868,
"step": 95,
"token_acc": 0.7561363636363636
},
{
"epoch": 0.10659560293137908,
"grad_norm": 5.007929386757231,
"learning_rate": 4.925274609182527e-06,
"loss": 0.9963092803955078,
"step": 100,
"token_acc": 0.7157873859579532
},
{
"epoch": 0.11192538307794804,
"grad_norm": 4.729475065378666,
"learning_rate": 4.914441865640636e-06,
"loss": 1.0125768661499024,
"step": 105,
"token_acc": 0.7199248120300752
},
{
"epoch": 0.11725516322451698,
"grad_norm": 5.352763340267538,
"learning_rate": 4.902889734513251e-06,
"loss": 0.977642822265625,
"step": 110,
"token_acc": 0.7219261890665087
},
{
"epoch": 0.12258494337108594,
"grad_norm": 5.01951337624599,
"learning_rate": 4.890621657780002e-06,
"loss": 0.9350076675415039,
"step": 115,
"token_acc": 0.7369072164948454
},
{
"epoch": 0.1279147235176549,
"grad_norm": 5.682732846054211,
"learning_rate": 4.8776412907378845e-06,
"loss": 0.9863363265991211,
"step": 120,
"token_acc": 0.7295660948536832
},
{
"epoch": 0.13324450366422386,
"grad_norm": 5.009947806421316,
"learning_rate": 4.86395250091216e-06,
"loss": 0.9168167114257812,
"step": 125,
"token_acc": 0.7459949409780776
},
{
"epoch": 0.13857428381079281,
"grad_norm": 5.16090001119418,
"learning_rate": 4.849559366904016e-06,
"loss": 0.9731523513793945,
"step": 130,
"token_acc": 0.7264408099688473
},
{
"epoch": 0.14390406395736177,
"grad_norm": 5.393302939264074,
"learning_rate": 4.83446617717534e-06,
"loss": 0.9265727996826172,
"step": 135,
"token_acc": 0.7382675211908207
},
{
"epoch": 0.1492338441039307,
"grad_norm": 5.1348529508995915,
"learning_rate": 4.818677428770965e-06,
"loss": 0.9645535469055175,
"step": 140,
"token_acc": 0.7365953109072375
},
{
"epoch": 0.15456362425049966,
"grad_norm": 5.274944058378017,
"learning_rate": 4.802197825978766e-06,
"loss": 0.9219735145568848,
"step": 145,
"token_acc": 0.7424147311152961
},
{
"epoch": 0.15989340439706862,
"grad_norm": 4.9265099733335225,
"learning_rate": 4.785032278928009e-06,
"loss": 0.9755061149597168,
"step": 150,
"token_acc": 0.7261881478584001
},
{
"epoch": 0.16522318454363757,
"grad_norm": 5.186158859969355,
"learning_rate": 4.7671859021263635e-06,
"loss": 0.8955172538757324,
"step": 155,
"token_acc": 0.7485970819304153
},
{
"epoch": 0.17055296469020653,
"grad_norm": 4.8553098725317,
"learning_rate": 4.748664012936035e-06,
"loss": 0.9992611885070801,
"step": 160,
"token_acc": 0.7210042432814711
},
{
"epoch": 0.1758827448367755,
"grad_norm": 5.607452928309844,
"learning_rate": 4.729472129989439e-06,
"loss": 0.8785256385803223,
"step": 165,
"token_acc": 0.7405417024935511
},
{
"epoch": 0.18121252498334445,
"grad_norm": 5.063309155216159,
"learning_rate": 4.709615971544912e-06,
"loss": 0.958376121520996,
"step": 170,
"token_acc": 0.7236842105263158
},
{
"epoch": 0.18654230512991338,
"grad_norm": 4.958090896706627,
"learning_rate": 4.689101453782954e-06,
"loss": 0.9511463165283203,
"step": 175,
"token_acc": 0.7302402347331745
},
{
"epoch": 0.19187208527648233,
"grad_norm": 5.554535034316316,
"learning_rate": 4.6679346890434864e-06,
"loss": 0.9487571716308594,
"step": 180,
"token_acc": 0.7357374546842206
},
{
"epoch": 0.1972018654230513,
"grad_norm": 5.0455487989596985,
"learning_rate": 4.646121984004666e-06,
"loss": 0.9357378005981445,
"step": 185,
"token_acc": 0.7466236645837533
},
{
"epoch": 0.20253164556962025,
"grad_norm": 5.102368295977931,
"learning_rate": 4.623669837803803e-06,
"loss": 0.9165685653686524,
"step": 190,
"token_acc": 0.7344439983942191
},
{
"epoch": 0.2078614257161892,
"grad_norm": 6.03242586164789,
"learning_rate": 4.600584940100929e-06,
"loss": 0.8400386810302735,
"step": 195,
"token_acc": 0.7588924387646432
},
{
"epoch": 0.21319120586275817,
"grad_norm": 6.046048176929514,
"learning_rate": 4.576874169085596e-06,
"loss": 1.0382144927978516,
"step": 200,
"token_acc": 0.7164823773324119
},
{
"epoch": 0.21852098600932712,
"grad_norm": 5.3128952327016306,
"learning_rate": 4.552544589427514e-06,
"loss": 0.9587564468383789,
"step": 205,
"token_acc": 0.7335751713018944
},
{
"epoch": 0.22385076615589608,
"grad_norm": 4.856245265355434,
"learning_rate": 4.527603450171614e-06,
"loss": 1.0106466293334961,
"step": 210,
"token_acc": 0.7285714285714285
},
{
"epoch": 0.229180546302465,
"grad_norm": 4.364039021197141,
"learning_rate": 4.502058182578181e-06,
"loss": 0.8945587158203125,
"step": 215,
"token_acc": 0.7414500683994528
},
{
"epoch": 0.23451032644903397,
"grad_norm": 5.794013669019053,
"learning_rate": 4.4759163979086985e-06,
"loss": 0.9250093460083008,
"step": 220,
"token_acc": 0.7438412408759124
},
{
"epoch": 0.23984010659560293,
"grad_norm": 4.9248548258612574,
"learning_rate": 4.449185885158056e-06,
"loss": 0.9222723007202148,
"step": 225,
"token_acc": 0.7404455344369507
},
{
"epoch": 0.24516988674217188,
"grad_norm": 5.088119921434173,
"learning_rate": 4.421874608733806e-06,
"loss": 0.927369213104248,
"step": 230,
"token_acc": 0.7411545623836127
},
{
"epoch": 0.2504996668887408,
"grad_norm": 5.261396967134003,
"learning_rate": 4.393990706083149e-06,
"loss": 0.9975498199462891,
"step": 235,
"token_acc": 0.7302452316076294
},
{
"epoch": 0.2558294470353098,
"grad_norm": 4.270065742658553,
"learning_rate": 4.3655424852683695e-06,
"loss": 0.9474044799804687,
"step": 240,
"token_acc": 0.7366733466933868
},
{
"epoch": 0.26115922718187873,
"grad_norm": 4.0368542213252745,
"learning_rate": 4.336538422491432e-06,
"loss": 0.9614317893981934,
"step": 245,
"token_acc": 0.7291335803776249
},
{
"epoch": 0.2664890073284477,
"grad_norm": 5.940590055777755,
"learning_rate": 4.3069871595684795e-06,
"loss": 0.9075183868408203,
"step": 250,
"token_acc": 0.7441300421432872
},
{
"epoch": 0.27181878747501664,
"grad_norm": 6.020932164981831,
"learning_rate": 4.276897501354987e-06,
"loss": 0.9187180519104003,
"step": 255,
"token_acc": 0.7450124688279302
},
{
"epoch": 0.27714856762158563,
"grad_norm": 5.007173976721824,
"learning_rate": 4.246278413122344e-06,
"loss": 0.9306771278381347,
"step": 260,
"token_acc": 0.7377112881290072
},
{
"epoch": 0.28247834776815456,
"grad_norm": 5.037382296205618,
"learning_rate": 4.21513901788662e-06,
"loss": 0.9223703384399414,
"step": 265,
"token_acc": 0.7390941348361512
},
{
"epoch": 0.28780812791472354,
"grad_norm": 4.579320525829714,
"learning_rate": 4.183488593690365e-06,
"loss": 0.8956701278686523,
"step": 270,
"token_acc": 0.7352545629202689
},
{
"epoch": 0.2931379080612925,
"grad_norm": 4.422357549054009,
"learning_rate": 4.151336570838182e-06,
"loss": 0.9767134666442872,
"step": 275,
"token_acc": 0.729864575908767
},
{
"epoch": 0.2984676882078614,
"grad_norm": 5.387070334048114,
"learning_rate": 4.118692529086956e-06,
"loss": 0.8453518867492675,
"step": 280,
"token_acc": 0.7639063153492263
},
{
"epoch": 0.3037974683544304,
"grad_norm": 5.218152944895714,
"learning_rate": 4.08556619479155e-06,
"loss": 0.9285115242004395,
"step": 285,
"token_acc": 0.7405669997960432
},
{
"epoch": 0.3091272485009993,
"grad_norm": 4.473635157614191,
"learning_rate": 4.051967438006812e-06,
"loss": 0.9505875587463379,
"step": 290,
"token_acc": 0.7344601962922573
},
{
"epoch": 0.3144570286475683,
"grad_norm": 4.630841152213125,
"learning_rate": 4.017906269546778e-06,
"loss": 0.9147586822509766,
"step": 295,
"token_acc": 0.7461659457333858
},
{
"epoch": 0.31978680879413723,
"grad_norm": 4.4381854725202805,
"learning_rate": 3.983392838001932e-06,
"loss": 0.8980813980102539,
"step": 300,
"token_acc": 0.7419038272816487
},
{
"epoch": 0.3251165889407062,
"grad_norm": 4.3925972764182735,
"learning_rate": 3.948437426715409e-06,
"loss": 0.9489370346069336,
"step": 305,
"token_acc": 0.7271723122238586
},
{
"epoch": 0.33044636908727515,
"grad_norm": 4.777819792705776,
"learning_rate": 3.913050450719057e-06,
"loss": 0.9575244903564453,
"step": 310,
"token_acc": 0.7353364519611371
},
{
"epoch": 0.3357761492338441,
"grad_norm": 4.690307974879278,
"learning_rate": 3.8772424536302565e-06,
"loss": 0.8590595245361328,
"step": 315,
"token_acc": 0.7531313131313131
},
{
"epoch": 0.34110592938041306,
"grad_norm": 5.548729734061873,
"learning_rate": 3.841024104510431e-06,
"loss": 0.78990797996521,
"step": 320,
"token_acc": 0.7674789341835573
},
{
"epoch": 0.346435709526982,
"grad_norm": 4.989871534387572,
"learning_rate": 3.8044061946861795e-06,
"loss": 0.9506110191345215,
"step": 325,
"token_acc": 0.7373412059402398
},
{
"epoch": 0.351765489673551,
"grad_norm": 4.676259945487118,
"learning_rate": 3.767399634533976e-06,
"loss": 0.9495532035827636,
"step": 330,
"token_acc": 0.7358214416392242
},
{
"epoch": 0.3570952698201199,
"grad_norm": 4.980936255741834,
"learning_rate": 3.7300154502294093e-06,
"loss": 0.9292638778686524,
"step": 335,
"token_acc": 0.7407978415879746
},
{
"epoch": 0.3624250499666889,
"grad_norm": 4.965729215770431,
"learning_rate": 3.6922647804619073e-06,
"loss": 0.9855228424072265,
"step": 340,
"token_acc": 0.7225995961079493
},
{
"epoch": 0.3677548301132578,
"grad_norm": 4.936075061330924,
"learning_rate": 3.654158873115944e-06,
"loss": 0.9076862335205078,
"step": 345,
"token_acc": 0.7436141822340832
},
{
"epoch": 0.37308461025982675,
"grad_norm": 5.231662158108306,
"learning_rate": 3.61570908191971e-06,
"loss": 0.8928030014038086,
"step": 350,
"token_acc": 0.7541051756391602
},
{
"epoch": 0.37841439040639574,
"grad_norm": 4.522556964501456,
"learning_rate": 3.5769268630622533e-06,
"loss": 0.9126054763793945,
"step": 355,
"token_acc": 0.7474881516587678
},
{
"epoch": 0.38374417055296467,
"grad_norm": 5.1995597965754685,
"learning_rate": 3.537823771780077e-06,
"loss": 0.8765750885009765,
"step": 360,
"token_acc": 0.7457895779671092
},
{
"epoch": 0.38907395069953365,
"grad_norm": 5.281242833316695,
"learning_rate": 3.4984114589142388e-06,
"loss": 0.9568130493164062,
"step": 365,
"token_acc": 0.737781383086595
},
{
"epoch": 0.3944037308461026,
"grad_norm": 4.5694389026717435,
"learning_rate": 3.458701667438965e-06,
"loss": 0.8665159225463868,
"step": 370,
"token_acc": 0.7548999797938978
},
{
"epoch": 0.39973351099267157,
"grad_norm": 4.332384453495391,
"learning_rate": 3.4187062289627983e-06,
"loss": 0.882652759552002,
"step": 375,
"token_acc": 0.7516012810248198
},
{
"epoch": 0.4050632911392405,
"grad_norm": 5.463888132443461,
"learning_rate": 3.3784370602033572e-06,
"loss": 0.7812636852264404,
"step": 380,
"token_acc": 0.769391062746006
},
{
"epoch": 0.4103930712858095,
"grad_norm": 4.6396298492906505,
"learning_rate": 3.337906159436721e-06,
"loss": 0.891010856628418,
"step": 385,
"token_acc": 0.752018908804412
},
{
"epoch": 0.4157228514323784,
"grad_norm": 5.11189175613832,
"learning_rate": 3.2971256029225253e-06,
"loss": 0.8738397598266602,
"step": 390,
"token_acc": 0.7532467532467533
},
{
"epoch": 0.42105263157894735,
"grad_norm": 4.9436874254737075,
"learning_rate": 3.2561075413058113e-06,
"loss": 0.8521583557128907,
"step": 395,
"token_acc": 0.7503179313268334
},
{
"epoch": 0.42638241172551633,
"grad_norm": 4.666360407842893,
"learning_rate": 3.214864195996723e-06,
"loss": 0.9074520111083985,
"step": 400,
"token_acc": 0.7434575733544806
},
{
"epoch": 0.43171219187208526,
"grad_norm": 4.750112035885002,
"learning_rate": 3.1734078555291137e-06,
"loss": 0.9123194694519043,
"step": 405,
"token_acc": 0.7403057419835943
},
{
"epoch": 0.43704197201865425,
"grad_norm": 4.771140489181497,
"learning_rate": 3.1317508718991496e-06,
"loss": 0.8552797317504883,
"step": 410,
"token_acc": 0.7574031890660592
},
{
"epoch": 0.4423717521652232,
"grad_norm": 4.692545746773149,
"learning_rate": 3.0899056568850148e-06,
"loss": 0.9358634948730469,
"step": 415,
"token_acc": 0.7455585052072697
},
{
"epoch": 0.44770153231179216,
"grad_norm": 5.0036012753115235,
"learning_rate": 3.047884678348795e-06,
"loss": 0.9225390434265137,
"step": 420,
"token_acc": 0.739247795573151
},
{
"epoch": 0.4530313124583611,
"grad_norm": 4.741320422434045,
"learning_rate": 3.0057004565216523e-06,
"loss": 0.8901847839355469,
"step": 425,
"token_acc": 0.7495299770210988
},
{
"epoch": 0.45836109260493,
"grad_norm": 4.703918003843587,
"learning_rate": 2.9633655602734035e-06,
"loss": 0.9226030349731446,
"step": 430,
"token_acc": 0.7433257055682685
},
{
"epoch": 0.463690872751499,
"grad_norm": 5.477224800396244,
"learning_rate": 2.920892603367596e-06,
"loss": 0.8314413070678711,
"step": 435,
"token_acc": 0.7559628975265018
},
{
"epoch": 0.46902065289806794,
"grad_norm": 4.8329386010913264,
"learning_rate": 2.8782942407032177e-06,
"loss": 0.8740703582763671,
"step": 440,
"token_acc": 0.7501027538018906
},
{
"epoch": 0.4743504330446369,
"grad_norm": 5.087948549401884,
"learning_rate": 2.835583164544139e-06,
"loss": 0.8511223793029785,
"step": 445,
"token_acc": 0.7644836272040302
},
{
"epoch": 0.47968021319120585,
"grad_norm": 4.99727487177206,
"learning_rate": 2.7927721007374347e-06,
"loss": 0.815243911743164,
"step": 450,
"token_acc": 0.766592647421663
},
{
"epoch": 0.48500999333777484,
"grad_norm": 5.576288110570715,
"learning_rate": 2.7498738049216904e-06,
"loss": 0.9731996536254883,
"step": 455,
"token_acc": 0.7280547752808989
},
{
"epoch": 0.49033977348434377,
"grad_norm": 4.652800298925416,
"learning_rate": 2.706901058726437e-06,
"loss": 0.8669706344604492,
"step": 460,
"token_acc": 0.7513238289205703
},
{
"epoch": 0.49566955363091275,
"grad_norm": 4.67527794119894,
"learning_rate": 2.6638666659638372e-06,
"loss": 0.8978883743286132,
"step": 465,
"token_acc": 0.7504095004095004
},
{
"epoch": 0.5009993337774816,
"grad_norm": 4.500058654431643,
"learning_rate": 2.620783448813768e-06,
"loss": 0.803309440612793,
"step": 470,
"token_acc": 0.7682926829268293
},
{
"epoch": 0.5063291139240507,
"grad_norm": 5.549718692193634,
"learning_rate": 2.577664244003426e-06,
"loss": 0.8737884521484375,
"step": 475,
"token_acc": 0.7512740101920815
},
{
"epoch": 0.5116588940706196,
"grad_norm": 4.633775513369992,
"learning_rate": 2.5345218989825985e-06,
"loss": 0.8945508003234863,
"step": 480,
"token_acc": 0.7538946953263657
},
{
"epoch": 0.5169886742171885,
"grad_norm": 4.457350960855039,
"learning_rate": 2.4913692680957472e-06,
"loss": 0.8398324012756347,
"step": 485,
"token_acc": 0.763268931823229
},
{
"epoch": 0.5223184543637575,
"grad_norm": 4.9918603168538365,
"learning_rate": 2.4482192087520225e-06,
"loss": 0.8864654541015625,
"step": 490,
"token_acc": 0.7518545994065282
},
{
"epoch": 0.5276482345103265,
"grad_norm": 4.279486984922963,
"learning_rate": 2.4050845775943816e-06,
"loss": 0.951543140411377,
"step": 495,
"token_acc": 0.7330591811192562
},
{
"epoch": 0.5329780146568954,
"grad_norm": 4.9236086215774115,
"learning_rate": 2.361978226668922e-06,
"loss": 0.9079221725463867,
"step": 500,
"token_acc": 0.7499020759890325
}
],
"logging_steps": 5,
"max_steps": 939,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 656384513933312.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}