MarianMT_UFAL_en_es / trainer_state.json
irenelizihui's picture
Upload trainer_state.json
a89e2e8
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9995835068721366,
"global_step": 168000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.9940500981733805e-05,
"loss": 1.1973,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 1.9881001963467605e-05,
"loss": 1.2168,
"step": 1000
},
{
"epoch": 0.01,
"learning_rate": 1.9821502945201405e-05,
"loss": 1.1454,
"step": 1500
},
{
"epoch": 0.01,
"learning_rate": 1.9762003926935208e-05,
"loss": 1.1647,
"step": 2000
},
{
"epoch": 0.01,
"learning_rate": 1.9702504908669008e-05,
"loss": 1.1795,
"step": 2500
},
{
"epoch": 0.02,
"learning_rate": 1.964300589040281e-05,
"loss": 1.1723,
"step": 3000
},
{
"epoch": 0.02,
"learning_rate": 1.958350687213661e-05,
"loss": 1.157,
"step": 3500
},
{
"epoch": 0.02,
"learning_rate": 1.952400785387041e-05,
"loss": 1.1699,
"step": 4000
},
{
"epoch": 0.03,
"learning_rate": 1.9464508835604215e-05,
"loss": 1.1593,
"step": 4500
},
{
"epoch": 0.03,
"learning_rate": 1.9405009817338018e-05,
"loss": 1.1421,
"step": 5000
},
{
"epoch": 0.03,
"learning_rate": 1.9345510799071814e-05,
"loss": 1.0648,
"step": 5500
},
{
"epoch": 0.04,
"learning_rate": 1.9286011780805618e-05,
"loss": 1.1345,
"step": 6000
},
{
"epoch": 0.04,
"learning_rate": 1.922651276253942e-05,
"loss": 1.1722,
"step": 6500
},
{
"epoch": 0.04,
"learning_rate": 1.916701374427322e-05,
"loss": 1.1588,
"step": 7000
},
{
"epoch": 0.04,
"learning_rate": 1.910751472600702e-05,
"loss": 1.1502,
"step": 7500
},
{
"epoch": 0.05,
"learning_rate": 1.9048015707740824e-05,
"loss": 1.1297,
"step": 8000
},
{
"epoch": 0.05,
"learning_rate": 1.8988516689474624e-05,
"loss": 1.1478,
"step": 8500
},
{
"epoch": 0.05,
"learning_rate": 1.8929017671208427e-05,
"loss": 1.1025,
"step": 9000
},
{
"epoch": 0.06,
"learning_rate": 1.8869518652942227e-05,
"loss": 1.126,
"step": 9500
},
{
"epoch": 0.06,
"learning_rate": 1.8810019634676027e-05,
"loss": 1.1438,
"step": 10000
},
{
"epoch": 0.06,
"learning_rate": 1.875052061640983e-05,
"loss": 1.1064,
"step": 10500
},
{
"epoch": 0.07,
"learning_rate": 1.8691021598143634e-05,
"loss": 1.1369,
"step": 11000
},
{
"epoch": 0.07,
"learning_rate": 1.8631522579877434e-05,
"loss": 1.1036,
"step": 11500
},
{
"epoch": 0.07,
"learning_rate": 1.8572023561611234e-05,
"loss": 1.0771,
"step": 12000
},
{
"epoch": 0.07,
"learning_rate": 1.8512524543345037e-05,
"loss": 1.0841,
"step": 12500
},
{
"epoch": 0.08,
"learning_rate": 1.8453025525078837e-05,
"loss": 1.0818,
"step": 13000
},
{
"epoch": 0.08,
"learning_rate": 1.839352650681264e-05,
"loss": 1.1462,
"step": 13500
},
{
"epoch": 0.08,
"learning_rate": 1.833402748854644e-05,
"loss": 1.1335,
"step": 14000
},
{
"epoch": 0.09,
"learning_rate": 1.827452847028024e-05,
"loss": 1.079,
"step": 14500
},
{
"epoch": 0.09,
"learning_rate": 1.8215029452014044e-05,
"loss": 1.1276,
"step": 15000
},
{
"epoch": 0.09,
"learning_rate": 1.8155530433747847e-05,
"loss": 1.1119,
"step": 15500
},
{
"epoch": 0.1,
"learning_rate": 1.8096031415481647e-05,
"loss": 1.1343,
"step": 16000
},
{
"epoch": 0.1,
"learning_rate": 1.8036532397215447e-05,
"loss": 1.0787,
"step": 16500
},
{
"epoch": 0.1,
"learning_rate": 1.797703337894925e-05,
"loss": 1.1122,
"step": 17000
},
{
"epoch": 0.1,
"learning_rate": 1.791753436068305e-05,
"loss": 1.0904,
"step": 17500
},
{
"epoch": 0.11,
"learning_rate": 1.785803534241685e-05,
"loss": 1.0882,
"step": 18000
},
{
"epoch": 0.11,
"learning_rate": 1.7798536324150653e-05,
"loss": 1.1311,
"step": 18500
},
{
"epoch": 0.11,
"learning_rate": 1.7739037305884453e-05,
"loss": 1.0936,
"step": 19000
},
{
"epoch": 0.12,
"learning_rate": 1.7679538287618257e-05,
"loss": 1.0959,
"step": 19500
},
{
"epoch": 0.12,
"learning_rate": 1.7620039269352056e-05,
"loss": 1.0879,
"step": 20000
},
{
"epoch": 0.12,
"learning_rate": 1.7560540251085856e-05,
"loss": 1.0763,
"step": 20500
},
{
"epoch": 0.12,
"learning_rate": 1.750104123281966e-05,
"loss": 1.0713,
"step": 21000
},
{
"epoch": 0.13,
"learning_rate": 1.7441542214553463e-05,
"loss": 1.1204,
"step": 21500
},
{
"epoch": 0.13,
"learning_rate": 1.7382043196287263e-05,
"loss": 1.0808,
"step": 22000
},
{
"epoch": 0.13,
"learning_rate": 1.7322544178021063e-05,
"loss": 1.1301,
"step": 22500
},
{
"epoch": 0.14,
"learning_rate": 1.7263045159754866e-05,
"loss": 1.1141,
"step": 23000
},
{
"epoch": 0.14,
"learning_rate": 1.7203546141488666e-05,
"loss": 1.109,
"step": 23500
},
{
"epoch": 0.14,
"learning_rate": 1.714404712322247e-05,
"loss": 1.0691,
"step": 24000
},
{
"epoch": 0.15,
"learning_rate": 1.708454810495627e-05,
"loss": 1.0805,
"step": 24500
},
{
"epoch": 0.15,
"learning_rate": 1.702504908669007e-05,
"loss": 1.0795,
"step": 25000
},
{
"epoch": 0.15,
"learning_rate": 1.6965550068423873e-05,
"loss": 1.0431,
"step": 25500
},
{
"epoch": 0.15,
"learning_rate": 1.6906051050157676e-05,
"loss": 1.0553,
"step": 26000
},
{
"epoch": 0.16,
"learning_rate": 1.6846552031891476e-05,
"loss": 1.0786,
"step": 26500
},
{
"epoch": 0.16,
"learning_rate": 1.6787053013625276e-05,
"loss": 1.1008,
"step": 27000
},
{
"epoch": 0.16,
"learning_rate": 1.672755399535908e-05,
"loss": 1.0776,
"step": 27500
},
{
"epoch": 0.17,
"learning_rate": 1.666805497709288e-05,
"loss": 1.0993,
"step": 28000
},
{
"epoch": 0.17,
"learning_rate": 1.660855595882668e-05,
"loss": 1.1101,
"step": 28500
},
{
"epoch": 0.17,
"learning_rate": 1.6549056940560482e-05,
"loss": 1.0791,
"step": 29000
},
{
"epoch": 0.18,
"learning_rate": 1.6489557922294282e-05,
"loss": 1.0705,
"step": 29500
},
{
"epoch": 0.18,
"learning_rate": 1.6430058904028086e-05,
"loss": 1.0618,
"step": 30000
},
{
"epoch": 0.18,
"learning_rate": 1.6370559885761885e-05,
"loss": 1.0408,
"step": 30500
},
{
"epoch": 0.18,
"learning_rate": 1.631106086749569e-05,
"loss": 1.1367,
"step": 31000
},
{
"epoch": 0.19,
"learning_rate": 1.625156184922949e-05,
"loss": 1.0727,
"step": 31500
},
{
"epoch": 0.19,
"learning_rate": 1.6192062830963292e-05,
"loss": 1.0444,
"step": 32000
},
{
"epoch": 0.19,
"learning_rate": 1.6132563812697092e-05,
"loss": 1.0723,
"step": 32500
},
{
"epoch": 0.2,
"learning_rate": 1.6073064794430892e-05,
"loss": 1.0627,
"step": 33000
},
{
"epoch": 0.2,
"learning_rate": 1.6013565776164695e-05,
"loss": 1.0835,
"step": 33500
},
{
"epoch": 0.2,
"learning_rate": 1.5954066757898495e-05,
"loss": 1.1624,
"step": 34000
},
{
"epoch": 0.21,
"learning_rate": 1.58945677396323e-05,
"loss": 1.0383,
"step": 34500
},
{
"epoch": 0.21,
"learning_rate": 1.58350687213661e-05,
"loss": 1.0805,
"step": 35000
},
{
"epoch": 0.21,
"learning_rate": 1.57755697030999e-05,
"loss": 1.0426,
"step": 35500
},
{
"epoch": 0.21,
"learning_rate": 1.57160706848337e-05,
"loss": 1.1281,
"step": 36000
},
{
"epoch": 0.22,
"learning_rate": 1.5656571666567505e-05,
"loss": 1.0695,
"step": 36500
},
{
"epoch": 0.22,
"learning_rate": 1.5597072648301305e-05,
"loss": 1.0617,
"step": 37000
},
{
"epoch": 0.22,
"learning_rate": 1.5537573630035105e-05,
"loss": 1.0849,
"step": 37500
},
{
"epoch": 0.23,
"learning_rate": 1.5478074611768908e-05,
"loss": 1.0923,
"step": 38000
},
{
"epoch": 0.23,
"learning_rate": 1.5418575593502708e-05,
"loss": 1.1421,
"step": 38500
},
{
"epoch": 0.23,
"learning_rate": 1.5359076575236508e-05,
"loss": 1.0981,
"step": 39000
},
{
"epoch": 0.24,
"learning_rate": 1.529957755697031e-05,
"loss": 1.0991,
"step": 39500
},
{
"epoch": 0.24,
"learning_rate": 1.5240078538704113e-05,
"loss": 1.0568,
"step": 40000
},
{
"epoch": 0.24,
"learning_rate": 1.5180579520437915e-05,
"loss": 1.1249,
"step": 40500
},
{
"epoch": 0.24,
"learning_rate": 1.5121080502171714e-05,
"loss": 1.0764,
"step": 41000
},
{
"epoch": 0.25,
"learning_rate": 1.5061581483905516e-05,
"loss": 1.0582,
"step": 41500
},
{
"epoch": 0.25,
"learning_rate": 1.5002082465639318e-05,
"loss": 1.0795,
"step": 42000
},
{
"epoch": 0.25,
"learning_rate": 1.494258344737312e-05,
"loss": 1.0514,
"step": 42500
},
{
"epoch": 0.26,
"learning_rate": 1.488308442910692e-05,
"loss": 1.0427,
"step": 43000
},
{
"epoch": 0.26,
"learning_rate": 1.4823585410840723e-05,
"loss": 1.0842,
"step": 43500
},
{
"epoch": 0.26,
"learning_rate": 1.4764086392574524e-05,
"loss": 1.0961,
"step": 44000
},
{
"epoch": 0.26,
"learning_rate": 1.4704587374308326e-05,
"loss": 1.1226,
"step": 44500
},
{
"epoch": 0.27,
"learning_rate": 1.4645088356042127e-05,
"loss": 1.0607,
"step": 45000
},
{
"epoch": 0.27,
"learning_rate": 1.4585589337775927e-05,
"loss": 1.0479,
"step": 45500
},
{
"epoch": 0.27,
"learning_rate": 1.4526090319509729e-05,
"loss": 1.0572,
"step": 46000
},
{
"epoch": 0.28,
"learning_rate": 1.446659130124353e-05,
"loss": 1.1465,
"step": 46500
},
{
"epoch": 0.28,
"learning_rate": 1.4407092282977332e-05,
"loss": 1.041,
"step": 47000
},
{
"epoch": 0.28,
"learning_rate": 1.4347593264711132e-05,
"loss": 1.0366,
"step": 47500
},
{
"epoch": 0.29,
"learning_rate": 1.4288094246444934e-05,
"loss": 1.0661,
"step": 48000
},
{
"epoch": 0.29,
"learning_rate": 1.4228595228178737e-05,
"loss": 1.0495,
"step": 48500
},
{
"epoch": 0.29,
"learning_rate": 1.4169096209912539e-05,
"loss": 1.0769,
"step": 49000
},
{
"epoch": 0.29,
"learning_rate": 1.4109597191646339e-05,
"loss": 1.0329,
"step": 49500
},
{
"epoch": 0.3,
"learning_rate": 1.405009817338014e-05,
"loss": 1.0702,
"step": 50000
},
{
"epoch": 0.3,
"learning_rate": 1.3990599155113942e-05,
"loss": 1.0354,
"step": 50500
},
{
"epoch": 0.3,
"learning_rate": 1.3931100136847744e-05,
"loss": 1.0817,
"step": 51000
},
{
"epoch": 0.31,
"learning_rate": 1.3871601118581544e-05,
"loss": 1.0362,
"step": 51500
},
{
"epoch": 0.31,
"learning_rate": 1.3812102100315345e-05,
"loss": 1.0761,
"step": 52000
},
{
"epoch": 0.31,
"learning_rate": 1.3752603082049147e-05,
"loss": 1.0295,
"step": 52500
},
{
"epoch": 0.32,
"learning_rate": 1.369310406378295e-05,
"loss": 1.0365,
"step": 53000
},
{
"epoch": 0.32,
"learning_rate": 1.3633605045516748e-05,
"loss": 1.0378,
"step": 53500
},
{
"epoch": 0.32,
"learning_rate": 1.3574106027250552e-05,
"loss": 1.0172,
"step": 54000
},
{
"epoch": 0.32,
"learning_rate": 1.3514607008984353e-05,
"loss": 1.0553,
"step": 54500
},
{
"epoch": 0.33,
"learning_rate": 1.3455107990718155e-05,
"loss": 1.0595,
"step": 55000
},
{
"epoch": 0.33,
"learning_rate": 1.3395608972451957e-05,
"loss": 1.0443,
"step": 55500
},
{
"epoch": 0.33,
"learning_rate": 1.3336109954185756e-05,
"loss": 1.0359,
"step": 56000
},
{
"epoch": 0.34,
"learning_rate": 1.3276610935919558e-05,
"loss": 1.0631,
"step": 56500
},
{
"epoch": 0.34,
"learning_rate": 1.321711191765336e-05,
"loss": 1.0293,
"step": 57000
},
{
"epoch": 0.34,
"learning_rate": 1.3157612899387161e-05,
"loss": 1.0878,
"step": 57500
},
{
"epoch": 0.35,
"learning_rate": 1.3098113881120961e-05,
"loss": 1.054,
"step": 58000
},
{
"epoch": 0.35,
"learning_rate": 1.3038614862854765e-05,
"loss": 1.0523,
"step": 58500
},
{
"epoch": 0.35,
"learning_rate": 1.2979115844588566e-05,
"loss": 1.0117,
"step": 59000
},
{
"epoch": 0.35,
"learning_rate": 1.2919616826322368e-05,
"loss": 1.0612,
"step": 59500
},
{
"epoch": 0.36,
"learning_rate": 1.2860117808056168e-05,
"loss": 1.0106,
"step": 60000
},
{
"epoch": 0.36,
"learning_rate": 1.280061878978997e-05,
"loss": 1.026,
"step": 60500
},
{
"epoch": 0.36,
"learning_rate": 1.2741119771523771e-05,
"loss": 1.0533,
"step": 61000
},
{
"epoch": 0.37,
"learning_rate": 1.2681620753257573e-05,
"loss": 1.0651,
"step": 61500
},
{
"epoch": 0.37,
"learning_rate": 1.2622121734991373e-05,
"loss": 1.0717,
"step": 62000
},
{
"epoch": 0.37,
"learning_rate": 1.2562622716725174e-05,
"loss": 1.0679,
"step": 62500
},
{
"epoch": 0.37,
"learning_rate": 1.2503123698458976e-05,
"loss": 1.0594,
"step": 63000
},
{
"epoch": 0.38,
"learning_rate": 1.2443624680192779e-05,
"loss": 1.0457,
"step": 63500
},
{
"epoch": 0.38,
"learning_rate": 1.2384125661926577e-05,
"loss": 1.0491,
"step": 64000
},
{
"epoch": 0.38,
"learning_rate": 1.232462664366038e-05,
"loss": 1.0212,
"step": 64500
},
{
"epoch": 0.39,
"learning_rate": 1.2265127625394182e-05,
"loss": 1.0079,
"step": 65000
},
{
"epoch": 0.39,
"learning_rate": 1.2205628607127984e-05,
"loss": 1.0513,
"step": 65500
},
{
"epoch": 0.39,
"learning_rate": 1.2146129588861786e-05,
"loss": 1.0651,
"step": 66000
},
{
"epoch": 0.4,
"learning_rate": 1.2086630570595585e-05,
"loss": 1.1196,
"step": 66500
},
{
"epoch": 0.4,
"learning_rate": 1.2027131552329387e-05,
"loss": 1.0522,
"step": 67000
},
{
"epoch": 0.4,
"learning_rate": 1.1967632534063189e-05,
"loss": 1.0775,
"step": 67500
},
{
"epoch": 0.4,
"learning_rate": 1.1908133515796992e-05,
"loss": 1.0494,
"step": 68000
},
{
"epoch": 0.41,
"learning_rate": 1.184863449753079e-05,
"loss": 1.0312,
"step": 68500
},
{
"epoch": 0.41,
"learning_rate": 1.1789135479264594e-05,
"loss": 0.9866,
"step": 69000
},
{
"epoch": 0.41,
"learning_rate": 1.1729636460998395e-05,
"loss": 1.0072,
"step": 69500
},
{
"epoch": 0.42,
"learning_rate": 1.1670137442732197e-05,
"loss": 1.0145,
"step": 70000
},
{
"epoch": 0.42,
"learning_rate": 1.1610638424465997e-05,
"loss": 1.0334,
"step": 70500
},
{
"epoch": 0.42,
"learning_rate": 1.1551139406199798e-05,
"loss": 1.1029,
"step": 71000
},
{
"epoch": 0.43,
"learning_rate": 1.14916403879336e-05,
"loss": 1.0556,
"step": 71500
},
{
"epoch": 0.43,
"learning_rate": 1.1432141369667402e-05,
"loss": 1.0046,
"step": 72000
},
{
"epoch": 0.43,
"learning_rate": 1.1372642351401202e-05,
"loss": 1.0827,
"step": 72500
},
{
"epoch": 0.43,
"learning_rate": 1.1313143333135003e-05,
"loss": 1.0392,
"step": 73000
},
{
"epoch": 0.44,
"learning_rate": 1.1253644314868807e-05,
"loss": 1.0635,
"step": 73500
},
{
"epoch": 0.44,
"learning_rate": 1.1194145296602608e-05,
"loss": 1.0246,
"step": 74000
},
{
"epoch": 0.44,
"learning_rate": 1.1134646278336408e-05,
"loss": 1.0271,
"step": 74500
},
{
"epoch": 0.45,
"learning_rate": 1.107514726007021e-05,
"loss": 0.9673,
"step": 75000
},
{
"epoch": 0.45,
"learning_rate": 1.1015648241804011e-05,
"loss": 1.0277,
"step": 75500
},
{
"epoch": 0.45,
"learning_rate": 1.0956149223537813e-05,
"loss": 1.0755,
"step": 76000
},
{
"epoch": 0.46,
"learning_rate": 1.0896650205271615e-05,
"loss": 1.017,
"step": 76500
},
{
"epoch": 0.46,
"learning_rate": 1.0837151187005414e-05,
"loss": 1.012,
"step": 77000
},
{
"epoch": 0.46,
"learning_rate": 1.0777652168739216e-05,
"loss": 1.0078,
"step": 77500
},
{
"epoch": 0.46,
"learning_rate": 1.0718153150473018e-05,
"loss": 1.0136,
"step": 78000
},
{
"epoch": 0.47,
"learning_rate": 1.0658654132206821e-05,
"loss": 1.0571,
"step": 78500
},
{
"epoch": 0.47,
"learning_rate": 1.059915511394062e-05,
"loss": 1.0558,
"step": 79000
},
{
"epoch": 0.47,
"learning_rate": 1.0539656095674423e-05,
"loss": 0.9997,
"step": 79500
},
{
"epoch": 0.48,
"learning_rate": 1.0480157077408224e-05,
"loss": 1.0411,
"step": 80000
},
{
"epoch": 0.48,
"learning_rate": 1.0420658059142026e-05,
"loss": 1.0569,
"step": 80500
},
{
"epoch": 0.48,
"learning_rate": 1.0361159040875826e-05,
"loss": 1.0182,
"step": 81000
},
{
"epoch": 0.48,
"learning_rate": 1.0301660022609627e-05,
"loss": 1.028,
"step": 81500
},
{
"epoch": 0.49,
"learning_rate": 1.0242161004343429e-05,
"loss": 1.0548,
"step": 82000
},
{
"epoch": 0.49,
"learning_rate": 1.018266198607723e-05,
"loss": 1.0717,
"step": 82500
},
{
"epoch": 0.49,
"learning_rate": 1.012316296781103e-05,
"loss": 1.0518,
"step": 83000
},
{
"epoch": 0.5,
"learning_rate": 1.0063663949544832e-05,
"loss": 1.025,
"step": 83500
},
{
"epoch": 0.5,
"learning_rate": 1.0004164931278636e-05,
"loss": 1.0131,
"step": 84000
},
{
"epoch": 0.5,
"learning_rate": 9.944665913012435e-06,
"loss": 1.0403,
"step": 84500
},
{
"epoch": 0.51,
"learning_rate": 9.885166894746237e-06,
"loss": 0.9678,
"step": 85000
},
{
"epoch": 0.51,
"learning_rate": 9.825667876480039e-06,
"loss": 1.0123,
"step": 85500
},
{
"epoch": 0.51,
"learning_rate": 9.76616885821384e-06,
"loss": 1.0033,
"step": 86000
},
{
"epoch": 0.51,
"learning_rate": 9.706669839947642e-06,
"loss": 1.0143,
"step": 86500
},
{
"epoch": 0.52,
"learning_rate": 9.647170821681444e-06,
"loss": 1.0296,
"step": 87000
},
{
"epoch": 0.52,
"learning_rate": 9.587671803415245e-06,
"loss": 1.0152,
"step": 87500
},
{
"epoch": 0.52,
"learning_rate": 9.528172785149045e-06,
"loss": 1.0504,
"step": 88000
},
{
"epoch": 0.53,
"learning_rate": 9.468673766882848e-06,
"loss": 1.0308,
"step": 88500
},
{
"epoch": 0.53,
"learning_rate": 9.409174748616648e-06,
"loss": 1.0092,
"step": 89000
},
{
"epoch": 0.53,
"learning_rate": 9.34967573035045e-06,
"loss": 1.0669,
"step": 89500
},
{
"epoch": 0.54,
"learning_rate": 9.290176712084252e-06,
"loss": 1.0227,
"step": 90000
},
{
"epoch": 0.54,
"learning_rate": 9.230677693818053e-06,
"loss": 1.0142,
"step": 90500
},
{
"epoch": 0.54,
"learning_rate": 9.171178675551853e-06,
"loss": 1.073,
"step": 91000
},
{
"epoch": 0.54,
"learning_rate": 9.111679657285657e-06,
"loss": 1.0564,
"step": 91500
},
{
"epoch": 0.55,
"learning_rate": 9.052180639019456e-06,
"loss": 1.0278,
"step": 92000
},
{
"epoch": 0.55,
"learning_rate": 8.992681620753258e-06,
"loss": 1.0213,
"step": 92500
},
{
"epoch": 0.55,
"learning_rate": 8.93318260248706e-06,
"loss": 1.0354,
"step": 93000
},
{
"epoch": 0.56,
"learning_rate": 8.873683584220861e-06,
"loss": 1.0315,
"step": 93500
},
{
"epoch": 0.56,
"learning_rate": 8.814184565954663e-06,
"loss": 1.0585,
"step": 94000
},
{
"epoch": 0.56,
"learning_rate": 8.754685547688465e-06,
"loss": 1.0345,
"step": 94500
},
{
"epoch": 0.57,
"learning_rate": 8.695186529422264e-06,
"loss": 1.0612,
"step": 95000
},
{
"epoch": 0.57,
"learning_rate": 8.635687511156066e-06,
"loss": 1.0734,
"step": 95500
},
{
"epoch": 0.57,
"learning_rate": 8.576188492889868e-06,
"loss": 1.0822,
"step": 96000
},
{
"epoch": 0.57,
"learning_rate": 8.51668947462367e-06,
"loss": 1.0095,
"step": 96500
},
{
"epoch": 0.58,
"learning_rate": 8.457190456357471e-06,
"loss": 1.0104,
"step": 97000
},
{
"epoch": 0.58,
"learning_rate": 8.397691438091273e-06,
"loss": 1.0637,
"step": 97500
},
{
"epoch": 0.58,
"learning_rate": 8.338192419825074e-06,
"loss": 1.0427,
"step": 98000
},
{
"epoch": 0.59,
"learning_rate": 8.278693401558874e-06,
"loss": 1.0073,
"step": 98500
},
{
"epoch": 0.59,
"learning_rate": 8.219194383292677e-06,
"loss": 0.9647,
"step": 99000
},
{
"epoch": 0.59,
"learning_rate": 8.159695365026477e-06,
"loss": 1.0151,
"step": 99500
},
{
"epoch": 0.59,
"learning_rate": 8.100196346760279e-06,
"loss": 1.0446,
"step": 100000
},
{
"epoch": 0.6,
"learning_rate": 8.04069732849408e-06,
"loss": 1.0598,
"step": 100500
},
{
"epoch": 0.6,
"learning_rate": 7.981198310227882e-06,
"loss": 1.0044,
"step": 101000
},
{
"epoch": 0.6,
"learning_rate": 7.921699291961684e-06,
"loss": 1.0272,
"step": 101500
},
{
"epoch": 0.61,
"learning_rate": 7.862200273695486e-06,
"loss": 1.009,
"step": 102000
},
{
"epoch": 0.61,
"learning_rate": 7.802701255429285e-06,
"loss": 1.0505,
"step": 102500
},
{
"epoch": 0.61,
"learning_rate": 7.743202237163087e-06,
"loss": 1.0559,
"step": 103000
},
{
"epoch": 0.62,
"learning_rate": 7.683703218896889e-06,
"loss": 1.0412,
"step": 103500
},
{
"epoch": 0.62,
"learning_rate": 7.62420420063069e-06,
"loss": 1.0651,
"step": 104000
},
{
"epoch": 0.62,
"learning_rate": 7.564705182364491e-06,
"loss": 0.9968,
"step": 104500
},
{
"epoch": 0.62,
"learning_rate": 7.505206164098294e-06,
"loss": 1.0395,
"step": 105000
},
{
"epoch": 0.63,
"learning_rate": 7.445707145832094e-06,
"loss": 1.0378,
"step": 105500
},
{
"epoch": 0.63,
"learning_rate": 7.386208127565896e-06,
"loss": 0.99,
"step": 106000
},
{
"epoch": 0.63,
"learning_rate": 7.326709109299697e-06,
"loss": 0.9792,
"step": 106500
},
{
"epoch": 0.64,
"learning_rate": 7.267210091033498e-06,
"loss": 1.0482,
"step": 107000
},
{
"epoch": 0.64,
"learning_rate": 7.207711072767299e-06,
"loss": 1.0143,
"step": 107500
},
{
"epoch": 0.64,
"learning_rate": 7.148212054501102e-06,
"loss": 1.0321,
"step": 108000
},
{
"epoch": 0.65,
"learning_rate": 7.088713036234903e-06,
"loss": 1.0312,
"step": 108500
},
{
"epoch": 0.65,
"learning_rate": 7.029214017968704e-06,
"loss": 1.022,
"step": 109000
},
{
"epoch": 0.65,
"learning_rate": 6.969714999702506e-06,
"loss": 1.0529,
"step": 109500
},
{
"epoch": 0.65,
"learning_rate": 6.9102159814363064e-06,
"loss": 1.0269,
"step": 110000
},
{
"epoch": 0.66,
"learning_rate": 6.850716963170109e-06,
"loss": 0.9678,
"step": 110500
},
{
"epoch": 0.66,
"learning_rate": 6.79121794490391e-06,
"loss": 1.0315,
"step": 111000
},
{
"epoch": 0.66,
"learning_rate": 6.731718926637711e-06,
"loss": 0.9849,
"step": 111500
},
{
"epoch": 0.67,
"learning_rate": 6.672219908371512e-06,
"loss": 1.0397,
"step": 112000
},
{
"epoch": 0.67,
"learning_rate": 6.6127208901053146e-06,
"loss": 1.0511,
"step": 112500
},
{
"epoch": 0.67,
"learning_rate": 6.553221871839115e-06,
"loss": 1.0381,
"step": 113000
},
{
"epoch": 0.68,
"learning_rate": 6.493722853572917e-06,
"loss": 1.0584,
"step": 113500
},
{
"epoch": 0.68,
"learning_rate": 6.434223835306718e-06,
"loss": 0.9933,
"step": 114000
},
{
"epoch": 0.68,
"learning_rate": 6.374724817040519e-06,
"loss": 1.0439,
"step": 114500
},
{
"epoch": 0.68,
"learning_rate": 6.31522579877432e-06,
"loss": 1.047,
"step": 115000
},
{
"epoch": 0.69,
"learning_rate": 6.255726780508123e-06,
"loss": 1.0349,
"step": 115500
},
{
"epoch": 0.69,
"learning_rate": 6.196227762241923e-06,
"loss": 0.9923,
"step": 116000
},
{
"epoch": 0.69,
"learning_rate": 6.136728743975725e-06,
"loss": 0.9953,
"step": 116500
},
{
"epoch": 0.7,
"learning_rate": 6.077229725709526e-06,
"loss": 1.0083,
"step": 117000
},
{
"epoch": 0.7,
"learning_rate": 6.017730707443327e-06,
"loss": 1.0258,
"step": 117500
},
{
"epoch": 0.7,
"learning_rate": 5.958231689177128e-06,
"loss": 0.9729,
"step": 118000
},
{
"epoch": 0.71,
"learning_rate": 5.898732670910931e-06,
"loss": 1.0351,
"step": 118500
},
{
"epoch": 0.71,
"learning_rate": 5.8392336526447314e-06,
"loss": 1.0298,
"step": 119000
},
{
"epoch": 0.71,
"learning_rate": 5.779734634378533e-06,
"loss": 1.0067,
"step": 119500
},
{
"epoch": 0.71,
"learning_rate": 5.7202356161123355e-06,
"loss": 1.0235,
"step": 120000
},
{
"epoch": 0.72,
"learning_rate": 5.660736597846136e-06,
"loss": 1.0018,
"step": 120500
},
{
"epoch": 0.72,
"learning_rate": 5.601237579579938e-06,
"loss": 1.0096,
"step": 121000
},
{
"epoch": 0.72,
"learning_rate": 5.541738561313739e-06,
"loss": 0.9732,
"step": 121500
},
{
"epoch": 0.73,
"learning_rate": 5.48223954304754e-06,
"loss": 1.0359,
"step": 122000
},
{
"epoch": 0.73,
"learning_rate": 5.422740524781341e-06,
"loss": 0.9913,
"step": 122500
},
{
"epoch": 0.73,
"learning_rate": 5.363241506515144e-06,
"loss": 1.0132,
"step": 123000
},
{
"epoch": 0.73,
"learning_rate": 5.303742488248944e-06,
"loss": 1.0526,
"step": 123500
},
{
"epoch": 0.74,
"learning_rate": 5.244243469982746e-06,
"loss": 0.9742,
"step": 124000
},
{
"epoch": 0.74,
"learning_rate": 5.184744451716547e-06,
"loss": 0.9988,
"step": 124500
},
{
"epoch": 0.74,
"learning_rate": 5.125245433450348e-06,
"loss": 0.9938,
"step": 125000
},
{
"epoch": 0.75,
"learning_rate": 5.065746415184149e-06,
"loss": 1.022,
"step": 125500
},
{
"epoch": 0.75,
"learning_rate": 5.006247396917952e-06,
"loss": 1.0214,
"step": 126000
},
{
"epoch": 0.75,
"learning_rate": 4.946748378651752e-06,
"loss": 0.9758,
"step": 126500
},
{
"epoch": 0.76,
"learning_rate": 4.887249360385554e-06,
"loss": 1.0235,
"step": 127000
},
{
"epoch": 0.76,
"learning_rate": 4.827750342119356e-06,
"loss": 1.0387,
"step": 127500
},
{
"epoch": 0.76,
"learning_rate": 4.768251323853157e-06,
"loss": 1.0319,
"step": 128000
},
{
"epoch": 0.76,
"learning_rate": 4.708752305586958e-06,
"loss": 0.9961,
"step": 128500
},
{
"epoch": 0.77,
"learning_rate": 4.64925328732076e-06,
"loss": 1.0318,
"step": 129000
},
{
"epoch": 0.77,
"learning_rate": 4.589754269054561e-06,
"loss": 1.1137,
"step": 129500
},
{
"epoch": 0.77,
"learning_rate": 4.530255250788362e-06,
"loss": 1.0398,
"step": 130000
},
{
"epoch": 0.78,
"learning_rate": 4.470756232522164e-06,
"loss": 0.9347,
"step": 130500
},
{
"epoch": 0.78,
"learning_rate": 4.411257214255965e-06,
"loss": 1.0618,
"step": 131000
},
{
"epoch": 0.78,
"learning_rate": 4.351758195989766e-06,
"loss": 0.9826,
"step": 131500
},
{
"epoch": 0.79,
"learning_rate": 4.292259177723568e-06,
"loss": 1.0422,
"step": 132000
},
{
"epoch": 0.79,
"learning_rate": 4.232760159457369e-06,
"loss": 0.9928,
"step": 132500
},
{
"epoch": 0.79,
"learning_rate": 4.17326114119117e-06,
"loss": 1.0487,
"step": 133000
},
{
"epoch": 0.79,
"learning_rate": 4.113762122924972e-06,
"loss": 1.0379,
"step": 133500
},
{
"epoch": 0.8,
"learning_rate": 4.054263104658773e-06,
"loss": 1.0225,
"step": 134000
},
{
"epoch": 0.8,
"learning_rate": 3.994764086392575e-06,
"loss": 0.951,
"step": 134500
},
{
"epoch": 0.8,
"learning_rate": 3.935265068126376e-06,
"loss": 1.0393,
"step": 135000
},
{
"epoch": 0.81,
"learning_rate": 3.875766049860177e-06,
"loss": 0.9851,
"step": 135500
},
{
"epoch": 0.81,
"learning_rate": 3.816267031593979e-06,
"loss": 0.9716,
"step": 136000
},
{
"epoch": 0.81,
"learning_rate": 3.7567680133277807e-06,
"loss": 0.9974,
"step": 136500
},
{
"epoch": 0.82,
"learning_rate": 3.697268995061582e-06,
"loss": 1.0052,
"step": 137000
},
{
"epoch": 0.82,
"learning_rate": 3.6377699767953835e-06,
"loss": 0.9949,
"step": 137500
},
{
"epoch": 0.82,
"learning_rate": 3.5782709585291847e-06,
"loss": 1.0279,
"step": 138000
},
{
"epoch": 0.82,
"learning_rate": 3.518771940262986e-06,
"loss": 0.9636,
"step": 138500
},
{
"epoch": 0.83,
"learning_rate": 3.4592729219967875e-06,
"loss": 1.0322,
"step": 139000
},
{
"epoch": 0.83,
"learning_rate": 3.3997739037305887e-06,
"loss": 1.0153,
"step": 139500
},
{
"epoch": 0.83,
"learning_rate": 3.3402748854643903e-06,
"loss": 0.9982,
"step": 140000
},
{
"epoch": 0.84,
"learning_rate": 3.2807758671981915e-06,
"loss": 1.022,
"step": 140500
},
{
"epoch": 0.84,
"learning_rate": 3.2212768489319927e-06,
"loss": 1.0208,
"step": 141000
},
{
"epoch": 0.84,
"learning_rate": 3.1617778306657944e-06,
"loss": 0.9761,
"step": 141500
},
{
"epoch": 0.84,
"learning_rate": 3.1022788123995956e-06,
"loss": 1.0125,
"step": 142000
},
{
"epoch": 0.85,
"learning_rate": 3.0427797941333968e-06,
"loss": 1.0531,
"step": 142500
},
{
"epoch": 0.85,
"learning_rate": 2.9832807758671984e-06,
"loss": 1.0232,
"step": 143000
},
{
"epoch": 0.85,
"learning_rate": 2.9237817576009996e-06,
"loss": 1.067,
"step": 143500
},
{
"epoch": 0.86,
"learning_rate": 2.8642827393348012e-06,
"loss": 0.9986,
"step": 144000
},
{
"epoch": 0.86,
"learning_rate": 2.8047837210686024e-06,
"loss": 1.0009,
"step": 144500
},
{
"epoch": 0.86,
"learning_rate": 2.7452847028024036e-06,
"loss": 1.0113,
"step": 145000
},
{
"epoch": 0.87,
"learning_rate": 2.6857856845362052e-06,
"loss": 0.9991,
"step": 145500
},
{
"epoch": 0.87,
"learning_rate": 2.6262866662700064e-06,
"loss": 1.0153,
"step": 146000
},
{
"epoch": 0.87,
"learning_rate": 2.5667876480038085e-06,
"loss": 1.0382,
"step": 146500
},
{
"epoch": 0.87,
"learning_rate": 2.5072886297376097e-06,
"loss": 0.993,
"step": 147000
},
{
"epoch": 0.88,
"learning_rate": 2.447789611471411e-06,
"loss": 0.9704,
"step": 147500
},
{
"epoch": 0.88,
"learning_rate": 2.388290593205212e-06,
"loss": 0.9935,
"step": 148000
},
{
"epoch": 0.88,
"learning_rate": 2.3287915749390137e-06,
"loss": 1.0239,
"step": 148500
},
{
"epoch": 0.89,
"learning_rate": 2.269292556672815e-06,
"loss": 0.9927,
"step": 149000
},
{
"epoch": 0.89,
"learning_rate": 2.2097935384066165e-06,
"loss": 1.0262,
"step": 149500
},
{
"epoch": 0.89,
"learning_rate": 2.150294520140418e-06,
"loss": 0.9762,
"step": 150000
},
{
"epoch": 0.9,
"learning_rate": 2.0907955018742194e-06,
"loss": 1.0298,
"step": 150500
},
{
"epoch": 0.9,
"learning_rate": 2.0312964836080206e-06,
"loss": 0.9869,
"step": 151000
},
{
"epoch": 0.9,
"learning_rate": 1.971797465341822e-06,
"loss": 1.0192,
"step": 151500
},
{
"epoch": 0.9,
"learning_rate": 1.9122984470756234e-06,
"loss": 1.0034,
"step": 152000
},
{
"epoch": 0.91,
"learning_rate": 1.8527994288094248e-06,
"loss": 1.0179,
"step": 152500
},
{
"epoch": 0.91,
"learning_rate": 1.7933004105432262e-06,
"loss": 1.0189,
"step": 153000
},
{
"epoch": 0.91,
"learning_rate": 1.7338013922770274e-06,
"loss": 0.9545,
"step": 153500
},
{
"epoch": 0.92,
"learning_rate": 1.6743023740108288e-06,
"loss": 1.0462,
"step": 154000
},
{
"epoch": 0.92,
"learning_rate": 1.6148033557446302e-06,
"loss": 1.0083,
"step": 154500
},
{
"epoch": 0.92,
"learning_rate": 1.5553043374784319e-06,
"loss": 0.9755,
"step": 155000
},
{
"epoch": 0.93,
"learning_rate": 1.4958053192122333e-06,
"loss": 1.0234,
"step": 155500
},
{
"epoch": 0.93,
"learning_rate": 1.4363063009460345e-06,
"loss": 1.0116,
"step": 156000
},
{
"epoch": 0.93,
"learning_rate": 1.3768072826798359e-06,
"loss": 1.0502,
"step": 156500
},
{
"epoch": 0.93,
"learning_rate": 1.3173082644136373e-06,
"loss": 0.9593,
"step": 157000
},
{
"epoch": 0.94,
"learning_rate": 1.2578092461474387e-06,
"loss": 1.0195,
"step": 157500
},
{
"epoch": 0.94,
"learning_rate": 1.19831022788124e-06,
"loss": 1.0212,
"step": 158000
},
{
"epoch": 0.94,
"learning_rate": 1.1388112096150413e-06,
"loss": 0.9523,
"step": 158500
},
{
"epoch": 0.95,
"learning_rate": 1.079312191348843e-06,
"loss": 0.9902,
"step": 159000
},
{
"epoch": 0.95,
"learning_rate": 1.0198131730826442e-06,
"loss": 1.0248,
"step": 159500
},
{
"epoch": 0.95,
"learning_rate": 9.603141548164456e-07,
"loss": 1.0062,
"step": 160000
},
{
"epoch": 0.95,
"learning_rate": 9.00815136550247e-07,
"loss": 1.0035,
"step": 160500
},
{
"epoch": 0.96,
"learning_rate": 8.413161182840483e-07,
"loss": 0.9755,
"step": 161000
},
{
"epoch": 0.96,
"learning_rate": 7.818171000178498e-07,
"loss": 1.0131,
"step": 161500
},
{
"epoch": 0.96,
"learning_rate": 7.223180817516512e-07,
"loss": 1.0353,
"step": 162000
},
{
"epoch": 0.97,
"learning_rate": 6.628190634854525e-07,
"loss": 1.0138,
"step": 162500
},
{
"epoch": 0.97,
"learning_rate": 6.033200452192539e-07,
"loss": 0.9738,
"step": 163000
},
{
"epoch": 0.97,
"learning_rate": 5.438210269530553e-07,
"loss": 0.9878,
"step": 163500
},
{
"epoch": 0.98,
"learning_rate": 4.843220086868568e-07,
"loss": 0.9787,
"step": 164000
},
{
"epoch": 0.98,
"learning_rate": 4.2482299042065806e-07,
"loss": 0.9903,
"step": 164500
},
{
"epoch": 0.98,
"learning_rate": 3.653239721544595e-07,
"loss": 1.032,
"step": 165000
},
{
"epoch": 0.98,
"learning_rate": 3.0582495388826084e-07,
"loss": 1.0436,
"step": 165500
},
{
"epoch": 0.99,
"learning_rate": 2.4632593562206225e-07,
"loss": 0.9956,
"step": 166000
},
{
"epoch": 0.99,
"learning_rate": 1.8682691735586366e-07,
"loss": 1.0102,
"step": 166500
},
{
"epoch": 0.99,
"learning_rate": 1.2732789908966505e-07,
"loss": 0.9634,
"step": 167000
},
{
"epoch": 1.0,
"learning_rate": 6.782888082346642e-08,
"loss": 1.004,
"step": 167500
},
{
"epoch": 1.0,
"learning_rate": 8.329862557267807e-09,
"loss": 0.9777,
"step": 168000
}
],
"max_steps": 168070,
"num_train_epochs": 1,
"total_flos": 6437535729057792.0,
"trial_name": null,
"trial_params": null
}