distilvert-complaints-subproduct / trainer_state.json
Kayvane's picture
Upload trainer_state.json
59b3f16
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9895366218236172,
"global_step": 64000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 2.5e-05,
"loss": 3.2479,
"step": 250
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 2.0411,
"step": 500
},
{
"epoch": 0.04,
"learning_rate": 4.980343439426343e-05,
"loss": 1.7301,
"step": 750
},
{
"epoch": 0.05,
"learning_rate": 4.9606868788526864e-05,
"loss": 1.5519,
"step": 1000
},
{
"epoch": 0.06,
"learning_rate": 4.941030318279029e-05,
"loss": 1.4656,
"step": 1250
},
{
"epoch": 0.07,
"learning_rate": 4.921373757705372e-05,
"loss": 1.4327,
"step": 1500
},
{
"epoch": 0.08,
"learning_rate": 4.901717197131715e-05,
"loss": 1.3757,
"step": 1750
},
{
"epoch": 0.09,
"learning_rate": 4.882060636558058e-05,
"loss": 1.3115,
"step": 2000
},
{
"epoch": 0.11,
"learning_rate": 4.862404075984401e-05,
"loss": 1.313,
"step": 2250
},
{
"epoch": 0.12,
"learning_rate": 4.8427475154107435e-05,
"loss": 1.255,
"step": 2500
},
{
"epoch": 0.13,
"learning_rate": 4.823090954837087e-05,
"loss": 1.2747,
"step": 2750
},
{
"epoch": 0.14,
"learning_rate": 4.8034343942634296e-05,
"loss": 1.2872,
"step": 3000
},
{
"epoch": 0.15,
"learning_rate": 4.7837778336897723e-05,
"loss": 1.1974,
"step": 3250
},
{
"epoch": 0.16,
"learning_rate": 4.764121273116116e-05,
"loss": 1.184,
"step": 3500
},
{
"epoch": 0.18,
"learning_rate": 4.7444647125424585e-05,
"loss": 1.1676,
"step": 3750
},
{
"epoch": 0.19,
"learning_rate": 4.724808151968801e-05,
"loss": 1.1646,
"step": 4000
},
{
"epoch": 0.2,
"learning_rate": 4.7051515913951446e-05,
"loss": 1.1572,
"step": 4250
},
{
"epoch": 0.21,
"learning_rate": 4.6854950308214874e-05,
"loss": 1.1324,
"step": 4500
},
{
"epoch": 0.22,
"learning_rate": 4.66583847024783e-05,
"loss": 1.1199,
"step": 4750
},
{
"epoch": 0.23,
"learning_rate": 4.6461819096741735e-05,
"loss": 1.1181,
"step": 5000
},
{
"epoch": 0.25,
"learning_rate": 4.626525349100516e-05,
"loss": 1.0963,
"step": 5250
},
{
"epoch": 0.26,
"learning_rate": 4.606868788526859e-05,
"loss": 1.0774,
"step": 5500
},
{
"epoch": 0.27,
"learning_rate": 4.587212227953202e-05,
"loss": 1.0936,
"step": 5750
},
{
"epoch": 0.28,
"learning_rate": 4.567555667379545e-05,
"loss": 1.0912,
"step": 6000
},
{
"epoch": 0.29,
"learning_rate": 4.547899106805888e-05,
"loss": 1.0929,
"step": 6250
},
{
"epoch": 0.3,
"learning_rate": 4.5282425462322306e-05,
"loss": 1.081,
"step": 6500
},
{
"epoch": 0.32,
"learning_rate": 4.508585985658574e-05,
"loss": 1.0755,
"step": 6750
},
{
"epoch": 0.33,
"learning_rate": 4.488929425084917e-05,
"loss": 1.0785,
"step": 7000
},
{
"epoch": 0.34,
"learning_rate": 4.4692728645112595e-05,
"loss": 1.0757,
"step": 7250
},
{
"epoch": 0.35,
"learning_rate": 4.449616303937603e-05,
"loss": 1.0703,
"step": 7500
},
{
"epoch": 0.36,
"learning_rate": 4.4299597433639456e-05,
"loss": 1.0694,
"step": 7750
},
{
"epoch": 0.37,
"learning_rate": 4.4103031827902884e-05,
"loss": 1.0665,
"step": 8000
},
{
"epoch": 0.39,
"learning_rate": 4.390646622216632e-05,
"loss": 1.0505,
"step": 8250
},
{
"epoch": 0.4,
"learning_rate": 4.3709900616429745e-05,
"loss": 1.0698,
"step": 8500
},
{
"epoch": 0.41,
"learning_rate": 4.351333501069317e-05,
"loss": 1.0681,
"step": 8750
},
{
"epoch": 0.42,
"learning_rate": 4.3316769404956606e-05,
"loss": 1.0433,
"step": 9000
},
{
"epoch": 0.43,
"learning_rate": 4.3120203799220034e-05,
"loss": 1.0236,
"step": 9250
},
{
"epoch": 0.44,
"learning_rate": 4.2923638193483454e-05,
"loss": 1.0457,
"step": 9500
},
{
"epoch": 0.46,
"learning_rate": 4.272707258774689e-05,
"loss": 1.04,
"step": 9750
},
{
"epoch": 0.47,
"learning_rate": 4.2530506982010316e-05,
"loss": 1.0164,
"step": 10000
},
{
"epoch": 0.48,
"learning_rate": 4.233394137627374e-05,
"loss": 1.0367,
"step": 10250
},
{
"epoch": 0.49,
"learning_rate": 4.213737577053718e-05,
"loss": 1.0207,
"step": 10500
},
{
"epoch": 0.5,
"learning_rate": 4.1940810164800605e-05,
"loss": 1.0309,
"step": 10750
},
{
"epoch": 0.51,
"learning_rate": 4.174424455906403e-05,
"loss": 1.0365,
"step": 11000
},
{
"epoch": 0.53,
"learning_rate": 4.154767895332746e-05,
"loss": 1.0386,
"step": 11250
},
{
"epoch": 0.54,
"learning_rate": 4.135111334759089e-05,
"loss": 1.0247,
"step": 11500
},
{
"epoch": 0.55,
"learning_rate": 4.115454774185432e-05,
"loss": 0.9685,
"step": 11750
},
{
"epoch": 0.56,
"learning_rate": 4.095798213611775e-05,
"loss": 1.0142,
"step": 12000
},
{
"epoch": 0.57,
"learning_rate": 4.076141653038118e-05,
"loss": 1.0273,
"step": 12250
},
{
"epoch": 0.59,
"learning_rate": 4.056485092464461e-05,
"loss": 1.0128,
"step": 12500
},
{
"epoch": 0.6,
"learning_rate": 4.036828531890804e-05,
"loss": 1.0145,
"step": 12750
},
{
"epoch": 0.61,
"learning_rate": 4.017171971317147e-05,
"loss": 1.018,
"step": 13000
},
{
"epoch": 0.62,
"learning_rate": 3.99751541074349e-05,
"loss": 1.0324,
"step": 13250
},
{
"epoch": 0.63,
"learning_rate": 3.9778588501698326e-05,
"loss": 1.0097,
"step": 13500
},
{
"epoch": 0.64,
"learning_rate": 3.958202289596176e-05,
"loss": 0.9895,
"step": 13750
},
{
"epoch": 0.66,
"learning_rate": 3.938545729022519e-05,
"loss": 1.008,
"step": 14000
},
{
"epoch": 0.67,
"learning_rate": 3.9188891684488614e-05,
"loss": 0.9985,
"step": 14250
},
{
"epoch": 0.68,
"learning_rate": 3.899232607875204e-05,
"loss": 1.0137,
"step": 14500
},
{
"epoch": 0.69,
"learning_rate": 3.8795760473015476e-05,
"loss": 1.0099,
"step": 14750
},
{
"epoch": 0.7,
"learning_rate": 3.85991948672789e-05,
"loss": 0.9924,
"step": 15000
},
{
"epoch": 0.71,
"learning_rate": 3.840262926154233e-05,
"loss": 0.9619,
"step": 15250
},
{
"epoch": 0.73,
"learning_rate": 3.8206063655805765e-05,
"loss": 0.9836,
"step": 15500
},
{
"epoch": 0.74,
"learning_rate": 3.800949805006919e-05,
"loss": 1.0207,
"step": 15750
},
{
"epoch": 0.75,
"learning_rate": 3.781293244433262e-05,
"loss": 0.9656,
"step": 16000
},
{
"epoch": 0.76,
"learning_rate": 3.7616366838596053e-05,
"loss": 0.9642,
"step": 16250
},
{
"epoch": 0.77,
"learning_rate": 3.741980123285948e-05,
"loss": 0.9653,
"step": 16500
},
{
"epoch": 0.78,
"learning_rate": 3.722323562712291e-05,
"loss": 0.9866,
"step": 16750
},
{
"epoch": 0.8,
"learning_rate": 3.702667002138634e-05,
"loss": 0.9737,
"step": 17000
},
{
"epoch": 0.81,
"learning_rate": 3.683010441564977e-05,
"loss": 0.979,
"step": 17250
},
{
"epoch": 0.82,
"learning_rate": 3.66335388099132e-05,
"loss": 1.0268,
"step": 17500
},
{
"epoch": 0.83,
"learning_rate": 3.643697320417663e-05,
"loss": 0.9858,
"step": 17750
},
{
"epoch": 0.84,
"learning_rate": 3.624040759844006e-05,
"loss": 0.9851,
"step": 18000
},
{
"epoch": 0.85,
"learning_rate": 3.6043841992703486e-05,
"loss": 0.9712,
"step": 18250
},
{
"epoch": 0.87,
"learning_rate": 3.584727638696691e-05,
"loss": 0.9768,
"step": 18500
},
{
"epoch": 0.88,
"learning_rate": 3.565071078123035e-05,
"loss": 0.9599,
"step": 18750
},
{
"epoch": 0.89,
"learning_rate": 3.5454145175493774e-05,
"loss": 0.9715,
"step": 19000
},
{
"epoch": 0.9,
"learning_rate": 3.52575795697572e-05,
"loss": 0.9628,
"step": 19250
},
{
"epoch": 0.91,
"learning_rate": 3.5061013964020636e-05,
"loss": 0.9597,
"step": 19500
},
{
"epoch": 0.92,
"learning_rate": 3.486444835828406e-05,
"loss": 0.9775,
"step": 19750
},
{
"epoch": 0.94,
"learning_rate": 3.466788275254749e-05,
"loss": 0.981,
"step": 20000
},
{
"epoch": 0.95,
"learning_rate": 3.4471317146810925e-05,
"loss": 0.9633,
"step": 20250
},
{
"epoch": 0.96,
"learning_rate": 3.427475154107435e-05,
"loss": 0.954,
"step": 20500
},
{
"epoch": 0.97,
"learning_rate": 3.407818593533778e-05,
"loss": 0.9485,
"step": 20750
},
{
"epoch": 0.98,
"learning_rate": 3.3881620329601213e-05,
"loss": 0.9707,
"step": 21000
},
{
"epoch": 0.99,
"learning_rate": 3.368505472386464e-05,
"loss": 0.9493,
"step": 21250
},
{
"epoch": 1.01,
"learning_rate": 3.348848911812807e-05,
"loss": 0.9319,
"step": 21500
},
{
"epoch": 1.02,
"learning_rate": 3.3291923512391496e-05,
"loss": 0.8635,
"step": 21750
},
{
"epoch": 1.03,
"learning_rate": 3.309535790665493e-05,
"loss": 0.8813,
"step": 22000
},
{
"epoch": 1.04,
"learning_rate": 3.289879230091836e-05,
"loss": 0.8519,
"step": 22250
},
{
"epoch": 1.05,
"learning_rate": 3.2702226695181784e-05,
"loss": 0.8565,
"step": 22500
},
{
"epoch": 1.06,
"learning_rate": 3.250566108944522e-05,
"loss": 0.8879,
"step": 22750
},
{
"epoch": 1.08,
"learning_rate": 3.2309095483708646e-05,
"loss": 0.8433,
"step": 23000
},
{
"epoch": 1.09,
"learning_rate": 3.211252987797207e-05,
"loss": 0.8814,
"step": 23250
},
{
"epoch": 1.1,
"learning_rate": 3.191596427223551e-05,
"loss": 0.8629,
"step": 23500
},
{
"epoch": 1.11,
"learning_rate": 3.1719398666498935e-05,
"loss": 0.912,
"step": 23750
},
{
"epoch": 1.12,
"learning_rate": 3.152283306076236e-05,
"loss": 0.8859,
"step": 24000
},
{
"epoch": 1.14,
"learning_rate": 3.1326267455025796e-05,
"loss": 0.8805,
"step": 24250
},
{
"epoch": 1.15,
"learning_rate": 3.112970184928922e-05,
"loss": 0.8974,
"step": 24500
},
{
"epoch": 1.16,
"learning_rate": 3.093313624355265e-05,
"loss": 0.8727,
"step": 24750
},
{
"epoch": 1.17,
"learning_rate": 3.073657063781608e-05,
"loss": 0.8894,
"step": 25000
},
{
"epoch": 1.18,
"learning_rate": 3.054000503207951e-05,
"loss": 0.8791,
"step": 25250
},
{
"epoch": 1.19,
"learning_rate": 3.034343942634294e-05,
"loss": 0.8723,
"step": 25500
},
{
"epoch": 1.21,
"learning_rate": 3.014687382060637e-05,
"loss": 0.8698,
"step": 25750
},
{
"epoch": 1.22,
"learning_rate": 2.9950308214869798e-05,
"loss": 0.8648,
"step": 26000
},
{
"epoch": 1.23,
"learning_rate": 2.9753742609133228e-05,
"loss": 0.8641,
"step": 26250
},
{
"epoch": 1.24,
"learning_rate": 2.955717700339666e-05,
"loss": 0.8708,
"step": 26500
},
{
"epoch": 1.25,
"learning_rate": 2.9360611397660086e-05,
"loss": 0.8715,
"step": 26750
},
{
"epoch": 1.26,
"learning_rate": 2.916404579192351e-05,
"loss": 0.8735,
"step": 27000
},
{
"epoch": 1.28,
"learning_rate": 2.896748018618694e-05,
"loss": 0.842,
"step": 27250
},
{
"epoch": 1.29,
"learning_rate": 2.877091458045037e-05,
"loss": 0.8573,
"step": 27500
},
{
"epoch": 1.3,
"learning_rate": 2.85743489747138e-05,
"loss": 0.879,
"step": 27750
},
{
"epoch": 1.31,
"learning_rate": 2.837778336897723e-05,
"loss": 0.848,
"step": 28000
},
{
"epoch": 1.32,
"learning_rate": 2.8181217763240657e-05,
"loss": 0.851,
"step": 28250
},
{
"epoch": 1.33,
"learning_rate": 2.7984652157504088e-05,
"loss": 0.8683,
"step": 28500
},
{
"epoch": 1.35,
"learning_rate": 2.778808655176752e-05,
"loss": 0.8472,
"step": 28750
},
{
"epoch": 1.36,
"learning_rate": 2.7591520946030946e-05,
"loss": 0.8723,
"step": 29000
},
{
"epoch": 1.37,
"learning_rate": 2.7394955340294377e-05,
"loss": 0.8873,
"step": 29250
},
{
"epoch": 1.38,
"learning_rate": 2.7198389734557804e-05,
"loss": 0.8726,
"step": 29500
},
{
"epoch": 1.39,
"learning_rate": 2.7001824128821235e-05,
"loss": 0.8662,
"step": 29750
},
{
"epoch": 1.4,
"learning_rate": 2.6805258523084665e-05,
"loss": 0.8773,
"step": 30000
},
{
"epoch": 1.42,
"learning_rate": 2.6608692917348093e-05,
"loss": 0.8719,
"step": 30250
},
{
"epoch": 1.43,
"learning_rate": 2.6412127311611523e-05,
"loss": 0.871,
"step": 30500
},
{
"epoch": 1.44,
"learning_rate": 2.6215561705874954e-05,
"loss": 0.8433,
"step": 30750
},
{
"epoch": 1.45,
"learning_rate": 2.601899610013838e-05,
"loss": 0.8528,
"step": 31000
},
{
"epoch": 1.46,
"learning_rate": 2.5822430494401812e-05,
"loss": 0.8892,
"step": 31250
},
{
"epoch": 1.47,
"learning_rate": 2.562586488866524e-05,
"loss": 0.8758,
"step": 31500
},
{
"epoch": 1.49,
"learning_rate": 2.542929928292867e-05,
"loss": 0.869,
"step": 31750
},
{
"epoch": 1.5,
"learning_rate": 2.52327336771921e-05,
"loss": 0.8661,
"step": 32000
},
{
"epoch": 1.51,
"learning_rate": 2.503616807145553e-05,
"loss": 0.8635,
"step": 32250
},
{
"epoch": 1.52,
"learning_rate": 2.483960246571896e-05,
"loss": 0.8431,
"step": 32500
},
{
"epoch": 1.53,
"learning_rate": 2.4643036859982386e-05,
"loss": 0.8526,
"step": 32750
},
{
"epoch": 1.54,
"learning_rate": 2.4446471254245817e-05,
"loss": 0.8702,
"step": 33000
},
{
"epoch": 1.56,
"learning_rate": 2.4249905648509248e-05,
"loss": 0.8743,
"step": 33250
},
{
"epoch": 1.57,
"learning_rate": 2.4053340042772675e-05,
"loss": 0.8704,
"step": 33500
},
{
"epoch": 1.58,
"learning_rate": 2.3856774437036106e-05,
"loss": 0.8458,
"step": 33750
},
{
"epoch": 1.59,
"learning_rate": 2.3660208831299537e-05,
"loss": 0.8745,
"step": 34000
},
{
"epoch": 1.6,
"learning_rate": 2.3463643225562964e-05,
"loss": 0.8626,
"step": 34250
},
{
"epoch": 1.61,
"learning_rate": 2.3267077619826395e-05,
"loss": 0.8421,
"step": 34500
},
{
"epoch": 1.63,
"learning_rate": 2.3070512014089822e-05,
"loss": 0.8734,
"step": 34750
},
{
"epoch": 1.64,
"learning_rate": 2.2873946408353253e-05,
"loss": 0.8465,
"step": 35000
},
{
"epoch": 1.65,
"learning_rate": 2.2677380802616684e-05,
"loss": 0.8516,
"step": 35250
},
{
"epoch": 1.66,
"learning_rate": 2.248081519688011e-05,
"loss": 0.8547,
"step": 35500
},
{
"epoch": 1.67,
"learning_rate": 2.228424959114354e-05,
"loss": 0.8833,
"step": 35750
},
{
"epoch": 1.69,
"learning_rate": 2.208768398540697e-05,
"loss": 0.8454,
"step": 36000
},
{
"epoch": 1.7,
"learning_rate": 2.18911183796704e-05,
"loss": 0.843,
"step": 36250
},
{
"epoch": 1.71,
"learning_rate": 2.169455277393383e-05,
"loss": 0.836,
"step": 36500
},
{
"epoch": 1.72,
"learning_rate": 2.1497987168197258e-05,
"loss": 0.8536,
"step": 36750
},
{
"epoch": 1.73,
"learning_rate": 2.130142156246069e-05,
"loss": 0.8717,
"step": 37000
},
{
"epoch": 1.74,
"learning_rate": 2.110485595672412e-05,
"loss": 0.8509,
"step": 37250
},
{
"epoch": 1.76,
"learning_rate": 2.0908290350987547e-05,
"loss": 0.8619,
"step": 37500
},
{
"epoch": 1.76,
"learning_rate": 2.0772393446739063e-05,
"loss": 0.8195,
"step": 37750
},
{
"epoch": 1.78,
"learning_rate": 2.0576235013495702e-05,
"loss": 0.7878,
"step": 38000
},
{
"epoch": 1.79,
"learning_rate": 2.0380076580252337e-05,
"loss": 0.7839,
"step": 38250
},
{
"epoch": 1.8,
"learning_rate": 2.0183918147008976e-05,
"loss": 0.7834,
"step": 38500
},
{
"epoch": 1.81,
"learning_rate": 1.9987759713765618e-05,
"loss": 0.787,
"step": 38750
},
{
"epoch": 1.82,
"learning_rate": 1.9791601280522253e-05,
"loss": 0.8008,
"step": 39000
},
{
"epoch": 1.83,
"learning_rate": 1.9595442847278892e-05,
"loss": 0.7967,
"step": 39250
},
{
"epoch": 1.85,
"learning_rate": 1.9399284414035527e-05,
"loss": 0.797,
"step": 39500
},
{
"epoch": 1.86,
"learning_rate": 1.9203125980792166e-05,
"loss": 0.7859,
"step": 39750
},
{
"epoch": 1.87,
"learning_rate": 1.9006967547548805e-05,
"loss": 0.807,
"step": 40000
},
{
"epoch": 1.88,
"learning_rate": 1.8810809114305444e-05,
"loss": 0.8051,
"step": 40250
},
{
"epoch": 1.89,
"learning_rate": 1.8614650681062082e-05,
"loss": 0.8176,
"step": 40500
},
{
"epoch": 1.9,
"learning_rate": 1.8418492247818718e-05,
"loss": 0.7946,
"step": 40750
},
{
"epoch": 1.92,
"learning_rate": 1.8222333814575357e-05,
"loss": 0.8024,
"step": 41000
},
{
"epoch": 1.93,
"learning_rate": 1.8026175381331995e-05,
"loss": 0.805,
"step": 41250
},
{
"epoch": 1.94,
"learning_rate": 1.783001694808863e-05,
"loss": 0.8075,
"step": 41500
},
{
"epoch": 1.95,
"learning_rate": 1.7633858514845273e-05,
"loss": 0.8016,
"step": 41750
},
{
"epoch": 1.96,
"learning_rate": 1.743770008160191e-05,
"loss": 0.7898,
"step": 42000
},
{
"epoch": 1.97,
"learning_rate": 1.7241541648358547e-05,
"loss": 0.808,
"step": 42250
},
{
"epoch": 1.99,
"learning_rate": 1.7045383215115186e-05,
"loss": 0.7606,
"step": 42500
},
{
"epoch": 2.0,
"learning_rate": 1.684922478187182e-05,
"loss": 0.7935,
"step": 42750
},
{
"epoch": 2.01,
"learning_rate": 1.665306634862846e-05,
"loss": 0.7706,
"step": 43000
},
{
"epoch": 2.02,
"learning_rate": 1.6456907915385102e-05,
"loss": 0.7628,
"step": 43250
},
{
"epoch": 2.03,
"learning_rate": 1.6260749482141737e-05,
"loss": 0.7503,
"step": 43500
},
{
"epoch": 2.04,
"learning_rate": 1.6064591048898376e-05,
"loss": 0.7384,
"step": 43750
},
{
"epoch": 2.06,
"learning_rate": 1.586843261565501e-05,
"loss": 0.7462,
"step": 44000
},
{
"epoch": 2.07,
"learning_rate": 1.567227418241165e-05,
"loss": 0.759,
"step": 44250
},
{
"epoch": 2.08,
"learning_rate": 1.547611574916829e-05,
"loss": 0.7626,
"step": 44500
},
{
"epoch": 2.09,
"learning_rate": 1.5279957315924927e-05,
"loss": 0.7602,
"step": 44750
},
{
"epoch": 2.1,
"learning_rate": 1.5083798882681566e-05,
"loss": 0.7631,
"step": 45000
},
{
"epoch": 2.11,
"learning_rate": 1.4887640449438203e-05,
"loss": 0.7511,
"step": 45250
},
{
"epoch": 2.13,
"learning_rate": 1.469148201619484e-05,
"loss": 0.7598,
"step": 45500
},
{
"epoch": 2.14,
"learning_rate": 1.4495323582951479e-05,
"loss": 0.7627,
"step": 45750
},
{
"epoch": 2.15,
"learning_rate": 1.4299165149708116e-05,
"loss": 0.7592,
"step": 46000
},
{
"epoch": 2.16,
"learning_rate": 1.4103006716464757e-05,
"loss": 0.7565,
"step": 46250
},
{
"epoch": 2.17,
"learning_rate": 1.3906848283221394e-05,
"loss": 0.7693,
"step": 46500
},
{
"epoch": 2.18,
"learning_rate": 1.371068984997803e-05,
"loss": 0.7757,
"step": 46750
},
{
"epoch": 2.2,
"learning_rate": 1.351453141673467e-05,
"loss": 0.7808,
"step": 47000
},
{
"epoch": 2.21,
"learning_rate": 1.3318372983491306e-05,
"loss": 0.7541,
"step": 47250
},
{
"epoch": 2.22,
"learning_rate": 1.3122214550247943e-05,
"loss": 0.7623,
"step": 47500
},
{
"epoch": 2.23,
"learning_rate": 1.2926056117004584e-05,
"loss": 0.7755,
"step": 47750
},
{
"epoch": 2.24,
"learning_rate": 1.2729897683761221e-05,
"loss": 0.776,
"step": 48000
},
{
"epoch": 2.25,
"learning_rate": 1.253373925051786e-05,
"loss": 0.7686,
"step": 48250
},
{
"epoch": 2.27,
"learning_rate": 1.2337580817274497e-05,
"loss": 0.7627,
"step": 48500
},
{
"epoch": 2.28,
"learning_rate": 1.2141422384031134e-05,
"loss": 0.7486,
"step": 48750
},
{
"epoch": 2.29,
"learning_rate": 1.1945263950787773e-05,
"loss": 0.7457,
"step": 49000
},
{
"epoch": 2.3,
"learning_rate": 1.1749105517544411e-05,
"loss": 0.7511,
"step": 49250
},
{
"epoch": 2.31,
"learning_rate": 1.1552947084301048e-05,
"loss": 0.7618,
"step": 49500
},
{
"epoch": 2.32,
"learning_rate": 1.1356788651057687e-05,
"loss": 0.7598,
"step": 49750
},
{
"epoch": 2.34,
"learning_rate": 1.1160630217814324e-05,
"loss": 0.7752,
"step": 50000
},
{
"epoch": 2.35,
"learning_rate": 1.0964471784570963e-05,
"loss": 0.7385,
"step": 50250
},
{
"epoch": 2.36,
"learning_rate": 1.0768313351327602e-05,
"loss": 0.7528,
"step": 50500
},
{
"epoch": 2.37,
"learning_rate": 1.0572154918084239e-05,
"loss": 0.7582,
"step": 50750
},
{
"epoch": 2.38,
"learning_rate": 1.0375996484840876e-05,
"loss": 0.7731,
"step": 51000
},
{
"epoch": 2.39,
"learning_rate": 1.0179838051597514e-05,
"loss": 0.7617,
"step": 51250
},
{
"epoch": 2.41,
"learning_rate": 9.983679618354153e-06,
"loss": 0.772,
"step": 51500
},
{
"epoch": 2.42,
"learning_rate": 9.78752118511079e-06,
"loss": 0.7681,
"step": 51750
},
{
"epoch": 2.43,
"learning_rate": 9.591362751867429e-06,
"loss": 0.7416,
"step": 52000
},
{
"epoch": 2.44,
"learning_rate": 9.395204318624066e-06,
"loss": 0.7732,
"step": 52250
},
{
"epoch": 2.45,
"learning_rate": 9.199045885380705e-06,
"loss": 0.7543,
"step": 52500
},
{
"epoch": 2.46,
"learning_rate": 9.002887452137343e-06,
"loss": 0.7649,
"step": 52750
},
{
"epoch": 2.48,
"learning_rate": 8.80672901889398e-06,
"loss": 0.756,
"step": 53000
},
{
"epoch": 2.49,
"learning_rate": 8.610570585650618e-06,
"loss": 0.7536,
"step": 53250
},
{
"epoch": 2.5,
"learning_rate": 8.414412152407256e-06,
"loss": 0.7274,
"step": 53500
},
{
"epoch": 2.51,
"learning_rate": 8.218253719163895e-06,
"loss": 0.7458,
"step": 53750
},
{
"epoch": 2.52,
"learning_rate": 8.022095285920532e-06,
"loss": 0.7617,
"step": 54000
},
{
"epoch": 2.53,
"learning_rate": 7.82593685267717e-06,
"loss": 0.755,
"step": 54250
},
{
"epoch": 2.55,
"learning_rate": 7.629778419433808e-06,
"loss": 0.7378,
"step": 54500
},
{
"epoch": 2.56,
"learning_rate": 7.433619986190446e-06,
"loss": 0.7586,
"step": 54750
},
{
"epoch": 2.57,
"learning_rate": 7.2374615529470845e-06,
"loss": 0.7345,
"step": 55000
},
{
"epoch": 2.58,
"learning_rate": 7.041303119703722e-06,
"loss": 0.7316,
"step": 55250
},
{
"epoch": 2.59,
"learning_rate": 6.84514468646036e-06,
"loss": 0.749,
"step": 55500
},
{
"epoch": 2.6,
"learning_rate": 6.648986253216999e-06,
"loss": 0.7583,
"step": 55750
},
{
"epoch": 2.62,
"learning_rate": 6.452827819973636e-06,
"loss": 0.7375,
"step": 56000
},
{
"epoch": 2.63,
"learning_rate": 6.256669386730274e-06,
"loss": 0.7384,
"step": 56250
},
{
"epoch": 2.64,
"learning_rate": 6.060510953486913e-06,
"loss": 0.7472,
"step": 56500
},
{
"epoch": 2.65,
"learning_rate": 5.864352520243551e-06,
"loss": 0.7351,
"step": 56750
},
{
"epoch": 2.66,
"learning_rate": 5.6681940870001885e-06,
"loss": 0.7223,
"step": 57000
},
{
"epoch": 2.67,
"learning_rate": 5.472035653756826e-06,
"loss": 0.7185,
"step": 57250
},
{
"epoch": 2.69,
"learning_rate": 5.275877220513464e-06,
"loss": 0.7496,
"step": 57500
},
{
"epoch": 2.7,
"learning_rate": 5.079718787270102e-06,
"loss": 0.7348,
"step": 57750
},
{
"epoch": 2.71,
"learning_rate": 4.883560354026741e-06,
"loss": 0.7377,
"step": 58000
},
{
"epoch": 2.72,
"learning_rate": 4.687401920783378e-06,
"loss": 0.7432,
"step": 58250
},
{
"epoch": 2.73,
"learning_rate": 4.491243487540017e-06,
"loss": 0.745,
"step": 58500
},
{
"epoch": 2.74,
"learning_rate": 4.295085054296655e-06,
"loss": 0.7545,
"step": 58750
},
{
"epoch": 2.76,
"learning_rate": 4.0989266210532925e-06,
"loss": 0.7346,
"step": 59000
},
{
"epoch": 2.77,
"learning_rate": 3.90276818780993e-06,
"loss": 0.7316,
"step": 59250
},
{
"epoch": 2.78,
"learning_rate": 3.7066097545665687e-06,
"loss": 0.7196,
"step": 59500
},
{
"epoch": 2.79,
"learning_rate": 3.510451321323206e-06,
"loss": 0.7554,
"step": 59750
},
{
"epoch": 2.8,
"learning_rate": 3.3142928880798445e-06,
"loss": 0.7478,
"step": 60000
},
{
"epoch": 2.81,
"learning_rate": 3.1181344548364824e-06,
"loss": 0.7269,
"step": 60250
},
{
"epoch": 2.83,
"learning_rate": 2.9219760215931207e-06,
"loss": 0.7501,
"step": 60500
},
{
"epoch": 2.84,
"learning_rate": 2.7258175883497586e-06,
"loss": 0.7427,
"step": 60750
},
{
"epoch": 2.85,
"learning_rate": 2.5296591551063965e-06,
"loss": 0.7461,
"step": 61000
},
{
"epoch": 2.86,
"learning_rate": 2.3335007218630344e-06,
"loss": 0.7658,
"step": 61250
},
{
"epoch": 2.87,
"learning_rate": 2.1373422886196723e-06,
"loss": 0.7382,
"step": 61500
},
{
"epoch": 2.88,
"learning_rate": 1.94118385537631e-06,
"loss": 0.7404,
"step": 61750
},
{
"epoch": 2.9,
"learning_rate": 1.7450254221329485e-06,
"loss": 0.7458,
"step": 62000
},
{
"epoch": 2.91,
"learning_rate": 1.5488669888895864e-06,
"loss": 0.7337,
"step": 62250
},
{
"epoch": 2.92,
"learning_rate": 1.3527085556462245e-06,
"loss": 0.7529,
"step": 62500
},
{
"epoch": 2.93,
"learning_rate": 1.1565501224028624e-06,
"loss": 0.7406,
"step": 62750
},
{
"epoch": 2.94,
"learning_rate": 9.603916891595003e-07,
"loss": 0.7396,
"step": 63000
},
{
"epoch": 2.95,
"learning_rate": 7.642332559161384e-07,
"loss": 0.7391,
"step": 63250
},
{
"epoch": 2.97,
"learning_rate": 5.680748226727764e-07,
"loss": 0.7235,
"step": 63500
},
{
"epoch": 2.98,
"learning_rate": 3.719163894294144e-07,
"loss": 0.7449,
"step": 63750
},
{
"epoch": 2.99,
"learning_rate": 1.7575795618605238e-07,
"loss": 0.7193,
"step": 64000
}
],
"max_steps": 64224,
"num_train_epochs": 3,
"total_flos": 2.716497252326277e+17,
"trial_name": null,
"trial_params": null
}