| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.21689621516104543, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999996372615291e-05, | |
| "loss": 1.4261, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999985490471688e-05, | |
| "loss": 1.3752, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999673536007715e-05, | |
| "loss": 1.3193, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999941962055173e-05, | |
| "loss": 1.2661, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9999093159085765e-05, | |
| "loss": 1.2613, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999869415255717e-05, | |
| "loss": 1.2233, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9998222602123825e-05, | |
| "loss": 1.2184, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9997678509154145e-05, | |
| "loss": 1.2031, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999706187522701e-05, | |
| "loss": 1.1744, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9996372702131866e-05, | |
| "loss": 1.1756, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9995610991868615e-05, | |
| "loss": 1.1611, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9994776746647656e-05, | |
| "loss": 1.1384, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999386996888991e-05, | |
| "loss": 1.1275, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999289066122676e-05, | |
| "loss": 1.1227, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999183882650006e-05, | |
| "loss": 1.1135, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999071446776215e-05, | |
| "loss": 1.1253, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998951758827579e-05, | |
| "loss": 1.11, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998824819151424e-05, | |
| "loss": 1.0942, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998690628116115e-05, | |
| "loss": 1.0963, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998549186111065e-05, | |
| "loss": 1.0676, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998400493546723e-05, | |
| "loss": 1.0891, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998244550854582e-05, | |
| "loss": 1.0612, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9980813584871734e-05, | |
| "loss": 1.0726, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9979109169180664e-05, | |
| "loss": 1.0721, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.997733226641867e-05, | |
| "loss": 1.0757, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9975482881742155e-05, | |
| "loss": 1.0606, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9973561020517854e-05, | |
| "loss": 1.0624, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9971566688322854e-05, | |
| "loss": 1.0503, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9969499890944505e-05, | |
| "loss": 1.0376, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996736063438047e-05, | |
| "loss": 1.0402, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996514892483866e-05, | |
| "loss": 1.0362, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996286476873727e-05, | |
| "loss": 1.0363, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.996050817270471e-05, | |
| "loss": 1.0354, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9958079143579594e-05, | |
| "loss": 1.0437, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9955577688410744e-05, | |
| "loss": 1.0223, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9953003814457154e-05, | |
| "loss": 1.017, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.995035752918797e-05, | |
| "loss": 1.013, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.994763884028246e-05, | |
| "loss": 1.0177, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.994484775563002e-05, | |
| "loss": 1.0445, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9941984283330115e-05, | |
| "loss": 1.0078, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.993904843169227e-05, | |
| "loss": 1.0033, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.993604020923607e-05, | |
| "loss": 1.0108, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.993295962469109e-05, | |
| "loss": 1.0085, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.99298066869969e-05, | |
| "loss": 1.0115, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.992658140530304e-05, | |
| "loss": 0.9952, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.992328378896898e-05, | |
| "loss": 1.0234, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.991991384756409e-05, | |
| "loss": 0.9995, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.991647159086764e-05, | |
| "loss": 0.9844, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.991295702886874e-05, | |
| "loss": 0.9989, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.990937017176632e-05, | |
| "loss": 1.0023, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.990571102996911e-05, | |
| "loss": 0.9729, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.990197961409561e-05, | |
| "loss": 0.9793, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.989817593497404e-05, | |
| "loss": 0.9876, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9894300003642304e-05, | |
| "loss": 0.9883, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.989035183134802e-05, | |
| "loss": 0.9997, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9886331429548434e-05, | |
| "loss": 0.9998, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.988223880991035e-05, | |
| "loss": 0.988, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9878073984310194e-05, | |
| "loss": 1.0108, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.987383696483389e-05, | |
| "loss": 0.9929, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.987126010519375e-05, | |
| "loss": 1.0691, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9866907601177285e-05, | |
| "loss": 0.9875, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.986248293568848e-05, | |
| "loss": 0.9738, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9857986121567325e-05, | |
| "loss": 0.9682, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.985341717186316e-05, | |
| "loss": 0.9955, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.984877609983465e-05, | |
| "loss": 0.9865, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.984406291894975e-05, | |
| "loss": 0.981, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9839277642885694e-05, | |
| "loss": 0.9725, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9834420285528904e-05, | |
| "loss": 0.9681, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9829490860974975e-05, | |
| "loss": 0.9632, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.982448938352865e-05, | |
| "loss": 0.9877, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.981941586770376e-05, | |
| "loss": 0.9808, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9814270328223166e-05, | |
| "loss": 0.9677, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.980905278001876e-05, | |
| "loss": 0.9699, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.980376323823139e-05, | |
| "loss": 0.9496, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.97984017182108e-05, | |
| "loss": 0.9669, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.979296823551565e-05, | |
| "loss": 0.9784, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.978746280591338e-05, | |
| "loss": 0.9626, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.978188544538026e-05, | |
| "loss": 0.9695, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.977623617010127e-05, | |
| "loss": 0.9575, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.977051499647008e-05, | |
| "loss": 0.9506, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9764721941089016e-05, | |
| "loss": 0.9489, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.975885702076898e-05, | |
| "loss": 0.9454, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.975292025252945e-05, | |
| "loss": 0.9427, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.974691165359836e-05, | |
| "loss": 0.944, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.974083124141211e-05, | |
| "loss": 0.9147, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9734679033615505e-05, | |
| "loss": 0.9487, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9728455048061686e-05, | |
| "loss": 0.9595, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9722159302812074e-05, | |
| "loss": 0.9522, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.971579181613635e-05, | |
| "loss": 0.9575, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.970935260651237e-05, | |
| "loss": 0.9474, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9702841692626125e-05, | |
| "loss": 0.9533, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.969625909337169e-05, | |
| "loss": 0.9674, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.968960482785116e-05, | |
| "loss": 0.9196, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.96828789153746e-05, | |
| "loss": 0.9326, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.967608137545998e-05, | |
| "loss": 0.9546, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.966921222783315e-05, | |
| "loss": 0.937, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.966227149242773e-05, | |
| "loss": 0.9298, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.96552591893851e-05, | |
| "loss": 0.944, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9648175339054314e-05, | |
| "loss": 0.9715, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.964101996199205e-05, | |
| "loss": 0.9465, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.963379307896255e-05, | |
| "loss": 0.9526, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9626494710937575e-05, | |
| "loss": 0.9149, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.96191248790963e-05, | |
| "loss": 0.9608, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.961168360482531e-05, | |
| "loss": 0.9236, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9604170909718494e-05, | |
| "loss": 0.9455, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9596586815577e-05, | |
| "loss": 0.9637, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.958893134440915e-05, | |
| "loss": 0.9477, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.958120451843045e-05, | |
| "loss": 0.9275, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.957340636006341e-05, | |
| "loss": 0.946, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.956553689193758e-05, | |
| "loss": 0.9226, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9557596136889425e-05, | |
| "loss": 0.9463, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.954958411796229e-05, | |
| "loss": 0.9514, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.95415008584063e-05, | |
| "loss": 0.9169, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9533346381678346e-05, | |
| "loss": 0.9345, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.952512071144195e-05, | |
| "loss": 0.9313, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.951682387156727e-05, | |
| "loss": 0.931, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.950845588613096e-05, | |
| "loss": 0.9292, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9500016779416124e-05, | |
| "loss": 0.9442, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9491506575912306e-05, | |
| "loss": 0.9307, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9482925300315296e-05, | |
| "loss": 0.9473, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9474272977527194e-05, | |
| "loss": 0.9279, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.946554963265623e-05, | |
| "loss": 0.9433, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9456755291016744e-05, | |
| "loss": 0.9286, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9447889978129105e-05, | |
| "loss": 0.9307, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.943895371971963e-05, | |
| "loss": 0.9207, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.942994654172053e-05, | |
| "loss": 0.9067, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.942086847026979e-05, | |
| "loss": 0.9253, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.941171953171115e-05, | |
| "loss": 0.9072, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9402499752593975e-05, | |
| "loss": 0.9051, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9393209159673217e-05, | |
| "loss": 0.9076, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.938384777990932e-05, | |
| "loss": 0.9214, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.937441564046814e-05, | |
| "loss": 0.9249, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.936491276872089e-05, | |
| "loss": 0.9379, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.935533919224401e-05, | |
| "loss": 0.933, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9345694938819146e-05, | |
| "loss": 0.9228, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.933598003643303e-05, | |
| "loss": 0.9327, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9326194513277415e-05, | |
| "loss": 0.9198, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.931633839774898e-05, | |
| "loss": 0.9349, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.930641171844926e-05, | |
| "loss": 0.9296, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.929641450418459e-05, | |
| "loss": 0.9283, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9286346783965925e-05, | |
| "loss": 0.9015, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9276208587008885e-05, | |
| "loss": 0.9227, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.926599994273358e-05, | |
| "loss": 0.9059, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.925572088076454e-05, | |
| "loss": 0.9038, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.924537143093068e-05, | |
| "loss": 0.9109, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.923495162326512e-05, | |
| "loss": 0.9094, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.92244614880052e-05, | |
| "loss": 0.9292, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.921390105559233e-05, | |
| "loss": 0.9266, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9203270356671894e-05, | |
| "loss": 0.8977, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.91925694220932e-05, | |
| "loss": 0.9151, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.918179828290939e-05, | |
| "loss": 0.9118, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.91709569703773e-05, | |
| "loss": 0.9259, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9160045515957435e-05, | |
| "loss": 0.9295, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.914906395131381e-05, | |
| "loss": 0.9152, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.913801230831392e-05, | |
| "loss": 0.9187, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9126890619028623e-05, | |
| "loss": 0.9371, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9115698915732025e-05, | |
| "loss": 0.909, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9104437230901424e-05, | |
| "loss": 0.9044, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.909310559721718e-05, | |
| "loss": 0.9031, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.908170404756267e-05, | |
| "loss": 0.9221, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9070232615024115e-05, | |
| "loss": 0.9227, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.905869133289056e-05, | |
| "loss": 0.928, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.904708023465375e-05, | |
| "loss": 0.8813, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.903539935400803e-05, | |
| "loss": 0.8944, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9023648724850214e-05, | |
| "loss": 0.8901, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.901182838127955e-05, | |
| "loss": 0.923, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8999938357597594e-05, | |
| "loss": 0.9103, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.89879786883081e-05, | |
| "loss": 0.9013, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8975949408116914e-05, | |
| "loss": 0.8921, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8963850551931904e-05, | |
| "loss": 0.9103, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8951682154862836e-05, | |
| "loss": 0.9057, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.893944425222128e-05, | |
| "loss": 0.9005, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.892713687952048e-05, | |
| "loss": 0.8894, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8914760072475316e-05, | |
| "loss": 0.9067, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.890231386700214e-05, | |
| "loss": 0.9206, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.888979829921869e-05, | |
| "loss": 0.9143, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.887721340544398e-05, | |
| "loss": 0.8992, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.886455922219822e-05, | |
| "loss": 0.9068, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.885183578620268e-05, | |
| "loss": 0.8978, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8839043134379604e-05, | |
| "loss": 0.9198, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8826181303852075e-05, | |
| "loss": 0.9042, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.881325033194395e-05, | |
| "loss": 0.8947, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.880025025617971e-05, | |
| "loss": 0.9095, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.878718111428438e-05, | |
| "loss": 0.9077, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8774042944183404e-05, | |
| "loss": 0.9092, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.876083578400255e-05, | |
| "loss": 0.894, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.874755967206776e-05, | |
| "loss": 0.9111, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8734214646905086e-05, | |
| "loss": 0.8819, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.872080074724058e-05, | |
| "loss": 0.8981, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.870731801200012e-05, | |
| "loss": 0.8966, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.869376648030938e-05, | |
| "loss": 0.8935, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.868014619149364e-05, | |
| "loss": 0.9013, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.866645718507772e-05, | |
| "loss": 0.9019, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.865269950078587e-05, | |
| "loss": 0.913, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.863887317854161e-05, | |
| "loss": 0.893, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8624978258467655e-05, | |
| "loss": 0.8725, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.861101478088578e-05, | |
| "loss": 0.9065, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.859698278631671e-05, | |
| "loss": 0.8794, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.858288231547999e-05, | |
| "loss": 0.8948, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.8568713409293906e-05, | |
| "loss": 0.9128, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 18442, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.0641966568819917e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |