diff --git "a/ranker_10/checkpoint-6200/trainer_state.json" "b/ranker_10/checkpoint-6200/trainer_state.json" new file mode 100644--- /dev/null +++ "b/ranker_10/checkpoint-6200/trainer_state.json" @@ -0,0 +1,7859 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.4949267192784665, + "global_step": 6200, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 1.999436302142052e-05, + "loss": 0.6976, + "step": 5 + }, + { + "epoch": 0.01, + "learning_rate": 1.9988726042841038e-05, + "loss": 0.6745, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 1.998308906426156e-05, + "loss": 0.661, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 1.9977452085682077e-05, + "loss": 0.6516, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 1.9971815107102595e-05, + "loss": 0.6658, + "step": 25 + }, + { + "epoch": 0.02, + "learning_rate": 1.9966178128523113e-05, + "loss": 0.6508, + "step": 30 + }, + { + "epoch": 0.02, + "learning_rate": 1.996054114994363e-05, + "loss": 0.6372, + "step": 35 + }, + { + "epoch": 0.02, + "learning_rate": 1.9954904171364152e-05, + "loss": 0.6267, + "step": 40 + }, + { + "epoch": 0.03, + "learning_rate": 1.994926719278467e-05, + "loss": 0.5938, + "step": 45 + }, + { + "epoch": 0.03, + "learning_rate": 1.9943630214205188e-05, + "loss": 0.5724, + "step": 50 + }, + { + "epoch": 0.03, + "learning_rate": 1.9937993235625706e-05, + "loss": 0.5756, + "step": 55 + }, + { + "epoch": 0.03, + "learning_rate": 1.9932356257046224e-05, + "loss": 0.5472, + "step": 60 + }, + { + "epoch": 0.04, + "learning_rate": 1.9926719278466746e-05, + "loss": 0.5244, + "step": 65 + }, + { + "epoch": 0.04, + "learning_rate": 1.9921082299887264e-05, + "loss": 0.5459, + "step": 70 + }, + { + "epoch": 0.04, + "learning_rate": 1.991544532130778e-05, + "loss": 0.4937, + "step": 75 + }, + { + "epoch": 0.05, + "learning_rate": 1.99098083427283e-05, + "loss": 0.4946, + "step": 80 + }, + { + "epoch": 0.05, + "learning_rate": 1.9904171364148818e-05, + "loss": 0.4974, + "step": 85 + }, + { + "epoch": 0.05, + "learning_rate": 1.9898534385569336e-05, + "loss": 0.4999, + "step": 90 + }, + { + "epoch": 0.05, + "learning_rate": 1.9892897406989854e-05, + "loss": 0.4863, + "step": 95 + }, + { + "epoch": 0.06, + "learning_rate": 1.988726042841037e-05, + "loss": 0.4896, + "step": 100 + }, + { + "epoch": 0.06, + "learning_rate": 1.9881623449830893e-05, + "loss": 0.4704, + "step": 105 + }, + { + "epoch": 0.06, + "learning_rate": 1.987598647125141e-05, + "loss": 0.4731, + "step": 110 + }, + { + "epoch": 0.06, + "learning_rate": 1.987034949267193e-05, + "loss": 0.476, + "step": 115 + }, + { + "epoch": 0.07, + "learning_rate": 1.9864712514092447e-05, + "loss": 0.4587, + "step": 120 + }, + { + "epoch": 0.07, + "learning_rate": 1.9859075535512965e-05, + "loss": 0.4682, + "step": 125 + }, + { + "epoch": 0.07, + "learning_rate": 1.9853438556933483e-05, + "loss": 0.4751, + "step": 130 + }, + { + "epoch": 0.08, + "learning_rate": 1.9847801578354004e-05, + "loss": 0.48, + "step": 135 + }, + { + "epoch": 0.08, + "learning_rate": 1.9842164599774522e-05, + "loss": 0.4626, + "step": 140 + }, + { + "epoch": 0.08, + "learning_rate": 1.983652762119504e-05, + "loss": 0.4496, + "step": 145 + }, + { + "epoch": 0.08, + "learning_rate": 1.9830890642615558e-05, + "loss": 0.4453, + "step": 150 + }, + { + "epoch": 0.09, + "learning_rate": 1.982525366403608e-05, + "loss": 0.4234, + "step": 155 + }, + { + "epoch": 0.09, + "learning_rate": 1.9819616685456598e-05, + "loss": 0.4603, + "step": 160 + }, + { + "epoch": 0.09, + "learning_rate": 1.9813979706877116e-05, + "loss": 0.4274, + "step": 165 + }, + { + "epoch": 0.1, + "learning_rate": 1.9808342728297634e-05, + "loss": 0.4563, + "step": 170 + }, + { + "epoch": 0.1, + "learning_rate": 1.980270574971815e-05, + "loss": 0.4626, + "step": 175 + }, + { + "epoch": 0.1, + "learning_rate": 1.9797068771138673e-05, + "loss": 0.45, + "step": 180 + }, + { + "epoch": 0.1, + "learning_rate": 1.979143179255919e-05, + "loss": 0.4417, + "step": 185 + }, + { + "epoch": 0.11, + "learning_rate": 1.978579481397971e-05, + "loss": 0.4141, + "step": 190 + }, + { + "epoch": 0.11, + "learning_rate": 1.9780157835400227e-05, + "loss": 0.4181, + "step": 195 + }, + { + "epoch": 0.11, + "learning_rate": 1.9774520856820745e-05, + "loss": 0.4299, + "step": 200 + }, + { + "epoch": 0.11, + "eval_accuracy": 0.7052037190283795, + "eval_accuracy_sklearn": 0.7052037190283795, + "eval_f1": 0.7746353081274189, + "eval_loss": 0.6019008755683899, + "eval_precision": 0.6594019260010137, + "eval_recall": 0.9386724386724387, + "eval_runtime": 117.5938, + "eval_samples_per_second": 174.695, + "eval_steps_per_second": 10.919, + "step": 200 + }, + { + "epoch": 0.12, + "learning_rate": 1.9768883878241266e-05, + "loss": 0.4399, + "step": 205 + }, + { + "epoch": 0.12, + "learning_rate": 1.9763246899661784e-05, + "loss": 0.4305, + "step": 210 + }, + { + "epoch": 0.12, + "learning_rate": 1.9757609921082302e-05, + "loss": 0.4226, + "step": 215 + }, + { + "epoch": 0.12, + "learning_rate": 1.975197294250282e-05, + "loss": 0.398, + "step": 220 + }, + { + "epoch": 0.13, + "learning_rate": 1.9746335963923338e-05, + "loss": 0.422, + "step": 225 + }, + { + "epoch": 0.13, + "learning_rate": 1.974069898534386e-05, + "loss": 0.4272, + "step": 230 + }, + { + "epoch": 0.13, + "learning_rate": 1.9735062006764378e-05, + "loss": 0.4178, + "step": 235 + }, + { + "epoch": 0.14, + "learning_rate": 1.9729425028184896e-05, + "loss": 0.4269, + "step": 240 + }, + { + "epoch": 0.14, + "learning_rate": 1.9723788049605414e-05, + "loss": 0.4188, + "step": 245 + }, + { + "epoch": 0.14, + "learning_rate": 1.971815107102593e-05, + "loss": 0.4306, + "step": 250 + }, + { + "epoch": 0.14, + "learning_rate": 1.971251409244645e-05, + "loss": 0.4119, + "step": 255 + }, + { + "epoch": 0.15, + "learning_rate": 1.9706877113866967e-05, + "loss": 0.4317, + "step": 260 + }, + { + "epoch": 0.15, + "learning_rate": 1.9701240135287485e-05, + "loss": 0.393, + "step": 265 + }, + { + "epoch": 0.15, + "learning_rate": 1.9695603156708007e-05, + "loss": 0.4155, + "step": 270 + }, + { + "epoch": 0.16, + "learning_rate": 1.9689966178128525e-05, + "loss": 0.4083, + "step": 275 + }, + { + "epoch": 0.16, + "learning_rate": 1.9684329199549043e-05, + "loss": 0.4159, + "step": 280 + }, + { + "epoch": 0.16, + "learning_rate": 1.967869222096956e-05, + "loss": 0.4387, + "step": 285 + }, + { + "epoch": 0.16, + "learning_rate": 1.967305524239008e-05, + "loss": 0.4012, + "step": 290 + }, + { + "epoch": 0.17, + "learning_rate": 1.9667418263810597e-05, + "loss": 0.3998, + "step": 295 + }, + { + "epoch": 0.17, + "learning_rate": 1.9661781285231118e-05, + "loss": 0.4234, + "step": 300 + }, + { + "epoch": 0.17, + "learning_rate": 1.9656144306651636e-05, + "loss": 0.3898, + "step": 305 + }, + { + "epoch": 0.17, + "learning_rate": 1.9650507328072154e-05, + "loss": 0.3925, + "step": 310 + }, + { + "epoch": 0.18, + "learning_rate": 1.9644870349492672e-05, + "loss": 0.4232, + "step": 315 + }, + { + "epoch": 0.18, + "learning_rate": 1.963923337091319e-05, + "loss": 0.3966, + "step": 320 + }, + { + "epoch": 0.18, + "learning_rate": 1.963359639233371e-05, + "loss": 0.4046, + "step": 325 + }, + { + "epoch": 0.19, + "learning_rate": 1.962795941375423e-05, + "loss": 0.4045, + "step": 330 + }, + { + "epoch": 0.19, + "learning_rate": 1.9622322435174747e-05, + "loss": 0.4014, + "step": 335 + }, + { + "epoch": 0.19, + "learning_rate": 1.9616685456595265e-05, + "loss": 0.3948, + "step": 340 + }, + { + "epoch": 0.19, + "learning_rate": 1.9611048478015787e-05, + "loss": 0.4133, + "step": 345 + }, + { + "epoch": 0.2, + "learning_rate": 1.9605411499436305e-05, + "loss": 0.3917, + "step": 350 + }, + { + "epoch": 0.2, + "learning_rate": 1.9599774520856823e-05, + "loss": 0.4036, + "step": 355 + }, + { + "epoch": 0.2, + "learning_rate": 1.959413754227734e-05, + "loss": 0.3996, + "step": 360 + }, + { + "epoch": 0.21, + "learning_rate": 1.958850056369786e-05, + "loss": 0.378, + "step": 365 + }, + { + "epoch": 0.21, + "learning_rate": 1.958286358511838e-05, + "loss": 0.3902, + "step": 370 + }, + { + "epoch": 0.21, + "learning_rate": 1.9577226606538898e-05, + "loss": 0.3728, + "step": 375 + }, + { + "epoch": 0.21, + "learning_rate": 1.9571589627959416e-05, + "loss": 0.4034, + "step": 380 + }, + { + "epoch": 0.22, + "learning_rate": 1.9565952649379934e-05, + "loss": 0.3945, + "step": 385 + }, + { + "epoch": 0.22, + "learning_rate": 1.9560315670800452e-05, + "loss": 0.3925, + "step": 390 + }, + { + "epoch": 0.22, + "learning_rate": 1.9554678692220973e-05, + "loss": 0.4005, + "step": 395 + }, + { + "epoch": 0.23, + "learning_rate": 1.954904171364149e-05, + "loss": 0.3852, + "step": 400 + }, + { + "epoch": 0.23, + "eval_accuracy": 0.7777345081049506, + "eval_accuracy_sklearn": 0.7777345081049506, + "eval_f1": 0.8089219953130231, + "eval_loss": 0.46637997031211853, + "eval_precision": 0.7546064959400375, + "eval_recall": 0.8716630591630592, + "eval_runtime": 122.6432, + "eval_samples_per_second": 167.502, + "eval_steps_per_second": 10.469, + "step": 400 + }, + { + "epoch": 0.23, + "learning_rate": 1.954340473506201e-05, + "loss": 0.3739, + "step": 405 + }, + { + "epoch": 0.23, + "learning_rate": 1.9537767756482527e-05, + "loss": 0.4123, + "step": 410 + }, + { + "epoch": 0.23, + "learning_rate": 1.9532130777903045e-05, + "loss": 0.4036, + "step": 415 + }, + { + "epoch": 0.24, + "learning_rate": 1.9526493799323563e-05, + "loss": 0.406, + "step": 420 + }, + { + "epoch": 0.24, + "learning_rate": 1.952085682074408e-05, + "loss": 0.4102, + "step": 425 + }, + { + "epoch": 0.24, + "learning_rate": 1.95152198421646e-05, + "loss": 0.4192, + "step": 430 + }, + { + "epoch": 0.25, + "learning_rate": 1.950958286358512e-05, + "loss": 0.3894, + "step": 435 + }, + { + "epoch": 0.25, + "learning_rate": 1.950394588500564e-05, + "loss": 0.3765, + "step": 440 + }, + { + "epoch": 0.25, + "learning_rate": 1.9498308906426157e-05, + "loss": 0.3818, + "step": 445 + }, + { + "epoch": 0.25, + "learning_rate": 1.9492671927846675e-05, + "loss": 0.3807, + "step": 450 + }, + { + "epoch": 0.26, + "learning_rate": 1.9487034949267193e-05, + "loss": 0.4095, + "step": 455 + }, + { + "epoch": 0.26, + "learning_rate": 1.948139797068771e-05, + "loss": 0.3792, + "step": 460 + }, + { + "epoch": 0.26, + "learning_rate": 1.9475760992108232e-05, + "loss": 0.3698, + "step": 465 + }, + { + "epoch": 0.26, + "learning_rate": 1.947012401352875e-05, + "loss": 0.3762, + "step": 470 + }, + { + "epoch": 0.27, + "learning_rate": 1.9464487034949268e-05, + "loss": 0.3603, + "step": 475 + }, + { + "epoch": 0.27, + "learning_rate": 1.9458850056369786e-05, + "loss": 0.3792, + "step": 480 + }, + { + "epoch": 0.27, + "learning_rate": 1.9453213077790304e-05, + "loss": 0.3774, + "step": 485 + }, + { + "epoch": 0.28, + "learning_rate": 1.9447576099210825e-05, + "loss": 0.3773, + "step": 490 + }, + { + "epoch": 0.28, + "learning_rate": 1.9441939120631343e-05, + "loss": 0.3917, + "step": 495 + }, + { + "epoch": 0.28, + "learning_rate": 1.943630214205186e-05, + "loss": 0.3806, + "step": 500 + }, + { + "epoch": 0.28, + "learning_rate": 1.943066516347238e-05, + "loss": 0.3757, + "step": 505 + }, + { + "epoch": 0.29, + "learning_rate": 1.94250281848929e-05, + "loss": 0.3813, + "step": 510 + }, + { + "epoch": 0.29, + "learning_rate": 1.941939120631342e-05, + "loss": 0.3818, + "step": 515 + }, + { + "epoch": 0.29, + "learning_rate": 1.9413754227733937e-05, + "loss": 0.374, + "step": 520 + }, + { + "epoch": 0.3, + "learning_rate": 1.9408117249154455e-05, + "loss": 0.3739, + "step": 525 + }, + { + "epoch": 0.3, + "learning_rate": 1.9402480270574973e-05, + "loss": 0.362, + "step": 530 + }, + { + "epoch": 0.3, + "learning_rate": 1.9396843291995494e-05, + "loss": 0.3626, + "step": 535 + }, + { + "epoch": 0.3, + "learning_rate": 1.9391206313416012e-05, + "loss": 0.373, + "step": 540 + }, + { + "epoch": 0.31, + "learning_rate": 1.938556933483653e-05, + "loss": 0.3833, + "step": 545 + }, + { + "epoch": 0.31, + "learning_rate": 1.9379932356257048e-05, + "loss": 0.3931, + "step": 550 + }, + { + "epoch": 0.31, + "learning_rate": 1.9374295377677566e-05, + "loss": 0.3715, + "step": 555 + }, + { + "epoch": 0.32, + "learning_rate": 1.9368658399098087e-05, + "loss": 0.3596, + "step": 560 + }, + { + "epoch": 0.32, + "learning_rate": 1.9363021420518605e-05, + "loss": 0.376, + "step": 565 + }, + { + "epoch": 0.32, + "learning_rate": 1.9357384441939123e-05, + "loss": 0.3936, + "step": 570 + }, + { + "epoch": 0.32, + "learning_rate": 1.935174746335964e-05, + "loss": 0.3561, + "step": 575 + }, + { + "epoch": 0.33, + "learning_rate": 1.934611048478016e-05, + "loss": 0.3757, + "step": 580 + }, + { + "epoch": 0.33, + "learning_rate": 1.9340473506200677e-05, + "loss": 0.3634, + "step": 585 + }, + { + "epoch": 0.33, + "learning_rate": 1.9334836527621195e-05, + "loss": 0.3743, + "step": 590 + }, + { + "epoch": 0.34, + "learning_rate": 1.9329199549041717e-05, + "loss": 0.3684, + "step": 595 + }, + { + "epoch": 0.34, + "learning_rate": 1.9323562570462235e-05, + "loss": 0.3383, + "step": 600 + }, + { + "epoch": 0.34, + "eval_accuracy": 0.7725745996203086, + "eval_accuracy_sklearn": 0.7725745996203086, + "eval_f1": 0.8170855845274451, + "eval_loss": 0.5046594142913818, + "eval_precision": 0.7219454822194549, + "eval_recall": 0.9411075036075036, + "eval_runtime": 122.8876, + "eval_samples_per_second": 167.169, + "eval_steps_per_second": 10.449, + "step": 600 + }, + { + "epoch": 0.34, + "learning_rate": 1.9317925591882753e-05, + "loss": 0.3639, + "step": 605 + }, + { + "epoch": 0.34, + "learning_rate": 1.931228861330327e-05, + "loss": 0.3759, + "step": 610 + }, + { + "epoch": 0.35, + "learning_rate": 1.930665163472379e-05, + "loss": 0.3858, + "step": 615 + }, + { + "epoch": 0.35, + "learning_rate": 1.9301014656144307e-05, + "loss": 0.3413, + "step": 620 + }, + { + "epoch": 0.35, + "learning_rate": 1.9295377677564825e-05, + "loss": 0.3691, + "step": 625 + }, + { + "epoch": 0.36, + "learning_rate": 1.9289740698985346e-05, + "loss": 0.3703, + "step": 630 + }, + { + "epoch": 0.36, + "learning_rate": 1.9284103720405864e-05, + "loss": 0.3535, + "step": 635 + }, + { + "epoch": 0.36, + "learning_rate": 1.9278466741826382e-05, + "loss": 0.3699, + "step": 640 + }, + { + "epoch": 0.36, + "learning_rate": 1.92728297632469e-05, + "loss": 0.3643, + "step": 645 + }, + { + "epoch": 0.37, + "learning_rate": 1.9267192784667418e-05, + "loss": 0.3646, + "step": 650 + }, + { + "epoch": 0.37, + "learning_rate": 1.926155580608794e-05, + "loss": 0.3706, + "step": 655 + }, + { + "epoch": 0.37, + "learning_rate": 1.9255918827508457e-05, + "loss": 0.3408, + "step": 660 + }, + { + "epoch": 0.37, + "learning_rate": 1.9250281848928975e-05, + "loss": 0.3665, + "step": 665 + }, + { + "epoch": 0.38, + "learning_rate": 1.9244644870349493e-05, + "loss": 0.3652, + "step": 670 + }, + { + "epoch": 0.38, + "learning_rate": 1.923900789177001e-05, + "loss": 0.3437, + "step": 675 + }, + { + "epoch": 0.38, + "learning_rate": 1.9233370913190533e-05, + "loss": 0.3361, + "step": 680 + }, + { + "epoch": 0.39, + "learning_rate": 1.922773393461105e-05, + "loss": 0.3604, + "step": 685 + }, + { + "epoch": 0.39, + "learning_rate": 1.922209695603157e-05, + "loss": 0.3682, + "step": 690 + }, + { + "epoch": 0.39, + "learning_rate": 1.9216459977452087e-05, + "loss": 0.3519, + "step": 695 + }, + { + "epoch": 0.39, + "learning_rate": 1.9210822998872608e-05, + "loss": 0.3646, + "step": 700 + }, + { + "epoch": 0.4, + "learning_rate": 1.9205186020293126e-05, + "loss": 0.3517, + "step": 705 + }, + { + "epoch": 0.4, + "learning_rate": 1.9199549041713644e-05, + "loss": 0.3577, + "step": 710 + }, + { + "epoch": 0.4, + "learning_rate": 1.9193912063134162e-05, + "loss": 0.3666, + "step": 715 + }, + { + "epoch": 0.41, + "learning_rate": 1.918827508455468e-05, + "loss": 0.343, + "step": 720 + }, + { + "epoch": 0.41, + "learning_rate": 1.91826381059752e-05, + "loss": 0.3315, + "step": 725 + }, + { + "epoch": 0.41, + "learning_rate": 1.917700112739572e-05, + "loss": 0.368, + "step": 730 + }, + { + "epoch": 0.41, + "learning_rate": 1.9171364148816237e-05, + "loss": 0.3566, + "step": 735 + }, + { + "epoch": 0.42, + "learning_rate": 1.9165727170236755e-05, + "loss": 0.358, + "step": 740 + }, + { + "epoch": 0.42, + "learning_rate": 1.9160090191657273e-05, + "loss": 0.3517, + "step": 745 + }, + { + "epoch": 0.42, + "learning_rate": 1.915445321307779e-05, + "loss": 0.3739, + "step": 750 + }, + { + "epoch": 0.43, + "learning_rate": 1.914881623449831e-05, + "loss": 0.3738, + "step": 755 + }, + { + "epoch": 0.43, + "learning_rate": 1.914317925591883e-05, + "loss": 0.3636, + "step": 760 + }, + { + "epoch": 0.43, + "learning_rate": 1.913754227733935e-05, + "loss": 0.3636, + "step": 765 + }, + { + "epoch": 0.43, + "learning_rate": 1.9131905298759867e-05, + "loss": 0.3592, + "step": 770 + }, + { + "epoch": 0.44, + "learning_rate": 1.9126268320180385e-05, + "loss": 0.3461, + "step": 775 + }, + { + "epoch": 0.44, + "learning_rate": 1.9120631341600903e-05, + "loss": 0.3218, + "step": 780 + }, + { + "epoch": 0.44, + "learning_rate": 1.911499436302142e-05, + "loss": 0.3709, + "step": 785 + }, + { + "epoch": 0.45, + "learning_rate": 1.910935738444194e-05, + "loss": 0.3428, + "step": 790 + }, + { + "epoch": 0.45, + "learning_rate": 1.910372040586246e-05, + "loss": 0.345, + "step": 795 + }, + { + "epoch": 0.45, + "learning_rate": 1.9098083427282978e-05, + "loss": 0.3204, + "step": 800 + }, + { + "epoch": 0.45, + "eval_accuracy": 0.7997371367375748, + "eval_accuracy_sklearn": 0.7997371367375748, + "eval_f1": 0.8352158936153168, + "eval_loss": 0.4629848897457123, + "eval_precision": 0.7512609886150742, + "eval_recall": 0.9402958152958153, + "eval_runtime": 122.8837, + "eval_samples_per_second": 167.174, + "eval_steps_per_second": 10.449, + "step": 800 + }, + { + "epoch": 0.45, + "learning_rate": 1.9092446448703496e-05, + "loss": 0.3527, + "step": 805 + }, + { + "epoch": 0.46, + "learning_rate": 1.9086809470124014e-05, + "loss": 0.3307, + "step": 810 + }, + { + "epoch": 0.46, + "learning_rate": 1.9081172491544532e-05, + "loss": 0.3243, + "step": 815 + }, + { + "epoch": 0.46, + "learning_rate": 1.9075535512965053e-05, + "loss": 0.3445, + "step": 820 + }, + { + "epoch": 0.47, + "learning_rate": 1.906989853438557e-05, + "loss": 0.3386, + "step": 825 + }, + { + "epoch": 0.47, + "learning_rate": 1.906426155580609e-05, + "loss": 0.3477, + "step": 830 + }, + { + "epoch": 0.47, + "learning_rate": 1.9058624577226607e-05, + "loss": 0.3353, + "step": 835 + }, + { + "epoch": 0.47, + "learning_rate": 1.9052987598647125e-05, + "loss": 0.3448, + "step": 840 + }, + { + "epoch": 0.48, + "learning_rate": 1.9047350620067647e-05, + "loss": 0.3399, + "step": 845 + }, + { + "epoch": 0.48, + "learning_rate": 1.9041713641488165e-05, + "loss": 0.3454, + "step": 850 + }, + { + "epoch": 0.48, + "learning_rate": 1.9036076662908683e-05, + "loss": 0.3347, + "step": 855 + }, + { + "epoch": 0.48, + "learning_rate": 1.90304396843292e-05, + "loss": 0.3544, + "step": 860 + }, + { + "epoch": 0.49, + "learning_rate": 1.902480270574972e-05, + "loss": 0.3301, + "step": 865 + }, + { + "epoch": 0.49, + "learning_rate": 1.901916572717024e-05, + "loss": 0.3536, + "step": 870 + }, + { + "epoch": 0.49, + "learning_rate": 1.9013528748590758e-05, + "loss": 0.3279, + "step": 875 + }, + { + "epoch": 0.5, + "learning_rate": 1.9007891770011276e-05, + "loss": 0.3345, + "step": 880 + }, + { + "epoch": 0.5, + "learning_rate": 1.9002254791431794e-05, + "loss": 0.3359, + "step": 885 + }, + { + "epoch": 0.5, + "learning_rate": 1.8996617812852315e-05, + "loss": 0.3357, + "step": 890 + }, + { + "epoch": 0.5, + "learning_rate": 1.8990980834272833e-05, + "loss": 0.3293, + "step": 895 + }, + { + "epoch": 0.51, + "learning_rate": 1.898534385569335e-05, + "loss": 0.3392, + "step": 900 + }, + { + "epoch": 0.51, + "learning_rate": 1.897970687711387e-05, + "loss": 0.3213, + "step": 905 + }, + { + "epoch": 0.51, + "learning_rate": 1.8974069898534387e-05, + "loss": 0.3431, + "step": 910 + }, + { + "epoch": 0.52, + "learning_rate": 1.8968432919954905e-05, + "loss": 0.3349, + "step": 915 + }, + { + "epoch": 0.52, + "learning_rate": 1.8962795941375423e-05, + "loss": 0.343, + "step": 920 + }, + { + "epoch": 0.52, + "learning_rate": 1.8957158962795945e-05, + "loss": 0.3426, + "step": 925 + }, + { + "epoch": 0.52, + "learning_rate": 1.8951521984216463e-05, + "loss": 0.332, + "step": 930 + }, + { + "epoch": 0.53, + "learning_rate": 1.894588500563698e-05, + "loss": 0.3466, + "step": 935 + }, + { + "epoch": 0.53, + "learning_rate": 1.89402480270575e-05, + "loss": 0.3342, + "step": 940 + }, + { + "epoch": 0.53, + "learning_rate": 1.8934611048478016e-05, + "loss": 0.3338, + "step": 945 + }, + { + "epoch": 0.54, + "learning_rate": 1.8928974069898534e-05, + "loss": 0.3274, + "step": 950 + }, + { + "epoch": 0.54, + "learning_rate": 1.8923337091319052e-05, + "loss": 0.3111, + "step": 955 + }, + { + "epoch": 0.54, + "learning_rate": 1.891770011273957e-05, + "loss": 0.3202, + "step": 960 + }, + { + "epoch": 0.54, + "learning_rate": 1.8912063134160092e-05, + "loss": 0.3266, + "step": 965 + }, + { + "epoch": 0.55, + "learning_rate": 1.890642615558061e-05, + "loss": 0.3275, + "step": 970 + }, + { + "epoch": 0.55, + "learning_rate": 1.8900789177001128e-05, + "loss": 0.3273, + "step": 975 + }, + { + "epoch": 0.55, + "learning_rate": 1.8895152198421646e-05, + "loss": 0.3276, + "step": 980 + }, + { + "epoch": 0.56, + "learning_rate": 1.8889515219842167e-05, + "loss": 0.3287, + "step": 985 + }, + { + "epoch": 0.56, + "learning_rate": 1.8883878241262685e-05, + "loss": 0.3291, + "step": 990 + }, + { + "epoch": 0.56, + "learning_rate": 1.8878241262683203e-05, + "loss": 0.3305, + "step": 995 + }, + { + "epoch": 0.56, + "learning_rate": 1.887260428410372e-05, + "loss": 0.3202, + "step": 1000 + }, + { + "epoch": 0.56, + "eval_accuracy": 0.8104463807623035, + "eval_accuracy_sklearn": 0.8104463807623035, + "eval_f1": 0.841229715404061, + "eval_loss": 0.44823434948921204, + "eval_precision": 0.7676737609763358, + "eval_recall": 0.9303751803751804, + "eval_runtime": 118.5276, + "eval_samples_per_second": 173.318, + "eval_steps_per_second": 10.833, + "step": 1000 + }, + { + "epoch": 0.57, + "learning_rate": 1.886696730552424e-05, + "loss": 0.3336, + "step": 1005 + }, + { + "epoch": 0.57, + "learning_rate": 1.886133032694476e-05, + "loss": 0.359, + "step": 1010 + }, + { + "epoch": 0.57, + "learning_rate": 1.885569334836528e-05, + "loss": 0.3248, + "step": 1015 + }, + { + "epoch": 0.57, + "learning_rate": 1.8850056369785796e-05, + "loss": 0.3286, + "step": 1020 + }, + { + "epoch": 0.58, + "learning_rate": 1.8844419391206314e-05, + "loss": 0.3257, + "step": 1025 + }, + { + "epoch": 0.58, + "learning_rate": 1.8838782412626832e-05, + "loss": 0.3386, + "step": 1030 + }, + { + "epoch": 0.58, + "learning_rate": 1.8833145434047354e-05, + "loss": 0.3032, + "step": 1035 + }, + { + "epoch": 0.59, + "learning_rate": 1.8827508455467872e-05, + "loss": 0.3149, + "step": 1040 + }, + { + "epoch": 0.59, + "learning_rate": 1.882187147688839e-05, + "loss": 0.3368, + "step": 1045 + }, + { + "epoch": 0.59, + "learning_rate": 1.8816234498308908e-05, + "loss": 0.3387, + "step": 1050 + }, + { + "epoch": 0.59, + "learning_rate": 1.8810597519729426e-05, + "loss": 0.3333, + "step": 1055 + }, + { + "epoch": 0.6, + "learning_rate": 1.8804960541149947e-05, + "loss": 0.3262, + "step": 1060 + }, + { + "epoch": 0.6, + "learning_rate": 1.8799323562570465e-05, + "loss": 0.328, + "step": 1065 + }, + { + "epoch": 0.6, + "learning_rate": 1.8793686583990983e-05, + "loss": 0.3139, + "step": 1070 + }, + { + "epoch": 0.61, + "learning_rate": 1.87880496054115e-05, + "loss": 0.3407, + "step": 1075 + }, + { + "epoch": 0.61, + "learning_rate": 1.878241262683202e-05, + "loss": 0.3124, + "step": 1080 + }, + { + "epoch": 0.61, + "learning_rate": 1.8776775648252537e-05, + "loss": 0.3124, + "step": 1085 + }, + { + "epoch": 0.61, + "learning_rate": 1.877113866967306e-05, + "loss": 0.3095, + "step": 1090 + }, + { + "epoch": 0.62, + "learning_rate": 1.8765501691093576e-05, + "loss": 0.332, + "step": 1095 + }, + { + "epoch": 0.62, + "learning_rate": 1.8759864712514094e-05, + "loss": 0.3198, + "step": 1100 + }, + { + "epoch": 0.62, + "learning_rate": 1.8754227733934612e-05, + "loss": 0.3046, + "step": 1105 + }, + { + "epoch": 0.63, + "learning_rate": 1.874859075535513e-05, + "loss": 0.3293, + "step": 1110 + }, + { + "epoch": 0.63, + "learning_rate": 1.874295377677565e-05, + "loss": 0.3019, + "step": 1115 + }, + { + "epoch": 0.63, + "learning_rate": 1.8737316798196166e-05, + "loss": 0.3168, + "step": 1120 + }, + { + "epoch": 0.63, + "learning_rate": 1.8731679819616684e-05, + "loss": 0.3115, + "step": 1125 + }, + { + "epoch": 0.64, + "learning_rate": 1.8726042841037206e-05, + "loss": 0.3298, + "step": 1130 + }, + { + "epoch": 0.64, + "learning_rate": 1.8720405862457724e-05, + "loss": 0.3109, + "step": 1135 + }, + { + "epoch": 0.64, + "learning_rate": 1.8714768883878242e-05, + "loss": 0.3099, + "step": 1140 + }, + { + "epoch": 0.65, + "learning_rate": 1.870913190529876e-05, + "loss": 0.2989, + "step": 1145 + }, + { + "epoch": 0.65, + "learning_rate": 1.8703494926719278e-05, + "loss": 0.3151, + "step": 1150 + }, + { + "epoch": 0.65, + "learning_rate": 1.86978579481398e-05, + "loss": 0.3004, + "step": 1155 + }, + { + "epoch": 0.65, + "learning_rate": 1.8692220969560317e-05, + "loss": 0.3284, + "step": 1160 + }, + { + "epoch": 0.66, + "learning_rate": 1.8686583990980835e-05, + "loss": 0.3469, + "step": 1165 + }, + { + "epoch": 0.66, + "learning_rate": 1.8680947012401353e-05, + "loss": 0.31, + "step": 1170 + }, + { + "epoch": 0.66, + "learning_rate": 1.8675310033821874e-05, + "loss": 0.3045, + "step": 1175 + }, + { + "epoch": 0.67, + "learning_rate": 1.8669673055242392e-05, + "loss": 0.3037, + "step": 1180 + }, + { + "epoch": 0.67, + "learning_rate": 1.866403607666291e-05, + "loss": 0.3086, + "step": 1185 + }, + { + "epoch": 0.67, + "learning_rate": 1.865839909808343e-05, + "loss": 0.3238, + "step": 1190 + }, + { + "epoch": 0.67, + "learning_rate": 1.8652762119503946e-05, + "loss": 0.3052, + "step": 1195 + }, + { + "epoch": 0.68, + "learning_rate": 1.8647125140924468e-05, + "loss": 0.3162, + "step": 1200 + }, + { + "epoch": 0.68, + "eval_accuracy": 0.8080124616657742, + "eval_accuracy_sklearn": 0.8080124616657742, + "eval_f1": 0.8425171697811851, + "eval_loss": 0.42917490005493164, + "eval_precision": 0.7559472628260246, + "eval_recall": 0.9514790764790765, + "eval_runtime": 122.7201, + "eval_samples_per_second": 167.397, + "eval_steps_per_second": 10.463, + "step": 1200 + }, + { + "epoch": 0.68, + "learning_rate": 1.8641488162344986e-05, + "loss": 0.3082, + "step": 1205 + }, + { + "epoch": 0.68, + "learning_rate": 1.8635851183765504e-05, + "loss": 0.3214, + "step": 1210 + }, + { + "epoch": 0.68, + "learning_rate": 1.863021420518602e-05, + "loss": 0.3292, + "step": 1215 + }, + { + "epoch": 0.69, + "learning_rate": 1.862457722660654e-05, + "loss": 0.3147, + "step": 1220 + }, + { + "epoch": 0.69, + "learning_rate": 1.861894024802706e-05, + "loss": 0.3237, + "step": 1225 + }, + { + "epoch": 0.69, + "learning_rate": 1.861330326944758e-05, + "loss": 0.3124, + "step": 1230 + }, + { + "epoch": 0.7, + "learning_rate": 1.8607666290868097e-05, + "loss": 0.3054, + "step": 1235 + }, + { + "epoch": 0.7, + "learning_rate": 1.8602029312288615e-05, + "loss": 0.2998, + "step": 1240 + }, + { + "epoch": 0.7, + "learning_rate": 1.8596392333709133e-05, + "loss": 0.309, + "step": 1245 + }, + { + "epoch": 0.7, + "learning_rate": 1.8590755355129654e-05, + "loss": 0.3014, + "step": 1250 + }, + { + "epoch": 0.71, + "learning_rate": 1.8585118376550172e-05, + "loss": 0.3002, + "step": 1255 + }, + { + "epoch": 0.71, + "learning_rate": 1.857948139797069e-05, + "loss": 0.3084, + "step": 1260 + }, + { + "epoch": 0.71, + "learning_rate": 1.857384441939121e-05, + "loss": 0.3113, + "step": 1265 + }, + { + "epoch": 0.72, + "learning_rate": 1.8568207440811726e-05, + "loss": 0.3008, + "step": 1270 + }, + { + "epoch": 0.72, + "learning_rate": 1.8562570462232244e-05, + "loss": 0.2833, + "step": 1275 + }, + { + "epoch": 0.72, + "learning_rate": 1.8556933483652762e-05, + "loss": 0.3165, + "step": 1280 + }, + { + "epoch": 0.72, + "learning_rate": 1.855129650507328e-05, + "loss": 0.3063, + "step": 1285 + }, + { + "epoch": 0.73, + "learning_rate": 1.8545659526493798e-05, + "loss": 0.3101, + "step": 1290 + }, + { + "epoch": 0.73, + "learning_rate": 1.854002254791432e-05, + "loss": 0.2924, + "step": 1295 + }, + { + "epoch": 0.73, + "learning_rate": 1.8534385569334838e-05, + "loss": 0.3136, + "step": 1300 + }, + { + "epoch": 0.74, + "learning_rate": 1.8528748590755356e-05, + "loss": 0.3208, + "step": 1305 + }, + { + "epoch": 0.74, + "learning_rate": 1.8523111612175874e-05, + "loss": 0.3168, + "step": 1310 + }, + { + "epoch": 0.74, + "learning_rate": 1.851747463359639e-05, + "loss": 0.3281, + "step": 1315 + }, + { + "epoch": 0.74, + "learning_rate": 1.8511837655016913e-05, + "loss": 0.3242, + "step": 1320 + }, + { + "epoch": 0.75, + "learning_rate": 1.850620067643743e-05, + "loss": 0.3111, + "step": 1325 + }, + { + "epoch": 0.75, + "learning_rate": 1.850056369785795e-05, + "loss": 0.3, + "step": 1330 + }, + { + "epoch": 0.75, + "learning_rate": 1.8494926719278467e-05, + "loss": 0.3119, + "step": 1335 + }, + { + "epoch": 0.76, + "learning_rate": 1.848928974069899e-05, + "loss": 0.2969, + "step": 1340 + }, + { + "epoch": 0.76, + "learning_rate": 1.8483652762119506e-05, + "loss": 0.2937, + "step": 1345 + }, + { + "epoch": 0.76, + "learning_rate": 1.8478015783540024e-05, + "loss": 0.2845, + "step": 1350 + }, + { + "epoch": 0.76, + "learning_rate": 1.8472378804960542e-05, + "loss": 0.3171, + "step": 1355 + }, + { + "epoch": 0.77, + "learning_rate": 1.846674182638106e-05, + "loss": 0.3079, + "step": 1360 + }, + { + "epoch": 0.77, + "learning_rate": 1.846110484780158e-05, + "loss": 0.3153, + "step": 1365 + }, + { + "epoch": 0.77, + "learning_rate": 1.84554678692221e-05, + "loss": 0.293, + "step": 1370 + }, + { + "epoch": 0.78, + "learning_rate": 1.8449830890642618e-05, + "loss": 0.2984, + "step": 1375 + }, + { + "epoch": 0.78, + "learning_rate": 1.8444193912063136e-05, + "loss": 0.3061, + "step": 1380 + }, + { + "epoch": 0.78, + "learning_rate": 1.8438556933483654e-05, + "loss": 0.3045, + "step": 1385 + }, + { + "epoch": 0.78, + "learning_rate": 1.8432919954904175e-05, + "loss": 0.3025, + "step": 1390 + }, + { + "epoch": 0.79, + "learning_rate": 1.8427282976324693e-05, + "loss": 0.2929, + "step": 1395 + }, + { + "epoch": 0.79, + "learning_rate": 1.842164599774521e-05, + "loss": 0.2857, + "step": 1400 + }, + { + "epoch": 0.79, + "eval_accuracy": 0.8094728131236918, + "eval_accuracy_sklearn": 0.8094728131236918, + "eval_f1": 0.8447810913705583, + "eval_loss": 0.4755173325538635, + "eval_precision": 0.7538929784824462, + "eval_recall": 0.960588023088023, + "eval_runtime": 123.7982, + "eval_samples_per_second": 165.939, + "eval_steps_per_second": 10.372, + "step": 1400 + }, + { + "epoch": 0.79, + "learning_rate": 1.841600901916573e-05, + "loss": 0.3155, + "step": 1405 + }, + { + "epoch": 0.79, + "learning_rate": 1.8410372040586247e-05, + "loss": 0.2855, + "step": 1410 + }, + { + "epoch": 0.8, + "learning_rate": 1.8404735062006768e-05, + "loss": 0.2869, + "step": 1415 + }, + { + "epoch": 0.8, + "learning_rate": 1.8399098083427286e-05, + "loss": 0.2985, + "step": 1420 + }, + { + "epoch": 0.8, + "learning_rate": 1.8393461104847804e-05, + "loss": 0.3076, + "step": 1425 + }, + { + "epoch": 0.81, + "learning_rate": 1.8387824126268322e-05, + "loss": 0.301, + "step": 1430 + }, + { + "epoch": 0.81, + "learning_rate": 1.838218714768884e-05, + "loss": 0.2993, + "step": 1435 + }, + { + "epoch": 0.81, + "learning_rate": 1.8376550169109358e-05, + "loss": 0.2739, + "step": 1440 + }, + { + "epoch": 0.81, + "learning_rate": 1.8370913190529876e-05, + "loss": 0.3116, + "step": 1445 + }, + { + "epoch": 0.82, + "learning_rate": 1.8365276211950394e-05, + "loss": 0.2897, + "step": 1450 + }, + { + "epoch": 0.82, + "learning_rate": 1.8359639233370916e-05, + "loss": 0.2957, + "step": 1455 + }, + { + "epoch": 0.82, + "learning_rate": 1.8354002254791434e-05, + "loss": 0.3094, + "step": 1460 + }, + { + "epoch": 0.83, + "learning_rate": 1.834836527621195e-05, + "loss": 0.3021, + "step": 1465 + }, + { + "epoch": 0.83, + "learning_rate": 1.834272829763247e-05, + "loss": 0.2892, + "step": 1470 + }, + { + "epoch": 0.83, + "learning_rate": 1.8337091319052988e-05, + "loss": 0.2956, + "step": 1475 + }, + { + "epoch": 0.83, + "learning_rate": 1.8331454340473506e-05, + "loss": 0.275, + "step": 1480 + }, + { + "epoch": 0.84, + "learning_rate": 1.8325817361894027e-05, + "loss": 0.3228, + "step": 1485 + }, + { + "epoch": 0.84, + "learning_rate": 1.8320180383314545e-05, + "loss": 0.2644, + "step": 1490 + }, + { + "epoch": 0.84, + "learning_rate": 1.8314543404735063e-05, + "loss": 0.284, + "step": 1495 + }, + { + "epoch": 0.85, + "learning_rate": 1.830890642615558e-05, + "loss": 0.3003, + "step": 1500 + }, + { + "epoch": 0.85, + "learning_rate": 1.83032694475761e-05, + "loss": 0.309, + "step": 1505 + }, + { + "epoch": 0.85, + "learning_rate": 1.829763246899662e-05, + "loss": 0.2749, + "step": 1510 + }, + { + "epoch": 0.85, + "learning_rate": 1.8291995490417138e-05, + "loss": 0.3065, + "step": 1515 + }, + { + "epoch": 0.86, + "learning_rate": 1.8286358511837656e-05, + "loss": 0.3006, + "step": 1520 + }, + { + "epoch": 0.86, + "learning_rate": 1.8280721533258174e-05, + "loss": 0.3009, + "step": 1525 + }, + { + "epoch": 0.86, + "learning_rate": 1.8275084554678696e-05, + "loss": 0.3006, + "step": 1530 + }, + { + "epoch": 0.87, + "learning_rate": 1.8269447576099214e-05, + "loss": 0.298, + "step": 1535 + }, + { + "epoch": 0.87, + "learning_rate": 1.826381059751973e-05, + "loss": 0.2972, + "step": 1540 + }, + { + "epoch": 0.87, + "learning_rate": 1.825817361894025e-05, + "loss": 0.3049, + "step": 1545 + }, + { + "epoch": 0.87, + "learning_rate": 1.8252536640360768e-05, + "loss": 0.2988, + "step": 1550 + }, + { + "epoch": 0.88, + "learning_rate": 1.824689966178129e-05, + "loss": 0.2848, + "step": 1555 + }, + { + "epoch": 0.88, + "learning_rate": 1.8241262683201807e-05, + "loss": 0.3006, + "step": 1560 + }, + { + "epoch": 0.88, + "learning_rate": 1.8235625704622325e-05, + "loss": 0.2962, + "step": 1565 + }, + { + "epoch": 0.89, + "learning_rate": 1.8229988726042843e-05, + "loss": 0.2953, + "step": 1570 + }, + { + "epoch": 0.89, + "learning_rate": 1.822435174746336e-05, + "loss": 0.2874, + "step": 1575 + }, + { + "epoch": 0.89, + "learning_rate": 1.8218714768883882e-05, + "loss": 0.2939, + "step": 1580 + }, + { + "epoch": 0.89, + "learning_rate": 1.82130777903044e-05, + "loss": 0.279, + "step": 1585 + }, + { + "epoch": 0.9, + "learning_rate": 1.8207440811724918e-05, + "loss": 0.2726, + "step": 1590 + }, + { + "epoch": 0.9, + "learning_rate": 1.8201803833145436e-05, + "loss": 0.2972, + "step": 1595 + }, + { + "epoch": 0.9, + "learning_rate": 1.8196166854565954e-05, + "loss": 0.2895, + "step": 1600 + }, + { + "epoch": 0.9, + "eval_accuracy": 0.8567395219782894, + "eval_accuracy_sklearn": 0.8567395219782894, + "eval_f1": 0.873966853667937, + "eval_loss": 0.3318902552127838, + "eval_precision": 0.8320965505993639, + "eval_recall": 0.9202741702741702, + "eval_runtime": 122.6077, + "eval_samples_per_second": 167.551, + "eval_steps_per_second": 10.472, + "step": 1600 + }, + { + "epoch": 0.9, + "learning_rate": 1.8190529875986472e-05, + "loss": 0.2886, + "step": 1605 + }, + { + "epoch": 0.91, + "learning_rate": 1.818489289740699e-05, + "loss": 0.2684, + "step": 1610 + }, + { + "epoch": 0.91, + "learning_rate": 1.8179255918827508e-05, + "loss": 0.2792, + "step": 1615 + }, + { + "epoch": 0.91, + "learning_rate": 1.817361894024803e-05, + "loss": 0.2817, + "step": 1620 + }, + { + "epoch": 0.92, + "learning_rate": 1.8167981961668547e-05, + "loss": 0.281, + "step": 1625 + }, + { + "epoch": 0.92, + "learning_rate": 1.8162344983089065e-05, + "loss": 0.2987, + "step": 1630 + }, + { + "epoch": 0.92, + "learning_rate": 1.8156708004509583e-05, + "loss": 0.2992, + "step": 1635 + }, + { + "epoch": 0.92, + "learning_rate": 1.81510710259301e-05, + "loss": 0.2869, + "step": 1640 + }, + { + "epoch": 0.93, + "learning_rate": 1.814543404735062e-05, + "loss": 0.2774, + "step": 1645 + }, + { + "epoch": 0.93, + "learning_rate": 1.813979706877114e-05, + "loss": 0.2876, + "step": 1650 + }, + { + "epoch": 0.93, + "learning_rate": 1.813416009019166e-05, + "loss": 0.2845, + "step": 1655 + }, + { + "epoch": 0.94, + "learning_rate": 1.8128523111612177e-05, + "loss": 0.2774, + "step": 1660 + }, + { + "epoch": 0.94, + "learning_rate": 1.8122886133032695e-05, + "loss": 0.2834, + "step": 1665 + }, + { + "epoch": 0.94, + "learning_rate": 1.8117249154453213e-05, + "loss": 0.2868, + "step": 1670 + }, + { + "epoch": 0.94, + "learning_rate": 1.8111612175873734e-05, + "loss": 0.2737, + "step": 1675 + }, + { + "epoch": 0.95, + "learning_rate": 1.8105975197294252e-05, + "loss": 0.2906, + "step": 1680 + }, + { + "epoch": 0.95, + "learning_rate": 1.810033821871477e-05, + "loss": 0.2897, + "step": 1685 + }, + { + "epoch": 0.95, + "learning_rate": 1.8094701240135288e-05, + "loss": 0.2814, + "step": 1690 + }, + { + "epoch": 0.96, + "learning_rate": 1.8089064261555806e-05, + "loss": 0.2856, + "step": 1695 + }, + { + "epoch": 0.96, + "learning_rate": 1.8083427282976327e-05, + "loss": 0.2939, + "step": 1700 + }, + { + "epoch": 0.96, + "learning_rate": 1.8077790304396845e-05, + "loss": 0.2705, + "step": 1705 + }, + { + "epoch": 0.96, + "learning_rate": 1.8072153325817363e-05, + "loss": 0.2853, + "step": 1710 + }, + { + "epoch": 0.97, + "learning_rate": 1.806651634723788e-05, + "loss": 0.2883, + "step": 1715 + }, + { + "epoch": 0.97, + "learning_rate": 1.8060879368658403e-05, + "loss": 0.2653, + "step": 1720 + }, + { + "epoch": 0.97, + "learning_rate": 1.805524239007892e-05, + "loss": 0.2732, + "step": 1725 + }, + { + "epoch": 0.98, + "learning_rate": 1.804960541149944e-05, + "loss": 0.2936, + "step": 1730 + }, + { + "epoch": 0.98, + "learning_rate": 1.8043968432919957e-05, + "loss": 0.2873, + "step": 1735 + }, + { + "epoch": 0.98, + "learning_rate": 1.8038331454340475e-05, + "loss": 0.28, + "step": 1740 + }, + { + "epoch": 0.98, + "learning_rate": 1.8032694475760996e-05, + "loss": 0.289, + "step": 1745 + }, + { + "epoch": 0.99, + "learning_rate": 1.8027057497181514e-05, + "loss": 0.2775, + "step": 1750 + }, + { + "epoch": 0.99, + "learning_rate": 1.8021420518602032e-05, + "loss": 0.2803, + "step": 1755 + }, + { + "epoch": 0.99, + "learning_rate": 1.801578354002255e-05, + "loss": 0.2924, + "step": 1760 + }, + { + "epoch": 0.99, + "learning_rate": 1.8010146561443068e-05, + "loss": 0.2728, + "step": 1765 + }, + { + "epoch": 1.0, + "learning_rate": 1.8004509582863586e-05, + "loss": 0.2782, + "step": 1770 + }, + { + "epoch": 1.0, + "learning_rate": 1.7998872604284104e-05, + "loss": 0.2585, + "step": 1775 + }, + { + "epoch": 1.0, + "learning_rate": 1.7993235625704622e-05, + "loss": 0.2767, + "step": 1780 + }, + { + "epoch": 1.01, + "learning_rate": 1.7987598647125143e-05, + "loss": 0.2651, + "step": 1785 + }, + { + "epoch": 1.01, + "learning_rate": 1.798196166854566e-05, + "loss": 0.2823, + "step": 1790 + }, + { + "epoch": 1.01, + "learning_rate": 1.797632468996618e-05, + "loss": 0.279, + "step": 1795 + }, + { + "epoch": 1.01, + "learning_rate": 1.7970687711386697e-05, + "loss": 0.2658, + "step": 1800 + }, + { + "epoch": 1.01, + "eval_accuracy": 0.8598062600399162, + "eval_accuracy_sklearn": 0.8598062600399162, + "eval_f1": 0.8798798798798798, + "eval_loss": 0.32673802971839905, + "eval_precision": 0.8184357541899442, + "eval_recall": 0.9512987012987013, + "eval_runtime": 124.2957, + "eval_samples_per_second": 165.275, + "eval_steps_per_second": 10.33, + "step": 1800 + }, + { + "epoch": 1.02, + "learning_rate": 1.7965050732807215e-05, + "loss": 0.2493, + "step": 1805 + }, + { + "epoch": 1.02, + "learning_rate": 1.7959413754227733e-05, + "loss": 0.2688, + "step": 1810 + }, + { + "epoch": 1.02, + "learning_rate": 1.7953776775648255e-05, + "loss": 0.2477, + "step": 1815 + }, + { + "epoch": 1.03, + "learning_rate": 1.7948139797068773e-05, + "loss": 0.2589, + "step": 1820 + }, + { + "epoch": 1.03, + "learning_rate": 1.794250281848929e-05, + "loss": 0.244, + "step": 1825 + }, + { + "epoch": 1.03, + "learning_rate": 1.793686583990981e-05, + "loss": 0.2779, + "step": 1830 + }, + { + "epoch": 1.03, + "learning_rate": 1.7931228861330327e-05, + "loss": 0.2559, + "step": 1835 + }, + { + "epoch": 1.04, + "learning_rate": 1.7925591882750848e-05, + "loss": 0.2595, + "step": 1840 + }, + { + "epoch": 1.04, + "learning_rate": 1.7919954904171366e-05, + "loss": 0.2536, + "step": 1845 + }, + { + "epoch": 1.04, + "learning_rate": 1.7914317925591884e-05, + "loss": 0.2785, + "step": 1850 + }, + { + "epoch": 1.05, + "learning_rate": 1.7908680947012402e-05, + "loss": 0.2525, + "step": 1855 + }, + { + "epoch": 1.05, + "learning_rate": 1.790304396843292e-05, + "loss": 0.2586, + "step": 1860 + }, + { + "epoch": 1.05, + "learning_rate": 1.789740698985344e-05, + "loss": 0.2629, + "step": 1865 + }, + { + "epoch": 1.05, + "learning_rate": 1.789177001127396e-05, + "loss": 0.2471, + "step": 1870 + }, + { + "epoch": 1.06, + "learning_rate": 1.7886133032694477e-05, + "loss": 0.2781, + "step": 1875 + }, + { + "epoch": 1.06, + "learning_rate": 1.7880496054114995e-05, + "loss": 0.2499, + "step": 1880 + }, + { + "epoch": 1.06, + "learning_rate": 1.7874859075535513e-05, + "loss": 0.2687, + "step": 1885 + }, + { + "epoch": 1.07, + "learning_rate": 1.7869222096956035e-05, + "loss": 0.268, + "step": 1890 + }, + { + "epoch": 1.07, + "learning_rate": 1.7863585118376553e-05, + "loss": 0.28, + "step": 1895 + }, + { + "epoch": 1.07, + "learning_rate": 1.785794813979707e-05, + "loss": 0.2611, + "step": 1900 + }, + { + "epoch": 1.07, + "learning_rate": 1.785231116121759e-05, + "loss": 0.2604, + "step": 1905 + }, + { + "epoch": 1.08, + "learning_rate": 1.784667418263811e-05, + "loss": 0.2661, + "step": 1910 + }, + { + "epoch": 1.08, + "learning_rate": 1.7841037204058628e-05, + "loss": 0.2469, + "step": 1915 + }, + { + "epoch": 1.08, + "learning_rate": 1.7835400225479146e-05, + "loss": 0.2618, + "step": 1920 + }, + { + "epoch": 1.09, + "learning_rate": 1.7829763246899664e-05, + "loss": 0.2434, + "step": 1925 + }, + { + "epoch": 1.09, + "learning_rate": 1.7824126268320182e-05, + "loss": 0.2736, + "step": 1930 + }, + { + "epoch": 1.09, + "learning_rate": 1.78184892897407e-05, + "loss": 0.2607, + "step": 1935 + }, + { + "epoch": 1.09, + "learning_rate": 1.7812852311161218e-05, + "loss": 0.2772, + "step": 1940 + }, + { + "epoch": 1.1, + "learning_rate": 1.7807215332581736e-05, + "loss": 0.2546, + "step": 1945 + }, + { + "epoch": 1.1, + "learning_rate": 1.7801578354002257e-05, + "loss": 0.2434, + "step": 1950 + }, + { + "epoch": 1.1, + "learning_rate": 1.7795941375422775e-05, + "loss": 0.2621, + "step": 1955 + }, + { + "epoch": 1.1, + "learning_rate": 1.7790304396843293e-05, + "loss": 0.2635, + "step": 1960 + }, + { + "epoch": 1.11, + "learning_rate": 1.778466741826381e-05, + "loss": 0.2634, + "step": 1965 + }, + { + "epoch": 1.11, + "learning_rate": 1.777903043968433e-05, + "loss": 0.2593, + "step": 1970 + }, + { + "epoch": 1.11, + "learning_rate": 1.7773393461104847e-05, + "loss": 0.2772, + "step": 1975 + }, + { + "epoch": 1.12, + "learning_rate": 1.7767756482525365e-05, + "loss": 0.2519, + "step": 1980 + }, + { + "epoch": 1.12, + "learning_rate": 1.7762119503945887e-05, + "loss": 0.2646, + "step": 1985 + }, + { + "epoch": 1.12, + "learning_rate": 1.7756482525366405e-05, + "loss": 0.2638, + "step": 1990 + }, + { + "epoch": 1.12, + "learning_rate": 1.7750845546786923e-05, + "loss": 0.2571, + "step": 1995 + }, + { + "epoch": 1.13, + "learning_rate": 1.774520856820744e-05, + "loss": 0.2564, + "step": 2000 + }, + { + "epoch": 1.13, + "eval_accuracy": 0.8555712408119555, + "eval_accuracy_sklearn": 0.8555712408119555, + "eval_f1": 0.8777049585754916, + "eval_loss": 0.350249707698822, + "eval_precision": 0.8082441357321795, + "eval_recall": 0.9602272727272727, + "eval_runtime": 123.1848, + "eval_samples_per_second": 166.766, + "eval_steps_per_second": 10.423, + "step": 2000 + }, + { + "epoch": 1.13, + "learning_rate": 1.7739571589627962e-05, + "loss": 0.2456, + "step": 2005 + }, + { + "epoch": 1.13, + "learning_rate": 1.773393461104848e-05, + "loss": 0.2568, + "step": 2010 + }, + { + "epoch": 1.14, + "learning_rate": 1.7728297632468998e-05, + "loss": 0.2454, + "step": 2015 + }, + { + "epoch": 1.14, + "learning_rate": 1.7722660653889516e-05, + "loss": 0.2641, + "step": 2020 + }, + { + "epoch": 1.14, + "learning_rate": 1.7717023675310034e-05, + "loss": 0.2571, + "step": 2025 + }, + { + "epoch": 1.14, + "learning_rate": 1.7711386696730555e-05, + "loss": 0.2707, + "step": 2030 + }, + { + "epoch": 1.15, + "learning_rate": 1.7705749718151073e-05, + "loss": 0.2477, + "step": 2035 + }, + { + "epoch": 1.15, + "learning_rate": 1.770011273957159e-05, + "loss": 0.2529, + "step": 2040 + }, + { + "epoch": 1.15, + "learning_rate": 1.769447576099211e-05, + "loss": 0.2504, + "step": 2045 + }, + { + "epoch": 1.16, + "learning_rate": 1.7688838782412627e-05, + "loss": 0.2594, + "step": 2050 + }, + { + "epoch": 1.16, + "learning_rate": 1.768320180383315e-05, + "loss": 0.2464, + "step": 2055 + }, + { + "epoch": 1.16, + "learning_rate": 1.7677564825253667e-05, + "loss": 0.2458, + "step": 2060 + }, + { + "epoch": 1.16, + "learning_rate": 1.7671927846674185e-05, + "loss": 0.24, + "step": 2065 + }, + { + "epoch": 1.17, + "learning_rate": 1.7666290868094703e-05, + "loss": 0.2496, + "step": 2070 + }, + { + "epoch": 1.17, + "learning_rate": 1.766065388951522e-05, + "loss": 0.2454, + "step": 2075 + }, + { + "epoch": 1.17, + "learning_rate": 1.7655016910935742e-05, + "loss": 0.2501, + "step": 2080 + }, + { + "epoch": 1.18, + "learning_rate": 1.764937993235626e-05, + "loss": 0.2333, + "step": 2085 + }, + { + "epoch": 1.18, + "learning_rate": 1.7643742953776778e-05, + "loss": 0.2624, + "step": 2090 + }, + { + "epoch": 1.18, + "learning_rate": 1.7638105975197296e-05, + "loss": 0.2439, + "step": 2095 + }, + { + "epoch": 1.18, + "learning_rate": 1.7632468996617814e-05, + "loss": 0.2492, + "step": 2100 + }, + { + "epoch": 1.19, + "learning_rate": 1.7626832018038332e-05, + "loss": 0.2589, + "step": 2105 + }, + { + "epoch": 1.19, + "learning_rate": 1.7621195039458853e-05, + "loss": 0.261, + "step": 2110 + }, + { + "epoch": 1.19, + "learning_rate": 1.761555806087937e-05, + "loss": 0.2605, + "step": 2115 + }, + { + "epoch": 1.2, + "learning_rate": 1.760992108229989e-05, + "loss": 0.2631, + "step": 2120 + }, + { + "epoch": 1.2, + "learning_rate": 1.7604284103720407e-05, + "loss": 0.2587, + "step": 2125 + }, + { + "epoch": 1.2, + "learning_rate": 1.7598647125140925e-05, + "loss": 0.2701, + "step": 2130 + }, + { + "epoch": 1.2, + "learning_rate": 1.7593010146561443e-05, + "loss": 0.2629, + "step": 2135 + }, + { + "epoch": 1.21, + "learning_rate": 1.758737316798196e-05, + "loss": 0.2534, + "step": 2140 + }, + { + "epoch": 1.21, + "learning_rate": 1.758173618940248e-05, + "loss": 0.2504, + "step": 2145 + }, + { + "epoch": 1.21, + "learning_rate": 1.7576099210823e-05, + "loss": 0.2585, + "step": 2150 + }, + { + "epoch": 1.21, + "learning_rate": 1.757046223224352e-05, + "loss": 0.2559, + "step": 2155 + }, + { + "epoch": 1.22, + "learning_rate": 1.7564825253664037e-05, + "loss": 0.2692, + "step": 2160 + }, + { + "epoch": 1.22, + "learning_rate": 1.7559188275084555e-05, + "loss": 0.26, + "step": 2165 + }, + { + "epoch": 1.22, + "learning_rate": 1.7553551296505073e-05, + "loss": 0.2602, + "step": 2170 + }, + { + "epoch": 1.23, + "learning_rate": 1.7547914317925594e-05, + "loss": 0.2633, + "step": 2175 + }, + { + "epoch": 1.23, + "learning_rate": 1.7542277339346112e-05, + "loss": 0.2494, + "step": 2180 + }, + { + "epoch": 1.23, + "learning_rate": 1.753664036076663e-05, + "loss": 0.2581, + "step": 2185 + }, + { + "epoch": 1.23, + "learning_rate": 1.7531003382187148e-05, + "loss": 0.2302, + "step": 2190 + }, + { + "epoch": 1.24, + "learning_rate": 1.752536640360767e-05, + "loss": 0.2364, + "step": 2195 + }, + { + "epoch": 1.24, + "learning_rate": 1.7519729425028187e-05, + "loss": 0.2384, + "step": 2200 + }, + { + "epoch": 1.24, + "eval_accuracy": 0.8757240909312174, + "eval_accuracy_sklearn": 0.8757240909312174, + "eval_f1": 0.8924735711578149, + "eval_loss": 0.30816930532455444, + "eval_precision": 0.8372184907151323, + "eval_recall": 0.9555375180375181, + "eval_runtime": 122.6824, + "eval_samples_per_second": 167.449, + "eval_steps_per_second": 10.466, + "step": 2200 + }, + { + "epoch": 1.24, + "learning_rate": 1.7514092446448705e-05, + "loss": 0.2734, + "step": 2205 + }, + { + "epoch": 1.25, + "learning_rate": 1.7508455467869223e-05, + "loss": 0.2515, + "step": 2210 + }, + { + "epoch": 1.25, + "learning_rate": 1.750281848928974e-05, + "loss": 0.2371, + "step": 2215 + }, + { + "epoch": 1.25, + "learning_rate": 1.7497181510710263e-05, + "loss": 0.2411, + "step": 2220 + }, + { + "epoch": 1.25, + "learning_rate": 1.749154453213078e-05, + "loss": 0.2704, + "step": 2225 + }, + { + "epoch": 1.26, + "learning_rate": 1.74859075535513e-05, + "loss": 0.2487, + "step": 2230 + }, + { + "epoch": 1.26, + "learning_rate": 1.7480270574971817e-05, + "loss": 0.2484, + "step": 2235 + }, + { + "epoch": 1.26, + "learning_rate": 1.7474633596392334e-05, + "loss": 0.2499, + "step": 2240 + }, + { + "epoch": 1.27, + "learning_rate": 1.7468996617812856e-05, + "loss": 0.2424, + "step": 2245 + }, + { + "epoch": 1.27, + "learning_rate": 1.7463359639233374e-05, + "loss": 0.2464, + "step": 2250 + }, + { + "epoch": 1.27, + "learning_rate": 1.7457722660653892e-05, + "loss": 0.2438, + "step": 2255 + }, + { + "epoch": 1.27, + "learning_rate": 1.745208568207441e-05, + "loss": 0.23, + "step": 2260 + }, + { + "epoch": 1.28, + "learning_rate": 1.7446448703494928e-05, + "loss": 0.2511, + "step": 2265 + }, + { + "epoch": 1.28, + "learning_rate": 1.7440811724915446e-05, + "loss": 0.2501, + "step": 2270 + }, + { + "epoch": 1.28, + "learning_rate": 1.7435174746335967e-05, + "loss": 0.2358, + "step": 2275 + }, + { + "epoch": 1.29, + "learning_rate": 1.7429537767756485e-05, + "loss": 0.2734, + "step": 2280 + }, + { + "epoch": 1.29, + "learning_rate": 1.7423900789177003e-05, + "loss": 0.2316, + "step": 2285 + }, + { + "epoch": 1.29, + "learning_rate": 1.741826381059752e-05, + "loss": 0.2282, + "step": 2290 + }, + { + "epoch": 1.29, + "learning_rate": 1.741262683201804e-05, + "loss": 0.2533, + "step": 2295 + }, + { + "epoch": 1.3, + "learning_rate": 1.7406989853438557e-05, + "loss": 0.251, + "step": 2300 + }, + { + "epoch": 1.3, + "learning_rate": 1.7401352874859075e-05, + "loss": 0.2358, + "step": 2305 + }, + { + "epoch": 1.3, + "learning_rate": 1.7395715896279593e-05, + "loss": 0.2412, + "step": 2310 + }, + { + "epoch": 1.3, + "learning_rate": 1.7390078917700114e-05, + "loss": 0.2685, + "step": 2315 + }, + { + "epoch": 1.31, + "learning_rate": 1.7384441939120632e-05, + "loss": 0.2508, + "step": 2320 + }, + { + "epoch": 1.31, + "learning_rate": 1.737880496054115e-05, + "loss": 0.2437, + "step": 2325 + }, + { + "epoch": 1.31, + "learning_rate": 1.737316798196167e-05, + "loss": 0.2283, + "step": 2330 + }, + { + "epoch": 1.32, + "learning_rate": 1.7367531003382186e-05, + "loss": 0.2398, + "step": 2335 + }, + { + "epoch": 1.32, + "learning_rate": 1.7361894024802708e-05, + "loss": 0.2574, + "step": 2340 + }, + { + "epoch": 1.32, + "learning_rate": 1.7356257046223226e-05, + "loss": 0.2403, + "step": 2345 + }, + { + "epoch": 1.32, + "learning_rate": 1.7350620067643744e-05, + "loss": 0.2504, + "step": 2350 + }, + { + "epoch": 1.33, + "learning_rate": 1.7344983089064262e-05, + "loss": 0.246, + "step": 2355 + }, + { + "epoch": 1.33, + "learning_rate": 1.7339346110484783e-05, + "loss": 0.2392, + "step": 2360 + }, + { + "epoch": 1.33, + "learning_rate": 1.73337091319053e-05, + "loss": 0.2405, + "step": 2365 + }, + { + "epoch": 1.34, + "learning_rate": 1.732807215332582e-05, + "loss": 0.2395, + "step": 2370 + }, + { + "epoch": 1.34, + "learning_rate": 1.7322435174746337e-05, + "loss": 0.2464, + "step": 2375 + }, + { + "epoch": 1.34, + "learning_rate": 1.7316798196166855e-05, + "loss": 0.2432, + "step": 2380 + }, + { + "epoch": 1.34, + "learning_rate": 1.7311161217587376e-05, + "loss": 0.2482, + "step": 2385 + }, + { + "epoch": 1.35, + "learning_rate": 1.7305524239007894e-05, + "loss": 0.2583, + "step": 2390 + }, + { + "epoch": 1.35, + "learning_rate": 1.7299887260428412e-05, + "loss": 0.2541, + "step": 2395 + }, + { + "epoch": 1.35, + "learning_rate": 1.729425028184893e-05, + "loss": 0.2402, + "step": 2400 + }, + { + "epoch": 1.35, + "eval_accuracy": 0.8860925862824319, + "eval_accuracy_sklearn": 0.8860925862824319, + "eval_f1": 0.9008222429431211, + "eval_loss": 0.27121710777282715, + "eval_precision": 0.8497521189828882, + "eval_recall": 0.9584235209235209, + "eval_runtime": 121.5197, + "eval_samples_per_second": 169.051, + "eval_steps_per_second": 10.566, + "step": 2400 + }, + { + "epoch": 1.36, + "learning_rate": 1.728861330326945e-05, + "loss": 0.256, + "step": 2405 + }, + { + "epoch": 1.36, + "learning_rate": 1.728297632468997e-05, + "loss": 0.2284, + "step": 2410 + }, + { + "epoch": 1.36, + "learning_rate": 1.7277339346110488e-05, + "loss": 0.2192, + "step": 2415 + }, + { + "epoch": 1.36, + "learning_rate": 1.7271702367531006e-05, + "loss": 0.2487, + "step": 2420 + }, + { + "epoch": 1.37, + "learning_rate": 1.7266065388951524e-05, + "loss": 0.2285, + "step": 2425 + }, + { + "epoch": 1.37, + "learning_rate": 1.7260428410372042e-05, + "loss": 0.234, + "step": 2430 + }, + { + "epoch": 1.37, + "learning_rate": 1.725479143179256e-05, + "loss": 0.2347, + "step": 2435 + }, + { + "epoch": 1.38, + "learning_rate": 1.724915445321308e-05, + "loss": 0.2382, + "step": 2440 + }, + { + "epoch": 1.38, + "learning_rate": 1.72435174746336e-05, + "loss": 0.2581, + "step": 2445 + }, + { + "epoch": 1.38, + "learning_rate": 1.7237880496054117e-05, + "loss": 0.2271, + "step": 2450 + }, + { + "epoch": 1.38, + "learning_rate": 1.7232243517474635e-05, + "loss": 0.2374, + "step": 2455 + }, + { + "epoch": 1.39, + "learning_rate": 1.7226606538895153e-05, + "loss": 0.2346, + "step": 2460 + }, + { + "epoch": 1.39, + "learning_rate": 1.722096956031567e-05, + "loss": 0.2613, + "step": 2465 + }, + { + "epoch": 1.39, + "learning_rate": 1.721533258173619e-05, + "loss": 0.2624, + "step": 2470 + }, + { + "epoch": 1.4, + "learning_rate": 1.7209695603156707e-05, + "loss": 0.219, + "step": 2475 + }, + { + "epoch": 1.4, + "learning_rate": 1.720405862457723e-05, + "loss": 0.2308, + "step": 2480 + }, + { + "epoch": 1.4, + "learning_rate": 1.7198421645997746e-05, + "loss": 0.2439, + "step": 2485 + }, + { + "epoch": 1.4, + "learning_rate": 1.7192784667418264e-05, + "loss": 0.2614, + "step": 2490 + }, + { + "epoch": 1.41, + "learning_rate": 1.7187147688838782e-05, + "loss": 0.2528, + "step": 2495 + }, + { + "epoch": 1.41, + "learning_rate": 1.71815107102593e-05, + "loss": 0.2454, + "step": 2500 + }, + { + "epoch": 1.41, + "learning_rate": 1.7175873731679822e-05, + "loss": 0.2288, + "step": 2505 + }, + { + "epoch": 1.41, + "learning_rate": 1.717023675310034e-05, + "loss": 0.2504, + "step": 2510 + }, + { + "epoch": 1.42, + "learning_rate": 1.7164599774520858e-05, + "loss": 0.2407, + "step": 2515 + }, + { + "epoch": 1.42, + "learning_rate": 1.7158962795941376e-05, + "loss": 0.2436, + "step": 2520 + }, + { + "epoch": 1.42, + "learning_rate": 1.7153325817361894e-05, + "loss": 0.229, + "step": 2525 + }, + { + "epoch": 1.43, + "learning_rate": 1.7147688838782415e-05, + "loss": 0.2377, + "step": 2530 + }, + { + "epoch": 1.43, + "learning_rate": 1.7142051860202933e-05, + "loss": 0.2402, + "step": 2535 + }, + { + "epoch": 1.43, + "learning_rate": 1.713641488162345e-05, + "loss": 0.2441, + "step": 2540 + }, + { + "epoch": 1.43, + "learning_rate": 1.713077790304397e-05, + "loss": 0.2452, + "step": 2545 + }, + { + "epoch": 1.44, + "learning_rate": 1.712514092446449e-05, + "loss": 0.2353, + "step": 2550 + }, + { + "epoch": 1.44, + "learning_rate": 1.711950394588501e-05, + "loss": 0.2478, + "step": 2555 + }, + { + "epoch": 1.44, + "learning_rate": 1.7113866967305526e-05, + "loss": 0.2508, + "step": 2560 + }, + { + "epoch": 1.45, + "learning_rate": 1.7108229988726044e-05, + "loss": 0.2623, + "step": 2565 + }, + { + "epoch": 1.45, + "learning_rate": 1.7102593010146562e-05, + "loss": 0.2588, + "step": 2570 + }, + { + "epoch": 1.45, + "learning_rate": 1.7096956031567084e-05, + "loss": 0.2397, + "step": 2575 + }, + { + "epoch": 1.45, + "learning_rate": 1.70913190529876e-05, + "loss": 0.2183, + "step": 2580 + }, + { + "epoch": 1.46, + "learning_rate": 1.708568207440812e-05, + "loss": 0.24, + "step": 2585 + }, + { + "epoch": 1.46, + "learning_rate": 1.7080045095828638e-05, + "loss": 0.2461, + "step": 2590 + }, + { + "epoch": 1.46, + "learning_rate": 1.7074408117249156e-05, + "loss": 0.2461, + "step": 2595 + }, + { + "epoch": 1.47, + "learning_rate": 1.7068771138669674e-05, + "loss": 0.2109, + "step": 2600 + }, + { + "epoch": 1.47, + "eval_accuracy": 0.8955361923769654, + "eval_accuracy_sklearn": 0.8955361923769654, + "eval_f1": 0.908501748102669, + "eval_loss": 0.2583935260772705, + "eval_precision": 0.8615558790231279, + "eval_recall": 0.9608585858585859, + "eval_runtime": 120.9295, + "eval_samples_per_second": 169.876, + "eval_steps_per_second": 10.618, + "step": 2600 + }, + { + "epoch": 1.47, + "learning_rate": 1.7063134160090195e-05, + "loss": 0.2278, + "step": 2605 + }, + { + "epoch": 1.47, + "learning_rate": 1.7057497181510713e-05, + "loss": 0.2465, + "step": 2610 + }, + { + "epoch": 1.47, + "learning_rate": 1.705186020293123e-05, + "loss": 0.2417, + "step": 2615 + }, + { + "epoch": 1.48, + "learning_rate": 1.704622322435175e-05, + "loss": 0.2226, + "step": 2620 + }, + { + "epoch": 1.48, + "learning_rate": 1.7040586245772267e-05, + "loss": 0.2404, + "step": 2625 + }, + { + "epoch": 1.48, + "learning_rate": 1.7034949267192785e-05, + "loss": 0.2433, + "step": 2630 + }, + { + "epoch": 1.49, + "learning_rate": 1.7029312288613303e-05, + "loss": 0.2266, + "step": 2635 + }, + { + "epoch": 1.49, + "learning_rate": 1.702367531003382e-05, + "loss": 0.2279, + "step": 2640 + }, + { + "epoch": 1.49, + "learning_rate": 1.7018038331454342e-05, + "loss": 0.2525, + "step": 2645 + }, + { + "epoch": 1.49, + "learning_rate": 1.701240135287486e-05, + "loss": 0.2074, + "step": 2650 + }, + { + "epoch": 1.5, + "learning_rate": 1.7006764374295378e-05, + "loss": 0.2459, + "step": 2655 + }, + { + "epoch": 1.5, + "learning_rate": 1.7001127395715896e-05, + "loss": 0.2161, + "step": 2660 + }, + { + "epoch": 1.5, + "learning_rate": 1.6995490417136414e-05, + "loss": 0.2401, + "step": 2665 + }, + { + "epoch": 1.51, + "learning_rate": 1.6989853438556936e-05, + "loss": 0.2142, + "step": 2670 + }, + { + "epoch": 1.51, + "learning_rate": 1.6984216459977454e-05, + "loss": 0.2341, + "step": 2675 + }, + { + "epoch": 1.51, + "learning_rate": 1.697857948139797e-05, + "loss": 0.2403, + "step": 2680 + }, + { + "epoch": 1.51, + "learning_rate": 1.697294250281849e-05, + "loss": 0.2376, + "step": 2685 + }, + { + "epoch": 1.52, + "learning_rate": 1.6967305524239008e-05, + "loss": 0.2368, + "step": 2690 + }, + { + "epoch": 1.52, + "learning_rate": 1.696166854565953e-05, + "loss": 0.2465, + "step": 2695 + }, + { + "epoch": 1.52, + "learning_rate": 1.6956031567080047e-05, + "loss": 0.2106, + "step": 2700 + }, + { + "epoch": 1.52, + "learning_rate": 1.6950394588500565e-05, + "loss": 0.2269, + "step": 2705 + }, + { + "epoch": 1.53, + "learning_rate": 1.6944757609921083e-05, + "loss": 0.228, + "step": 2710 + }, + { + "epoch": 1.53, + "learning_rate": 1.69391206313416e-05, + "loss": 0.2355, + "step": 2715 + }, + { + "epoch": 1.53, + "learning_rate": 1.6933483652762122e-05, + "loss": 0.2141, + "step": 2720 + }, + { + "epoch": 1.54, + "learning_rate": 1.692784667418264e-05, + "loss": 0.223, + "step": 2725 + }, + { + "epoch": 1.54, + "learning_rate": 1.6922209695603158e-05, + "loss": 0.2165, + "step": 2730 + }, + { + "epoch": 1.54, + "learning_rate": 1.6916572717023676e-05, + "loss": 0.2428, + "step": 2735 + }, + { + "epoch": 1.54, + "learning_rate": 1.6910935738444198e-05, + "loss": 0.2391, + "step": 2740 + }, + { + "epoch": 1.55, + "learning_rate": 1.6905298759864716e-05, + "loss": 0.2261, + "step": 2745 + }, + { + "epoch": 1.55, + "learning_rate": 1.6899661781285234e-05, + "loss": 0.2504, + "step": 2750 + }, + { + "epoch": 1.55, + "learning_rate": 1.689402480270575e-05, + "loss": 0.2335, + "step": 2755 + }, + { + "epoch": 1.56, + "learning_rate": 1.688838782412627e-05, + "loss": 0.2174, + "step": 2760 + }, + { + "epoch": 1.56, + "learning_rate": 1.6882750845546788e-05, + "loss": 0.2116, + "step": 2765 + }, + { + "epoch": 1.56, + "learning_rate": 1.687711386696731e-05, + "loss": 0.2186, + "step": 2770 + }, + { + "epoch": 1.56, + "learning_rate": 1.6871476888387827e-05, + "loss": 0.2297, + "step": 2775 + }, + { + "epoch": 1.57, + "learning_rate": 1.6865839909808345e-05, + "loss": 0.2253, + "step": 2780 + }, + { + "epoch": 1.57, + "learning_rate": 1.6860202931228863e-05, + "loss": 0.258, + "step": 2785 + }, + { + "epoch": 1.57, + "learning_rate": 1.685456595264938e-05, + "loss": 0.232, + "step": 2790 + }, + { + "epoch": 1.58, + "learning_rate": 1.68489289740699e-05, + "loss": 0.2331, + "step": 2795 + }, + { + "epoch": 1.58, + "learning_rate": 1.6843291995490417e-05, + "loss": 0.2039, + "step": 2800 + }, + { + "epoch": 1.58, + "eval_accuracy": 0.8755293774034951, + "eval_accuracy_sklearn": 0.8755293774034951, + "eval_f1": 0.894369397281778, + "eval_loss": 0.3111112415790558, + "eval_precision": 0.8251391112127449, + "eval_recall": 0.9762806637806638, + "eval_runtime": 122.1939, + "eval_samples_per_second": 168.118, + "eval_steps_per_second": 10.508, + "step": 2800 + }, + { + "epoch": 1.58, + "learning_rate": 1.6837655016910935e-05, + "loss": 0.2369, + "step": 2805 + }, + { + "epoch": 1.58, + "learning_rate": 1.6832018038331456e-05, + "loss": 0.2138, + "step": 2810 + }, + { + "epoch": 1.59, + "learning_rate": 1.6826381059751974e-05, + "loss": 0.2351, + "step": 2815 + }, + { + "epoch": 1.59, + "learning_rate": 1.6820744081172492e-05, + "loss": 0.2268, + "step": 2820 + }, + { + "epoch": 1.59, + "learning_rate": 1.681510710259301e-05, + "loss": 0.2442, + "step": 2825 + }, + { + "epoch": 1.6, + "learning_rate": 1.6809470124013528e-05, + "loss": 0.2442, + "step": 2830 + }, + { + "epoch": 1.6, + "learning_rate": 1.680383314543405e-05, + "loss": 0.2224, + "step": 2835 + }, + { + "epoch": 1.6, + "learning_rate": 1.6798196166854568e-05, + "loss": 0.2417, + "step": 2840 + }, + { + "epoch": 1.6, + "learning_rate": 1.6792559188275086e-05, + "loss": 0.2242, + "step": 2845 + }, + { + "epoch": 1.61, + "learning_rate": 1.6786922209695604e-05, + "loss": 0.2301, + "step": 2850 + }, + { + "epoch": 1.61, + "learning_rate": 1.678128523111612e-05, + "loss": 0.215, + "step": 2855 + }, + { + "epoch": 1.61, + "learning_rate": 1.6775648252536643e-05, + "loss": 0.2352, + "step": 2860 + }, + { + "epoch": 1.61, + "learning_rate": 1.677001127395716e-05, + "loss": 0.2192, + "step": 2865 + }, + { + "epoch": 1.62, + "learning_rate": 1.676437429537768e-05, + "loss": 0.2183, + "step": 2870 + }, + { + "epoch": 1.62, + "learning_rate": 1.6758737316798197e-05, + "loss": 0.2603, + "step": 2875 + }, + { + "epoch": 1.62, + "learning_rate": 1.6753100338218715e-05, + "loss": 0.2258, + "step": 2880 + }, + { + "epoch": 1.63, + "learning_rate": 1.6747463359639236e-05, + "loss": 0.218, + "step": 2885 + }, + { + "epoch": 1.63, + "learning_rate": 1.6741826381059754e-05, + "loss": 0.2201, + "step": 2890 + }, + { + "epoch": 1.63, + "learning_rate": 1.6736189402480272e-05, + "loss": 0.2364, + "step": 2895 + }, + { + "epoch": 1.63, + "learning_rate": 1.673055242390079e-05, + "loss": 0.242, + "step": 2900 + }, + { + "epoch": 1.64, + "learning_rate": 1.6724915445321308e-05, + "loss": 0.2257, + "step": 2905 + }, + { + "epoch": 1.64, + "learning_rate": 1.671927846674183e-05, + "loss": 0.2223, + "step": 2910 + }, + { + "epoch": 1.64, + "learning_rate": 1.6713641488162348e-05, + "loss": 0.2024, + "step": 2915 + }, + { + "epoch": 1.65, + "learning_rate": 1.6708004509582866e-05, + "loss": 0.2364, + "step": 2920 + }, + { + "epoch": 1.65, + "learning_rate": 1.6702367531003383e-05, + "loss": 0.2279, + "step": 2925 + }, + { + "epoch": 1.65, + "learning_rate": 1.6696730552423905e-05, + "loss": 0.2301, + "step": 2930 + }, + { + "epoch": 1.65, + "learning_rate": 1.6691093573844423e-05, + "loss": 0.2165, + "step": 2935 + }, + { + "epoch": 1.66, + "learning_rate": 1.668545659526494e-05, + "loss": 0.2343, + "step": 2940 + }, + { + "epoch": 1.66, + "learning_rate": 1.667981961668546e-05, + "loss": 0.2475, + "step": 2945 + }, + { + "epoch": 1.66, + "learning_rate": 1.6674182638105977e-05, + "loss": 0.2329, + "step": 2950 + }, + { + "epoch": 1.67, + "learning_rate": 1.6668545659526495e-05, + "loss": 0.2097, + "step": 2955 + }, + { + "epoch": 1.67, + "learning_rate": 1.6662908680947013e-05, + "loss": 0.2111, + "step": 2960 + }, + { + "epoch": 1.67, + "learning_rate": 1.665727170236753e-05, + "loss": 0.2222, + "step": 2965 + }, + { + "epoch": 1.67, + "learning_rate": 1.665163472378805e-05, + "loss": 0.2313, + "step": 2970 + }, + { + "epoch": 1.68, + "learning_rate": 1.664599774520857e-05, + "loss": 0.224, + "step": 2975 + }, + { + "epoch": 1.68, + "learning_rate": 1.6640360766629088e-05, + "loss": 0.2268, + "step": 2980 + }, + { + "epoch": 1.68, + "learning_rate": 1.6634723788049606e-05, + "loss": 0.2301, + "step": 2985 + }, + { + "epoch": 1.69, + "learning_rate": 1.6629086809470124e-05, + "loss": 0.2136, + "step": 2990 + }, + { + "epoch": 1.69, + "learning_rate": 1.6623449830890642e-05, + "loss": 0.2129, + "step": 2995 + }, + { + "epoch": 1.69, + "learning_rate": 1.661781285231116e-05, + "loss": 0.2254, + "step": 3000 + }, + { + "epoch": 1.69, + "eval_accuracy": 0.9032273767219977, + "eval_accuracy_sklearn": 0.9032273767219977, + "eval_f1": 0.9145974740097945, + "eval_loss": 0.23222507536411285, + "eval_precision": 0.8732567678424938, + "eval_recall": 0.9600468975468975, + "eval_runtime": 122.7968, + "eval_samples_per_second": 167.293, + "eval_steps_per_second": 10.456, + "step": 3000 + }, + { + "epoch": 1.69, + "learning_rate": 1.661217587373168e-05, + "loss": 0.2196, + "step": 3005 + }, + { + "epoch": 1.7, + "learning_rate": 1.66065388951522e-05, + "loss": 0.2302, + "step": 3010 + }, + { + "epoch": 1.7, + "learning_rate": 1.6600901916572717e-05, + "loss": 0.2368, + "step": 3015 + }, + { + "epoch": 1.7, + "learning_rate": 1.6595264937993235e-05, + "loss": 0.2373, + "step": 3020 + }, + { + "epoch": 1.71, + "learning_rate": 1.6589627959413757e-05, + "loss": 0.2194, + "step": 3025 + }, + { + "epoch": 1.71, + "learning_rate": 1.6583990980834275e-05, + "loss": 0.2172, + "step": 3030 + }, + { + "epoch": 1.71, + "learning_rate": 1.6578354002254793e-05, + "loss": 0.2252, + "step": 3035 + }, + { + "epoch": 1.71, + "learning_rate": 1.657271702367531e-05, + "loss": 0.2212, + "step": 3040 + }, + { + "epoch": 1.72, + "learning_rate": 1.656708004509583e-05, + "loss": 0.2356, + "step": 3045 + }, + { + "epoch": 1.72, + "learning_rate": 1.656144306651635e-05, + "loss": 0.2232, + "step": 3050 + }, + { + "epoch": 1.72, + "learning_rate": 1.6555806087936868e-05, + "loss": 0.2149, + "step": 3055 + }, + { + "epoch": 1.72, + "learning_rate": 1.6550169109357386e-05, + "loss": 0.2145, + "step": 3060 + }, + { + "epoch": 1.73, + "learning_rate": 1.6544532130777904e-05, + "loss": 0.2309, + "step": 3065 + }, + { + "epoch": 1.73, + "learning_rate": 1.6538895152198422e-05, + "loss": 0.2154, + "step": 3070 + }, + { + "epoch": 1.73, + "learning_rate": 1.6533258173618943e-05, + "loss": 0.2327, + "step": 3075 + }, + { + "epoch": 1.74, + "learning_rate": 1.652762119503946e-05, + "loss": 0.2221, + "step": 3080 + }, + { + "epoch": 1.74, + "learning_rate": 1.652198421645998e-05, + "loss": 0.2153, + "step": 3085 + }, + { + "epoch": 1.74, + "learning_rate": 1.6516347237880497e-05, + "loss": 0.2088, + "step": 3090 + }, + { + "epoch": 1.74, + "learning_rate": 1.6510710259301015e-05, + "loss": 0.2207, + "step": 3095 + }, + { + "epoch": 1.75, + "learning_rate": 1.6505073280721537e-05, + "loss": 0.2175, + "step": 3100 + }, + { + "epoch": 1.75, + "learning_rate": 1.6499436302142055e-05, + "loss": 0.2431, + "step": 3105 + }, + { + "epoch": 1.75, + "learning_rate": 1.6493799323562573e-05, + "loss": 0.2203, + "step": 3110 + }, + { + "epoch": 1.76, + "learning_rate": 1.648816234498309e-05, + "loss": 0.2213, + "step": 3115 + }, + { + "epoch": 1.76, + "learning_rate": 1.648252536640361e-05, + "loss": 0.2238, + "step": 3120 + }, + { + "epoch": 1.76, + "learning_rate": 1.6476888387824127e-05, + "loss": 0.2192, + "step": 3125 + }, + { + "epoch": 1.76, + "learning_rate": 1.6471251409244645e-05, + "loss": 0.2298, + "step": 3130 + }, + { + "epoch": 1.77, + "learning_rate": 1.6465614430665166e-05, + "loss": 0.2144, + "step": 3135 + }, + { + "epoch": 1.77, + "learning_rate": 1.6459977452085684e-05, + "loss": 0.2292, + "step": 3140 + }, + { + "epoch": 1.77, + "learning_rate": 1.6454340473506202e-05, + "loss": 0.2187, + "step": 3145 + }, + { + "epoch": 1.78, + "learning_rate": 1.644870349492672e-05, + "loss": 0.2168, + "step": 3150 + }, + { + "epoch": 1.78, + "learning_rate": 1.6443066516347238e-05, + "loss": 0.2231, + "step": 3155 + }, + { + "epoch": 1.78, + "learning_rate": 1.6437429537767756e-05, + "loss": 0.2104, + "step": 3160 + }, + { + "epoch": 1.78, + "learning_rate": 1.6431792559188274e-05, + "loss": 0.2235, + "step": 3165 + }, + { + "epoch": 1.79, + "learning_rate": 1.6426155580608795e-05, + "loss": 0.2176, + "step": 3170 + }, + { + "epoch": 1.79, + "learning_rate": 1.6420518602029313e-05, + "loss": 0.2462, + "step": 3175 + }, + { + "epoch": 1.79, + "learning_rate": 1.641488162344983e-05, + "loss": 0.2228, + "step": 3180 + }, + { + "epoch": 1.8, + "learning_rate": 1.640924464487035e-05, + "loss": 0.2108, + "step": 3185 + }, + { + "epoch": 1.8, + "learning_rate": 1.640360766629087e-05, + "loss": 0.2149, + "step": 3190 + }, + { + "epoch": 1.8, + "learning_rate": 1.639797068771139e-05, + "loss": 0.2277, + "step": 3195 + }, + { + "epoch": 1.8, + "learning_rate": 1.6392333709131907e-05, + "loss": 0.205, + "step": 3200 + }, + { + "epoch": 1.8, + "eval_accuracy": 0.911989485469503, + "eval_accuracy_sklearn": 0.911989485469503, + "eval_f1": 0.9214049730481654, + "eval_loss": 0.2187822312116623, + "eval_precision": 0.8893924135615978, + "eval_recall": 0.9558080808080808, + "eval_runtime": 136.2717, + "eval_samples_per_second": 150.75, + "eval_steps_per_second": 9.422, + "step": 3200 + }, + { + "epoch": 1.81, + "learning_rate": 1.6386696730552425e-05, + "loss": 0.2238, + "step": 3205 + }, + { + "epoch": 1.81, + "learning_rate": 1.6381059751972943e-05, + "loss": 0.2241, + "step": 3210 + }, + { + "epoch": 1.81, + "learning_rate": 1.6375422773393464e-05, + "loss": 0.226, + "step": 3215 + }, + { + "epoch": 1.82, + "learning_rate": 1.6369785794813982e-05, + "loss": 0.2323, + "step": 3220 + }, + { + "epoch": 1.82, + "learning_rate": 1.63641488162345e-05, + "loss": 0.2252, + "step": 3225 + }, + { + "epoch": 1.82, + "learning_rate": 1.6358511837655018e-05, + "loss": 0.2279, + "step": 3230 + }, + { + "epoch": 1.82, + "learning_rate": 1.6352874859075536e-05, + "loss": 0.2262, + "step": 3235 + }, + { + "epoch": 1.83, + "learning_rate": 1.6347237880496057e-05, + "loss": 0.2215, + "step": 3240 + }, + { + "epoch": 1.83, + "learning_rate": 1.6341600901916575e-05, + "loss": 0.2186, + "step": 3245 + }, + { + "epoch": 1.83, + "learning_rate": 1.6335963923337093e-05, + "loss": 0.2202, + "step": 3250 + }, + { + "epoch": 1.83, + "learning_rate": 1.633032694475761e-05, + "loss": 0.2088, + "step": 3255 + }, + { + "epoch": 1.84, + "learning_rate": 1.632468996617813e-05, + "loss": 0.2184, + "step": 3260 + }, + { + "epoch": 1.84, + "learning_rate": 1.631905298759865e-05, + "loss": 0.2229, + "step": 3265 + }, + { + "epoch": 1.84, + "learning_rate": 1.631341600901917e-05, + "loss": 0.2322, + "step": 3270 + }, + { + "epoch": 1.85, + "learning_rate": 1.6307779030439687e-05, + "loss": 0.2027, + "step": 3275 + }, + { + "epoch": 1.85, + "learning_rate": 1.6302142051860205e-05, + "loss": 0.2139, + "step": 3280 + }, + { + "epoch": 1.85, + "learning_rate": 1.6296505073280723e-05, + "loss": 0.2209, + "step": 3285 + }, + { + "epoch": 1.85, + "learning_rate": 1.629086809470124e-05, + "loss": 0.2164, + "step": 3290 + }, + { + "epoch": 1.86, + "learning_rate": 1.628523111612176e-05, + "loss": 0.2255, + "step": 3295 + }, + { + "epoch": 1.86, + "learning_rate": 1.627959413754228e-05, + "loss": 0.2052, + "step": 3300 + }, + { + "epoch": 1.86, + "learning_rate": 1.6273957158962798e-05, + "loss": 0.2173, + "step": 3305 + }, + { + "epoch": 1.87, + "learning_rate": 1.6268320180383316e-05, + "loss": 0.2211, + "step": 3310 + }, + { + "epoch": 1.87, + "learning_rate": 1.6262683201803834e-05, + "loss": 0.2168, + "step": 3315 + }, + { + "epoch": 1.87, + "learning_rate": 1.6257046223224352e-05, + "loss": 0.2297, + "step": 3320 + }, + { + "epoch": 1.87, + "learning_rate": 1.625140924464487e-05, + "loss": 0.2161, + "step": 3325 + }, + { + "epoch": 1.88, + "learning_rate": 1.6245772266065388e-05, + "loss": 0.2149, + "step": 3330 + }, + { + "epoch": 1.88, + "learning_rate": 1.624013528748591e-05, + "loss": 0.2282, + "step": 3335 + }, + { + "epoch": 1.88, + "learning_rate": 1.6234498308906427e-05, + "loss": 0.1988, + "step": 3340 + }, + { + "epoch": 1.89, + "learning_rate": 1.6228861330326945e-05, + "loss": 0.2223, + "step": 3345 + }, + { + "epoch": 1.89, + "learning_rate": 1.6223224351747463e-05, + "loss": 0.2037, + "step": 3350 + }, + { + "epoch": 1.89, + "learning_rate": 1.621758737316798e-05, + "loss": 0.2127, + "step": 3355 + }, + { + "epoch": 1.89, + "learning_rate": 1.6211950394588503e-05, + "loss": 0.2027, + "step": 3360 + }, + { + "epoch": 1.9, + "learning_rate": 1.620631341600902e-05, + "loss": 0.2116, + "step": 3365 + }, + { + "epoch": 1.9, + "learning_rate": 1.620067643742954e-05, + "loss": 0.2018, + "step": 3370 + }, + { + "epoch": 1.9, + "learning_rate": 1.6195039458850057e-05, + "loss": 0.2131, + "step": 3375 + }, + { + "epoch": 1.91, + "learning_rate": 1.6189402480270578e-05, + "loss": 0.2189, + "step": 3380 + }, + { + "epoch": 1.91, + "learning_rate": 1.6183765501691096e-05, + "loss": 0.2151, + "step": 3385 + }, + { + "epoch": 1.91, + "learning_rate": 1.6178128523111614e-05, + "loss": 0.1961, + "step": 3390 + }, + { + "epoch": 1.91, + "learning_rate": 1.6172491544532132e-05, + "loss": 0.2249, + "step": 3395 + }, + { + "epoch": 1.92, + "learning_rate": 1.616685456595265e-05, + "loss": 0.2208, + "step": 3400 + }, + { + "epoch": 1.92, + "eval_accuracy": 0.8987976439663146, + "eval_accuracy_sklearn": 0.8987976439663146, + "eval_f1": 0.9118731719723624, + "eval_loss": 0.24518540501594543, + "eval_precision": 0.8602735343517556, + "eval_recall": 0.9700577200577201, + "eval_runtime": 122.3997, + "eval_samples_per_second": 167.835, + "eval_steps_per_second": 10.49, + "step": 3400 + }, + { + "epoch": 1.92, + "learning_rate": 1.616121758737317e-05, + "loss": 0.1993, + "step": 3405 + }, + { + "epoch": 1.92, + "learning_rate": 1.615558060879369e-05, + "loss": 0.2106, + "step": 3410 + }, + { + "epoch": 1.93, + "learning_rate": 1.6149943630214207e-05, + "loss": 0.1996, + "step": 3415 + }, + { + "epoch": 1.93, + "learning_rate": 1.6144306651634725e-05, + "loss": 0.2002, + "step": 3420 + }, + { + "epoch": 1.93, + "learning_rate": 1.6138669673055243e-05, + "loss": 0.2058, + "step": 3425 + }, + { + "epoch": 1.93, + "learning_rate": 1.6133032694475765e-05, + "loss": 0.2111, + "step": 3430 + }, + { + "epoch": 1.94, + "learning_rate": 1.6127395715896283e-05, + "loss": 0.2095, + "step": 3435 + }, + { + "epoch": 1.94, + "learning_rate": 1.61217587373168e-05, + "loss": 0.2034, + "step": 3440 + }, + { + "epoch": 1.94, + "learning_rate": 1.611612175873732e-05, + "loss": 0.2136, + "step": 3445 + }, + { + "epoch": 1.94, + "learning_rate": 1.6110484780157837e-05, + "loss": 0.2317, + "step": 3450 + }, + { + "epoch": 1.95, + "learning_rate": 1.6104847801578355e-05, + "loss": 0.2258, + "step": 3455 + }, + { + "epoch": 1.95, + "learning_rate": 1.6099210822998873e-05, + "loss": 0.2183, + "step": 3460 + }, + { + "epoch": 1.95, + "learning_rate": 1.6093573844419394e-05, + "loss": 0.1888, + "step": 3465 + }, + { + "epoch": 1.96, + "learning_rate": 1.6087936865839912e-05, + "loss": 0.2154, + "step": 3470 + }, + { + "epoch": 1.96, + "learning_rate": 1.608229988726043e-05, + "loss": 0.2211, + "step": 3475 + }, + { + "epoch": 1.96, + "learning_rate": 1.6076662908680948e-05, + "loss": 0.2292, + "step": 3480 + }, + { + "epoch": 1.96, + "learning_rate": 1.6071025930101466e-05, + "loss": 0.2214, + "step": 3485 + }, + { + "epoch": 1.97, + "learning_rate": 1.6065388951521984e-05, + "loss": 0.2165, + "step": 3490 + }, + { + "epoch": 1.97, + "learning_rate": 1.6059751972942502e-05, + "loss": 0.2248, + "step": 3495 + }, + { + "epoch": 1.97, + "learning_rate": 1.6054114994363023e-05, + "loss": 0.186, + "step": 3500 + }, + { + "epoch": 1.98, + "learning_rate": 1.604847801578354e-05, + "loss": 0.203, + "step": 3505 + }, + { + "epoch": 1.98, + "learning_rate": 1.604284103720406e-05, + "loss": 0.2094, + "step": 3510 + }, + { + "epoch": 1.98, + "learning_rate": 1.6037204058624577e-05, + "loss": 0.2249, + "step": 3515 + }, + { + "epoch": 1.98, + "learning_rate": 1.6031567080045095e-05, + "loss": 0.215, + "step": 3520 + }, + { + "epoch": 1.99, + "learning_rate": 1.6025930101465617e-05, + "loss": 0.2165, + "step": 3525 + }, + { + "epoch": 1.99, + "learning_rate": 1.6020293122886135e-05, + "loss": 0.2152, + "step": 3530 + }, + { + "epoch": 1.99, + "learning_rate": 1.6014656144306653e-05, + "loss": 0.197, + "step": 3535 + }, + { + "epoch": 2.0, + "learning_rate": 1.600901916572717e-05, + "loss": 0.192, + "step": 3540 + }, + { + "epoch": 2.0, + "learning_rate": 1.600338218714769e-05, + "loss": 0.2032, + "step": 3545 + }, + { + "epoch": 2.0, + "learning_rate": 1.599774520856821e-05, + "loss": 0.2057, + "step": 3550 + }, + { + "epoch": 2.0, + "learning_rate": 1.5992108229988728e-05, + "loss": 0.1876, + "step": 3555 + }, + { + "epoch": 2.01, + "learning_rate": 1.5986471251409246e-05, + "loss": 0.211, + "step": 3560 + }, + { + "epoch": 2.01, + "learning_rate": 1.5980834272829764e-05, + "loss": 0.1902, + "step": 3565 + }, + { + "epoch": 2.01, + "learning_rate": 1.5975197294250285e-05, + "loss": 0.18, + "step": 3570 + }, + { + "epoch": 2.02, + "learning_rate": 1.5969560315670803e-05, + "loss": 0.2015, + "step": 3575 + }, + { + "epoch": 2.02, + "learning_rate": 1.596392333709132e-05, + "loss": 0.1849, + "step": 3580 + }, + { + "epoch": 2.02, + "learning_rate": 1.595828635851184e-05, + "loss": 0.1955, + "step": 3585 + }, + { + "epoch": 2.02, + "learning_rate": 1.5952649379932357e-05, + "loss": 0.1871, + "step": 3590 + }, + { + "epoch": 2.03, + "learning_rate": 1.594701240135288e-05, + "loss": 0.203, + "step": 3595 + }, + { + "epoch": 2.03, + "learning_rate": 1.5941375422773397e-05, + "loss": 0.196, + "step": 3600 + }, + { + "epoch": 2.03, + "eval_accuracy": 0.9003066738061627, + "eval_accuracy_sklearn": 0.9003066738061627, + "eval_f1": 0.9132864764162927, + "eval_loss": 0.24654868245124817, + "eval_precision": 0.8607342378292099, + "eval_recall": 0.9726731601731602, + "eval_runtime": 121.9395, + "eval_samples_per_second": 168.469, + "eval_steps_per_second": 10.53, + "step": 3600 + }, + { + "epoch": 2.03, + "learning_rate": 1.5935738444193914e-05, + "loss": 0.179, + "step": 3605 + }, + { + "epoch": 2.03, + "learning_rate": 1.5930101465614432e-05, + "loss": 0.1959, + "step": 3610 + }, + { + "epoch": 2.04, + "learning_rate": 1.592446448703495e-05, + "loss": 0.1958, + "step": 3615 + }, + { + "epoch": 2.04, + "learning_rate": 1.591882750845547e-05, + "loss": 0.1984, + "step": 3620 + }, + { + "epoch": 2.04, + "learning_rate": 1.5913190529875986e-05, + "loss": 0.2122, + "step": 3625 + }, + { + "epoch": 2.05, + "learning_rate": 1.5907553551296508e-05, + "loss": 0.1962, + "step": 3630 + }, + { + "epoch": 2.05, + "learning_rate": 1.5901916572717026e-05, + "loss": 0.2054, + "step": 3635 + }, + { + "epoch": 2.05, + "learning_rate": 1.5896279594137544e-05, + "loss": 0.1946, + "step": 3640 + }, + { + "epoch": 2.05, + "learning_rate": 1.5890642615558062e-05, + "loss": 0.199, + "step": 3645 + }, + { + "epoch": 2.06, + "learning_rate": 1.588500563697858e-05, + "loss": 0.1812, + "step": 3650 + }, + { + "epoch": 2.06, + "learning_rate": 1.5879368658399098e-05, + "loss": 0.1969, + "step": 3655 + }, + { + "epoch": 2.06, + "learning_rate": 1.5873731679819616e-05, + "loss": 0.1847, + "step": 3660 + }, + { + "epoch": 2.07, + "learning_rate": 1.5868094701240137e-05, + "loss": 0.2084, + "step": 3665 + }, + { + "epoch": 2.07, + "learning_rate": 1.5862457722660655e-05, + "loss": 0.1985, + "step": 3670 + }, + { + "epoch": 2.07, + "learning_rate": 1.5856820744081173e-05, + "loss": 0.1929, + "step": 3675 + }, + { + "epoch": 2.07, + "learning_rate": 1.585118376550169e-05, + "loss": 0.1871, + "step": 3680 + }, + { + "epoch": 2.08, + "learning_rate": 1.584554678692221e-05, + "loss": 0.1935, + "step": 3685 + }, + { + "epoch": 2.08, + "learning_rate": 1.583990980834273e-05, + "loss": 0.2035, + "step": 3690 + }, + { + "epoch": 2.08, + "learning_rate": 1.583427282976325e-05, + "loss": 0.1842, + "step": 3695 + }, + { + "epoch": 2.09, + "learning_rate": 1.5828635851183766e-05, + "loss": 0.2124, + "step": 3700 + }, + { + "epoch": 2.09, + "learning_rate": 1.5822998872604284e-05, + "loss": 0.1707, + "step": 3705 + }, + { + "epoch": 2.09, + "learning_rate": 1.5817361894024802e-05, + "loss": 0.1868, + "step": 3710 + }, + { + "epoch": 2.09, + "learning_rate": 1.5811724915445324e-05, + "loss": 0.1911, + "step": 3715 + }, + { + "epoch": 2.1, + "learning_rate": 1.5806087936865842e-05, + "loss": 0.1991, + "step": 3720 + }, + { + "epoch": 2.1, + "learning_rate": 1.580045095828636e-05, + "loss": 0.2025, + "step": 3725 + }, + { + "epoch": 2.1, + "learning_rate": 1.5794813979706878e-05, + "loss": 0.203, + "step": 3730 + }, + { + "epoch": 2.11, + "learning_rate": 1.5789177001127396e-05, + "loss": 0.1946, + "step": 3735 + }, + { + "epoch": 2.11, + "learning_rate": 1.5783540022547917e-05, + "loss": 0.1966, + "step": 3740 + }, + { + "epoch": 2.11, + "learning_rate": 1.5777903043968435e-05, + "loss": 0.1902, + "step": 3745 + }, + { + "epoch": 2.11, + "learning_rate": 1.5772266065388953e-05, + "loss": 0.1981, + "step": 3750 + }, + { + "epoch": 2.12, + "learning_rate": 1.576662908680947e-05, + "loss": 0.2008, + "step": 3755 + }, + { + "epoch": 2.12, + "learning_rate": 1.5760992108229992e-05, + "loss": 0.1595, + "step": 3760 + }, + { + "epoch": 2.12, + "learning_rate": 1.575535512965051e-05, + "loss": 0.2039, + "step": 3765 + }, + { + "epoch": 2.13, + "learning_rate": 1.574971815107103e-05, + "loss": 0.2026, + "step": 3770 + }, + { + "epoch": 2.13, + "learning_rate": 1.5744081172491546e-05, + "loss": 0.1841, + "step": 3775 + }, + { + "epoch": 2.13, + "learning_rate": 1.5738444193912064e-05, + "loss": 0.2036, + "step": 3780 + }, + { + "epoch": 2.13, + "learning_rate": 1.5732807215332582e-05, + "loss": 0.206, + "step": 3785 + }, + { + "epoch": 2.14, + "learning_rate": 1.5727170236753104e-05, + "loss": 0.1966, + "step": 3790 + }, + { + "epoch": 2.14, + "learning_rate": 1.5721533258173622e-05, + "loss": 0.1933, + "step": 3795 + }, + { + "epoch": 2.14, + "learning_rate": 1.571589627959414e-05, + "loss": 0.1819, + "step": 3800 + }, + { + "epoch": 2.14, + "eval_accuracy": 0.9086793554982232, + "eval_accuracy_sklearn": 0.9086793554982232, + "eval_f1": 0.9200136437281488, + "eval_loss": 0.24143485724925995, + "eval_precision": 0.8724729095908135, + "eval_recall": 0.9730339105339105, + "eval_runtime": 122.9645, + "eval_samples_per_second": 167.064, + "eval_steps_per_second": 10.442, + "step": 3800 + }, + { + "epoch": 2.14, + "learning_rate": 1.5710259301014658e-05, + "loss": 0.1746, + "step": 3805 + }, + { + "epoch": 2.15, + "learning_rate": 1.5704622322435176e-05, + "loss": 0.2166, + "step": 3810 + }, + { + "epoch": 2.15, + "learning_rate": 1.5698985343855694e-05, + "loss": 0.2068, + "step": 3815 + }, + { + "epoch": 2.15, + "learning_rate": 1.569334836527621e-05, + "loss": 0.1826, + "step": 3820 + }, + { + "epoch": 2.16, + "learning_rate": 1.568771138669673e-05, + "loss": 0.1926, + "step": 3825 + }, + { + "epoch": 2.16, + "learning_rate": 1.5682074408117248e-05, + "loss": 0.2156, + "step": 3830 + }, + { + "epoch": 2.16, + "learning_rate": 1.567643742953777e-05, + "loss": 0.1941, + "step": 3835 + }, + { + "epoch": 2.16, + "learning_rate": 1.5670800450958287e-05, + "loss": 0.1956, + "step": 3840 + }, + { + "epoch": 2.17, + "learning_rate": 1.5665163472378805e-05, + "loss": 0.1847, + "step": 3845 + }, + { + "epoch": 2.17, + "learning_rate": 1.5659526493799323e-05, + "loss": 0.185, + "step": 3850 + }, + { + "epoch": 2.17, + "learning_rate": 1.5653889515219844e-05, + "loss": 0.1799, + "step": 3855 + }, + { + "epoch": 2.18, + "learning_rate": 1.5648252536640362e-05, + "loss": 0.2, + "step": 3860 + }, + { + "epoch": 2.18, + "learning_rate": 1.564261555806088e-05, + "loss": 0.1784, + "step": 3865 + }, + { + "epoch": 2.18, + "learning_rate": 1.56369785794814e-05, + "loss": 0.1912, + "step": 3870 + }, + { + "epoch": 2.18, + "learning_rate": 1.5631341600901916e-05, + "loss": 0.1859, + "step": 3875 + }, + { + "epoch": 2.19, + "learning_rate": 1.5625704622322438e-05, + "loss": 0.2022, + "step": 3880 + }, + { + "epoch": 2.19, + "learning_rate": 1.5620067643742956e-05, + "loss": 0.1983, + "step": 3885 + }, + { + "epoch": 2.19, + "learning_rate": 1.5614430665163474e-05, + "loss": 0.1873, + "step": 3890 + }, + { + "epoch": 2.2, + "learning_rate": 1.560879368658399e-05, + "loss": 0.189, + "step": 3895 + }, + { + "epoch": 2.2, + "learning_rate": 1.560315670800451e-05, + "loss": 0.1806, + "step": 3900 + }, + { + "epoch": 2.2, + "learning_rate": 1.559751972942503e-05, + "loss": 0.1779, + "step": 3905 + }, + { + "epoch": 2.2, + "learning_rate": 1.559188275084555e-05, + "loss": 0.1923, + "step": 3910 + }, + { + "epoch": 2.21, + "learning_rate": 1.5586245772266067e-05, + "loss": 0.1985, + "step": 3915 + }, + { + "epoch": 2.21, + "learning_rate": 1.5580608793686585e-05, + "loss": 0.1983, + "step": 3920 + }, + { + "epoch": 2.21, + "learning_rate": 1.5574971815107103e-05, + "loss": 0.2024, + "step": 3925 + }, + { + "epoch": 2.22, + "learning_rate": 1.5569334836527624e-05, + "loss": 0.1739, + "step": 3930 + }, + { + "epoch": 2.22, + "learning_rate": 1.5563697857948142e-05, + "loss": 0.1939, + "step": 3935 + }, + { + "epoch": 2.22, + "learning_rate": 1.555806087936866e-05, + "loss": 0.2087, + "step": 3940 + }, + { + "epoch": 2.22, + "learning_rate": 1.555242390078918e-05, + "loss": 0.2018, + "step": 3945 + }, + { + "epoch": 2.23, + "learning_rate": 1.5546786922209696e-05, + "loss": 0.1876, + "step": 3950 + }, + { + "epoch": 2.23, + "learning_rate": 1.5541149943630218e-05, + "loss": 0.1731, + "step": 3955 + }, + { + "epoch": 2.23, + "learning_rate": 1.5535512965050736e-05, + "loss": 0.1862, + "step": 3960 + }, + { + "epoch": 2.24, + "learning_rate": 1.5529875986471254e-05, + "loss": 0.1919, + "step": 3965 + }, + { + "epoch": 2.24, + "learning_rate": 1.552423900789177e-05, + "loss": 0.2005, + "step": 3970 + }, + { + "epoch": 2.24, + "learning_rate": 1.551860202931229e-05, + "loss": 0.1932, + "step": 3975 + }, + { + "epoch": 2.24, + "learning_rate": 1.5512965050732808e-05, + "loss": 0.1883, + "step": 3980 + }, + { + "epoch": 2.25, + "learning_rate": 1.5507328072153326e-05, + "loss": 0.2155, + "step": 3985 + }, + { + "epoch": 2.25, + "learning_rate": 1.5501691093573844e-05, + "loss": 0.2018, + "step": 3990 + }, + { + "epoch": 2.25, + "learning_rate": 1.5496054114994365e-05, + "loss": 0.1871, + "step": 3995 + }, + { + "epoch": 2.25, + "learning_rate": 1.5490417136414883e-05, + "loss": 0.1842, + "step": 4000 + }, + { + "epoch": 2.25, + "eval_accuracy": 0.9229421214038845, + "eval_accuracy_sklearn": 0.9229421214038845, + "eval_f1": 0.9310990206746463, + "eval_loss": 0.1880422979593277, + "eval_precision": 0.8998065113148818, + "eval_recall": 0.9646464646464646, + "eval_runtime": 123.0461, + "eval_samples_per_second": 166.954, + "eval_steps_per_second": 10.435, + "step": 4000 + }, + { + "epoch": 2.26, + "learning_rate": 1.54847801578354e-05, + "loss": 0.1989, + "step": 4005 + }, + { + "epoch": 2.26, + "learning_rate": 1.547914317925592e-05, + "loss": 0.1781, + "step": 4010 + }, + { + "epoch": 2.26, + "learning_rate": 1.5473506200676437e-05, + "loss": 0.1957, + "step": 4015 + }, + { + "epoch": 2.27, + "learning_rate": 1.5467869222096958e-05, + "loss": 0.1786, + "step": 4020 + }, + { + "epoch": 2.27, + "learning_rate": 1.5462232243517476e-05, + "loss": 0.1838, + "step": 4025 + }, + { + "epoch": 2.27, + "learning_rate": 1.5456595264937994e-05, + "loss": 0.1835, + "step": 4030 + }, + { + "epoch": 2.27, + "learning_rate": 1.5450958286358512e-05, + "loss": 0.1988, + "step": 4035 + }, + { + "epoch": 2.28, + "learning_rate": 1.544532130777903e-05, + "loss": 0.1917, + "step": 4040 + }, + { + "epoch": 2.28, + "learning_rate": 1.543968432919955e-05, + "loss": 0.1779, + "step": 4045 + }, + { + "epoch": 2.28, + "learning_rate": 1.543404735062007e-05, + "loss": 0.1759, + "step": 4050 + }, + { + "epoch": 2.29, + "learning_rate": 1.5428410372040588e-05, + "loss": 0.1988, + "step": 4055 + }, + { + "epoch": 2.29, + "learning_rate": 1.5422773393461106e-05, + "loss": 0.1883, + "step": 4060 + }, + { + "epoch": 2.29, + "learning_rate": 1.5417136414881624e-05, + "loss": 0.1783, + "step": 4065 + }, + { + "epoch": 2.29, + "learning_rate": 1.5411499436302145e-05, + "loss": 0.2045, + "step": 4070 + }, + { + "epoch": 2.3, + "learning_rate": 1.5405862457722663e-05, + "loss": 0.1857, + "step": 4075 + }, + { + "epoch": 2.3, + "learning_rate": 1.540022547914318e-05, + "loss": 0.1885, + "step": 4080 + }, + { + "epoch": 2.3, + "learning_rate": 1.53945885005637e-05, + "loss": 0.1777, + "step": 4085 + }, + { + "epoch": 2.31, + "learning_rate": 1.5388951521984217e-05, + "loss": 0.1885, + "step": 4090 + }, + { + "epoch": 2.31, + "learning_rate": 1.5383314543404738e-05, + "loss": 0.1764, + "step": 4095 + }, + { + "epoch": 2.31, + "learning_rate": 1.5377677564825256e-05, + "loss": 0.1858, + "step": 4100 + }, + { + "epoch": 2.31, + "learning_rate": 1.5372040586245774e-05, + "loss": 0.1945, + "step": 4105 + }, + { + "epoch": 2.32, + "learning_rate": 1.5366403607666292e-05, + "loss": 0.1845, + "step": 4110 + }, + { + "epoch": 2.32, + "learning_rate": 1.536076662908681e-05, + "loss": 0.184, + "step": 4115 + }, + { + "epoch": 2.32, + "learning_rate": 1.535512965050733e-05, + "loss": 0.1781, + "step": 4120 + }, + { + "epoch": 2.33, + "learning_rate": 1.534949267192785e-05, + "loss": 0.1931, + "step": 4125 + }, + { + "epoch": 2.33, + "learning_rate": 1.5343855693348368e-05, + "loss": 0.1862, + "step": 4130 + }, + { + "epoch": 2.33, + "learning_rate": 1.5338218714768886e-05, + "loss": 0.1817, + "step": 4135 + }, + { + "epoch": 2.33, + "learning_rate": 1.5332581736189404e-05, + "loss": 0.1897, + "step": 4140 + }, + { + "epoch": 2.34, + "learning_rate": 1.532694475760992e-05, + "loss": 0.1719, + "step": 4145 + }, + { + "epoch": 2.34, + "learning_rate": 1.532130777903044e-05, + "loss": 0.2006, + "step": 4150 + }, + { + "epoch": 2.34, + "learning_rate": 1.5315670800450958e-05, + "loss": 0.2004, + "step": 4155 + }, + { + "epoch": 2.34, + "learning_rate": 1.531003382187148e-05, + "loss": 0.1794, + "step": 4160 + }, + { + "epoch": 2.35, + "learning_rate": 1.5304396843291997e-05, + "loss": 0.1978, + "step": 4165 + }, + { + "epoch": 2.35, + "learning_rate": 1.5298759864712515e-05, + "loss": 0.1922, + "step": 4170 + }, + { + "epoch": 2.35, + "learning_rate": 1.5293122886133033e-05, + "loss": 0.1926, + "step": 4175 + }, + { + "epoch": 2.36, + "learning_rate": 1.528748590755355e-05, + "loss": 0.1702, + "step": 4180 + }, + { + "epoch": 2.36, + "learning_rate": 1.528184892897407e-05, + "loss": 0.2034, + "step": 4185 + }, + { + "epoch": 2.36, + "learning_rate": 1.527621195039459e-05, + "loss": 0.1727, + "step": 4190 + }, + { + "epoch": 2.36, + "learning_rate": 1.5270574971815108e-05, + "loss": 0.1762, + "step": 4195 + }, + { + "epoch": 2.37, + "learning_rate": 1.5264937993235626e-05, + "loss": 0.2029, + "step": 4200 + }, + { + "epoch": 2.37, + "eval_accuracy": 0.9053205471450129, + "eval_accuracy_sklearn": 0.9053205471450129, + "eval_f1": 0.9177694161417157, + "eval_loss": 0.23918241262435913, + "eval_precision": 0.8638280939116594, + "eval_recall": 0.9788961038961039, + "eval_runtime": 121.4115, + "eval_samples_per_second": 169.201, + "eval_steps_per_second": 10.576, + "step": 4200 + }, + { + "epoch": 2.37, + "learning_rate": 1.5259301014656144e-05, + "loss": 0.1627, + "step": 4205 + }, + { + "epoch": 2.37, + "learning_rate": 1.5253664036076666e-05, + "loss": 0.1838, + "step": 4210 + }, + { + "epoch": 2.38, + "learning_rate": 1.5248027057497184e-05, + "loss": 0.1808, + "step": 4215 + }, + { + "epoch": 2.38, + "learning_rate": 1.5242390078917702e-05, + "loss": 0.1723, + "step": 4220 + }, + { + "epoch": 2.38, + "learning_rate": 1.523675310033822e-05, + "loss": 0.1697, + "step": 4225 + }, + { + "epoch": 2.38, + "learning_rate": 1.5231116121758737e-05, + "loss": 0.1694, + "step": 4230 + }, + { + "epoch": 2.39, + "learning_rate": 1.5225479143179259e-05, + "loss": 0.1738, + "step": 4235 + }, + { + "epoch": 2.39, + "learning_rate": 1.5219842164599777e-05, + "loss": 0.1828, + "step": 4240 + }, + { + "epoch": 2.39, + "learning_rate": 1.5214205186020295e-05, + "loss": 0.1825, + "step": 4245 + }, + { + "epoch": 2.4, + "learning_rate": 1.5208568207440813e-05, + "loss": 0.1775, + "step": 4250 + }, + { + "epoch": 2.4, + "learning_rate": 1.520293122886133e-05, + "loss": 0.1928, + "step": 4255 + }, + { + "epoch": 2.4, + "learning_rate": 1.519729425028185e-05, + "loss": 0.1783, + "step": 4260 + }, + { + "epoch": 2.4, + "learning_rate": 1.5191657271702368e-05, + "loss": 0.1667, + "step": 4265 + }, + { + "epoch": 2.41, + "learning_rate": 1.5186020293122886e-05, + "loss": 0.1832, + "step": 4270 + }, + { + "epoch": 2.41, + "learning_rate": 1.5180383314543406e-05, + "loss": 0.1915, + "step": 4275 + }, + { + "epoch": 2.41, + "learning_rate": 1.5174746335963924e-05, + "loss": 0.2001, + "step": 4280 + }, + { + "epoch": 2.42, + "learning_rate": 1.5169109357384444e-05, + "loss": 0.187, + "step": 4285 + }, + { + "epoch": 2.42, + "learning_rate": 1.5163472378804962e-05, + "loss": 0.2092, + "step": 4290 + }, + { + "epoch": 2.42, + "learning_rate": 1.515783540022548e-05, + "loss": 0.1713, + "step": 4295 + }, + { + "epoch": 2.42, + "learning_rate": 1.5152198421645998e-05, + "loss": 0.1958, + "step": 4300 + }, + { + "epoch": 2.43, + "learning_rate": 1.5146561443066519e-05, + "loss": 0.1782, + "step": 4305 + }, + { + "epoch": 2.43, + "learning_rate": 1.5140924464487037e-05, + "loss": 0.179, + "step": 4310 + }, + { + "epoch": 2.43, + "learning_rate": 1.5135287485907555e-05, + "loss": 0.1669, + "step": 4315 + }, + { + "epoch": 2.44, + "learning_rate": 1.5129650507328073e-05, + "loss": 0.198, + "step": 4320 + }, + { + "epoch": 2.44, + "learning_rate": 1.5124013528748591e-05, + "loss": 0.1842, + "step": 4325 + }, + { + "epoch": 2.44, + "learning_rate": 1.511837655016911e-05, + "loss": 0.193, + "step": 4330 + }, + { + "epoch": 2.44, + "learning_rate": 1.511273957158963e-05, + "loss": 0.1893, + "step": 4335 + }, + { + "epoch": 2.45, + "learning_rate": 1.5107102593010148e-05, + "loss": 0.1812, + "step": 4340 + }, + { + "epoch": 2.45, + "learning_rate": 1.5101465614430666e-05, + "loss": 0.1951, + "step": 4345 + }, + { + "epoch": 2.45, + "learning_rate": 1.5095828635851184e-05, + "loss": 0.1783, + "step": 4350 + }, + { + "epoch": 2.45, + "learning_rate": 1.5090191657271704e-05, + "loss": 0.181, + "step": 4355 + }, + { + "epoch": 2.46, + "learning_rate": 1.5084554678692222e-05, + "loss": 0.18, + "step": 4360 + }, + { + "epoch": 2.46, + "learning_rate": 1.507891770011274e-05, + "loss": 0.1816, + "step": 4365 + }, + { + "epoch": 2.46, + "learning_rate": 1.5073280721533258e-05, + "loss": 0.1911, + "step": 4370 + }, + { + "epoch": 2.47, + "learning_rate": 1.5067643742953778e-05, + "loss": 0.173, + "step": 4375 + }, + { + "epoch": 2.47, + "learning_rate": 1.5062006764374297e-05, + "loss": 0.1947, + "step": 4380 + }, + { + "epoch": 2.47, + "learning_rate": 1.5056369785794815e-05, + "loss": 0.1983, + "step": 4385 + }, + { + "epoch": 2.47, + "learning_rate": 1.5050732807215333e-05, + "loss": 0.1832, + "step": 4390 + }, + { + "epoch": 2.48, + "learning_rate": 1.5045095828635851e-05, + "loss": 0.176, + "step": 4395 + }, + { + "epoch": 2.48, + "learning_rate": 1.5039458850056373e-05, + "loss": 0.1731, + "step": 4400 + }, + { + "epoch": 2.48, + "eval_accuracy": 0.9001606386603709, + "eval_accuracy_sklearn": 0.9001606386603709, + "eval_f1": 0.9142092274229304, + "eval_loss": 0.26414573192596436, + "eval_precision": 0.8524845931819954, + "eval_recall": 0.9855699855699855, + "eval_runtime": 122.8598, + "eval_samples_per_second": 167.207, + "eval_steps_per_second": 10.451, + "step": 4400 + }, + { + "epoch": 2.48, + "learning_rate": 1.503382187147689e-05, + "loss": 0.1813, + "step": 4405 + }, + { + "epoch": 2.49, + "learning_rate": 1.5028184892897409e-05, + "loss": 0.1805, + "step": 4410 + }, + { + "epoch": 2.49, + "learning_rate": 1.5022547914317927e-05, + "loss": 0.1939, + "step": 4415 + }, + { + "epoch": 2.49, + "learning_rate": 1.5016910935738445e-05, + "loss": 0.1708, + "step": 4420 + }, + { + "epoch": 2.49, + "learning_rate": 1.5011273957158964e-05, + "loss": 0.19, + "step": 4425 + }, + { + "epoch": 2.5, + "learning_rate": 1.5005636978579482e-05, + "loss": 0.1878, + "step": 4430 + }, + { + "epoch": 2.5, + "learning_rate": 1.5000000000000002e-05, + "loss": 0.1811, + "step": 4435 + }, + { + "epoch": 2.5, + "learning_rate": 1.499436302142052e-05, + "loss": 0.1874, + "step": 4440 + }, + { + "epoch": 2.51, + "learning_rate": 1.4988726042841038e-05, + "loss": 0.174, + "step": 4445 + }, + { + "epoch": 2.51, + "learning_rate": 1.4983089064261558e-05, + "loss": 0.1693, + "step": 4450 + }, + { + "epoch": 2.51, + "learning_rate": 1.4977452085682076e-05, + "loss": 0.1651, + "step": 4455 + }, + { + "epoch": 2.51, + "learning_rate": 1.4971815107102594e-05, + "loss": 0.1823, + "step": 4460 + }, + { + "epoch": 2.52, + "learning_rate": 1.4966178128523112e-05, + "loss": 0.1883, + "step": 4465 + }, + { + "epoch": 2.52, + "learning_rate": 1.496054114994363e-05, + "loss": 0.1918, + "step": 4470 + }, + { + "epoch": 2.52, + "learning_rate": 1.4954904171364151e-05, + "loss": 0.1714, + "step": 4475 + }, + { + "epoch": 2.53, + "learning_rate": 1.4949267192784669e-05, + "loss": 0.1663, + "step": 4480 + }, + { + "epoch": 2.53, + "learning_rate": 1.4943630214205187e-05, + "loss": 0.1862, + "step": 4485 + }, + { + "epoch": 2.53, + "learning_rate": 1.4937993235625705e-05, + "loss": 0.1649, + "step": 4490 + }, + { + "epoch": 2.53, + "learning_rate": 1.4932356257046225e-05, + "loss": 0.1597, + "step": 4495 + }, + { + "epoch": 2.54, + "learning_rate": 1.4926719278466744e-05, + "loss": 0.1918, + "step": 4500 + }, + { + "epoch": 2.54, + "learning_rate": 1.4921082299887262e-05, + "loss": 0.1775, + "step": 4505 + }, + { + "epoch": 2.54, + "learning_rate": 1.491544532130778e-05, + "loss": 0.1775, + "step": 4510 + }, + { + "epoch": 2.55, + "learning_rate": 1.4909808342728298e-05, + "loss": 0.1841, + "step": 4515 + }, + { + "epoch": 2.55, + "learning_rate": 1.4904171364148818e-05, + "loss": 0.1678, + "step": 4520 + }, + { + "epoch": 2.55, + "learning_rate": 1.4898534385569336e-05, + "loss": 0.1925, + "step": 4525 + }, + { + "epoch": 2.55, + "learning_rate": 1.4892897406989854e-05, + "loss": 0.1759, + "step": 4530 + }, + { + "epoch": 2.56, + "learning_rate": 1.4887260428410372e-05, + "loss": 0.181, + "step": 4535 + }, + { + "epoch": 2.56, + "learning_rate": 1.4881623449830892e-05, + "loss": 0.1679, + "step": 4540 + }, + { + "epoch": 2.56, + "learning_rate": 1.4875986471251411e-05, + "loss": 0.1967, + "step": 4545 + }, + { + "epoch": 2.56, + "learning_rate": 1.487034949267193e-05, + "loss": 0.1978, + "step": 4550 + }, + { + "epoch": 2.57, + "learning_rate": 1.4864712514092447e-05, + "loss": 0.1667, + "step": 4555 + }, + { + "epoch": 2.57, + "learning_rate": 1.4859075535512965e-05, + "loss": 0.1845, + "step": 4560 + }, + { + "epoch": 2.57, + "learning_rate": 1.4853438556933483e-05, + "loss": 0.1875, + "step": 4565 + }, + { + "epoch": 2.58, + "learning_rate": 1.4847801578354005e-05, + "loss": 0.1812, + "step": 4570 + }, + { + "epoch": 2.58, + "learning_rate": 1.4842164599774523e-05, + "loss": 0.1692, + "step": 4575 + }, + { + "epoch": 2.58, + "learning_rate": 1.483652762119504e-05, + "loss": 0.1683, + "step": 4580 + }, + { + "epoch": 2.58, + "learning_rate": 1.4830890642615559e-05, + "loss": 0.1836, + "step": 4585 + }, + { + "epoch": 2.59, + "learning_rate": 1.4825253664036078e-05, + "loss": 0.1808, + "step": 4590 + }, + { + "epoch": 2.59, + "learning_rate": 1.4819616685456596e-05, + "loss": 0.1591, + "step": 4595 + }, + { + "epoch": 2.59, + "learning_rate": 1.4813979706877116e-05, + "loss": 0.1973, + "step": 4600 + }, + { + "epoch": 2.59, + "eval_accuracy": 0.9007447792435379, + "eval_accuracy_sklearn": 0.9007447792435379, + "eval_f1": 0.9145897038495372, + "eval_loss": 0.25396761298179626, + "eval_precision": 0.8538912788423935, + "eval_recall": 0.984577922077922, + "eval_runtime": 122.6752, + "eval_samples_per_second": 167.458, + "eval_steps_per_second": 10.467, + "step": 4600 + }, + { + "epoch": 2.6, + "learning_rate": 1.4808342728297634e-05, + "loss": 0.184, + "step": 4605 + }, + { + "epoch": 2.6, + "learning_rate": 1.4802705749718152e-05, + "loss": 0.1786, + "step": 4610 + }, + { + "epoch": 2.6, + "learning_rate": 1.4797068771138672e-05, + "loss": 0.1775, + "step": 4615 + }, + { + "epoch": 2.6, + "learning_rate": 1.479143179255919e-05, + "loss": 0.19, + "step": 4620 + }, + { + "epoch": 2.61, + "learning_rate": 1.4785794813979708e-05, + "loss": 0.1755, + "step": 4625 + }, + { + "epoch": 2.61, + "learning_rate": 1.4780157835400226e-05, + "loss": 0.1742, + "step": 4630 + }, + { + "epoch": 2.61, + "learning_rate": 1.4774520856820744e-05, + "loss": 0.1842, + "step": 4635 + }, + { + "epoch": 2.62, + "learning_rate": 1.4768883878241265e-05, + "loss": 0.1846, + "step": 4640 + }, + { + "epoch": 2.62, + "learning_rate": 1.4763246899661783e-05, + "loss": 0.1806, + "step": 4645 + }, + { + "epoch": 2.62, + "learning_rate": 1.4757609921082301e-05, + "loss": 0.1713, + "step": 4650 + }, + { + "epoch": 2.62, + "learning_rate": 1.4751972942502819e-05, + "loss": 0.1876, + "step": 4655 + }, + { + "epoch": 2.63, + "learning_rate": 1.4746335963923337e-05, + "loss": 0.1817, + "step": 4660 + }, + { + "epoch": 2.63, + "learning_rate": 1.4740698985343858e-05, + "loss": 0.1779, + "step": 4665 + }, + { + "epoch": 2.63, + "learning_rate": 1.4735062006764376e-05, + "loss": 0.193, + "step": 4670 + }, + { + "epoch": 2.64, + "learning_rate": 1.4729425028184894e-05, + "loss": 0.187, + "step": 4675 + }, + { + "epoch": 2.64, + "learning_rate": 1.4723788049605412e-05, + "loss": 0.1752, + "step": 4680 + }, + { + "epoch": 2.64, + "learning_rate": 1.4718151071025932e-05, + "loss": 0.1894, + "step": 4685 + }, + { + "epoch": 2.64, + "learning_rate": 1.471251409244645e-05, + "loss": 0.1832, + "step": 4690 + }, + { + "epoch": 2.65, + "learning_rate": 1.4706877113866968e-05, + "loss": 0.1751, + "step": 4695 + }, + { + "epoch": 2.65, + "learning_rate": 1.4701240135287486e-05, + "loss": 0.1622, + "step": 4700 + }, + { + "epoch": 2.65, + "learning_rate": 1.4695603156708006e-05, + "loss": 0.1724, + "step": 4705 + }, + { + "epoch": 2.66, + "learning_rate": 1.4689966178128525e-05, + "loss": 0.1685, + "step": 4710 + }, + { + "epoch": 2.66, + "learning_rate": 1.4684329199549043e-05, + "loss": 0.1817, + "step": 4715 + }, + { + "epoch": 2.66, + "learning_rate": 1.4678692220969561e-05, + "loss": 0.1613, + "step": 4720 + }, + { + "epoch": 2.66, + "learning_rate": 1.467305524239008e-05, + "loss": 0.1784, + "step": 4725 + }, + { + "epoch": 2.67, + "learning_rate": 1.4667418263810597e-05, + "loss": 0.1873, + "step": 4730 + }, + { + "epoch": 2.67, + "learning_rate": 1.4661781285231119e-05, + "loss": 0.1791, + "step": 4735 + }, + { + "epoch": 2.67, + "learning_rate": 1.4656144306651637e-05, + "loss": 0.1955, + "step": 4740 + }, + { + "epoch": 2.67, + "learning_rate": 1.4650507328072155e-05, + "loss": 0.2023, + "step": 4745 + }, + { + "epoch": 2.68, + "learning_rate": 1.4644870349492673e-05, + "loss": 0.1712, + "step": 4750 + }, + { + "epoch": 2.68, + "learning_rate": 1.463923337091319e-05, + "loss": 0.1816, + "step": 4755 + }, + { + "epoch": 2.68, + "learning_rate": 1.463359639233371e-05, + "loss": 0.1687, + "step": 4760 + }, + { + "epoch": 2.69, + "learning_rate": 1.462795941375423e-05, + "loss": 0.173, + "step": 4765 + }, + { + "epoch": 2.69, + "learning_rate": 1.4622322435174748e-05, + "loss": 0.1713, + "step": 4770 + }, + { + "epoch": 2.69, + "learning_rate": 1.4616685456595266e-05, + "loss": 0.1661, + "step": 4775 + }, + { + "epoch": 2.69, + "learning_rate": 1.4611048478015786e-05, + "loss": 0.1795, + "step": 4780 + }, + { + "epoch": 2.7, + "learning_rate": 1.4605411499436304e-05, + "loss": 0.193, + "step": 4785 + }, + { + "epoch": 2.7, + "learning_rate": 1.4599774520856822e-05, + "loss": 0.1958, + "step": 4790 + }, + { + "epoch": 2.7, + "learning_rate": 1.459413754227734e-05, + "loss": 0.1674, + "step": 4795 + }, + { + "epoch": 2.71, + "learning_rate": 1.4588500563697858e-05, + "loss": 0.1682, + "step": 4800 + }, + { + "epoch": 2.71, + "eval_accuracy": 0.9116487367959889, + "eval_accuracy_sklearn": 0.9116487367959889, + "eval_f1": 0.9232330922471768, + "eval_loss": 0.23131902515888214, + "eval_precision": 0.869295101553166, + "eval_recall": 0.9843073593073594, + "eval_runtime": 121.9789, + "eval_samples_per_second": 168.414, + "eval_steps_per_second": 10.526, + "step": 4800 + }, + { + "epoch": 2.71, + "learning_rate": 1.4582863585118379e-05, + "loss": 0.1694, + "step": 4805 + }, + { + "epoch": 2.71, + "learning_rate": 1.4577226606538897e-05, + "loss": 0.1732, + "step": 4810 + }, + { + "epoch": 2.71, + "learning_rate": 1.4571589627959415e-05, + "loss": 0.1717, + "step": 4815 + }, + { + "epoch": 2.72, + "learning_rate": 1.4565952649379933e-05, + "loss": 0.1644, + "step": 4820 + }, + { + "epoch": 2.72, + "learning_rate": 1.456031567080045e-05, + "loss": 0.1687, + "step": 4825 + }, + { + "epoch": 2.72, + "learning_rate": 1.4554678692220972e-05, + "loss": 0.1848, + "step": 4830 + }, + { + "epoch": 2.73, + "learning_rate": 1.454904171364149e-05, + "loss": 0.183, + "step": 4835 + }, + { + "epoch": 2.73, + "learning_rate": 1.4543404735062008e-05, + "loss": 0.1819, + "step": 4840 + }, + { + "epoch": 2.73, + "learning_rate": 1.4537767756482526e-05, + "loss": 0.1792, + "step": 4845 + }, + { + "epoch": 2.73, + "learning_rate": 1.4532130777903044e-05, + "loss": 0.1809, + "step": 4850 + }, + { + "epoch": 2.74, + "learning_rate": 1.4526493799323564e-05, + "loss": 0.1778, + "step": 4855 + }, + { + "epoch": 2.74, + "learning_rate": 1.4520856820744082e-05, + "loss": 0.1897, + "step": 4860 + }, + { + "epoch": 2.74, + "learning_rate": 1.45152198421646e-05, + "loss": 0.1695, + "step": 4865 + }, + { + "epoch": 2.75, + "learning_rate": 1.450958286358512e-05, + "loss": 0.1634, + "step": 4870 + }, + { + "epoch": 2.75, + "learning_rate": 1.450394588500564e-05, + "loss": 0.1628, + "step": 4875 + }, + { + "epoch": 2.75, + "learning_rate": 1.4498308906426157e-05, + "loss": 0.1763, + "step": 4880 + }, + { + "epoch": 2.75, + "learning_rate": 1.4492671927846675e-05, + "loss": 0.1948, + "step": 4885 + }, + { + "epoch": 2.76, + "learning_rate": 1.4487034949267193e-05, + "loss": 0.1981, + "step": 4890 + }, + { + "epoch": 2.76, + "learning_rate": 1.4481397970687711e-05, + "loss": 0.2072, + "step": 4895 + }, + { + "epoch": 2.76, + "learning_rate": 1.4475760992108233e-05, + "loss": 0.2001, + "step": 4900 + }, + { + "epoch": 2.76, + "learning_rate": 1.447012401352875e-05, + "loss": 0.1824, + "step": 4905 + }, + { + "epoch": 2.77, + "learning_rate": 1.4464487034949268e-05, + "loss": 0.1794, + "step": 4910 + }, + { + "epoch": 2.77, + "learning_rate": 1.4458850056369786e-05, + "loss": 0.1905, + "step": 4915 + }, + { + "epoch": 2.77, + "learning_rate": 1.4453213077790304e-05, + "loss": 0.1902, + "step": 4920 + }, + { + "epoch": 2.78, + "learning_rate": 1.4447576099210824e-05, + "loss": 0.1952, + "step": 4925 + }, + { + "epoch": 2.78, + "learning_rate": 1.4441939120631344e-05, + "loss": 0.1732, + "step": 4930 + }, + { + "epoch": 2.78, + "learning_rate": 1.4436302142051862e-05, + "loss": 0.1665, + "step": 4935 + }, + { + "epoch": 2.78, + "learning_rate": 1.443066516347238e-05, + "loss": 0.1833, + "step": 4940 + }, + { + "epoch": 2.79, + "learning_rate": 1.44250281848929e-05, + "loss": 0.1835, + "step": 4945 + }, + { + "epoch": 2.79, + "learning_rate": 1.4419391206313417e-05, + "loss": 0.1634, + "step": 4950 + }, + { + "epoch": 2.79, + "learning_rate": 1.4413754227733935e-05, + "loss": 0.1757, + "step": 4955 + }, + { + "epoch": 2.8, + "learning_rate": 1.4408117249154453e-05, + "loss": 0.1926, + "step": 4960 + }, + { + "epoch": 2.8, + "learning_rate": 1.4402480270574971e-05, + "loss": 0.1772, + "step": 4965 + }, + { + "epoch": 2.8, + "learning_rate": 1.4396843291995493e-05, + "loss": 0.1866, + "step": 4970 + }, + { + "epoch": 2.8, + "learning_rate": 1.439120631341601e-05, + "loss": 0.1727, + "step": 4975 + }, + { + "epoch": 2.81, + "learning_rate": 1.4385569334836529e-05, + "loss": 0.1665, + "step": 4980 + }, + { + "epoch": 2.81, + "learning_rate": 1.4379932356257047e-05, + "loss": 0.1704, + "step": 4985 + }, + { + "epoch": 2.81, + "learning_rate": 1.4374295377677565e-05, + "loss": 0.1752, + "step": 4990 + }, + { + "epoch": 2.82, + "learning_rate": 1.4368658399098086e-05, + "loss": 0.1722, + "step": 4995 + }, + { + "epoch": 2.82, + "learning_rate": 1.4363021420518604e-05, + "loss": 0.1898, + "step": 5000 + }, + { + "epoch": 2.82, + "eval_accuracy": 0.9163705398432556, + "eval_accuracy_sklearn": 0.9163705398432556, + "eval_f1": 0.9268438085505025, + "eval_loss": 0.20223605632781982, + "eval_precision": 0.8779444982252339, + "eval_recall": 0.9815115440115441, + "eval_runtime": 122.3884, + "eval_samples_per_second": 167.851, + "eval_steps_per_second": 10.491, + "step": 5000 + }, + { + "epoch": 2.82, + "learning_rate": 1.4357384441939122e-05, + "loss": 0.1741, + "step": 5005 + }, + { + "epoch": 2.82, + "learning_rate": 1.435174746335964e-05, + "loss": 0.1666, + "step": 5010 + }, + { + "epoch": 2.83, + "learning_rate": 1.4346110484780158e-05, + "loss": 0.175, + "step": 5015 + }, + { + "epoch": 2.83, + "learning_rate": 1.4340473506200678e-05, + "loss": 0.1711, + "step": 5020 + }, + { + "epoch": 2.83, + "learning_rate": 1.4334836527621196e-05, + "loss": 0.164, + "step": 5025 + }, + { + "epoch": 2.84, + "learning_rate": 1.4329199549041715e-05, + "loss": 0.1597, + "step": 5030 + }, + { + "epoch": 2.84, + "learning_rate": 1.4323562570462233e-05, + "loss": 0.1914, + "step": 5035 + }, + { + "epoch": 2.84, + "learning_rate": 1.4317925591882753e-05, + "loss": 0.17, + "step": 5040 + }, + { + "epoch": 2.84, + "learning_rate": 1.4312288613303271e-05, + "loss": 0.168, + "step": 5045 + }, + { + "epoch": 2.85, + "learning_rate": 1.4306651634723789e-05, + "loss": 0.1796, + "step": 5050 + }, + { + "epoch": 2.85, + "learning_rate": 1.4301014656144307e-05, + "loss": 0.1735, + "step": 5055 + }, + { + "epoch": 2.85, + "learning_rate": 1.4295377677564825e-05, + "loss": 0.1693, + "step": 5060 + }, + { + "epoch": 2.86, + "learning_rate": 1.4289740698985346e-05, + "loss": 0.1634, + "step": 5065 + }, + { + "epoch": 2.86, + "learning_rate": 1.4284103720405864e-05, + "loss": 0.1809, + "step": 5070 + }, + { + "epoch": 2.86, + "learning_rate": 1.4278466741826382e-05, + "loss": 0.165, + "step": 5075 + }, + { + "epoch": 2.86, + "learning_rate": 1.42728297632469e-05, + "loss": 0.1701, + "step": 5080 + }, + { + "epoch": 2.87, + "learning_rate": 1.4267192784667418e-05, + "loss": 0.181, + "step": 5085 + }, + { + "epoch": 2.87, + "learning_rate": 1.4261555806087938e-05, + "loss": 0.1717, + "step": 5090 + }, + { + "epoch": 2.87, + "learning_rate": 1.4255918827508458e-05, + "loss": 0.156, + "step": 5095 + }, + { + "epoch": 2.87, + "learning_rate": 1.4250281848928976e-05, + "loss": 0.1771, + "step": 5100 + }, + { + "epoch": 2.88, + "learning_rate": 1.4244644870349494e-05, + "loss": 0.1771, + "step": 5105 + }, + { + "epoch": 2.88, + "learning_rate": 1.4239007891770012e-05, + "loss": 0.1799, + "step": 5110 + }, + { + "epoch": 2.88, + "learning_rate": 1.4233370913190531e-05, + "loss": 0.1703, + "step": 5115 + }, + { + "epoch": 2.89, + "learning_rate": 1.422773393461105e-05, + "loss": 0.1658, + "step": 5120 + }, + { + "epoch": 2.89, + "learning_rate": 1.4222096956031567e-05, + "loss": 0.1768, + "step": 5125 + }, + { + "epoch": 2.89, + "learning_rate": 1.4216459977452085e-05, + "loss": 0.1757, + "step": 5130 + }, + { + "epoch": 2.89, + "learning_rate": 1.4210822998872607e-05, + "loss": 0.1722, + "step": 5135 + }, + { + "epoch": 2.9, + "learning_rate": 1.4205186020293125e-05, + "loss": 0.1886, + "step": 5140 + }, + { + "epoch": 2.9, + "learning_rate": 1.4199549041713643e-05, + "loss": 0.1724, + "step": 5145 + }, + { + "epoch": 2.9, + "learning_rate": 1.419391206313416e-05, + "loss": 0.1829, + "step": 5150 + }, + { + "epoch": 2.91, + "learning_rate": 1.4188275084554679e-05, + "loss": 0.176, + "step": 5155 + }, + { + "epoch": 2.91, + "learning_rate": 1.41826381059752e-05, + "loss": 0.1721, + "step": 5160 + }, + { + "epoch": 2.91, + "learning_rate": 1.4177001127395718e-05, + "loss": 0.174, + "step": 5165 + }, + { + "epoch": 2.91, + "learning_rate": 1.4171364148816236e-05, + "loss": 0.1804, + "step": 5170 + }, + { + "epoch": 2.92, + "learning_rate": 1.4165727170236754e-05, + "loss": 0.181, + "step": 5175 + }, + { + "epoch": 2.92, + "learning_rate": 1.4160090191657272e-05, + "loss": 0.1833, + "step": 5180 + }, + { + "epoch": 2.92, + "learning_rate": 1.4154453213077792e-05, + "loss": 0.1768, + "step": 5185 + }, + { + "epoch": 2.93, + "learning_rate": 1.414881623449831e-05, + "loss": 0.1835, + "step": 5190 + }, + { + "epoch": 2.93, + "learning_rate": 1.414317925591883e-05, + "loss": 0.1686, + "step": 5195 + }, + { + "epoch": 2.93, + "learning_rate": 1.4137542277339347e-05, + "loss": 0.1731, + "step": 5200 + }, + { + "epoch": 2.93, + "eval_accuracy": 0.9128656963442535, + "eval_accuracy_sklearn": 0.9128656963442535, + "eval_f1": 0.9243448858833473, + "eval_loss": 0.2073081135749817, + "eval_precision": 0.8697900095450207, + "eval_recall": 0.9862012987012987, + "eval_runtime": 118.9558, + "eval_samples_per_second": 172.694, + "eval_steps_per_second": 10.794, + "step": 5200 + }, + { + "epoch": 2.93, + "learning_rate": 1.4131905298759865e-05, + "loss": 0.1636, + "step": 5205 + }, + { + "epoch": 2.94, + "learning_rate": 1.4126268320180385e-05, + "loss": 0.1641, + "step": 5210 + }, + { + "epoch": 2.94, + "learning_rate": 1.4120631341600903e-05, + "loss": 0.1818, + "step": 5215 + }, + { + "epoch": 2.94, + "learning_rate": 1.4114994363021421e-05, + "loss": 0.1727, + "step": 5220 + }, + { + "epoch": 2.95, + "learning_rate": 1.4109357384441939e-05, + "loss": 0.1702, + "step": 5225 + }, + { + "epoch": 2.95, + "learning_rate": 1.410372040586246e-05, + "loss": 0.1701, + "step": 5230 + }, + { + "epoch": 2.95, + "learning_rate": 1.4098083427282978e-05, + "loss": 0.1726, + "step": 5235 + }, + { + "epoch": 2.95, + "learning_rate": 1.4092446448703496e-05, + "loss": 0.1658, + "step": 5240 + }, + { + "epoch": 2.96, + "learning_rate": 1.4086809470124014e-05, + "loss": 0.1833, + "step": 5245 + }, + { + "epoch": 2.96, + "learning_rate": 1.4081172491544532e-05, + "loss": 0.1727, + "step": 5250 + }, + { + "epoch": 2.96, + "learning_rate": 1.4075535512965054e-05, + "loss": 0.1568, + "step": 5255 + }, + { + "epoch": 2.97, + "learning_rate": 1.4069898534385572e-05, + "loss": 0.185, + "step": 5260 + }, + { + "epoch": 2.97, + "learning_rate": 1.406426155580609e-05, + "loss": 0.1843, + "step": 5265 + }, + { + "epoch": 2.97, + "learning_rate": 1.4058624577226608e-05, + "loss": 0.1643, + "step": 5270 + }, + { + "epoch": 2.97, + "learning_rate": 1.4052987598647126e-05, + "loss": 0.1773, + "step": 5275 + }, + { + "epoch": 2.98, + "learning_rate": 1.4047350620067645e-05, + "loss": 0.1873, + "step": 5280 + }, + { + "epoch": 2.98, + "learning_rate": 1.4041713641488163e-05, + "loss": 0.1682, + "step": 5285 + }, + { + "epoch": 2.98, + "learning_rate": 1.4036076662908681e-05, + "loss": 0.1539, + "step": 5290 + }, + { + "epoch": 2.98, + "learning_rate": 1.40304396843292e-05, + "loss": 0.1786, + "step": 5295 + }, + { + "epoch": 2.99, + "learning_rate": 1.4024802705749719e-05, + "loss": 0.1658, + "step": 5300 + }, + { + "epoch": 2.99, + "learning_rate": 1.4019165727170239e-05, + "loss": 0.1561, + "step": 5305 + }, + { + "epoch": 2.99, + "learning_rate": 1.4013528748590757e-05, + "loss": 0.1693, + "step": 5310 + }, + { + "epoch": 3.0, + "learning_rate": 1.4007891770011275e-05, + "loss": 0.1623, + "step": 5315 + }, + { + "epoch": 3.0, + "learning_rate": 1.4002254791431793e-05, + "loss": 0.1837, + "step": 5320 + }, + { + "epoch": 3.0, + "learning_rate": 1.3996617812852314e-05, + "loss": 0.1628, + "step": 5325 + }, + { + "epoch": 3.0, + "learning_rate": 1.3990980834272832e-05, + "loss": 0.1455, + "step": 5330 + }, + { + "epoch": 3.01, + "learning_rate": 1.398534385569335e-05, + "loss": 0.1656, + "step": 5335 + }, + { + "epoch": 3.01, + "learning_rate": 1.3979706877113868e-05, + "loss": 0.1609, + "step": 5340 + }, + { + "epoch": 3.01, + "learning_rate": 1.3974069898534386e-05, + "loss": 0.1573, + "step": 5345 + }, + { + "epoch": 3.02, + "learning_rate": 1.3968432919954906e-05, + "loss": 0.1569, + "step": 5350 + }, + { + "epoch": 3.02, + "learning_rate": 1.3962795941375424e-05, + "loss": 0.1518, + "step": 5355 + }, + { + "epoch": 3.02, + "learning_rate": 1.3957158962795943e-05, + "loss": 0.1489, + "step": 5360 + }, + { + "epoch": 3.02, + "learning_rate": 1.3951521984216461e-05, + "loss": 0.1543, + "step": 5365 + }, + { + "epoch": 3.03, + "learning_rate": 1.394588500563698e-05, + "loss": 0.1554, + "step": 5370 + }, + { + "epoch": 3.03, + "learning_rate": 1.3940248027057499e-05, + "loss": 0.168, + "step": 5375 + }, + { + "epoch": 3.03, + "learning_rate": 1.3934611048478017e-05, + "loss": 0.1747, + "step": 5380 + }, + { + "epoch": 3.04, + "learning_rate": 1.3928974069898535e-05, + "loss": 0.1627, + "step": 5385 + }, + { + "epoch": 3.04, + "learning_rate": 1.3923337091319053e-05, + "loss": 0.1603, + "step": 5390 + }, + { + "epoch": 3.04, + "learning_rate": 1.3917700112739571e-05, + "loss": 0.148, + "step": 5395 + }, + { + "epoch": 3.04, + "learning_rate": 1.3912063134160092e-05, + "loss": 0.1618, + "step": 5400 + }, + { + "epoch": 3.04, + "eval_accuracy": 0.9301465219296111, + "eval_accuracy_sklearn": 0.9301465219296111, + "eval_f1": 0.9378707191410139, + "eval_loss": 0.18178626894950867, + "eval_precision": 0.9019069031559663, + "eval_recall": 0.9768217893217893, + "eval_runtime": 117.2665, + "eval_samples_per_second": 175.182, + "eval_steps_per_second": 10.949, + "step": 5400 + }, + { + "epoch": 3.05, + "learning_rate": 1.390642615558061e-05, + "loss": 0.1488, + "step": 5405 + }, + { + "epoch": 3.05, + "learning_rate": 1.3900789177001128e-05, + "loss": 0.1492, + "step": 5410 + }, + { + "epoch": 3.05, + "learning_rate": 1.3895152198421646e-05, + "loss": 0.1541, + "step": 5415 + }, + { + "epoch": 3.06, + "learning_rate": 1.3889515219842168e-05, + "loss": 0.1691, + "step": 5420 + }, + { + "epoch": 3.06, + "learning_rate": 1.3883878241262686e-05, + "loss": 0.1593, + "step": 5425 + }, + { + "epoch": 3.06, + "learning_rate": 1.3878241262683204e-05, + "loss": 0.1697, + "step": 5430 + }, + { + "epoch": 3.06, + "learning_rate": 1.3872604284103722e-05, + "loss": 0.1689, + "step": 5435 + }, + { + "epoch": 3.07, + "learning_rate": 1.386696730552424e-05, + "loss": 0.1506, + "step": 5440 + }, + { + "epoch": 3.07, + "learning_rate": 1.386133032694476e-05, + "loss": 0.1576, + "step": 5445 + }, + { + "epoch": 3.07, + "learning_rate": 1.3855693348365277e-05, + "loss": 0.1515, + "step": 5450 + }, + { + "epoch": 3.07, + "learning_rate": 1.3850056369785795e-05, + "loss": 0.1515, + "step": 5455 + }, + { + "epoch": 3.08, + "learning_rate": 1.3844419391206315e-05, + "loss": 0.1615, + "step": 5460 + }, + { + "epoch": 3.08, + "learning_rate": 1.3838782412626833e-05, + "loss": 0.1511, + "step": 5465 + }, + { + "epoch": 3.08, + "learning_rate": 1.3833145434047353e-05, + "loss": 0.1591, + "step": 5470 + }, + { + "epoch": 3.09, + "learning_rate": 1.382750845546787e-05, + "loss": 0.1592, + "step": 5475 + }, + { + "epoch": 3.09, + "learning_rate": 1.3821871476888389e-05, + "loss": 0.1553, + "step": 5480 + }, + { + "epoch": 3.09, + "learning_rate": 1.3816234498308907e-05, + "loss": 0.1384, + "step": 5485 + }, + { + "epoch": 3.09, + "learning_rate": 1.3810597519729425e-05, + "loss": 0.16, + "step": 5490 + }, + { + "epoch": 3.1, + "learning_rate": 1.3804960541149946e-05, + "loss": 0.1503, + "step": 5495 + }, + { + "epoch": 3.1, + "learning_rate": 1.3799323562570464e-05, + "loss": 0.161, + "step": 5500 + }, + { + "epoch": 3.1, + "learning_rate": 1.3793686583990982e-05, + "loss": 0.1762, + "step": 5505 + }, + { + "epoch": 3.11, + "learning_rate": 1.37880496054115e-05, + "loss": 0.1538, + "step": 5510 + }, + { + "epoch": 3.11, + "learning_rate": 1.378241262683202e-05, + "loss": 0.1602, + "step": 5515 + }, + { + "epoch": 3.11, + "learning_rate": 1.3776775648252538e-05, + "loss": 0.159, + "step": 5520 + }, + { + "epoch": 3.11, + "learning_rate": 1.3771138669673057e-05, + "loss": 0.1616, + "step": 5525 + }, + { + "epoch": 3.12, + "learning_rate": 1.3765501691093575e-05, + "loss": 0.1618, + "step": 5530 + }, + { + "epoch": 3.12, + "learning_rate": 1.3759864712514093e-05, + "loss": 0.1772, + "step": 5535 + }, + { + "epoch": 3.12, + "learning_rate": 1.3754227733934613e-05, + "loss": 0.1831, + "step": 5540 + }, + { + "epoch": 3.13, + "learning_rate": 1.374859075535513e-05, + "loss": 0.1499, + "step": 5545 + }, + { + "epoch": 3.13, + "learning_rate": 1.3742953776775649e-05, + "loss": 0.1726, + "step": 5550 + }, + { + "epoch": 3.13, + "learning_rate": 1.3737316798196167e-05, + "loss": 0.1471, + "step": 5555 + }, + { + "epoch": 3.13, + "learning_rate": 1.3731679819616685e-05, + "loss": 0.1385, + "step": 5560 + }, + { + "epoch": 3.14, + "learning_rate": 1.3726042841037206e-05, + "loss": 0.1572, + "step": 5565 + }, + { + "epoch": 3.14, + "learning_rate": 1.3720405862457724e-05, + "loss": 0.1539, + "step": 5570 + }, + { + "epoch": 3.14, + "learning_rate": 1.3714768883878242e-05, + "loss": 0.1582, + "step": 5575 + }, + { + "epoch": 3.15, + "learning_rate": 1.370913190529876e-05, + "loss": 0.1491, + "step": 5580 + }, + { + "epoch": 3.15, + "learning_rate": 1.3703494926719278e-05, + "loss": 0.1706, + "step": 5585 + }, + { + "epoch": 3.15, + "learning_rate": 1.36978579481398e-05, + "loss": 0.1409, + "step": 5590 + }, + { + "epoch": 3.15, + "learning_rate": 1.3692220969560317e-05, + "loss": 0.1551, + "step": 5595 + }, + { + "epoch": 3.16, + "learning_rate": 1.3686583990980835e-05, + "loss": 0.1613, + "step": 5600 + }, + { + "epoch": 3.16, + "eval_accuracy": 0.912817017962323, + "eval_accuracy_sklearn": 0.912817017962323, + "eval_f1": 0.9243250095069083, + "eval_loss": 0.22527408599853516, + "eval_precision": 0.8695444788933937, + "eval_recall": 0.9864718614718615, + "eval_runtime": 122.512, + "eval_samples_per_second": 167.682, + "eval_steps_per_second": 10.481, + "step": 5600 + }, + { + "epoch": 3.16, + "learning_rate": 1.3680947012401353e-05, + "loss": 0.1635, + "step": 5605 + }, + { + "epoch": 3.16, + "learning_rate": 1.3675310033821873e-05, + "loss": 0.142, + "step": 5610 + }, + { + "epoch": 3.17, + "learning_rate": 1.3669673055242391e-05, + "loss": 0.1469, + "step": 5615 + }, + { + "epoch": 3.17, + "learning_rate": 1.3664036076662909e-05, + "loss": 0.1667, + "step": 5620 + }, + { + "epoch": 3.17, + "learning_rate": 1.3658399098083429e-05, + "loss": 0.1622, + "step": 5625 + }, + { + "epoch": 3.17, + "learning_rate": 1.3652762119503947e-05, + "loss": 0.1677, + "step": 5630 + }, + { + "epoch": 3.18, + "learning_rate": 1.3647125140924466e-05, + "loss": 0.1556, + "step": 5635 + }, + { + "epoch": 3.18, + "learning_rate": 1.3641488162344984e-05, + "loss": 0.1451, + "step": 5640 + }, + { + "epoch": 3.18, + "learning_rate": 1.3635851183765502e-05, + "loss": 0.1532, + "step": 5645 + }, + { + "epoch": 3.18, + "learning_rate": 1.363021420518602e-05, + "loss": 0.1531, + "step": 5650 + }, + { + "epoch": 3.19, + "learning_rate": 1.3624577226606538e-05, + "loss": 0.1559, + "step": 5655 + }, + { + "epoch": 3.19, + "learning_rate": 1.361894024802706e-05, + "loss": 0.143, + "step": 5660 + }, + { + "epoch": 3.19, + "learning_rate": 1.3613303269447578e-05, + "loss": 0.1522, + "step": 5665 + }, + { + "epoch": 3.2, + "learning_rate": 1.3607666290868096e-05, + "loss": 0.168, + "step": 5670 + }, + { + "epoch": 3.2, + "learning_rate": 1.3602029312288614e-05, + "loss": 0.1596, + "step": 5675 + }, + { + "epoch": 3.2, + "learning_rate": 1.3596392333709132e-05, + "loss": 0.158, + "step": 5680 + }, + { + "epoch": 3.2, + "learning_rate": 1.3590755355129653e-05, + "loss": 0.1654, + "step": 5685 + }, + { + "epoch": 3.21, + "learning_rate": 1.3585118376550171e-05, + "loss": 0.1466, + "step": 5690 + }, + { + "epoch": 3.21, + "learning_rate": 1.3579481397970689e-05, + "loss": 0.1572, + "step": 5695 + }, + { + "epoch": 3.21, + "learning_rate": 1.3573844419391207e-05, + "loss": 0.1612, + "step": 5700 + }, + { + "epoch": 3.22, + "learning_rate": 1.3568207440811727e-05, + "loss": 0.1423, + "step": 5705 + }, + { + "epoch": 3.22, + "learning_rate": 1.3562570462232245e-05, + "loss": 0.144, + "step": 5710 + }, + { + "epoch": 3.22, + "learning_rate": 1.3556933483652763e-05, + "loss": 0.1792, + "step": 5715 + }, + { + "epoch": 3.22, + "learning_rate": 1.355129650507328e-05, + "loss": 0.1545, + "step": 5720 + }, + { + "epoch": 3.23, + "learning_rate": 1.3545659526493799e-05, + "loss": 0.1481, + "step": 5725 + }, + { + "epoch": 3.23, + "learning_rate": 1.354002254791432e-05, + "loss": 0.1634, + "step": 5730 + }, + { + "epoch": 3.23, + "learning_rate": 1.3534385569334838e-05, + "loss": 0.1511, + "step": 5735 + }, + { + "epoch": 3.24, + "learning_rate": 1.3528748590755356e-05, + "loss": 0.1371, + "step": 5740 + }, + { + "epoch": 3.24, + "learning_rate": 1.3523111612175874e-05, + "loss": 0.1479, + "step": 5745 + }, + { + "epoch": 3.24, + "learning_rate": 1.3517474633596392e-05, + "loss": 0.1535, + "step": 5750 + }, + { + "epoch": 3.24, + "learning_rate": 1.3511837655016913e-05, + "loss": 0.1364, + "step": 5755 + }, + { + "epoch": 3.25, + "learning_rate": 1.3506200676437431e-05, + "loss": 0.1659, + "step": 5760 + }, + { + "epoch": 3.25, + "learning_rate": 1.350056369785795e-05, + "loss": 0.1484, + "step": 5765 + }, + { + "epoch": 3.25, + "learning_rate": 1.3494926719278467e-05, + "loss": 0.1511, + "step": 5770 + }, + { + "epoch": 3.26, + "learning_rate": 1.3489289740698987e-05, + "loss": 0.1585, + "step": 5775 + }, + { + "epoch": 3.26, + "learning_rate": 1.3483652762119505e-05, + "loss": 0.1663, + "step": 5780 + }, + { + "epoch": 3.26, + "learning_rate": 1.3478015783540023e-05, + "loss": 0.1548, + "step": 5785 + }, + { + "epoch": 3.26, + "learning_rate": 1.3472378804960543e-05, + "loss": 0.167, + "step": 5790 + }, + { + "epoch": 3.27, + "learning_rate": 1.346674182638106e-05, + "loss": 0.1583, + "step": 5795 + }, + { + "epoch": 3.27, + "learning_rate": 1.346110484780158e-05, + "loss": 0.1573, + "step": 5800 + }, + { + "epoch": 3.27, + "eval_accuracy": 0.9257167891739279, + "eval_accuracy_sklearn": 0.9257167891739279, + "eval_f1": 0.9346914319952068, + "eval_loss": 0.1873285174369812, + "eval_precision": 0.8893956670467503, + "eval_recall": 0.9848484848484849, + "eval_runtime": 122.6111, + "eval_samples_per_second": 167.546, + "eval_steps_per_second": 10.472, + "step": 5800 + }, + { + "epoch": 3.27, + "learning_rate": 1.3455467869222098e-05, + "loss": 0.148, + "step": 5805 + }, + { + "epoch": 3.28, + "learning_rate": 1.3449830890642616e-05, + "loss": 0.1528, + "step": 5810 + }, + { + "epoch": 3.28, + "learning_rate": 1.3444193912063134e-05, + "loss": 0.1448, + "step": 5815 + }, + { + "epoch": 3.28, + "learning_rate": 1.3438556933483652e-05, + "loss": 0.1731, + "step": 5820 + }, + { + "epoch": 3.28, + "learning_rate": 1.3432919954904174e-05, + "loss": 0.1591, + "step": 5825 + }, + { + "epoch": 3.29, + "learning_rate": 1.3427282976324692e-05, + "loss": 0.138, + "step": 5830 + }, + { + "epoch": 3.29, + "learning_rate": 1.342164599774521e-05, + "loss": 0.1605, + "step": 5835 + }, + { + "epoch": 3.29, + "learning_rate": 1.3416009019165728e-05, + "loss": 0.1568, + "step": 5840 + }, + { + "epoch": 3.29, + "learning_rate": 1.3410372040586246e-05, + "loss": 0.159, + "step": 5845 + }, + { + "epoch": 3.3, + "learning_rate": 1.3404735062006767e-05, + "loss": 0.1632, + "step": 5850 + }, + { + "epoch": 3.3, + "learning_rate": 1.3399098083427285e-05, + "loss": 0.1597, + "step": 5855 + }, + { + "epoch": 3.3, + "learning_rate": 1.3393461104847803e-05, + "loss": 0.1641, + "step": 5860 + }, + { + "epoch": 3.31, + "learning_rate": 1.3387824126268321e-05, + "loss": 0.1567, + "step": 5865 + }, + { + "epoch": 3.31, + "learning_rate": 1.338218714768884e-05, + "loss": 0.1418, + "step": 5870 + }, + { + "epoch": 3.31, + "learning_rate": 1.3376550169109359e-05, + "loss": 0.1479, + "step": 5875 + }, + { + "epoch": 3.31, + "learning_rate": 1.3370913190529877e-05, + "loss": 0.1582, + "step": 5880 + }, + { + "epoch": 3.32, + "learning_rate": 1.3365276211950395e-05, + "loss": 0.1544, + "step": 5885 + }, + { + "epoch": 3.32, + "learning_rate": 1.3359639233370914e-05, + "loss": 0.1386, + "step": 5890 + }, + { + "epoch": 3.32, + "learning_rate": 1.3354002254791434e-05, + "loss": 0.148, + "step": 5895 + }, + { + "epoch": 3.33, + "learning_rate": 1.3348365276211952e-05, + "loss": 0.1657, + "step": 5900 + }, + { + "epoch": 3.33, + "learning_rate": 1.334272829763247e-05, + "loss": 0.1354, + "step": 5905 + }, + { + "epoch": 3.33, + "learning_rate": 1.3337091319052988e-05, + "loss": 0.16, + "step": 5910 + }, + { + "epoch": 3.33, + "learning_rate": 1.3331454340473506e-05, + "loss": 0.1588, + "step": 5915 + }, + { + "epoch": 3.34, + "learning_rate": 1.3325817361894027e-05, + "loss": 0.1575, + "step": 5920 + }, + { + "epoch": 3.34, + "learning_rate": 1.3320180383314545e-05, + "loss": 0.1678, + "step": 5925 + }, + { + "epoch": 3.34, + "learning_rate": 1.3314543404735063e-05, + "loss": 0.156, + "step": 5930 + }, + { + "epoch": 3.35, + "learning_rate": 1.3308906426155581e-05, + "loss": 0.1669, + "step": 5935 + }, + { + "epoch": 3.35, + "learning_rate": 1.33032694475761e-05, + "loss": 0.1426, + "step": 5940 + }, + { + "epoch": 3.35, + "learning_rate": 1.3297632468996619e-05, + "loss": 0.1614, + "step": 5945 + }, + { + "epoch": 3.35, + "learning_rate": 1.3291995490417137e-05, + "loss": 0.1411, + "step": 5950 + }, + { + "epoch": 3.36, + "learning_rate": 1.3286358511837657e-05, + "loss": 0.1575, + "step": 5955 + }, + { + "epoch": 3.36, + "learning_rate": 1.3280721533258175e-05, + "loss": 0.1622, + "step": 5960 + }, + { + "epoch": 3.36, + "learning_rate": 1.3275084554678694e-05, + "loss": 0.1509, + "step": 5965 + }, + { + "epoch": 3.37, + "learning_rate": 1.3269447576099212e-05, + "loss": 0.1474, + "step": 5970 + }, + { + "epoch": 3.37, + "learning_rate": 1.326381059751973e-05, + "loss": 0.1517, + "step": 5975 + }, + { + "epoch": 3.37, + "learning_rate": 1.3258173618940248e-05, + "loss": 0.1508, + "step": 5980 + }, + { + "epoch": 3.37, + "learning_rate": 1.3252536640360766e-05, + "loss": 0.1562, + "step": 5985 + }, + { + "epoch": 3.38, + "learning_rate": 1.3246899661781288e-05, + "loss": 0.1619, + "step": 5990 + }, + { + "epoch": 3.38, + "learning_rate": 1.3241262683201806e-05, + "loss": 0.1476, + "step": 5995 + }, + { + "epoch": 3.38, + "learning_rate": 1.3235625704622324e-05, + "loss": 0.1536, + "step": 6000 + }, + { + "epoch": 3.38, + "eval_accuracy": 0.9334566519008908, + "eval_accuracy_sklearn": 0.9334566519008908, + "eval_f1": 0.9408915985644485, + "eval_loss": 0.1667889952659607, + "eval_precision": 0.9037295456433259, + "eval_recall": 0.9812409812409812, + "eval_runtime": 117.3242, + "eval_samples_per_second": 175.096, + "eval_steps_per_second": 10.944, + "step": 6000 + }, + { + "epoch": 3.39, + "learning_rate": 1.3229988726042842e-05, + "loss": 0.1606, + "step": 6005 + }, + { + "epoch": 3.39, + "learning_rate": 1.322435174746336e-05, + "loss": 0.135, + "step": 6010 + }, + { + "epoch": 3.39, + "learning_rate": 1.3218714768883881e-05, + "loss": 0.154, + "step": 6015 + }, + { + "epoch": 3.39, + "learning_rate": 1.3213077790304399e-05, + "loss": 0.1588, + "step": 6020 + }, + { + "epoch": 3.4, + "learning_rate": 1.3207440811724917e-05, + "loss": 0.1241, + "step": 6025 + }, + { + "epoch": 3.4, + "learning_rate": 1.3201803833145435e-05, + "loss": 0.1563, + "step": 6030 + }, + { + "epoch": 3.4, + "learning_rate": 1.3196166854565953e-05, + "loss": 0.139, + "step": 6035 + }, + { + "epoch": 3.4, + "learning_rate": 1.3190529875986473e-05, + "loss": 0.1504, + "step": 6040 + }, + { + "epoch": 3.41, + "learning_rate": 1.318489289740699e-05, + "loss": 0.1663, + "step": 6045 + }, + { + "epoch": 3.41, + "learning_rate": 1.3179255918827509e-05, + "loss": 0.1468, + "step": 6050 + }, + { + "epoch": 3.41, + "learning_rate": 1.3173618940248028e-05, + "loss": 0.1494, + "step": 6055 + }, + { + "epoch": 3.42, + "learning_rate": 1.3167981961668548e-05, + "loss": 0.148, + "step": 6060 + }, + { + "epoch": 3.42, + "learning_rate": 1.3162344983089066e-05, + "loss": 0.1451, + "step": 6065 + }, + { + "epoch": 3.42, + "learning_rate": 1.3156708004509584e-05, + "loss": 0.1407, + "step": 6070 + }, + { + "epoch": 3.42, + "learning_rate": 1.3151071025930102e-05, + "loss": 0.1423, + "step": 6075 + }, + { + "epoch": 3.43, + "learning_rate": 1.314543404735062e-05, + "loss": 0.1452, + "step": 6080 + }, + { + "epoch": 3.43, + "learning_rate": 1.3139797068771141e-05, + "loss": 0.1553, + "step": 6085 + }, + { + "epoch": 3.43, + "learning_rate": 1.313416009019166e-05, + "loss": 0.1543, + "step": 6090 + }, + { + "epoch": 3.44, + "learning_rate": 1.3128523111612177e-05, + "loss": 0.1469, + "step": 6095 + }, + { + "epoch": 3.44, + "learning_rate": 1.3122886133032695e-05, + "loss": 0.1492, + "step": 6100 + }, + { + "epoch": 3.44, + "learning_rate": 1.3117249154453213e-05, + "loss": 0.1479, + "step": 6105 + }, + { + "epoch": 3.44, + "learning_rate": 1.3111612175873733e-05, + "loss": 0.1449, + "step": 6110 + }, + { + "epoch": 3.45, + "learning_rate": 1.3105975197294253e-05, + "loss": 0.154, + "step": 6115 + }, + { + "epoch": 3.45, + "learning_rate": 1.310033821871477e-05, + "loss": 0.1431, + "step": 6120 + }, + { + "epoch": 3.45, + "learning_rate": 1.3094701240135289e-05, + "loss": 0.1506, + "step": 6125 + }, + { + "epoch": 3.46, + "learning_rate": 1.3089064261555807e-05, + "loss": 0.1513, + "step": 6130 + }, + { + "epoch": 3.46, + "learning_rate": 1.3083427282976326e-05, + "loss": 0.1508, + "step": 6135 + }, + { + "epoch": 3.46, + "learning_rate": 1.3077790304396844e-05, + "loss": 0.1442, + "step": 6140 + }, + { + "epoch": 3.46, + "learning_rate": 1.3072153325817362e-05, + "loss": 0.1621, + "step": 6145 + }, + { + "epoch": 3.47, + "learning_rate": 1.306651634723788e-05, + "loss": 0.1518, + "step": 6150 + }, + { + "epoch": 3.47, + "learning_rate": 1.3060879368658402e-05, + "loss": 0.1442, + "step": 6155 + }, + { + "epoch": 3.47, + "learning_rate": 1.305524239007892e-05, + "loss": 0.1437, + "step": 6160 + }, + { + "epoch": 3.48, + "learning_rate": 1.3049605411499438e-05, + "loss": 0.1432, + "step": 6165 + }, + { + "epoch": 3.48, + "learning_rate": 1.3043968432919956e-05, + "loss": 0.1515, + "step": 6170 + }, + { + "epoch": 3.48, + "learning_rate": 1.3038331454340473e-05, + "loss": 0.1535, + "step": 6175 + }, + { + "epoch": 3.48, + "learning_rate": 1.3032694475760995e-05, + "loss": 0.1423, + "step": 6180 + }, + { + "epoch": 3.49, + "learning_rate": 1.3027057497181513e-05, + "loss": 0.1529, + "step": 6185 + }, + { + "epoch": 3.49, + "learning_rate": 1.3021420518602031e-05, + "loss": 0.161, + "step": 6190 + }, + { + "epoch": 3.49, + "learning_rate": 1.3015783540022549e-05, + "loss": 0.1628, + "step": 6195 + }, + { + "epoch": 3.49, + "learning_rate": 1.3010146561443067e-05, + "loss": 0.1456, + "step": 6200 + }, + { + "epoch": 3.49, + "eval_accuracy": 0.9362313196709341, + "eval_accuracy_sklearn": 0.9362313196709341, + "eval_f1": 0.9432162982228003, + "eval_loss": 0.16281718015670776, + "eval_precision": 0.9080287097312636, + "eval_recall": 0.9812409812409812, + "eval_runtime": 136.2073, + "eval_samples_per_second": 150.822, + "eval_steps_per_second": 9.427, + "step": 6200 + } + ], + "max_steps": 17740, + "num_train_epochs": 10, + "total_flos": 8.351676381458227e+17, + "trial_name": null, + "trial_params": null +}