{ "best_metric": 0.15952083840732592, "best_model_checkpoint": "esm2_t30_150M_qlora_ptm_sites_2023-10-19_22-31-34/checkpoint-48919", "epoch": 1.0, "eval_steps": 500, "global_step": 48919, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0003701416918864124, "loss": 0.2635, "step": 200 }, { "epoch": 0.01, "learning_rate": 0.00037009604913071327, "loss": 0.0714, "step": 400 }, { "epoch": 0.01, "learning_rate": 0.00037001988446013227, "loss": 0.056, "step": 600 }, { "epoch": 0.02, "learning_rate": 0.00036991381967446495, "loss": 0.0526, "step": 800 }, { "epoch": 0.02, "learning_rate": 0.00036977680630794575, "loss": 0.0411, "step": 1000 }, { "epoch": 0.02, "learning_rate": 0.00036960932369126316, "loss": 0.0415, "step": 1200 }, { "epoch": 0.03, "learning_rate": 0.00036941139945362056, "loss": 0.0338, "step": 1400 }, { "epoch": 0.03, "learning_rate": 0.00036918306624610206, "loss": 0.029, "step": 1600 }, { "epoch": 0.04, "learning_rate": 0.0003689243617362863, "loss": 0.0333, "step": 1800 }, { "epoch": 0.04, "learning_rate": 0.0003686353286020322, "loss": 0.0308, "step": 2000 }, { "epoch": 0.04, "learning_rate": 0.0003683160145244388, "loss": 0.0314, "step": 2200 }, { "epoch": 0.05, "learning_rate": 0.00036796647217997937, "loss": 0.0243, "step": 2400 }, { "epoch": 0.05, "learning_rate": 0.00036758675923181147, "loss": 0.027, "step": 2600 }, { "epoch": 0.06, "learning_rate": 0.00036717693832026435, "loss": 0.0283, "step": 2800 }, { "epoch": 0.06, "learning_rate": 0.0003667370770525053, "loss": 0.0235, "step": 3000 }, { "epoch": 0.07, "learning_rate": 0.0003662672479913868, "loss": 0.0223, "step": 3200 }, { "epoch": 0.07, "learning_rate": 0.0003657701014579829, "loss": 0.0208, "step": 3400 }, { "epoch": 0.07, "learning_rate": 0.00036524072308681836, "loss": 0.0251, "step": 3600 }, { "epoch": 0.08, "learning_rate": 0.00036468162377219915, "loss": 0.0198, "step": 3800 }, { "epoch": 0.08, "learning_rate": 0.00036409289574739075, "loss": 0.0231, "step": 4000 }, { "epoch": 0.09, "learning_rate": 0.0003634746361334354, "loss": 0.0212, "step": 4200 }, { "epoch": 0.09, "learning_rate": 0.0003628269469231305, "loss": 0.0203, "step": 4400 }, { "epoch": 0.09, "learning_rate": 0.000362149934964203, "loss": 0.021, "step": 4600 }, { "epoch": 0.1, "learning_rate": 0.0003614437119416829, "loss": 0.0241, "step": 4800 }, { "epoch": 0.1, "learning_rate": 0.00036070839435947883, "loss": 0.0205, "step": 5000 }, { "epoch": 0.11, "learning_rate": 0.0003599441035211589, "loss": 0.0197, "step": 5200 }, { "epoch": 0.11, "learning_rate": 0.00035915096550993934, "loss": 0.0201, "step": 5400 }, { "epoch": 0.11, "learning_rate": 0.000358329111167885, "loss": 0.0214, "step": 5600 }, { "epoch": 0.12, "learning_rate": 0.00035747867607432445, "loss": 0.0194, "step": 5800 }, { "epoch": 0.12, "learning_rate": 0.0003565998005234839, "loss": 0.0167, "step": 6000 }, { "epoch": 0.13, "learning_rate": 0.0003556926295013433, "loss": 0.0234, "step": 6200 }, { "epoch": 0.13, "learning_rate": 0.00035475731266171827, "loss": 0.0171, "step": 6400 }, { "epoch": 0.13, "learning_rate": 0.00035379400430157195, "loss": 0.0181, "step": 6600 }, { "epoch": 0.14, "learning_rate": 0.00035280286333556107, "loss": 0.0194, "step": 6800 }, { "epoch": 0.14, "learning_rate": 0.00035178405326982047, "loss": 0.0244, "step": 7000 }, { "epoch": 0.15, "learning_rate": 0.0003507377421749893, "loss": 0.0188, "step": 7200 }, { "epoch": 0.15, "learning_rate": 0.00034966410265848546, "loss": 0.0186, "step": 7400 }, { "epoch": 0.16, "learning_rate": 0.00034856331183603055, "loss": 0.0164, "step": 7600 }, { "epoch": 0.16, "learning_rate": 0.0003474355513024315, "loss": 0.0184, "step": 7800 }, { "epoch": 0.16, "learning_rate": 0.0003462810071016236, "loss": 0.0154, "step": 8000 }, { "epoch": 0.17, "learning_rate": 0.00034509986969597905, "loss": 0.0203, "step": 8200 }, { "epoch": 0.17, "learning_rate": 0.0003438984369518822, "loss": 0.0214, "step": 8400 }, { "epoch": 0.18, "learning_rate": 0.00034266483253267676, "loss": 0.0166, "step": 8600 }, { "epoch": 0.18, "learning_rate": 0.0003414052314602316, "loss": 0.017, "step": 8800 }, { "epoch": 0.18, "learning_rate": 0.0003401198415278977, "loss": 0.0182, "step": 9000 }, { "epoch": 0.19, "learning_rate": 0.0003388088747833521, "loss": 0.017, "step": 9200 }, { "epoch": 0.19, "learning_rate": 0.0003374725474936167, "loss": 0.0183, "step": 9400 }, { "epoch": 0.2, "learning_rate": 0.0003361110801093815, "loss": 0.0168, "step": 9600 }, { "epoch": 0.2, "learning_rate": 0.0003347246972286373, "loss": 0.0172, "step": 9800 }, { "epoch": 0.2, "learning_rate": 0.000333313627559624, "loss": 0.0197, "step": 10000 }, { "epoch": 0.21, "learning_rate": 0.0003318781038831018, "loss": 0.0139, "step": 10200 }, { "epoch": 0.21, "learning_rate": 0.0003304183630139492, "loss": 0.016, "step": 10400 }, { "epoch": 0.22, "learning_rate": 0.0003289346457620967, "loss": 0.0205, "step": 10600 }, { "epoch": 0.22, "learning_rate": 0.00032742719689280103, "loss": 0.0179, "step": 10800 }, { "epoch": 0.22, "learning_rate": 0.00032589626508626625, "loss": 0.0143, "step": 11000 }, { "epoch": 0.23, "learning_rate": 0.00032434210289662013, "loss": 0.0175, "step": 11200 }, { "epoch": 0.23, "learning_rate": 0.0003227649667102507, "loss": 0.0156, "step": 11400 }, { "epoch": 0.24, "learning_rate": 0.0003211651167035108, "loss": 0.0206, "step": 11600 }, { "epoch": 0.24, "learning_rate": 0.0003195428167997974, "loss": 0.0176, "step": 11800 }, { "epoch": 0.25, "learning_rate": 0.00031789833462601304, "loss": 0.0151, "step": 12000 }, { "epoch": 0.25, "learning_rate": 0.0003162319414684155, "loss": 0.0159, "step": 12200 }, { "epoch": 0.25, "learning_rate": 0.00031455240573465427, "loss": 0.0168, "step": 12400 }, { "epoch": 0.26, "learning_rate": 0.0003128431249707856, "loss": 0.0148, "step": 12600 }, { "epoch": 0.26, "learning_rate": 0.0003111127671688482, "loss": 0.0236, "step": 12800 }, { "epoch": 0.27, "learning_rate": 0.00030937042477134576, "loss": 0.0146, "step": 13000 }, { "epoch": 0.27, "learning_rate": 0.00030759887447136833, "loss": 0.0157, "step": 13200 }, { "epoch": 0.27, "learning_rate": 0.0003058071122646787, "loss": 0.0216, "step": 13400 }, { "epoch": 0.28, "learning_rate": 0.00030399543373396665, "loss": 0.0171, "step": 13600 }, { "epoch": 0.28, "learning_rate": 0.0003021641377474699, "loss": 0.0174, "step": 13800 }, { "epoch": 0.29, "learning_rate": 0.0003003135264096706, "loss": 0.0174, "step": 14000 }, { "epoch": 0.29, "learning_rate": 0.0002984439050114576, "loss": 0.0155, "step": 14200 }, { "epoch": 0.29, "learning_rate": 0.00029655558197976386, "loss": 0.0181, "step": 14400 }, { "epoch": 0.3, "learning_rate": 0.0002946488688266855, "loss": 0.0165, "step": 14600 }, { "epoch": 0.3, "learning_rate": 0.00029272408009809267, "loss": 0.0186, "step": 14800 }, { "epoch": 0.31, "learning_rate": 0.00029078153332173963, "loss": 0.0159, "step": 15000 }, { "epoch": 0.31, "learning_rate": 0.00028882154895488294, "loss": 0.0148, "step": 15200 }, { "epoch": 0.31, "learning_rate": 0.00028684445033141644, "loss": 0.0159, "step": 15400 }, { "epoch": 0.32, "learning_rate": 0.00028485056360853155, "loss": 0.015, "step": 15600 }, { "epoch": 0.32, "learning_rate": 0.00028284021771291196, "loss": 0.0146, "step": 15800 }, { "epoch": 0.33, "learning_rate": 0.0002808137442864714, "loss": 0.0156, "step": 16000 }, { "epoch": 0.33, "learning_rate": 0.0002787817276943337, "loss": 0.0154, "step": 16200 }, { "epoch": 0.34, "learning_rate": 0.0002767240811582267, "loss": 0.0146, "step": 16400 }, { "epoch": 0.34, "learning_rate": 0.0002746513160555949, "loss": 0.0138, "step": 16600 }, { "epoch": 0.34, "learning_rate": 0.0002725637743254982, "loss": 0.0159, "step": 16800 }, { "epoch": 0.35, "learning_rate": 0.00027046180034466093, "loss": 0.0149, "step": 17000 }, { "epoch": 0.35, "learning_rate": 0.00026834574087066106, "loss": 0.0131, "step": 17200 }, { "epoch": 0.36, "learning_rate": 0.0002662159449847264, "loss": 0.0143, "step": 17400 }, { "epoch": 0.36, "learning_rate": 0.00026407276403414754, "loss": 0.0209, "step": 17600 }, { "epoch": 0.36, "learning_rate": 0.0002619165515743168, "loss": 0.0136, "step": 17800 }, { "epoch": 0.37, "learning_rate": 0.0002597476633104033, "loss": 0.0148, "step": 18000 }, { "epoch": 0.37, "learning_rate": 0.0002575664570386728, "loss": 0.0169, "step": 18200 }, { "epoch": 0.38, "learning_rate": 0.0002553732925874632, "loss": 0.0116, "step": 18400 }, { "epoch": 0.38, "learning_rate": 0.00025316853175782447, "loss": 0.0164, "step": 18600 }, { "epoch": 0.38, "learning_rate": 0.00025096364556859434, "loss": 0.0154, "step": 18800 }, { "epoch": 0.39, "learning_rate": 0.00024873683840049996, "loss": 0.0141, "step": 19000 }, { "epoch": 0.39, "learning_rate": 0.0002464995296538175, "loss": 0.0144, "step": 19200 }, { "epoch": 0.4, "learning_rate": 0.00024425208841197493, "loss": 0.0122, "step": 19400 }, { "epoch": 0.4, "learning_rate": 0.0002419948854299333, "loss": 0.0129, "step": 19600 }, { "epoch": 0.4, "learning_rate": 0.00023972829307302465, "loss": 0.0133, "step": 19800 }, { "epoch": 0.41, "learning_rate": 0.00023745268525552336, "loss": 0.013, "step": 20000 }, { "epoch": 0.41, "learning_rate": 0.00023516843737896262, "loss": 0.0162, "step": 20200 }, { "epoch": 0.42, "learning_rate": 0.00023287592627020543, "loss": 0.0156, "step": 20400 }, { "epoch": 0.42, "learning_rate": 0.00023057553011928013, "loss": 0.0137, "step": 20600 }, { "epoch": 0.43, "learning_rate": 0.0002282676284169916, "loss": 0.0146, "step": 20800 }, { "epoch": 0.43, "learning_rate": 0.00022595260189231734, "loss": 0.0143, "step": 21000 }, { "epoch": 0.43, "learning_rate": 0.00022363083244959957, "loss": 0.0173, "step": 21200 }, { "epoch": 0.44, "learning_rate": 0.00022131435893762175, "loss": 0.0141, "step": 21400 }, { "epoch": 0.44, "learning_rate": 0.00021899196729468126, "loss": 0.0152, "step": 21600 }, { "epoch": 0.45, "learning_rate": 0.00021665232533016668, "loss": 0.0132, "step": 21800 }, { "epoch": 0.45, "learning_rate": 0.00021430747469180223, "loss": 0.0138, "step": 22000 }, { "epoch": 0.45, "learning_rate": 0.00021195780220394222, "loss": 0.0144, "step": 22200 }, { "epoch": 0.46, "learning_rate": 0.00020960369548639, "loss": 0.017, "step": 22400 }, { "epoch": 0.46, "learning_rate": 0.00020724554289045297, "loss": 0.0131, "step": 22600 }, { "epoch": 0.47, "learning_rate": 0.00020488373343487728, "loss": 0.0151, "step": 22800 }, { "epoch": 0.47, "learning_rate": 0.0002025186567416726, "loss": 0.0134, "step": 23000 }, { "epoch": 0.47, "learning_rate": 0.00020015070297183673, "loss": 0.015, "step": 23200 }, { "epoch": 0.48, "learning_rate": 0.000197780262760992, "loss": 0.0153, "step": 23400 }, { "epoch": 0.48, "learning_rate": 0.00019540772715494277, "loss": 0.0151, "step": 23600 }, { "epoch": 0.49, "learning_rate": 0.00019303348754516576, "loss": 0.0153, "step": 23800 }, { "epoch": 0.49, "learning_rate": 0.00019065793560424322, "loss": 0.0125, "step": 24000 }, { "epoch": 0.49, "learning_rate": 0.00018828146322124917, "loss": 0.0126, "step": 24200 }, { "epoch": 0.5, "learning_rate": 0.00018590446243710106, "loss": 0.0146, "step": 24400 }, { "epoch": 0.5, "learning_rate": 0.0001835273253798852, "loss": 0.0123, "step": 24600 }, { "epoch": 0.51, "learning_rate": 0.00018115044420016874, "loss": 0.0173, "step": 24800 }, { "epoch": 0.51, "learning_rate": 0.00017877421100630705, "loss": 0.0151, "step": 25000 }, { "epoch": 0.52, "learning_rate": 0.00017639901779975916, "loss": 0.0174, "step": 25200 }, { "epoch": 0.52, "learning_rate": 0.00017402525641041958, "loss": 0.0122, "step": 25400 }, { "epoch": 0.52, "learning_rate": 0.00017166517293872442, "loss": 0.0193, "step": 25600 }, { "epoch": 0.53, "learning_rate": 0.00016929543761778923, "loss": 0.0182, "step": 25800 }, { "epoch": 0.53, "learning_rate": 0.00016692830597455233, "loss": 0.0136, "step": 26000 }, { "epoch": 0.54, "learning_rate": 0.0001645641685090116, "loss": 0.0133, "step": 26200 }, { "epoch": 0.54, "learning_rate": 0.000162203415227223, "loss": 0.0107, "step": 26400 }, { "epoch": 0.54, "learning_rate": 0.00015984643557696138, "loss": 0.0128, "step": 26600 }, { "epoch": 0.55, "learning_rate": 0.00015749361838347518, "loss": 0.0134, "step": 26800 }, { "epoch": 0.55, "learning_rate": 0.000155145351785342, "loss": 0.0122, "step": 27000 }, { "epoch": 0.56, "learning_rate": 0.000152802023170439, "loss": 0.0145, "step": 27200 }, { "epoch": 0.56, "learning_rate": 0.0001504640191120359, "loss": 0.0145, "step": 27400 }, { "epoch": 0.56, "learning_rate": 0.00014813172530502337, "loss": 0.0163, "step": 27600 }, { "epoch": 0.57, "learning_rate": 0.00014580552650228558, "loss": 0.0106, "step": 27800 }, { "epoch": 0.57, "learning_rate": 0.00014348580645122863, "loss": 0.0132, "step": 28000 }, { "epoch": 0.58, "learning_rate": 0.00014117294783047438, "loss": 0.0117, "step": 28200 }, { "epoch": 0.58, "learning_rate": 0.00013886733218673142, "loss": 0.0129, "step": 28400 }, { "epoch": 0.58, "learning_rate": 0.0001365693398718515, "loss": 0.0134, "step": 28600 }, { "epoch": 0.59, "learning_rate": 0.00013427934998008455, "loss": 0.0127, "step": 28800 }, { "epoch": 0.59, "learning_rate": 0.00013199774028553978, "loss": 0.0112, "step": 29000 }, { "epoch": 0.6, "learning_rate": 0.0001297248871798657, "loss": 0.0137, "step": 29200 }, { "epoch": 0.6, "learning_rate": 0.00012746116561015725, "loss": 0.0167, "step": 29400 }, { "epoch": 0.61, "learning_rate": 0.0001252069490171022, "loss": 0.0135, "step": 29600 }, { "epoch": 0.61, "learning_rate": 0.00012296260927337492, "loss": 0.0103, "step": 29800 }, { "epoch": 0.61, "learning_rate": 0.00012072851662228994, "loss": 0.0121, "step": 30000 }, { "epoch": 0.62, "learning_rate": 0.00011850503961672364, "loss": 0.0134, "step": 30200 }, { "epoch": 0.62, "learning_rate": 0.00011630357960773998, "loss": 0.0126, "step": 30400 }, { "epoch": 0.63, "learning_rate": 0.000114102374844354, "loss": 0.0126, "step": 30600 }, { "epoch": 0.63, "learning_rate": 0.00011191287882511448, "loss": 0.0131, "step": 30800 }, { "epoch": 0.63, "learning_rate": 0.0001097354527458991, "loss": 0.0135, "step": 31000 }, { "epoch": 0.64, "learning_rate": 0.00010757045581143638, "loss": 0.0162, "step": 31200 }, { "epoch": 0.64, "learning_rate": 0.00010541824517604906, "loss": 0.012, "step": 31400 }, { "epoch": 0.65, "learning_rate": 0.00010327917588473477, "loss": 0.0134, "step": 31600 }, { "epoch": 0.65, "learning_rate": 0.00010115360081459545, "loss": 0.0144, "step": 31800 }, { "epoch": 0.65, "learning_rate": 9.904187061662376e-05, "loss": 0.0195, "step": 32000 }, { "epoch": 0.66, "learning_rate": 9.69443336578574e-05, "loss": 0.0157, "step": 32200 }, { "epoch": 0.66, "learning_rate": 9.486133596390939e-05, "loss": 0.0131, "step": 32400 }, { "epoch": 0.67, "learning_rate": 9.279322116188555e-05, "loss": 0.0146, "step": 32600 }, { "epoch": 0.67, "learning_rate": 9.074033042369668e-05, "loss": 0.0118, "step": 32800 }, { "epoch": 0.67, "learning_rate": 8.870300240977657e-05, "loss": 0.0153, "step": 33000 }, { "epoch": 0.68, "learning_rate": 8.668157321321404e-05, "loss": 0.0107, "step": 33200 }, { "epoch": 0.68, "learning_rate": 8.467637630430823e-05, "loss": 0.0145, "step": 33400 }, { "epoch": 0.69, "learning_rate": 8.268774247555691e-05, "loss": 0.0124, "step": 33600 }, { "epoch": 0.69, "learning_rate": 8.07159997870866e-05, "loss": 0.0117, "step": 33800 }, { "epoch": 0.7, "learning_rate": 7.876147351253296e-05, "loss": 0.0153, "step": 34000 }, { "epoch": 0.7, "learning_rate": 7.682448608538163e-05, "loss": 0.0118, "step": 34200 }, { "epoch": 0.7, "learning_rate": 7.49149077432921e-05, "loss": 0.0133, "step": 34400 }, { "epoch": 0.71, "learning_rate": 7.301386202782465e-05, "loss": 0.0129, "step": 34600 }, { "epoch": 0.71, "learning_rate": 7.114066962139441e-05, "loss": 0.0105, "step": 34800 }, { "epoch": 0.72, "learning_rate": 6.927681374580455e-05, "loss": 0.0132, "step": 35000 }, { "epoch": 0.72, "learning_rate": 6.74320613790631e-05, "loss": 0.0148, "step": 35200 }, { "epoch": 0.72, "learning_rate": 6.560671684551868e-05, "loss": 0.0135, "step": 35400 }, { "epoch": 0.73, "learning_rate": 6.380108126785653e-05, "loss": 0.0147, "step": 35600 }, { "epoch": 0.73, "learning_rate": 6.20154525174229e-05, "loss": 0.0139, "step": 35800 }, { "epoch": 0.74, "learning_rate": 6.025012516508621e-05, "loss": 0.0125, "step": 36000 }, { "epoch": 0.74, "learning_rate": 5.850539043264194e-05, "loss": 0.0105, "step": 36200 }, { "epoch": 0.74, "learning_rate": 5.6781536144770985e-05, "loss": 0.0162, "step": 36400 }, { "epoch": 0.75, "learning_rate": 5.5078846681557415e-05, "loss": 0.0111, "step": 36600 }, { "epoch": 0.75, "learning_rate": 5.339760293157545e-05, "loss": 0.0111, "step": 36800 }, { "epoch": 0.76, "learning_rate": 5.173808224555153e-05, "loss": 0.0108, "step": 37000 }, { "epoch": 0.76, "learning_rate": 5.010055839061078e-05, "loss": 0.0124, "step": 37200 }, { "epoch": 0.76, "learning_rate": 4.849332195813418e-05, "loss": 0.013, "step": 37400 }, { "epoch": 0.77, "learning_rate": 4.690048518324095e-05, "loss": 0.0128, "step": 37600 }, { "epoch": 0.77, "learning_rate": 4.5338236363359446e-05, "loss": 0.0156, "step": 37800 }, { "epoch": 0.78, "learning_rate": 4.3791132442674044e-05, "loss": 0.0116, "step": 38000 }, { "epoch": 0.78, "learning_rate": 4.226733634221389e-05, "loss": 0.0131, "step": 38200 }, { "epoch": 0.78, "learning_rate": 4.0767099438946455e-05, "loss": 0.0115, "step": 38400 }, { "epoch": 0.79, "learning_rate": 3.929066922333545e-05, "loss": 0.013, "step": 38600 }, { "epoch": 0.79, "learning_rate": 3.78382892585125e-05, "loss": 0.0126, "step": 38800 }, { "epoch": 0.8, "learning_rate": 3.6410199140097565e-05, "loss": 0.0115, "step": 39000 }, { "epoch": 0.8, "learning_rate": 3.500663445667306e-05, "loss": 0.0137, "step": 39200 }, { "epoch": 0.81, "learning_rate": 3.3627826750919713e-05, "loss": 0.012, "step": 39400 }, { "epoch": 0.81, "learning_rate": 3.2274003481419156e-05, "loss": 0.0139, "step": 39600 }, { "epoch": 0.81, "learning_rate": 3.094538798513103e-05, "loss": 0.0124, "step": 39800 }, { "epoch": 0.82, "learning_rate": 2.964219944054926e-05, "loss": 0.0159, "step": 40000 }, { "epoch": 0.82, "learning_rate": 2.8364652831545036e-05, "loss": 0.0136, "step": 40200 }, { "epoch": 0.83, "learning_rate": 2.7112958911901083e-05, "loss": 0.0159, "step": 40400 }, { "epoch": 0.83, "learning_rate": 2.588732417054455e-05, "loss": 0.011, "step": 40600 }, { "epoch": 0.83, "learning_rate": 2.4693882010091674e-05, "loss": 0.0152, "step": 40800 }, { "epoch": 0.84, "learning_rate": 2.3520835081962702e-05, "loss": 0.0161, "step": 41000 }, { "epoch": 0.84, "learning_rate": 2.2374439916125642e-05, "loss": 0.0117, "step": 41200 }, { "epoch": 0.85, "learning_rate": 2.1254885630626012e-05, "loss": 0.0137, "step": 41400 }, { "epoch": 0.85, "learning_rate": 2.016235691563421e-05, "loss": 0.0105, "step": 41600 }, { "epoch": 0.85, "learning_rate": 1.909703400297725e-05, "loss": 0.0125, "step": 41800 }, { "epoch": 0.86, "learning_rate": 1.8059092636406624e-05, "loss": 0.0124, "step": 42000 }, { "epoch": 0.86, "learning_rate": 1.704870404260628e-05, "loss": 0.0119, "step": 42200 }, { "epoch": 0.87, "learning_rate": 1.6066034902945553e-05, "loss": 0.0096, "step": 42400 }, { "epoch": 0.87, "learning_rate": 1.5111247325982417e-05, "loss": 0.0114, "step": 42600 }, { "epoch": 0.87, "learning_rate": 1.4184498820720606e-05, "loss": 0.0129, "step": 42800 }, { "epoch": 0.88, "learning_rate": 1.3285942270625904e-05, "loss": 0.0143, "step": 43000 }, { "epoch": 0.88, "learning_rate": 1.2415725908405126e-05, "loss": 0.0135, "step": 43200 }, { "epoch": 0.89, "learning_rate": 1.1573993291552778e-05, "loss": 0.0111, "step": 43400 }, { "epoch": 0.89, "learning_rate": 1.0760883278668474e-05, "loss": 0.012, "step": 43600 }, { "epoch": 0.9, "learning_rate": 9.976530006549986e-06, "loss": 0.0152, "step": 43800 }, { "epoch": 0.9, "learning_rate": 9.221062868064821e-06, "loss": 0.0124, "step": 44000 }, { "epoch": 0.9, "learning_rate": 8.494606490804811e-06, "loss": 0.0163, "step": 44200 }, { "epoch": 0.91, "learning_rate": 7.800694690729383e-06, "loss": 0.012, "step": 44400 }, { "epoch": 0.91, "learning_rate": 7.132468049208598e-06, "loss": 0.0112, "step": 44600 }, { "epoch": 0.92, "learning_rate": 6.493596719362824e-06, "loss": 0.0106, "step": 44800 }, { "epoch": 0.92, "learning_rate": 5.884186094254458e-06, "loss": 0.0132, "step": 45000 }, { "epoch": 0.92, "learning_rate": 5.304336706884555e-06, "loss": 0.0137, "step": 45200 }, { "epoch": 0.93, "learning_rate": 4.754144213608023e-06, "loss": 0.0101, "step": 45400 }, { "epoch": 0.93, "learning_rate": 4.233699378353457e-06, "loss": 0.009, "step": 45600 }, { "epoch": 0.94, "learning_rate": 3.743088057650132e-06, "loss": 0.014, "step": 45800 }, { "epoch": 0.94, "learning_rate": 3.2823911864643436e-06, "loss": 0.0129, "step": 46000 }, { "epoch": 0.94, "learning_rate": 2.8516847648477812e-06, "loss": 0.0132, "step": 46200 }, { "epoch": 0.95, "learning_rate": 2.4510398454000396e-06, "loss": 0.0121, "step": 46400 }, { "epoch": 0.95, "learning_rate": 2.0805225215471503e-06, "loss": 0.0115, "step": 46600 }, { "epoch": 0.96, "learning_rate": 1.740193916638432e-06, "loss": 0.0121, "step": 46800 }, { "epoch": 0.96, "learning_rate": 1.430110173862991e-06, "loss": 0.0145, "step": 47000 }, { "epoch": 0.96, "learning_rate": 1.1503224469880126e-06, "loss": 0.013, "step": 47200 }, { "epoch": 0.97, "learning_rate": 9.008768919199934e-07, "loss": 0.0127, "step": 47400 }, { "epoch": 0.97, "learning_rate": 6.818146590905163e-07, "loss": 0.0138, "step": 47600 }, { "epoch": 0.98, "learning_rate": 4.931718866678032e-07, "loss": 0.0113, "step": 47800 }, { "epoch": 0.98, "learning_rate": 3.356948644792021e-07, "loss": 0.0169, "step": 48000 }, { "epoch": 0.99, "learning_rate": 2.0782690935592252e-07, "loss": 0.0109, "step": 48200 }, { "epoch": 0.99, "learning_rate": 1.1045660725531198e-07, "loss": 0.0119, "step": 48400 }, { "epoch": 0.99, "learning_rate": 4.360002112121604e-08, "loss": 0.0102, "step": 48600 }, { "epoch": 1.0, "learning_rate": 7.2681801234168665e-09, "loss": 0.0114, "step": 48800 }, { "epoch": 1.0, "eval_accuracy": 0.9751466987432912, "eval_auc": 0.9180107216187284, "eval_f1": 0.15952083840732592, "eval_loss": 0.2542668879032135, "eval_mcc": 0.27048987280764264, "eval_precision": 0.0879081257328702, "eval_recall": 0.8605598317425242, "eval_runtime": 27823.2385, "eval_samples_per_second": 16.313, "eval_steps_per_second": 0.453, "step": 48919 } ], "logging_steps": 200, "max_steps": 48919, "num_train_epochs": 1, "save_steps": 500, "total_flos": 1.56321913484268e+18, "trial_name": null, "trial_params": null }