{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9911172880536587, "global_step": 148500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.983214830033369e-05, "loss": 2.0208, "step": 500 }, { "epoch": 0.02, "learning_rate": 4.9664296600667384e-05, "loss": 1.9132, "step": 1000 }, { "epoch": 0.03, "learning_rate": 4.949644490100107e-05, "loss": 1.9868, "step": 1500 }, { "epoch": 0.04, "learning_rate": 4.932859320133476e-05, "loss": 1.9393, "step": 2000 }, { "epoch": 0.05, "learning_rate": 4.9160741501668454e-05, "loss": 1.9249, "step": 2500 }, { "epoch": 0.06, "learning_rate": 4.899288980200214e-05, "loss": 1.9124, "step": 3000 }, { "epoch": 0.07, "learning_rate": 4.882503810233582e-05, "loss": 1.9182, "step": 3500 }, { "epoch": 0.08, "learning_rate": 4.8657186402669516e-05, "loss": 1.8824, "step": 4000 }, { "epoch": 0.09, "learning_rate": 4.8489334703003204e-05, "loss": 1.8849, "step": 4500 }, { "epoch": 0.1, "learning_rate": 4.832148300333689e-05, "loss": 1.912, "step": 5000 }, { "epoch": 0.11, "learning_rate": 4.815363130367058e-05, "loss": 1.9117, "step": 5500 }, { "epoch": 0.12, "learning_rate": 4.798577960400427e-05, "loss": 1.9266, "step": 6000 }, { "epoch": 0.13, "learning_rate": 4.781792790433796e-05, "loss": 1.8955, "step": 6500 }, { "epoch": 0.14, "learning_rate": 4.765007620467165e-05, "loss": 1.8364, "step": 7000 }, { "epoch": 0.15, "learning_rate": 4.748222450500534e-05, "loss": 1.9024, "step": 7500 }, { "epoch": 0.16, "learning_rate": 4.731437280533903e-05, "loss": 1.9295, "step": 8000 }, { "epoch": 0.17, "learning_rate": 4.714652110567272e-05, "loss": 1.8972, "step": 8500 }, { "epoch": 0.18, "learning_rate": 4.697866940600641e-05, "loss": 1.8787, "step": 9000 }, { "epoch": 0.19, "learning_rate": 4.68108177063401e-05, "loss": 1.8565, "step": 9500 }, { "epoch": 0.2, "learning_rate": 4.664296600667379e-05, "loss": 1.8623, "step": 10000 }, { "epoch": 0.21, "learning_rate": 4.6475114307007474e-05, "loss": 1.9118, "step": 10500 }, { "epoch": 0.22, "learning_rate": 4.630726260734116e-05, "loss": 1.8846, "step": 11000 }, { "epoch": 0.23, "learning_rate": 4.613941090767485e-05, "loss": 1.806, "step": 11500 }, { "epoch": 0.24, "learning_rate": 4.5971559208008544e-05, "loss": 1.8755, "step": 12000 }, { "epoch": 0.25, "learning_rate": 4.580370750834223e-05, "loss": 1.8432, "step": 12500 }, { "epoch": 0.26, "learning_rate": 4.563585580867592e-05, "loss": 1.8495, "step": 13000 }, { "epoch": 0.27, "learning_rate": 4.5468004109009606e-05, "loss": 1.8765, "step": 13500 }, { "epoch": 0.28, "learning_rate": 4.53001524093433e-05, "loss": 1.881, "step": 14000 }, { "epoch": 0.29, "learning_rate": 4.513230070967699e-05, "loss": 1.8373, "step": 14500 }, { "epoch": 0.3, "learning_rate": 4.4964449010010676e-05, "loss": 1.8529, "step": 15000 }, { "epoch": 0.31, "learning_rate": 4.479659731034437e-05, "loss": 1.8757, "step": 15500 }, { "epoch": 0.32, "learning_rate": 4.462874561067806e-05, "loss": 1.8826, "step": 16000 }, { "epoch": 0.33, "learning_rate": 4.4460893911011745e-05, "loss": 1.824, "step": 16500 }, { "epoch": 0.34, "learning_rate": 4.429304221134544e-05, "loss": 1.8273, "step": 17000 }, { "epoch": 0.35, "learning_rate": 4.412519051167912e-05, "loss": 1.8063, "step": 17500 }, { "epoch": 0.36, "learning_rate": 4.395733881201281e-05, "loss": 1.8627, "step": 18000 }, { "epoch": 0.37, "learning_rate": 4.37894871123465e-05, "loss": 1.8668, "step": 18500 }, { "epoch": 0.38, "learning_rate": 4.362163541268019e-05, "loss": 1.8519, "step": 19000 }, { "epoch": 0.39, "learning_rate": 4.345378371301388e-05, "loss": 1.8254, "step": 19500 }, { "epoch": 0.4, "learning_rate": 4.328593201334757e-05, "loss": 1.8253, "step": 20000 }, { "epoch": 0.41, "learning_rate": 4.311808031368126e-05, "loss": 1.8707, "step": 20500 }, { "epoch": 0.42, "learning_rate": 4.2950228614014947e-05, "loss": 1.865, "step": 21000 }, { "epoch": 0.43, "learning_rate": 4.2782376914348634e-05, "loss": 1.8111, "step": 21500 }, { "epoch": 0.44, "learning_rate": 4.261452521468233e-05, "loss": 1.8106, "step": 22000 }, { "epoch": 0.45, "learning_rate": 4.2446673515016016e-05, "loss": 1.8437, "step": 22500 }, { "epoch": 0.46, "learning_rate": 4.2278821815349704e-05, "loss": 1.7636, "step": 23000 }, { "epoch": 0.47, "learning_rate": 4.21109701156834e-05, "loss": 1.8333, "step": 23500 }, { "epoch": 0.48, "learning_rate": 4.1943118416017085e-05, "loss": 1.7956, "step": 24000 }, { "epoch": 0.49, "learning_rate": 4.1775266716350766e-05, "loss": 1.7986, "step": 24500 }, { "epoch": 0.5, "learning_rate": 4.160741501668446e-05, "loss": 1.8407, "step": 25000 }, { "epoch": 0.51, "learning_rate": 4.143956331701815e-05, "loss": 1.7847, "step": 25500 }, { "epoch": 0.52, "learning_rate": 4.1271711617351835e-05, "loss": 1.8141, "step": 26000 }, { "epoch": 0.53, "learning_rate": 4.110385991768553e-05, "loss": 1.8165, "step": 26500 }, { "epoch": 0.54, "learning_rate": 4.093600821801922e-05, "loss": 1.7665, "step": 27000 }, { "epoch": 0.55, "learning_rate": 4.0768156518352905e-05, "loss": 1.7716, "step": 27500 }, { "epoch": 0.56, "learning_rate": 4.06003048186866e-05, "loss": 1.8061, "step": 28000 }, { "epoch": 0.57, "learning_rate": 4.043245311902029e-05, "loss": 1.7838, "step": 28500 }, { "epoch": 0.58, "learning_rate": 4.0264601419353974e-05, "loss": 1.8049, "step": 29000 }, { "epoch": 0.59, "learning_rate": 4.009674971968766e-05, "loss": 1.8444, "step": 29500 }, { "epoch": 0.6, "learning_rate": 3.9928898020021356e-05, "loss": 1.7829, "step": 30000 }, { "epoch": 0.61, "learning_rate": 3.9761046320355044e-05, "loss": 1.7886, "step": 30500 }, { "epoch": 0.62, "learning_rate": 3.959319462068873e-05, "loss": 1.7421, "step": 31000 }, { "epoch": 0.63, "learning_rate": 3.942534292102242e-05, "loss": 1.7796, "step": 31500 }, { "epoch": 0.64, "learning_rate": 3.9257491221356106e-05, "loss": 1.7857, "step": 32000 }, { "epoch": 0.65, "learning_rate": 3.9089639521689794e-05, "loss": 1.7642, "step": 32500 }, { "epoch": 0.66, "learning_rate": 3.892178782202349e-05, "loss": 1.7624, "step": 33000 }, { "epoch": 0.67, "learning_rate": 3.8753936122357176e-05, "loss": 1.7658, "step": 33500 }, { "epoch": 0.68, "learning_rate": 3.858608442269086e-05, "loss": 1.7744, "step": 34000 }, { "epoch": 0.69, "learning_rate": 3.841823272302456e-05, "loss": 1.7797, "step": 34500 }, { "epoch": 0.7, "learning_rate": 3.8250381023358245e-05, "loss": 1.7507, "step": 35000 }, { "epoch": 0.72, "learning_rate": 3.808252932369193e-05, "loss": 1.7936, "step": 35500 }, { "epoch": 0.73, "learning_rate": 3.791467762402563e-05, "loss": 1.8028, "step": 36000 }, { "epoch": 0.74, "learning_rate": 3.7746825924359314e-05, "loss": 1.7356, "step": 36500 }, { "epoch": 0.75, "learning_rate": 3.7578974224693e-05, "loss": 1.7545, "step": 37000 }, { "epoch": 0.76, "learning_rate": 3.741112252502669e-05, "loss": 1.7676, "step": 37500 }, { "epoch": 0.77, "learning_rate": 3.724327082536038e-05, "loss": 1.7377, "step": 38000 }, { "epoch": 0.78, "learning_rate": 3.7075419125694065e-05, "loss": 1.796, "step": 38500 }, { "epoch": 0.79, "learning_rate": 3.690756742602776e-05, "loss": 1.7323, "step": 39000 }, { "epoch": 0.8, "learning_rate": 3.6739715726361446e-05, "loss": 1.7567, "step": 39500 }, { "epoch": 0.81, "learning_rate": 3.6571864026695134e-05, "loss": 1.7604, "step": 40000 }, { "epoch": 0.82, "learning_rate": 3.640401232702882e-05, "loss": 1.769, "step": 40500 }, { "epoch": 0.83, "learning_rate": 3.6236160627362516e-05, "loss": 1.7287, "step": 41000 }, { "epoch": 0.84, "learning_rate": 3.60683089276962e-05, "loss": 1.6988, "step": 41500 }, { "epoch": 0.85, "learning_rate": 3.590045722802989e-05, "loss": 1.6647, "step": 42000 }, { "epoch": 0.86, "learning_rate": 3.5732605528363585e-05, "loss": 1.7291, "step": 42500 }, { "epoch": 0.87, "learning_rate": 3.556475382869727e-05, "loss": 1.7041, "step": 43000 }, { "epoch": 0.88, "learning_rate": 3.539690212903096e-05, "loss": 1.7678, "step": 43500 }, { "epoch": 0.89, "learning_rate": 3.5229050429364654e-05, "loss": 1.7404, "step": 44000 }, { "epoch": 0.9, "learning_rate": 3.506119872969834e-05, "loss": 1.7057, "step": 44500 }, { "epoch": 0.91, "learning_rate": 3.489334703003202e-05, "loss": 1.7606, "step": 45000 }, { "epoch": 0.92, "learning_rate": 3.472549533036572e-05, "loss": 1.7058, "step": 45500 }, { "epoch": 0.93, "learning_rate": 3.4557643630699405e-05, "loss": 1.7202, "step": 46000 }, { "epoch": 0.94, "learning_rate": 3.438979193103309e-05, "loss": 1.7365, "step": 46500 }, { "epoch": 0.95, "learning_rate": 3.4221940231366786e-05, "loss": 1.7607, "step": 47000 }, { "epoch": 0.96, "learning_rate": 3.4054088531700474e-05, "loss": 1.6973, "step": 47500 }, { "epoch": 0.97, "learning_rate": 3.388623683203416e-05, "loss": 1.7425, "step": 48000 }, { "epoch": 0.98, "learning_rate": 3.371838513236785e-05, "loss": 1.7351, "step": 48500 }, { "epoch": 0.99, "learning_rate": 3.355053343270154e-05, "loss": 1.7134, "step": 49000 }, { "epoch": 1.0, "learning_rate": 3.338268173303523e-05, "loss": 1.7319, "step": 49500 }, { "epoch": 1.01, "learning_rate": 3.321483003336892e-05, "loss": 1.674, "step": 50000 }, { "epoch": 1.02, "learning_rate": 3.304697833370261e-05, "loss": 1.6833, "step": 50500 }, { "epoch": 1.03, "learning_rate": 3.28791266340363e-05, "loss": 1.7056, "step": 51000 }, { "epoch": 1.04, "learning_rate": 3.271127493436998e-05, "loss": 1.6541, "step": 51500 }, { "epoch": 1.05, "learning_rate": 3.2543423234703675e-05, "loss": 1.6648, "step": 52000 }, { "epoch": 1.06, "learning_rate": 3.237557153503736e-05, "loss": 1.6497, "step": 52500 }, { "epoch": 1.07, "learning_rate": 3.220771983537105e-05, "loss": 1.623, "step": 53000 }, { "epoch": 1.08, "learning_rate": 3.2039868135704745e-05, "loss": 1.6813, "step": 53500 }, { "epoch": 1.09, "learning_rate": 3.187201643603843e-05, "loss": 1.6686, "step": 54000 }, { "epoch": 1.1, "learning_rate": 3.170416473637212e-05, "loss": 1.6269, "step": 54500 }, { "epoch": 1.11, "learning_rate": 3.1536313036705814e-05, "loss": 1.656, "step": 55000 }, { "epoch": 1.12, "learning_rate": 3.13684613370395e-05, "loss": 1.7025, "step": 55500 }, { "epoch": 1.13, "learning_rate": 3.120060963737319e-05, "loss": 1.6969, "step": 56000 }, { "epoch": 1.14, "learning_rate": 3.103275793770688e-05, "loss": 1.6252, "step": 56500 }, { "epoch": 1.15, "learning_rate": 3.086490623804057e-05, "loss": 1.6621, "step": 57000 }, { "epoch": 1.16, "learning_rate": 3.069705453837426e-05, "loss": 1.6941, "step": 57500 }, { "epoch": 1.17, "learning_rate": 3.0529202838707946e-05, "loss": 1.6198, "step": 58000 }, { "epoch": 1.18, "learning_rate": 3.0361351139041634e-05, "loss": 1.6934, "step": 58500 }, { "epoch": 1.19, "learning_rate": 3.019349943937532e-05, "loss": 1.6778, "step": 59000 }, { "epoch": 1.2, "learning_rate": 3.0025647739709012e-05, "loss": 1.5945, "step": 59500 }, { "epoch": 1.21, "learning_rate": 2.9857796040042703e-05, "loss": 1.6752, "step": 60000 }, { "epoch": 1.22, "learning_rate": 2.968994434037639e-05, "loss": 1.6683, "step": 60500 }, { "epoch": 1.23, "learning_rate": 2.952209264071008e-05, "loss": 1.6261, "step": 61000 }, { "epoch": 1.24, "learning_rate": 2.935424094104377e-05, "loss": 1.5955, "step": 61500 }, { "epoch": 1.25, "learning_rate": 2.918638924137746e-05, "loss": 1.6483, "step": 62000 }, { "epoch": 1.26, "learning_rate": 2.901853754171115e-05, "loss": 1.5902, "step": 62500 }, { "epoch": 1.27, "learning_rate": 2.885068584204484e-05, "loss": 1.6174, "step": 63000 }, { "epoch": 1.28, "learning_rate": 2.868283414237853e-05, "loss": 1.6423, "step": 63500 }, { "epoch": 1.29, "learning_rate": 2.8514982442712217e-05, "loss": 1.6085, "step": 64000 }, { "epoch": 1.3, "learning_rate": 2.8347130743045908e-05, "loss": 1.6304, "step": 64500 }, { "epoch": 1.31, "learning_rate": 2.81792790433796e-05, "loss": 1.6473, "step": 65000 }, { "epoch": 1.32, "learning_rate": 2.8011427343713283e-05, "loss": 1.6614, "step": 65500 }, { "epoch": 1.33, "learning_rate": 2.784357564404697e-05, "loss": 1.678, "step": 66000 }, { "epoch": 1.34, "learning_rate": 2.767572394438066e-05, "loss": 1.5877, "step": 66500 }, { "epoch": 1.35, "learning_rate": 2.750787224471435e-05, "loss": 1.6678, "step": 67000 }, { "epoch": 1.36, "learning_rate": 2.734002054504804e-05, "loss": 1.5927, "step": 67500 }, { "epoch": 1.37, "learning_rate": 2.717216884538173e-05, "loss": 1.637, "step": 68000 }, { "epoch": 1.38, "learning_rate": 2.7004317145715418e-05, "loss": 1.6597, "step": 68500 }, { "epoch": 1.39, "learning_rate": 2.683646544604911e-05, "loss": 1.6144, "step": 69000 }, { "epoch": 1.4, "learning_rate": 2.6668613746382797e-05, "loss": 1.6186, "step": 69500 }, { "epoch": 1.41, "learning_rate": 2.6500762046716488e-05, "loss": 1.6057, "step": 70000 }, { "epoch": 1.42, "learning_rate": 2.633291034705018e-05, "loss": 1.6342, "step": 70500 }, { "epoch": 1.43, "learning_rate": 2.6165058647383866e-05, "loss": 1.6287, "step": 71000 }, { "epoch": 1.44, "learning_rate": 2.5997206947717557e-05, "loss": 1.5951, "step": 71500 }, { "epoch": 1.45, "learning_rate": 2.5829355248051244e-05, "loss": 1.5532, "step": 72000 }, { "epoch": 1.46, "learning_rate": 2.566150354838493e-05, "loss": 1.6008, "step": 72500 }, { "epoch": 1.47, "learning_rate": 2.549365184871862e-05, "loss": 1.6344, "step": 73000 }, { "epoch": 1.48, "learning_rate": 2.532580014905231e-05, "loss": 1.6064, "step": 73500 }, { "epoch": 1.49, "learning_rate": 2.5157948449385998e-05, "loss": 1.5704, "step": 74000 }, { "epoch": 1.5, "learning_rate": 2.499009674971969e-05, "loss": 1.651, "step": 74500 }, { "epoch": 1.51, "learning_rate": 2.4822245050053376e-05, "loss": 1.5544, "step": 75000 }, { "epoch": 1.52, "learning_rate": 2.4654393350387067e-05, "loss": 1.6534, "step": 75500 }, { "epoch": 1.53, "learning_rate": 2.448654165072076e-05, "loss": 1.5917, "step": 76000 }, { "epoch": 1.54, "learning_rate": 2.4318689951054446e-05, "loss": 1.641, "step": 76500 }, { "epoch": 1.55, "learning_rate": 2.4150838251388137e-05, "loss": 1.5987, "step": 77000 }, { "epoch": 1.56, "learning_rate": 2.3982986551721824e-05, "loss": 1.5428, "step": 77500 }, { "epoch": 1.57, "learning_rate": 2.3815134852055512e-05, "loss": 1.5564, "step": 78000 }, { "epoch": 1.58, "learning_rate": 2.3647283152389203e-05, "loss": 1.5471, "step": 78500 }, { "epoch": 1.59, "learning_rate": 2.347943145272289e-05, "loss": 1.5965, "step": 79000 }, { "epoch": 1.6, "learning_rate": 2.331157975305658e-05, "loss": 1.5883, "step": 79500 }, { "epoch": 1.61, "learning_rate": 2.3143728053390272e-05, "loss": 1.6109, "step": 80000 }, { "epoch": 1.62, "learning_rate": 2.297587635372396e-05, "loss": 1.5431, "step": 80500 }, { "epoch": 1.63, "learning_rate": 2.2808024654057647e-05, "loss": 1.565, "step": 81000 }, { "epoch": 1.64, "learning_rate": 2.2640172954391338e-05, "loss": 1.5316, "step": 81500 }, { "epoch": 1.65, "learning_rate": 2.2472321254725026e-05, "loss": 1.5565, "step": 82000 }, { "epoch": 1.66, "learning_rate": 2.2304469555058717e-05, "loss": 1.5611, "step": 82500 }, { "epoch": 1.67, "learning_rate": 2.2136617855392404e-05, "loss": 1.5737, "step": 83000 }, { "epoch": 1.68, "learning_rate": 2.1968766155726095e-05, "loss": 1.5443, "step": 83500 }, { "epoch": 1.69, "learning_rate": 2.1800914456059786e-05, "loss": 1.5886, "step": 84000 }, { "epoch": 1.7, "learning_rate": 2.163306275639347e-05, "loss": 1.5162, "step": 84500 }, { "epoch": 1.71, "learning_rate": 2.146521105672716e-05, "loss": 1.5889, "step": 85000 }, { "epoch": 1.72, "learning_rate": 2.1297359357060852e-05, "loss": 1.5115, "step": 85500 }, { "epoch": 1.73, "learning_rate": 2.112950765739454e-05, "loss": 1.554, "step": 86000 }, { "epoch": 1.74, "learning_rate": 2.096165595772823e-05, "loss": 1.5168, "step": 86500 }, { "epoch": 1.75, "learning_rate": 2.0793804258061918e-05, "loss": 1.5798, "step": 87000 }, { "epoch": 1.76, "learning_rate": 2.0625952558395605e-05, "loss": 1.5888, "step": 87500 }, { "epoch": 1.77, "learning_rate": 2.0458100858729296e-05, "loss": 1.5502, "step": 88000 }, { "epoch": 1.78, "learning_rate": 2.0290249159062984e-05, "loss": 1.5765, "step": 88500 }, { "epoch": 1.79, "learning_rate": 2.0122397459396675e-05, "loss": 1.6176, "step": 89000 }, { "epoch": 1.8, "learning_rate": 1.9954545759730366e-05, "loss": 1.584, "step": 89500 }, { "epoch": 1.81, "learning_rate": 1.9786694060064053e-05, "loss": 1.5765, "step": 90000 }, { "epoch": 1.82, "learning_rate": 1.9618842360397744e-05, "loss": 1.571, "step": 90500 }, { "epoch": 1.83, "learning_rate": 1.9450990660731432e-05, "loss": 1.5622, "step": 91000 }, { "epoch": 1.84, "learning_rate": 1.928313896106512e-05, "loss": 1.5403, "step": 91500 }, { "epoch": 1.85, "learning_rate": 1.911528726139881e-05, "loss": 1.5757, "step": 92000 }, { "epoch": 1.86, "learning_rate": 1.8947435561732498e-05, "loss": 1.5464, "step": 92500 }, { "epoch": 1.87, "learning_rate": 1.877958386206619e-05, "loss": 1.5551, "step": 93000 }, { "epoch": 1.88, "learning_rate": 1.861173216239988e-05, "loss": 1.6047, "step": 93500 }, { "epoch": 1.89, "learning_rate": 1.8443880462733567e-05, "loss": 1.5463, "step": 94000 }, { "epoch": 1.9, "learning_rate": 1.8276028763067255e-05, "loss": 1.5441, "step": 94500 }, { "epoch": 1.91, "learning_rate": 1.8108177063400946e-05, "loss": 1.4773, "step": 95000 }, { "epoch": 1.92, "learning_rate": 1.7940325363734633e-05, "loss": 1.5374, "step": 95500 }, { "epoch": 1.93, "learning_rate": 1.7772473664068324e-05, "loss": 1.5019, "step": 96000 }, { "epoch": 1.94, "learning_rate": 1.760462196440201e-05, "loss": 1.5133, "step": 96500 }, { "epoch": 1.95, "learning_rate": 1.7436770264735703e-05, "loss": 1.5296, "step": 97000 }, { "epoch": 1.96, "learning_rate": 1.7268918565069393e-05, "loss": 1.5244, "step": 97500 }, { "epoch": 1.97, "learning_rate": 1.7101066865403078e-05, "loss": 1.5128, "step": 98000 }, { "epoch": 1.98, "learning_rate": 1.693321516573677e-05, "loss": 1.5492, "step": 98500 }, { "epoch": 1.99, "learning_rate": 1.676536346607046e-05, "loss": 1.5296, "step": 99000 }, { "epoch": 2.0, "learning_rate": 1.6597511766404147e-05, "loss": 1.5301, "step": 99500 }, { "epoch": 2.01, "learning_rate": 1.6429660066737838e-05, "loss": 1.437, "step": 100000 }, { "epoch": 2.02, "learning_rate": 1.6261808367071525e-05, "loss": 1.5212, "step": 100500 }, { "epoch": 2.03, "learning_rate": 1.6093956667405216e-05, "loss": 1.5229, "step": 101000 }, { "epoch": 2.04, "learning_rate": 1.5926104967738904e-05, "loss": 1.5179, "step": 101500 }, { "epoch": 2.05, "learning_rate": 1.575825326807259e-05, "loss": 1.4308, "step": 102000 }, { "epoch": 2.06, "learning_rate": 1.5590401568406282e-05, "loss": 1.4449, "step": 102500 }, { "epoch": 2.07, "learning_rate": 1.5422549868739973e-05, "loss": 1.4243, "step": 103000 }, { "epoch": 2.08, "learning_rate": 1.525469816907366e-05, "loss": 1.4568, "step": 103500 }, { "epoch": 2.09, "learning_rate": 1.508684646940735e-05, "loss": 1.4311, "step": 104000 }, { "epoch": 2.1, "learning_rate": 1.4918994769741041e-05, "loss": 1.4512, "step": 104500 }, { "epoch": 2.11, "learning_rate": 1.4751143070074727e-05, "loss": 1.4654, "step": 105000 }, { "epoch": 2.13, "learning_rate": 1.4583291370408416e-05, "loss": 1.5063, "step": 105500 }, { "epoch": 2.14, "learning_rate": 1.4415439670742107e-05, "loss": 1.4677, "step": 106000 }, { "epoch": 2.15, "learning_rate": 1.4247587971075796e-05, "loss": 1.42, "step": 106500 }, { "epoch": 2.16, "learning_rate": 1.4079736271409485e-05, "loss": 1.4649, "step": 107000 }, { "epoch": 2.17, "learning_rate": 1.3911884571743175e-05, "loss": 1.457, "step": 107500 }, { "epoch": 2.18, "learning_rate": 1.3744032872076864e-05, "loss": 1.4483, "step": 108000 }, { "epoch": 2.19, "learning_rate": 1.3576181172410551e-05, "loss": 1.4556, "step": 108500 }, { "epoch": 2.2, "learning_rate": 1.340832947274424e-05, "loss": 1.4821, "step": 109000 }, { "epoch": 2.21, "learning_rate": 1.324047777307793e-05, "loss": 1.4516, "step": 109500 }, { "epoch": 2.22, "learning_rate": 1.307262607341162e-05, "loss": 1.4931, "step": 110000 }, { "epoch": 2.23, "learning_rate": 1.290477437374531e-05, "loss": 1.486, "step": 110500 }, { "epoch": 2.24, "learning_rate": 1.2736922674079e-05, "loss": 1.4215, "step": 111000 }, { "epoch": 2.25, "learning_rate": 1.2569070974412688e-05, "loss": 1.4592, "step": 111500 }, { "epoch": 2.26, "learning_rate": 1.2401219274746378e-05, "loss": 1.4665, "step": 112000 }, { "epoch": 2.27, "learning_rate": 1.2233367575080065e-05, "loss": 1.4869, "step": 112500 }, { "epoch": 2.28, "learning_rate": 1.2065515875413754e-05, "loss": 1.4399, "step": 113000 }, { "epoch": 2.29, "learning_rate": 1.1897664175747444e-05, "loss": 1.4149, "step": 113500 }, { "epoch": 2.3, "learning_rate": 1.1729812476081135e-05, "loss": 1.4397, "step": 114000 }, { "epoch": 2.31, "learning_rate": 1.1561960776414822e-05, "loss": 1.4257, "step": 114500 }, { "epoch": 2.32, "learning_rate": 1.1394109076748511e-05, "loss": 1.4216, "step": 115000 }, { "epoch": 2.33, "learning_rate": 1.12262573770822e-05, "loss": 1.4762, "step": 115500 }, { "epoch": 2.34, "learning_rate": 1.105840567741589e-05, "loss": 1.4572, "step": 116000 }, { "epoch": 2.35, "learning_rate": 1.0890553977749579e-05, "loss": 1.4371, "step": 116500 }, { "epoch": 2.36, "learning_rate": 1.0722702278083268e-05, "loss": 1.373, "step": 117000 }, { "epoch": 2.37, "learning_rate": 1.0554850578416957e-05, "loss": 1.4095, "step": 117500 }, { "epoch": 2.38, "learning_rate": 1.0386998878750647e-05, "loss": 1.428, "step": 118000 }, { "epoch": 2.39, "learning_rate": 1.0219147179084336e-05, "loss": 1.3898, "step": 118500 }, { "epoch": 2.4, "learning_rate": 1.0051295479418025e-05, "loss": 1.3816, "step": 119000 }, { "epoch": 2.41, "learning_rate": 9.883443779751714e-06, "loss": 1.4071, "step": 119500 }, { "epoch": 2.42, "learning_rate": 9.715592080085404e-06, "loss": 1.4318, "step": 120000 }, { "epoch": 2.43, "learning_rate": 9.547740380419093e-06, "loss": 1.4176, "step": 120500 }, { "epoch": 2.44, "learning_rate": 9.37988868075278e-06, "loss": 1.4008, "step": 121000 }, { "epoch": 2.45, "learning_rate": 9.212036981086471e-06, "loss": 1.3841, "step": 121500 }, { "epoch": 2.46, "learning_rate": 9.04418528142016e-06, "loss": 1.3971, "step": 122000 }, { "epoch": 2.47, "learning_rate": 8.87633358175385e-06, "loss": 1.4213, "step": 122500 }, { "epoch": 2.48, "learning_rate": 8.708481882087537e-06, "loss": 1.4593, "step": 123000 }, { "epoch": 2.49, "learning_rate": 8.540630182421228e-06, "loss": 1.4426, "step": 123500 }, { "epoch": 2.5, "learning_rate": 8.372778482754917e-06, "loss": 1.4344, "step": 124000 }, { "epoch": 2.51, "learning_rate": 8.204926783088605e-06, "loss": 1.4315, "step": 124500 }, { "epoch": 2.52, "learning_rate": 8.037075083422294e-06, "loss": 1.3943, "step": 125000 }, { "epoch": 2.53, "learning_rate": 7.869223383755985e-06, "loss": 1.4208, "step": 125500 }, { "epoch": 2.54, "learning_rate": 7.701371684089674e-06, "loss": 1.3591, "step": 126000 }, { "epoch": 2.55, "learning_rate": 7.533519984423362e-06, "loss": 1.3961, "step": 126500 }, { "epoch": 2.56, "learning_rate": 7.365668284757052e-06, "loss": 1.4032, "step": 127000 }, { "epoch": 2.57, "learning_rate": 7.197816585090741e-06, "loss": 1.4005, "step": 127500 }, { "epoch": 2.58, "learning_rate": 7.02996488542443e-06, "loss": 1.3813, "step": 128000 }, { "epoch": 2.59, "learning_rate": 6.862113185758119e-06, "loss": 1.3699, "step": 128500 }, { "epoch": 2.6, "learning_rate": 6.694261486091809e-06, "loss": 1.3379, "step": 129000 }, { "epoch": 2.61, "learning_rate": 6.526409786425498e-06, "loss": 1.4578, "step": 129500 }, { "epoch": 2.62, "learning_rate": 6.3585580867591865e-06, "loss": 1.3731, "step": 130000 }, { "epoch": 2.63, "learning_rate": 6.190706387092876e-06, "loss": 1.3846, "step": 130500 }, { "epoch": 2.64, "learning_rate": 6.022854687426565e-06, "loss": 1.3794, "step": 131000 }, { "epoch": 2.65, "learning_rate": 5.855002987760254e-06, "loss": 1.3652, "step": 131500 }, { "epoch": 2.66, "learning_rate": 5.6871512880939434e-06, "loss": 1.3766, "step": 132000 }, { "epoch": 2.67, "learning_rate": 5.519299588427633e-06, "loss": 1.3596, "step": 132500 }, { "epoch": 2.68, "learning_rate": 5.351447888761322e-06, "loss": 1.3424, "step": 133000 }, { "epoch": 2.69, "learning_rate": 5.183596189095011e-06, "loss": 1.3692, "step": 133500 }, { "epoch": 2.7, "learning_rate": 5.0157444894287e-06, "loss": 1.3868, "step": 134000 }, { "epoch": 2.71, "learning_rate": 4.8478927897623896e-06, "loss": 1.3785, "step": 134500 }, { "epoch": 2.72, "learning_rate": 4.680041090096079e-06, "loss": 1.4073, "step": 135000 }, { "epoch": 2.73, "learning_rate": 4.512189390429768e-06, "loss": 1.4196, "step": 135500 }, { "epoch": 2.74, "learning_rate": 4.344337690763456e-06, "loss": 1.36, "step": 136000 }, { "epoch": 2.75, "learning_rate": 4.1764859910971465e-06, "loss": 1.3424, "step": 136500 }, { "epoch": 2.76, "learning_rate": 4.008634291430835e-06, "loss": 1.3667, "step": 137000 }, { "epoch": 2.77, "learning_rate": 3.840782591764525e-06, "loss": 1.3806, "step": 137500 }, { "epoch": 2.78, "learning_rate": 3.6729308920982133e-06, "loss": 1.3193, "step": 138000 }, { "epoch": 2.79, "learning_rate": 3.505079192431903e-06, "loss": 1.3346, "step": 138500 }, { "epoch": 2.8, "learning_rate": 3.3372274927655918e-06, "loss": 1.3369, "step": 139000 }, { "epoch": 2.81, "learning_rate": 3.169375793099281e-06, "loss": 1.3824, "step": 139500 }, { "epoch": 2.82, "learning_rate": 3.0015240934329702e-06, "loss": 1.3799, "step": 140000 }, { "epoch": 2.83, "learning_rate": 2.8336723937666595e-06, "loss": 1.3645, "step": 140500 }, { "epoch": 2.84, "learning_rate": 2.6658206941003487e-06, "loss": 1.343, "step": 141000 }, { "epoch": 2.85, "learning_rate": 2.497968994434038e-06, "loss": 1.3412, "step": 141500 }, { "epoch": 2.86, "learning_rate": 2.330117294767727e-06, "loss": 1.3874, "step": 142000 }, { "epoch": 2.87, "learning_rate": 2.1622655951014164e-06, "loss": 1.4018, "step": 142500 }, { "epoch": 2.88, "learning_rate": 1.9944138954351056e-06, "loss": 1.3285, "step": 143000 }, { "epoch": 2.89, "learning_rate": 1.8265621957687942e-06, "loss": 1.349, "step": 143500 }, { "epoch": 2.9, "learning_rate": 1.6587104961024834e-06, "loss": 1.3412, "step": 144000 }, { "epoch": 2.91, "learning_rate": 1.4908587964361727e-06, "loss": 1.4176, "step": 144500 }, { "epoch": 2.92, "learning_rate": 1.3230070967698619e-06, "loss": 1.3716, "step": 145000 }, { "epoch": 2.93, "learning_rate": 1.1551553971035511e-06, "loss": 1.3065, "step": 145500 }, { "epoch": 2.94, "learning_rate": 9.873036974372403e-07, "loss": 1.377, "step": 146000 }, { "epoch": 2.95, "learning_rate": 8.194519977709295e-07, "loss": 1.3855, "step": 146500 }, { "epoch": 2.96, "learning_rate": 6.516002981046186e-07, "loss": 1.3627, "step": 147000 }, { "epoch": 2.97, "learning_rate": 4.837485984383078e-07, "loss": 1.3446, "step": 147500 }, { "epoch": 2.98, "learning_rate": 3.15896898771997e-07, "loss": 1.3152, "step": 148000 }, { "epoch": 2.99, "learning_rate": 1.4804519910568613e-07, "loss": 1.3785, "step": 148500 } ], "max_steps": 148941, "num_train_epochs": 3, "total_flos": 41584533903906840, "trial_name": null, "trial_params": null }