| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.550335570469799, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.8169934667141895e-06, | |
| "loss": 1.2113, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.101496404594247e-06, | |
| "loss": 1.1368, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.771212547196624e-06, | |
| "loss": 1.0866, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.227339080223317e-06, | |
| "loss": 1.0541, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.5736595264523914e-06, | |
| "loss": 1.0277, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.852916185574972e-06, | |
| "loss": 1.0367, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 6.086916009002755e-06, | |
| "loss": 1.0106, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 6.288302417241607e-06, | |
| "loss": 1.0122, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.465064175395396e-06, | |
| "loss": 0.9955, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 6.622572447554151e-06, | |
| "loss": 0.9957, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.7646125922840325e-06, | |
| "loss": 1.0047, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.893952872487206e-06, | |
| "loss": 1.0056, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.01267906985319e-06, | |
| "loss": 0.9908, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.122401890521357e-06, | |
| "loss": 0.9841, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.224391115827254e-06, | |
| "loss": 0.9865, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.31966550803078e-06, | |
| "loss": 0.9963, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.409054903825279e-06, | |
| "loss": 0.9818, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.493244221206023e-06, | |
| "loss": 0.9678, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.572805355121664e-06, | |
| "loss": 0.986, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 7.648220753871978e-06, | |
| "loss": 0.984, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.719901151523059e-06, | |
| "loss": 0.9686, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.788199112828433e-06, | |
| "loss": 0.9692, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.853419523977077e-06, | |
| "loss": 0.965, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.91582782003368e-06, | |
| "loss": 0.9816, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.975656510865554e-06, | |
| "loss": 0.9602, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.033110411104317e-06, | |
| "loss": 0.9757, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.088370871215252e-06, | |
| "loss": 0.9854, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.14159923021483e-06, | |
| "loss": 0.9651, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.192939655779976e-06, | |
| "loss": 0.9641, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.242521497724043e-06, | |
| "loss": 0.9602, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.290461251590623e-06, | |
| "loss": 0.9496, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.336864207392505e-06, | |
| "loss": 0.9578, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.381825842200955e-06, | |
| "loss": 0.9678, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.425433002904463e-06, | |
| "loss": 0.9609, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.467764915969581e-06, | |
| "loss": 0.9628, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.508894053707311e-06, | |
| "loss": 0.9496, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.5488868808403e-06, | |
| "loss": 0.9577, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.587804500685978e-06, | |
| "loss": 0.9535, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.625703216729706e-06, | |
| "loss": 0.9452, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.662635022543719e-06, | |
| "loss": 0.9341, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.698648030750733e-06, | |
| "loss": 0.9228, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.733786849912527e-06, | |
| "loss": 0.9315, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.76809291675008e-06, | |
| "loss": 0.9127, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.801604789901407e-06, | |
| "loss": 0.9084, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.834358410439789e-06, | |
| "loss": 0.9157, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.866387333566169e-06, | |
| "loss": 0.9143, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.897722935220375e-06, | |
| "loss": 0.9221, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.928394596800381e-06, | |
| "loss": 0.911, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.958429870715448e-06, | |
| "loss": 0.9176, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 8.987854629111107e-06, | |
| "loss": 0.9122, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_code_evol_accuracy": 0.7536964329038056, | |
| "eval_code_evol_loss": 0.9287109375, | |
| "eval_code_evol_runtime": 27.3209, | |
| "eval_code_evol_samples_per_second": 28.22, | |
| "eval_code_evol_steps_per_second": 3.55, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.016693197777789e-06, | |
| "loss": 0.9191, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.044968476979809e-06, | |
| "loss": 0.9176, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.072702050708489e-06, | |
| "loss": 0.9235, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.099914285665186e-06, | |
| "loss": 0.899, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.126624421111438e-06, | |
| "loss": 0.9364, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9.152850650579098e-06, | |
| "loss": 0.9136, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.17861019630969e-06, | |
| "loss": 0.9204, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 9.203919377185773e-06, | |
| "loss": 0.9167, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.228793670825383e-06, | |
| "loss": 0.9092, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.253247770431231e-06, | |
| "loss": 0.9183, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.277295636917526e-06, | |
| "loss": 0.9113, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.300950546777473e-06, | |
| "loss": 0.8988, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.324225136102389e-06, | |
| "loss": 0.9229, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.347131441117836e-06, | |
| "loss": 0.9193, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.369680935562337e-06, | |
| "loss": 0.9085, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.39188456519927e-06, | |
| "loss": 0.9128, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.41375277972183e-06, | |
| "loss": 0.9079, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 9.435295562283815e-06, | |
| "loss": 0.9097, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 9.456522456865169e-06, | |
| "loss": 0.9077, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.477442593660033e-06, | |
| "loss": 0.9147, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.498064712656333e-06, | |
| "loss": 0.9096, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 9.518397185559335e-06, | |
| "loss": 0.9009, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.538448036196794e-06, | |
| "loss": 0.9214, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 9.558224959530173e-06, | |
| "loss": 0.9198, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 9.577735339384664e-06, | |
| "loss": 0.9162, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.596986265000242e-06, | |
| "loss": 0.8982, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.615984546496603e-06, | |
| "loss": 0.9075, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.634736729336382e-06, | |
| "loss": 0.9104, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.653249107863549e-06, | |
| "loss": 0.8939, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.671527737987041e-06, | |
| "loss": 0.8777, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.68957844907357e-06, | |
| "loss": 0.8674, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.707406855108054e-06, | |
| "loss": 0.8647, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.725018365175157e-06, | |
| "loss": 0.8778, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.742418193310868e-06, | |
| "loss": 0.8665, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.759611367769025e-06, | |
| "loss": 0.8717, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.776602739743996e-06, | |
| "loss": 0.8663, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.793396991587369e-06, | |
| "loss": 0.8599, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.80999864455347e-06, | |
| "loss": 0.8634, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 9.826412066105755e-06, | |
| "loss": 0.8725, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 9.842641476813642e-06, | |
| "loss": 0.8699, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.858690956866985e-06, | |
| "loss": 0.8634, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.874564452233405e-06, | |
| "loss": 0.8753, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 9.890265780481658e-06, | |
| "loss": 0.8694, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 9.905798636292595e-06, | |
| "loss": 0.8701, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.92116659667758e-06, | |
| "loss": 0.88, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.936373125922813e-06, | |
| "loss": 0.8711, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.951421580276673e-06, | |
| "loss": 0.8564, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.966315212395911e-06, | |
| "loss": 0.879, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 9.981057175565457e-06, | |
| "loss": 0.8744, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 9.99565052770552e-06, | |
| "loss": 0.8517, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_code_evol_accuracy": 0.7573287954429739, | |
| "eval_code_evol_loss": 0.908203125, | |
| "eval_code_evol_runtime": 9.3653, | |
| "eval_code_evol_samples_per_second": 82.325, | |
| "eval_code_evol_steps_per_second": 10.357, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.972426470588235e-06, | |
| "loss": 0.8765, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 9.926470588235295e-06, | |
| "loss": 0.877, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 9.880514705882353e-06, | |
| "loss": 0.8678, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 9.834558823529412e-06, | |
| "loss": 0.8656, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 9.788602941176472e-06, | |
| "loss": 0.8811, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 9.74264705882353e-06, | |
| "loss": 0.8484, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 9.69669117647059e-06, | |
| "loss": 0.8701, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 9.650735294117649e-06, | |
| "loss": 0.8549, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 9.604779411764707e-06, | |
| "loss": 0.8762, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 9.558823529411766e-06, | |
| "loss": 0.8597, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 9.512867647058824e-06, | |
| "loss": 0.8579, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 9.466911764705882e-06, | |
| "loss": 0.8698, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 9.420955882352942e-06, | |
| "loss": 0.8696, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 9.375000000000001e-06, | |
| "loss": 0.8796, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 9.32904411764706e-06, | |
| "loss": 0.8689, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 9.28308823529412e-06, | |
| "loss": 0.8763, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.237132352941178e-06, | |
| "loss": 0.8639, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 9.191176470588236e-06, | |
| "loss": 0.8658, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 9.145220588235295e-06, | |
| "loss": 0.8741, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.099264705882353e-06, | |
| "loss": 0.8233, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 9.053308823529411e-06, | |
| "loss": 0.8298, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.007352941176471e-06, | |
| "loss": 0.8378, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 8.96139705882353e-06, | |
| "loss": 0.8387, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 8.91544117647059e-06, | |
| "loss": 0.8277, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 8.869485294117648e-06, | |
| "loss": 0.8276, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 8.823529411764707e-06, | |
| "loss": 0.8398, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 8.777573529411765e-06, | |
| "loss": 0.8344, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 8.731617647058824e-06, | |
| "loss": 0.8274, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 8.685661764705882e-06, | |
| "loss": 0.8296, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 8.639705882352942e-06, | |
| "loss": 0.8286, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 8.59375e-06, | |
| "loss": 0.8342, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 8.547794117647059e-06, | |
| "loss": 0.8265, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 8.501838235294119e-06, | |
| "loss": 0.8284, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 8.455882352941177e-06, | |
| "loss": 0.8316, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 8.409926470588236e-06, | |
| "loss": 0.8145, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 8.363970588235294e-06, | |
| "loss": 0.8182, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 8.318014705882354e-06, | |
| "loss": 0.8239, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 8.272058823529413e-06, | |
| "loss": 0.8319, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.226102941176471e-06, | |
| "loss": 0.8332, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 8.18014705882353e-06, | |
| "loss": 0.8107, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 8.13419117647059e-06, | |
| "loss": 0.8289, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.088235294117648e-06, | |
| "loss": 0.8384, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 8.042279411764706e-06, | |
| "loss": 0.8264, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.996323529411767e-06, | |
| "loss": 0.8234, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 7.950367647058825e-06, | |
| "loss": 0.8188, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 7.904411764705883e-06, | |
| "loss": 0.8393, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 7.858455882352942e-06, | |
| "loss": 0.8183, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 7.8125e-06, | |
| "loss": 0.8543, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 7.766544117647059e-06, | |
| "loss": 0.8216, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 7.720588235294119e-06, | |
| "loss": 0.8392, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "eval_code_evol_accuracy": 0.7596694994310552, | |
| "eval_code_evol_loss": 0.90087890625, | |
| "eval_code_evol_runtime": 10.7536, | |
| "eval_code_evol_samples_per_second": 71.697, | |
| "eval_code_evol_steps_per_second": 9.02, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 7.674632352941177e-06, | |
| "loss": 0.8201, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 7.6286764705882355e-06, | |
| "loss": 0.8245, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.582720588235295e-06, | |
| "loss": 0.832, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 7.536764705882353e-06, | |
| "loss": 0.8208, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 7.4908088235294115e-06, | |
| "loss": 0.8362, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 7.444852941176472e-06, | |
| "loss": 0.8405, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 7.39889705882353e-06, | |
| "loss": 0.8319, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 7.352941176470589e-06, | |
| "loss": 0.8327, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 7.306985294117648e-06, | |
| "loss": 0.8244, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 7.261029411764706e-06, | |
| "loss": 0.7933, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 7.215073529411765e-06, | |
| "loss": 0.7896, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 7.169117647058825e-06, | |
| "loss": 0.7918, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 7.123161764705883e-06, | |
| "loss": 0.7935, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 7.077205882352942e-06, | |
| "loss": 0.7995, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 7.031250000000001e-06, | |
| "loss": 0.8056, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 6.985294117647059e-06, | |
| "loss": 0.798, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 6.939338235294118e-06, | |
| "loss": 0.7755, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 6.893382352941177e-06, | |
| "loss": 0.7891, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 6.847426470588235e-06, | |
| "loss": 0.7904, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 6.801470588235295e-06, | |
| "loss": 0.8003, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 6.755514705882354e-06, | |
| "loss": 0.7922, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 6.709558823529412e-06, | |
| "loss": 0.7832, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 6.663602941176471e-06, | |
| "loss": 0.7901, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 6.61764705882353e-06, | |
| "loss": 0.8005, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 6.571691176470589e-06, | |
| "loss": 0.7988, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 6.525735294117647e-06, | |
| "loss": 0.7879, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 6.479779411764706e-06, | |
| "loss": 0.8045, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 6.433823529411766e-06, | |
| "loss": 0.8042, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.387867647058824e-06, | |
| "loss": 0.7951, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 6.341911764705883e-06, | |
| "loss": 0.7858, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 6.295955882352942e-06, | |
| "loss": 0.8009, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.8055, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 6.204044117647059e-06, | |
| "loss": 0.8041, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 6.158088235294119e-06, | |
| "loss": 0.7989, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 6.112132352941177e-06, | |
| "loss": 0.8029, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 6.066176470588236e-06, | |
| "loss": 0.8021, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 6.020220588235295e-06, | |
| "loss": 0.8085, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.974264705882353e-06, | |
| "loss": 0.7904, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 5.928308823529412e-06, | |
| "loss": 0.8014, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 0.8114, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 5.836397058823529e-06, | |
| "loss": 0.8067, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 5.790441176470589e-06, | |
| "loss": 0.7983, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.744485294117648e-06, | |
| "loss": 0.7998, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 5.698529411764706e-06, | |
| "loss": 0.7973, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 5.6525735294117655e-06, | |
| "loss": 0.8186, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 5.606617647058824e-06, | |
| "loss": 0.7907, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 5.560661764705882e-06, | |
| "loss": 0.8019, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.514705882352942e-06, | |
| "loss": 0.8082, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 5.468750000000001e-06, | |
| "loss": 0.7778, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.422794117647059e-06, | |
| "loss": 0.775, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "eval_code_evol_accuracy": 0.7592867857862886, | |
| "eval_code_evol_loss": 0.904296875, | |
| "eval_code_evol_runtime": 11.4335, | |
| "eval_code_evol_samples_per_second": 67.433, | |
| "eval_code_evol_steps_per_second": 8.484, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.3768382352941184e-06, | |
| "loss": 0.765, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 5.330882352941177e-06, | |
| "loss": 0.7735, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 5.284926470588235e-06, | |
| "loss": 0.76, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 5.2389705882352945e-06, | |
| "loss": 0.7769, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 5.193014705882353e-06, | |
| "loss": 0.7735, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 5.147058823529411e-06, | |
| "loss": 0.7655, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 5.101102941176471e-06, | |
| "loss": 0.7742, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.05514705882353e-06, | |
| "loss": 0.7749, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 5.009191176470589e-06, | |
| "loss": 0.782, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 4.9632352941176475e-06, | |
| "loss": 0.7806, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 4.917279411764706e-06, | |
| "loss": 0.7796, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 4.871323529411765e-06, | |
| "loss": 0.7685, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 4.825367647058824e-06, | |
| "loss": 0.7681, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 4.779411764705883e-06, | |
| "loss": 0.7729, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 4.733455882352941e-06, | |
| "loss": 0.7634, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 4.6875000000000004e-06, | |
| "loss": 0.7732, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 4.64154411764706e-06, | |
| "loss": 0.7703, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 4.595588235294118e-06, | |
| "loss": 0.7815, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.5496323529411765e-06, | |
| "loss": 0.7811, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 4.503676470588236e-06, | |
| "loss": 0.7702, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 4.457720588235295e-06, | |
| "loss": 0.7753, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.411764705882353e-06, | |
| "loss": 0.7751, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 4.365808823529412e-06, | |
| "loss": 0.7785, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 4.319852941176471e-06, | |
| "loss": 0.7793, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.2738970588235295e-06, | |
| "loss": 0.7861, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 4.227941176470589e-06, | |
| "loss": 0.7811, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 4.181985294117647e-06, | |
| "loss": 0.7728, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 4.136029411764706e-06, | |
| "loss": 0.7803, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 4.090073529411765e-06, | |
| "loss": 0.7672, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 4.044117647058824e-06, | |
| "loss": 0.7781, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 3.998161764705883e-06, | |
| "loss": 0.7861, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 3.952205882352942e-06, | |
| "loss": 0.7833, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 3.90625e-06, | |
| "loss": 0.7852, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 3.860294117647059e-06, | |
| "loss": 0.7662, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 3.8143382352941177e-06, | |
| "loss": 0.7777, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 3.7683823529411766e-06, | |
| "loss": 0.7646, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 3.722426470588236e-06, | |
| "loss": 0.7703, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 3.6764705882352946e-06, | |
| "loss": 0.7803, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 3.630514705882353e-06, | |
| "loss": 0.7677, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 3.5845588235294123e-06, | |
| "loss": 0.747, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 3.538602941176471e-06, | |
| "loss": 0.753, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 3.4926470588235295e-06, | |
| "loss": 0.7575, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 3.4466911764705884e-06, | |
| "loss": 0.7409, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 3.4007352941176476e-06, | |
| "loss": 0.7591, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 3.354779411764706e-06, | |
| "loss": 0.7659, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 3.308823529411765e-06, | |
| "loss": 0.7479, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 3.2628676470588237e-06, | |
| "loss": 0.7453, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.216911764705883e-06, | |
| "loss": 0.752, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 3.1709558823529413e-06, | |
| "loss": 0.7508, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 3.125e-06, | |
| "loss": 0.7585, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "eval_code_evol_accuracy": 0.7601342231425574, | |
| "eval_code_evol_loss": 0.90576171875, | |
| "eval_code_evol_runtime": 11.4001, | |
| "eval_code_evol_samples_per_second": 67.631, | |
| "eval_code_evol_steps_per_second": 8.509, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 3.0790441176470594e-06, | |
| "loss": 0.7617, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 3.033088235294118e-06, | |
| "loss": 0.7565, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 2.9871323529411766e-06, | |
| "loss": 0.7588, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 0.7637, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 2.8952205882352947e-06, | |
| "loss": 0.7585, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 2.849264705882353e-06, | |
| "loss": 0.77, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 2.803308823529412e-06, | |
| "loss": 0.7599, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 2.757352941176471e-06, | |
| "loss": 0.7527, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 2.7113970588235296e-06, | |
| "loss": 0.7476, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 2.6654411764705884e-06, | |
| "loss": 0.7531, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 2.6194852941176473e-06, | |
| "loss": 0.7595, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 2.5735294117647057e-06, | |
| "loss": 0.7596, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 2.527573529411765e-06, | |
| "loss": 0.7578, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 2.4816176470588237e-06, | |
| "loss": 0.7661, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 2.4356617647058826e-06, | |
| "loss": 0.7527, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 2.3897058823529414e-06, | |
| "loss": 0.7614, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 2.3437500000000002e-06, | |
| "loss": 0.7637, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 2.297794117647059e-06, | |
| "loss": 0.7699, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 2.251838235294118e-06, | |
| "loss": 0.7465, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 2.2058823529411767e-06, | |
| "loss": 0.7528, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 2.1599264705882355e-06, | |
| "loss": 0.7604, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 2.1139705882352944e-06, | |
| "loss": 0.7535, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 2.068014705882353e-06, | |
| "loss": 0.7626, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 2.022058823529412e-06, | |
| "loss": 0.7517, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 1.976102941176471e-06, | |
| "loss": 0.7618, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 1.9301470588235297e-06, | |
| "loss": 0.7545, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 1.8841911764705883e-06, | |
| "loss": 0.7725, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.8382352941176473e-06, | |
| "loss": 0.7489, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.7922794117647061e-06, | |
| "loss": 0.7553, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.7463235294117648e-06, | |
| "loss": 0.753, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 1.7003676470588238e-06, | |
| "loss": 0.7504, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 1.6544117647058824e-06, | |
| "loss": 0.751, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 1.6084558823529415e-06, | |
| "loss": 0.7473, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 1.5625e-06, | |
| "loss": 0.7398, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 1.516544117647059e-06, | |
| "loss": 0.7412, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 1.4705882352941177e-06, | |
| "loss": 0.7465, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 1.4246323529411766e-06, | |
| "loss": 0.7315, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 1.3786764705882356e-06, | |
| "loss": 0.7316, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 1.3327205882352942e-06, | |
| "loss": 0.7457, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 1.2867647058823528e-06, | |
| "loss": 0.7397, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.2408088235294119e-06, | |
| "loss": 0.7498, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 1.1948529411764707e-06, | |
| "loss": 0.7477, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 1.1488970588235295e-06, | |
| "loss": 0.7433, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 1.1029411764705884e-06, | |
| "loss": 0.743, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 1.0569852941176472e-06, | |
| "loss": 0.7393, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 1.011029411764706e-06, | |
| "loss": 0.7483, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 7.47, | |
| "learning_rate": 9.650735294117648e-07, | |
| "loss": 0.7466, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 9.191176470588237e-07, | |
| "loss": 0.7509, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 8.731617647058824e-07, | |
| "loss": 0.7554, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 8.272058823529412e-07, | |
| "loss": 0.7326, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "eval_code_evol_accuracy": 0.759983871353542, | |
| "eval_code_evol_loss": 0.90576171875, | |
| "eval_code_evol_runtime": 10.3923, | |
| "eval_code_evol_samples_per_second": 74.19, | |
| "eval_code_evol_steps_per_second": 9.334, | |
| "step": 3000 | |
| } | |
| ], | |
| "max_steps": 3176, | |
| "num_train_epochs": 8, | |
| "total_flos": 3.32565039055215e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |