| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 705, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 1.2041, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 1.3097, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 1.2029, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 1.1238, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 1.3468, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.0357, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 1.1403, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 1.3239, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 1.0573, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 1.268, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-05, | |
| "loss": 1.2789, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 1.152, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.181818181818182e-05, | |
| "loss": 1.2186, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 1.1533, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.3636363636363637e-05, | |
| "loss": 1.0094, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 0.9916, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.5454545454545454e-05, | |
| "loss": 1.0282, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 1.0948, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.7272727272727274e-05, | |
| "loss": 1.304, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 1.1626, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9090909090909094e-05, | |
| "loss": 1.3438, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2e-05, | |
| "loss": 1.0418, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.999989421415082e-05, | |
| "loss": 1.2165, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9999576858841395e-05, | |
| "loss": 1.2404, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9999047940786076e-05, | |
| "loss": 1.151, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9998307471175264e-05, | |
| "loss": 0.946, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9997355465675207e-05, | |
| "loss": 1.1605, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.999619194442764e-05, | |
| "loss": 1.0594, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9994816932049384e-05, | |
| "loss": 1.1498, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9993230457631812e-05, | |
| "loss": 1.1866, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9991432554740228e-05, | |
| "loss": 1.0144, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.998942326141317e-05, | |
| "loss": 1.0463, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.99872026201616e-05, | |
| "loss": 1.1274, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9984770677968e-05, | |
| "loss": 0.9874, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9982127486285386e-05, | |
| "loss": 1.2072, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.997927310103621e-05, | |
| "loss": 1.0499, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.997620758261119e-05, | |
| "loss": 1.0916, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9972930995868015e-05, | |
| "loss": 1.1274, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.996944341012999e-05, | |
| "loss": 1.09, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.996574489918456e-05, | |
| "loss": 1.0397, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9961835541281746e-05, | |
| "loss": 1.076, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.99577154191325e-05, | |
| "loss": 1.256, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9953384619906945e-05, | |
| "loss": 1.0442, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9948843235232534e-05, | |
| "loss": 1.0692, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.994409136119212e-05, | |
| "loss": 0.9714, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9939129098321904e-05, | |
| "loss": 1.1995, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.993395655160932e-05, | |
| "loss": 1.0976, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9928573830490828e-05, | |
| "loss": 1.0683, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9922981048849563e-05, | |
| "loss": 1.042, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9917178325012962e-05, | |
| "loss": 0.9205, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9911165781750235e-05, | |
| "loss": 1.0892, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9904943546269787e-05, | |
| "loss": 1.1622, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9898511750216505e-05, | |
| "loss": 1.0148, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.989187052966899e-05, | |
| "loss": 1.0344, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9885020025136677e-05, | |
| "loss": 1.0166, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9877960381556852e-05, | |
| "loss": 1.0739, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9870691748291594e-05, | |
| "loss": 1.0504, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.986321427912461e-05, | |
| "loss": 1.0437, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9855528132257984e-05, | |
| "loss": 1.0806, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9847633470308833e-05, | |
| "loss": 1.0201, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9839530460305863e-05, | |
| "loss": 1.1955, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9831219273685828e-05, | |
| "loss": 1.1367, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9822700086289915e-05, | |
| "loss": 1.2063, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9813973078360024e-05, | |
| "loss": 1.1051, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.980503843453494e-05, | |
| "loss": 1.0723, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.979589634384644e-05, | |
| "loss": 0.9828, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9786546999715285e-05, | |
| "loss": 0.8472, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9776990599947148e-05, | |
| "loss": 1.0304, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9767227346728393e-05, | |
| "loss": 1.0447, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.975725744662183e-05, | |
| "loss": 1.1636, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9747081110562324e-05, | |
| "loss": 1.0379, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.973669855385235e-05, | |
| "loss": 1.1352, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9726109996157423e-05, | |
| "loss": 1.0338, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9715315661501454e-05, | |
| "loss": 1.0593, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.970431577826202e-05, | |
| "loss": 1.2188, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9693110579165514e-05, | |
| "loss": 1.1663, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9681700301282235e-05, | |
| "loss": 1.0257, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9670085186021377e-05, | |
| "loss": 1.1147, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.96582654791259e-05, | |
| "loss": 1.0613, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9646241430667353e-05, | |
| "loss": 1.0762, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.963401329504057e-05, | |
| "loss": 1.122, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9621581330958293e-05, | |
| "loss": 0.9268, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.96089458014457e-05, | |
| "loss": 0.9894, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9596106973834833e-05, | |
| "loss": 1.0181, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.958306511975895e-05, | |
| "loss": 0.9888, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9569820515146768e-05, | |
| "loss": 0.9074, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.955637344021664e-05, | |
| "loss": 0.8809, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9542724179470616e-05, | |
| "loss": 1.009, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9528873021688422e-05, | |
| "loss": 1.2569, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9514820259921353e-05, | |
| "loss": 1.2034, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9500566191486077e-05, | |
| "loss": 1.1256, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9486111117958343e-05, | |
| "loss": 1.075, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9471455345166595e-05, | |
| "loss": 0.8509, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9456599183185508e-05, | |
| "loss": 1.1977, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9441542946329422e-05, | |
| "loss": 1.1116, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9426286953145707e-05, | |
| "loss": 1.0384, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9410831526407984e-05, | |
| "loss": 1.1884, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9395176993109357e-05, | |
| "loss": 0.9846, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.937932368445544e-05, | |
| "loss": 1.008, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9363271935857374e-05, | |
| "loss": 0.9739, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9347022086924733e-05, | |
| "loss": 0.9542, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.933057448145833e-05, | |
| "loss": 1.0802, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9313929467442953e-05, | |
| "loss": 1.0577, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9297087397039985e-05, | |
| "loss": 0.9401, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9280048626579964e-05, | |
| "loss": 1.2293, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.926281351655506e-05, | |
| "loss": 1.0088, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.924538243161142e-05, | |
| "loss": 1.1677, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.922775574054147e-05, | |
| "loss": 1.1176, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9209933816276104e-05, | |
| "loss": 0.9818, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.91919170358768e-05, | |
| "loss": 1.0263, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9173705780527643e-05, | |
| "loss": 1.1362, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9155300435527255e-05, | |
| "loss": 1.0157, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9136701390280644e-05, | |
| "loss": 1.0391, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9117909038290975e-05, | |
| "loss": 1.0368, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9098923777151222e-05, | |
| "loss": 1.1465, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9079746008535784e-05, | |
| "loss": 1.0191, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.906037613819197e-05, | |
| "loss": 0.9189, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9040814575931413e-05, | |
| "loss": 0.947, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9021061735621413e-05, | |
| "loss": 1.1261, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9001118035176163e-05, | |
| "loss": 0.8959, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8980983896547922e-05, | |
| "loss": 1.0372, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.896065974571808e-05, | |
| "loss": 0.982, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8940146012688148e-05, | |
| "loss": 1.0937, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.891944313147066e-05, | |
| "loss": 1.0602, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.889855154007999e-05, | |
| "loss": 0.9134, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8877471680523083e-05, | |
| "loss": 0.9775, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8856203998790113e-05, | |
| "loss": 1.1114, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8834748944845028e-05, | |
| "loss": 0.9003, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8813106972616055e-05, | |
| "loss": 0.9752, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.879127853998607e-05, | |
| "loss": 1.0671, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8769264108782935e-05, | |
| "loss": 0.9768, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8747064144769705e-05, | |
| "loss": 1.0939, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.872467911763479e-05, | |
| "loss": 0.864, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.870210950098201e-05, | |
| "loss": 0.9725, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8679355772320585e-05, | |
| "loss": 0.9786, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8656418413055006e-05, | |
| "loss": 1.0435, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8633297908474883e-05, | |
| "loss": 1.0748, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8609994747744663e-05, | |
| "loss": 0.9074, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8586509423893268e-05, | |
| "loss": 1.079, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8562842433803688e-05, | |
| "loss": 0.9117, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.853899427820245e-05, | |
| "loss": 1.1194, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8514965461649032e-05, | |
| "loss": 1.0018, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8490756492525185e-05, | |
| "loss": 0.9685, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8466367883024187e-05, | |
| "loss": 1.0195, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8441800149139988e-05, | |
| "loss": 0.9216, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8417053810656302e-05, | |
| "loss": 1.0363, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.839212939113562e-05, | |
| "loss": 1.1621, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8367027417908115e-05, | |
| "loss": 1.0108, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8341748422060504e-05, | |
| "loss": 1.0736, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8316292938424788e-05, | |
| "loss": 1.13, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8290661505566964e-05, | |
| "loss": 0.8874, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8264854665775607e-05, | |
| "loss": 0.8437, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.823887296505041e-05, | |
| "loss": 1.0343, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8212716953090627e-05, | |
| "loss": 1.1008, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8186387183283446e-05, | |
| "loss": 0.9401, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8159884212692275e-05, | |
| "loss": 1.0816, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.813320860204497e-05, | |
| "loss": 1.0325, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8106360915721956e-05, | |
| "loss": 1.0924, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.807934172174429e-05, | |
| "loss": 0.9261, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8052151591761644e-05, | |
| "loss": 0.9983, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.802479110104022e-05, | |
| "loss": 1.0472, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7997260828450568e-05, | |
| "loss": 1.1403, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.796956135645534e-05, | |
| "loss": 1.0501, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7941693271096966e-05, | |
| "loss": 1.0481, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7913657161985266e-05, | |
| "loss": 1.0142, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7885453622284957e-05, | |
| "loss": 1.1758, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7857083248703126e-05, | |
| "loss": 1.1482, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7828546641476577e-05, | |
| "loss": 1.1445, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7799844404359162e-05, | |
| "loss": 1.1811, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.777097714460898e-05, | |
| "loss": 0.9937, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.774194547297555e-05, | |
| "loss": 1.151, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7712750003686884e-05, | |
| "loss": 1.0487, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.768339135443648e-05, | |
| "loss": 1.0659, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.765387014637027e-05, | |
| "loss": 0.8877, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7624187004073464e-05, | |
| "loss": 1.1443, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7594342555557343e-05, | |
| "loss": 0.9694, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7564337432245977e-05, | |
| "loss": 1.1203, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7534172268962853e-05, | |
| "loss": 0.9387, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7503847703917456e-05, | |
| "loss": 0.9995, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.747336437869176e-05, | |
| "loss": 0.9974, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7442722938226647e-05, | |
| "loss": 0.8692, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7411924030808287e-05, | |
| "loss": 0.9464, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7380968308054385e-05, | |
| "loss": 0.9365, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.734985642490043e-05, | |
| "loss": 1.0132, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7318589039585818e-05, | |
| "loss": 1.0511, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.728716681363993e-05, | |
| "loss": 1.0418, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7255590411868138e-05, | |
| "loss": 1.0608, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7223860502337735e-05, | |
| "loss": 1.0386, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7191977756363808e-05, | |
| "loss": 0.973, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7159942848495027e-05, | |
| "loss": 1.1204, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7127756456499373e-05, | |
| "loss": 1.0082, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7095419261349818e-05, | |
| "loss": 1.0809, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.706293194720989e-05, | |
| "loss": 0.7784, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.703029520141921e-05, | |
| "loss": 0.9853, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6997509714478944e-05, | |
| "loss": 1.1248, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6964576180037217e-05, | |
| "loss": 1.0447, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.693149529487441e-05, | |
| "loss": 0.9837, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6898267758888422e-05, | |
| "loss": 0.9668, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.686489427507988e-05, | |
| "loss": 1.094, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.683137554953725e-05, | |
| "loss": 1.071, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6797712291421905e-05, | |
| "loss": 1.0331, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6763905212953103e-05, | |
| "loss": 0.9158, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.672995502939295e-05, | |
| "loss": 1.0372, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.669586245903125e-05, | |
| "loss": 0.9151, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6661628223170298e-05, | |
| "loss": 1.006, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.662725304610964e-05, | |
| "loss": 0.9668, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.659273765513073e-05, | |
| "loss": 1.0059, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6558082780481562e-05, | |
| "loss": 0.8644, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6523289155361206e-05, | |
| "loss": 0.9563, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6488357515904297e-05, | |
| "loss": 0.9226, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.645328860116546e-05, | |
| "loss": 1.0188, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6418083153103683e-05, | |
| "loss": 0.9819, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.638274191656661e-05, | |
| "loss": 1.1677, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.634726563927478e-05, | |
| "loss": 1.0698, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6311655071805824e-05, | |
| "loss": 0.9551, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6275910967578558e-05, | |
| "loss": 1.1529, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6240034082837078e-05, | |
| "loss": 1.0159, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6204025176634712e-05, | |
| "loss": 1.2224, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6167885010818017e-05, | |
| "loss": 1.0017, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6131614350010614e-05, | |
| "loss": 0.8603, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6095213961597034e-05, | |
| "loss": 1.0294, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6058684615706476e-05, | |
| "loss": 1.1921, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6022027085196516e-05, | |
| "loss": 0.983, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5985242145636753e-05, | |
| "loss": 1.089, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.59483305752924e-05, | |
| "loss": 0.8742, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.591129315510782e-05, | |
| "loss": 1.0083, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5874130668690004e-05, | |
| "loss": 0.9337, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5836843902291986e-05, | |
| "loss": 1.0195, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5799433644796215e-05, | |
| "loss": 0.9243, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5761900687697867e-05, | |
| "loss": 0.9224, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5724245825088086e-05, | |
| "loss": 1.0919, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.568646985363719e-05, | |
| "loss": 0.9274, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5648573572577837e-05, | |
| "loss": 0.9783, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.561055778368807e-05, | |
| "loss": 1.2035, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5572423291274393e-05, | |
| "loss": 0.9525, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.553417090215474e-05, | |
| "loss": 1.2295, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.549580142564141e-05, | |
| "loss": 1.0563, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.545731567352392e-05, | |
| "loss": 0.8898, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5418714460051875e-05, | |
| "loss": 1.1597, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5379998601917706e-05, | |
| "loss": 0.9869, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.534116891823939e-05, | |
| "loss": 0.9286, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5302226230543146e-05, | |
| "loss": 0.8408, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5263171362746028e-05, | |
| "loss": 0.8621, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5224005141138509e-05, | |
| "loss": 0.9568, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.518472839436699e-05, | |
| "loss": 1.0426, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5145341953416273e-05, | |
| "loss": 0.9256, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5105846651591981e-05, | |
| "loss": 1.1305, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5066243324502919e-05, | |
| "loss": 1.0253, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5026532810043408e-05, | |
| "loss": 1.0201, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4986715948375543e-05, | |
| "loss": 0.8552, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.494679358191143e-05, | |
| "loss": 0.971, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4906766555295357e-05, | |
| "loss": 1.1717, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4866635715385926e-05, | |
| "loss": 1.0087, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4826401911238131e-05, | |
| "loss": 1.0643, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4786065994085397e-05, | |
| "loss": 1.0314, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4745628817321577e-05, | |
| "loss": 1.1085, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4705091236482888e-05, | |
| "loss": 0.9997, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4664454109229809e-05, | |
| "loss": 1.1218, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4623718295328946e-05, | |
| "loss": 0.983, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4582884656634828e-05, | |
| "loss": 1.0985, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4541954057071692e-05, | |
| "loss": 1.1434, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4500927362615178e-05, | |
| "loss": 0.9975, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4459805441274028e-05, | |
| "loss": 0.9655, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4418589163071721e-05, | |
| "loss": 1.0507, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4377279400028053e-05, | |
| "loss": 1.0641, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.433587702614069e-05, | |
| "loss": 0.86, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4294382917366698e-05, | |
| "loss": 0.9321, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4252797951603978e-05, | |
| "loss": 0.9124, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4211123008672714e-05, | |
| "loss": 0.9226, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4169358970296751e-05, | |
| "loss": 1.0869, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4127506720084943e-05, | |
| "loss": 0.9645, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4085567143512458e-05, | |
| "loss": 0.94, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4043541127902037e-05, | |
| "loss": 0.9829, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4001429562405227e-05, | |
| "loss": 1.0159, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3959233337983582e-05, | |
| "loss": 1.0578, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3916953347389776e-05, | |
| "loss": 0.9338, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.387459048514876e-05, | |
| "loss": 1.0826, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.38321456475388e-05, | |
| "loss": 0.8876, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.378961973257254e-05, | |
| "loss": 1.0341, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3747013639977972e-05, | |
| "loss": 0.961, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.370432827117945e-05, | |
| "loss": 0.898, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3661564529278562e-05, | |
| "loss": 0.9889, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3618723319035056e-05, | |
| "loss": 1.096, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.35758055468477e-05, | |
| "loss": 1.044, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3532812120735088e-05, | |
| "loss": 0.9632, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3489743950316431e-05, | |
| "loss": 0.8849, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3446601946792334e-05, | |
| "loss": 1.0345, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3403387022925488e-05, | |
| "loss": 0.8672, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3360100093021378e-05, | |
| "loss": 1.2281, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3316742072908927e-05, | |
| "loss": 0.9715, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3273313879921139e-05, | |
| "loss": 1.0987, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.3229816432875665e-05, | |
| "loss": 0.9382, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.318625065205538e-05, | |
| "loss": 0.815, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.31426174591889e-05, | |
| "loss": 1.075, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.309891777743111e-05, | |
| "loss": 1.0681, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3055152531343592e-05, | |
| "loss": 0.9761, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3011322646875088e-05, | |
| "loss": 0.9582, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2967429051341913e-05, | |
| "loss": 0.9192, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2923472673408321e-05, | |
| "loss": 1.1883, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2879454443066862e-05, | |
| "loss": 1.0281, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2835375291618716e-05, | |
| "loss": 1.1208, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2791236151653974e-05, | |
| "loss": 0.9879, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2747037957031917e-05, | |
| "loss": 1.0591, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2702781642861252e-05, | |
| "loss": 1.0364, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2658468145480338e-05, | |
| "loss": 1.0091, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2614098402437367e-05, | |
| "loss": 1.0409, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2569673352470523e-05, | |
| "loss": 0.9112, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2525193935488138e-05, | |
| "loss": 1.0983, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2480661092548787e-05, | |
| "loss": 1.2059, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2436075765841396e-05, | |
| "loss": 0.9998, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.239143889866529e-05, | |
| "loss": 0.888, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.2346751435410249e-05, | |
| "loss": 0.8673, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.230201432153653e-05, | |
| "loss": 1.0609, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2257228503554834e-05, | |
| "loss": 0.8979, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2212394929006337e-05, | |
| "loss": 0.8436, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2167514546442576e-05, | |
| "loss": 1.0598, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2122588305405434e-05, | |
| "loss": 1.2273, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.207761715640702e-05, | |
| "loss": 1.0679, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2032602050909575e-05, | |
| "loss": 1.037, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1987543941305322e-05, | |
| "loss": 0.9155, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.194244378089635e-05, | |
| "loss": 1.0046, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1897302523874405e-05, | |
| "loss": 0.9412, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.185212112530073e-05, | |
| "loss": 1.0702, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.180690054108585e-05, | |
| "loss": 0.9705, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1761641727969344e-05, | |
| "loss": 1.0439, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1716345643499608e-05, | |
| "loss": 1.1034, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1671013246013595e-05, | |
| "loss": 0.954, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1625645494616535e-05, | |
| "loss": 1.1571, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.158024334916165e-05, | |
| "loss": 0.9643, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1534807770229845e-05, | |
| "loss": 0.8641, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1489339719109378e-05, | |
| "loss": 0.9144, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1443840157775528e-05, | |
| "loss": 1.0252, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1398310048870247e-05, | |
| "loss": 0.8798, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1352750355681771e-05, | |
| "loss": 0.9269, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1307162042124278e-05, | |
| "loss": 0.82, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1261546072717455e-05, | |
| "loss": 0.8315, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1215903412566112e-05, | |
| "loss": 0.9339, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1170235027339766e-05, | |
| "loss": 1.1432, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1124541883252199e-05, | |
| "loss": 0.8138, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1078824947041016e-05, | |
| "loss": 0.9904, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1033085185947208e-05, | |
| "loss": 0.9693, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0987323567694661e-05, | |
| "loss": 1.0318, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0941541060469712e-05, | |
| "loss": 0.9994, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0895738632900637e-05, | |
| "loss": 1.0541, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0849917254037174e-05, | |
| "loss": 0.9852, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0804077893330022e-05, | |
| "loss": 0.8855, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0758221520610321e-05, | |
| "loss": 1.0008, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0712349106069131e-05, | |
| "loss": 0.9588, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0666461620236923e-05, | |
| "loss": 0.9101, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0620560033963026e-05, | |
| "loss": 0.825, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0574645318395096e-05, | |
| "loss": 1.0322, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0528718444958568e-05, | |
| "loss": 1.0302, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0482780385336107e-05, | |
| "loss": 1.1374, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0436832111447034e-05, | |
| "loss": 0.941, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0390874595426794e-05, | |
| "loss": 0.9547, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0344908809606353e-05, | |
| "loss": 1.0572, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0298935726491648e-05, | |
| "loss": 0.9117, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0252956318743006e-05, | |
| "loss": 0.9152, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.020697155915457e-05, | |
| "loss": 0.8955, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0160982420633701e-05, | |
| "loss": 0.9301, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0114989876180424e-05, | |
| "loss": 0.9871, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0068994898866805e-05, | |
| "loss": 0.9394, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.002299846181639e-05, | |
| "loss": 1.0454, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.977001538183615e-06, | |
| "loss": 1.0332, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.9310051011332e-06, | |
| "loss": 1.0111, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.88501012381958e-06, | |
| "loss": 0.9455, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.839017579366299e-06, | |
| "loss": 0.9485, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.793028440845435e-06, | |
| "loss": 0.9813, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.747043681256996e-06, | |
| "loss": 0.9804, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.701064273508356e-06, | |
| "loss": 0.9842, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.65509119039365e-06, | |
| "loss": 1.0643, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.609125404573211e-06, | |
| "loss": 0.8455, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.563167888552969e-06, | |
| "loss": 0.999, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.517219614663896e-06, | |
| "loss": 0.8631, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.471281555041432e-06, | |
| "loss": 1.0162, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.425354681604908e-06, | |
| "loss": 1.0233, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.379439966036977e-06, | |
| "loss": 0.8583, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.33353837976308e-06, | |
| "loss": 1.0218, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.28765089393087e-06, | |
| "loss": 0.9836, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.241778479389684e-06, | |
| "loss": 0.8345, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.195922106669981e-06, | |
| "loss": 0.9187, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.150082745962828e-06, | |
| "loss": 0.9211, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.104261367099365e-06, | |
| "loss": 1.0165, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.058458939530295e-06, | |
| "loss": 0.9564, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.01267643230534e-06, | |
| "loss": 0.9146, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.966914814052797e-06, | |
| "loss": 0.9635, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.921175052958985e-06, | |
| "loss": 1.0199, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.875458116747807e-06, | |
| "loss": 1.2024, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.829764972660237e-06, | |
| "loss": 0.9747, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.78409658743389e-06, | |
| "loss": 0.9822, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.738453927282549e-06, | |
| "loss": 0.9269, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.692837957875727e-06, | |
| "loss": 0.9547, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.647249644318232e-06, | |
| "loss": 0.8801, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.601689951129757e-06, | |
| "loss": 1.006, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.556159842224472e-06, | |
| "loss": 1.1163, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.510660280890625e-06, | |
| "loss": 1.154, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.465192229770157e-06, | |
| "loss": 0.8659, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.419756650838351e-06, | |
| "loss": 1.062, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.374354505383468e-06, | |
| "loss": 0.9176, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.32898675398641e-06, | |
| "loss": 0.8966, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.283654356500395e-06, | |
| "loss": 1.033, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.238358272030658e-06, | |
| "loss": 0.9129, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.193099458914148e-06, | |
| "loss": 1.0298, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.147878874699273e-06, | |
| "loss": 0.8082, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.102697476125596e-06, | |
| "loss": 0.93, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.057556219103653e-06, | |
| "loss": 1.0115, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.012456058694678e-06, | |
| "loss": 1.0265, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.967397949090432e-06, | |
| "loss": 0.9531, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.922382843592983e-06, | |
| "loss": 0.9911, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.87741169459457e-06, | |
| "loss": 0.9252, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.832485453557424e-06, | |
| "loss": 0.8949, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.787605070993668e-06, | |
| "loss": 0.9604, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.742771496445167e-06, | |
| "loss": 0.9617, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.697985678463476e-06, | |
| "loss": 1.2224, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.653248564589751e-06, | |
| "loss": 1.0475, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.608561101334714e-06, | |
| "loss": 1.1681, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.563924234158608e-06, | |
| "loss": 1.0791, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.519338907451214e-06, | |
| "loss": 1.0108, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.474806064511864e-06, | |
| "loss": 0.9222, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.43032664752948e-06, | |
| "loss": 1.1035, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.385901597562637e-06, | |
| "loss": 1.0648, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.341531854519664e-06, | |
| "loss": 0.9768, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.297218357138749e-06, | |
| "loss": 1.0155, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.25296204296809e-06, | |
| "loss": 0.951, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.2087638483460295e-06, | |
| "loss": 1.0709, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.164624708381286e-06, | |
| "loss": 0.9438, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.1205455569331384e-06, | |
| "loss": 0.987, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.076527326591682e-06, | |
| "loss": 0.877, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.03257094865809e-06, | |
| "loss": 1.0672, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.988677353124913e-06, | |
| "loss": 0.9273, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.94484746865641e-06, | |
| "loss": 1.0268, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.901082222568894e-06, | |
| "loss": 0.8797, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.857382540811101e-06, | |
| "loss": 1.013, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.813749347944625e-06, | |
| "loss": 1.0319, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.770183567124336e-06, | |
| "loss": 1.0675, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.726686120078862e-06, | |
| "loss": 0.9384, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.683257927091073e-06, | |
| "loss": 0.9433, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.639899906978626e-06, | |
| "loss": 1.0475, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.596612977074515e-06, | |
| "loss": 0.9955, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.5533980532076716e-06, | |
| "loss": 0.9284, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.510256049683572e-06, | |
| "loss": 1.0718, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.467187879264916e-06, | |
| "loss": 0.7599, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.4241944531523e-06, | |
| "loss": 1.0618, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.381276680964947e-06, | |
| "loss": 0.823, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.338435470721442e-06, | |
| "loss": 1.1501, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.295671728820553e-06, | |
| "loss": 1.0543, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.2529863600220285e-06, | |
| "loss": 0.967, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.210380267427467e-06, | |
| "loss": 0.954, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.167854352461202e-06, | |
| "loss": 0.9743, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.125409514851244e-06, | |
| "loss": 1.1063, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.083046652610224e-06, | |
| "loss": 0.9986, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.040766662016424e-06, | |
| "loss": 0.9286, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.998570437594775e-06, | |
| "loss": 0.9988, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.956458872097966e-06, | |
| "loss": 0.907, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.914432856487544e-06, | |
| "loss": 0.9555, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.872493279915059e-06, | |
| "loss": 0.949, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.830641029703254e-06, | |
| "loss": 0.8961, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.788876991327288e-06, | |
| "loss": 0.9688, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.747202048396023e-06, | |
| "loss": 0.991, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.705617082633305e-06, | |
| "loss": 1.0318, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.664122973859313e-06, | |
| "loss": 0.9646, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.622720599971953e-06, | |
| "loss": 0.9297, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.5814108369282824e-06, | |
| "loss": 0.9207, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.540194558725973e-06, | |
| "loss": 0.8339, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.499072637384825e-06, | |
| "loss": 1.0934, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.458045942928309e-06, | |
| "loss": 0.9421, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.417115343365172e-06, | |
| "loss": 1.1037, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.37628170467106e-06, | |
| "loss": 1.1366, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.335545890770193e-06, | |
| "loss": 0.9912, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.294908763517115e-06, | |
| "loss": 0.9681, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.254371182678424e-06, | |
| "loss": 0.9724, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.213934005914607e-06, | |
| "loss": 0.8659, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.1735980887618745e-06, | |
| "loss": 0.9341, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.133364284614077e-06, | |
| "loss": 0.9018, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.093233444704642e-06, | |
| "loss": 0.99, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.053206418088572e-06, | |
| "loss": 1.1519, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.01328405162446e-06, | |
| "loss": 1.0575, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.973467189956596e-06, | |
| "loss": 1.0781, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.933756675497082e-06, | |
| "loss": 0.9561, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.894153348408022e-06, | |
| "loss": 1.1172, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.854658046583728e-06, | |
| "loss": 1.0779, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.815271605633012e-06, | |
| "loss": 0.8225, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.775994858861492e-06, | |
| "loss": 0.931, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7368286372539775e-06, | |
| "loss": 1.0002, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.697773769456859e-06, | |
| "loss": 1.0456, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.658831081760614e-06, | |
| "loss": 1.1681, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.620001398082295e-06, | |
| "loss": 1.1472, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.581285539948126e-06, | |
| "loss": 0.9188, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.542684326476082e-06, | |
| "loss": 1.0009, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.504198574358596e-06, | |
| "loss": 0.9792, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.4658290978452615e-06, | |
| "loss": 1.0876, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.4275767087256095e-06, | |
| "loss": 1.076, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.389442216311933e-06, | |
| "loss": 0.9606, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.351426427422165e-06, | |
| "loss": 0.9863, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.313530146362809e-06, | |
| "loss": 0.9644, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.275754174911921e-06, | |
| "loss": 1.0194, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.2380993123021385e-06, | |
| "loss": 0.8884, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.200566355203784e-06, | |
| "loss": 1.0964, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.163156097708014e-06, | |
| "loss": 0.8013, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.125869331309999e-06, | |
| "loss": 0.9615, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.088706844892182e-06, | |
| "loss": 0.9682, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.051669424707603e-06, | |
| "loss": 0.8881, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.014757854363249e-06, | |
| "loss": 1.0597, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.977972914803487e-06, | |
| "loss": 1.0211, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.9413153842935255e-06, | |
| "loss": 0.9069, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.904786038402968e-06, | |
| "loss": 0.9595, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.868385649989388e-06, | |
| "loss": 1.0338, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.832114989181988e-06, | |
| "loss": 1.0379, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.795974823365287e-06, | |
| "loss": 0.937, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.7599659171629254e-06, | |
| "loss": 0.8212, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.724089032421441e-06, | |
| "loss": 0.7433, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.688344928194181e-06, | |
| "loss": 1.0898, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.652734360725224e-06, | |
| "loss": 1.0453, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.617258083433396e-06, | |
| "loss": 1.1607, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.5819168468963183e-06, | |
| "loss": 0.8873, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.5467113988345438e-06, | |
| "loss": 0.9415, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.5116424840957066e-06, | |
| "loss": 1.067, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.4767108446387955e-06, | |
| "loss": 1.0038, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.441917219518438e-06, | |
| "loss": 0.9977, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.407262344869272e-06, | |
| "loss": 0.946, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.3727469538903645e-06, | |
| "loss": 0.9236, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.3383717768297054e-06, | |
| "loss": 1.148, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.3041375409687525e-06, | |
| "loss": 1.0021, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.2700449706070536e-06, | |
| "loss": 0.9826, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.2360947870469016e-06, | |
| "loss": 0.9472, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.2022877085780967e-06, | |
| "loss": 1.0949, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.168624450462746e-06, | |
| "loss": 0.8969, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.13510572492012e-06, | |
| "loss": 0.9965, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.1017322411115803e-06, | |
| "loss": 1.1017, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.0685047051255946e-06, | |
| "loss": 1.1201, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.035423819962785e-06, | |
| "loss": 0.9899, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.002490285521059e-06, | |
| "loss": 0.8676, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.969704798580796e-06, | |
| "loss": 1.0323, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.937068052790112e-06, | |
| "loss": 1.0097, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.904580738650181e-06, | |
| "loss": 1.1074, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.872243543500629e-06, | |
| "loss": 1.0416, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.8400571515049793e-06, | |
| "loss": 0.8266, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.8080222436361937e-06, | |
| "loss": 1.063, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.776139497662266e-06, | |
| "loss": 1.0304, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.7444095881318656e-06, | |
| "loss": 1.0025, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.712833186360072e-06, | |
| "loss": 1.0514, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.6814109604141845e-06, | |
| "loss": 1.0438, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.650143575099573e-06, | |
| "loss": 0.9465, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.6190316919456184e-06, | |
| "loss": 1.0099, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.588075969191718e-06, | |
| "loss": 1.0305, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.5572770617733544e-06, | |
| "loss": 0.9918, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.5266356213082434e-06, | |
| "loss": 0.8579, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.496152296082548e-06, | |
| "loss": 1.2437, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.465827731037147e-06, | |
| "loss": 1.1401, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.4356625677540237e-06, | |
| "loss": 1.0769, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.405657444442657e-06, | |
| "loss": 0.9141, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.375812995926541e-06, | |
| "loss": 0.9615, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3461298536297326e-06, | |
| "loss": 0.7358, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.3166086455635216e-06, | |
| "loss": 1.048, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2872499963131155e-06, | |
| "loss": 1.1585, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2580545270244512e-06, | |
| "loss": 1.0178, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.229022855391024e-06, | |
| "loss": 0.9983, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2001555956408428e-06, | |
| "loss": 0.899, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1714533585234243e-06, | |
| "loss": 0.8212, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1429167512968763e-06, | |
| "loss": 0.9148, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.114546377715042e-06, | |
| "loss": 0.9587, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.0863428380147346e-06, | |
| "loss": 1.0275, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.0583067289030334e-06, | |
| "loss": 0.8267, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.030438643544663e-06, | |
| "loss": 0.9677, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.002739171549435e-06, | |
| "loss": 0.9609, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9752088989597794e-06, | |
| "loss": 1.074, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9478484082383564e-06, | |
| "loss": 0.9023, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.920658278255714e-06, | |
| "loss": 0.9831, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.893639084278046e-06, | |
| "loss": 0.966, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8667913979550302e-06, | |
| "loss": 0.9048, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8401157873077258e-06, | |
| "loss": 0.9463, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.813612816716558e-06, | |
| "loss": 0.9245, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.787283046909376e-06, | |
| "loss": 0.9813, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7611270349495923e-06, | |
| "loss": 0.9172, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.735145334224394e-06, | |
| "loss": 0.9771, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7093384944330394e-06, | |
| "loss": 0.9921, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.6837070615752116e-06, | |
| "loss": 1.0297, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.658251577939497e-06, | |
| "loss": 0.9288, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.632972582091884e-06, | |
| "loss": 0.8836, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.6078706088643836e-06, | |
| "loss": 0.8562, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.5829461893437015e-06, | |
| "loss": 0.8242, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.5581998508600161e-06, | |
| "loss": 0.9472, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.533632116975814e-06, | |
| "loss": 1.026, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.5092435074748145e-06, | |
| "loss": 0.8626, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4850345383509701e-06, | |
| "loss": 0.869, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4610057217975526e-06, | |
| "loss": 0.9362, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4371575661963145e-06, | |
| "loss": 0.9731, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.413490576106733e-06, | |
| "loss": 1.0407, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3900052522553398e-06, | |
| "loss": 0.9148, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3667020915251173e-06, | |
| "loss": 1.1184, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3435815869449964e-06, | |
| "loss": 0.9784, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3206442276794207e-06, | |
| "loss": 1.0372, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2978904990179919e-06, | |
| "loss": 0.9861, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2753208823652142e-06, | |
| "loss": 1.1687, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2529358552302973e-06, | |
| "loss": 1.1031, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2307358912170687e-06, | |
| "loss": 0.8472, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2087214600139308e-06, | |
| "loss": 1.0433, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1868930273839474e-06, | |
| "loss": 0.9519, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1652510551549722e-06, | |
| "loss": 0.9413, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1437960012098892e-06, | |
| "loss": 1.0189, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1225283194769176e-06, | |
| "loss": 0.9054, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1014484599200125e-06, | |
| "loss": 1.0996, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0805568685293421e-06, | |
| "loss": 0.9392, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0598539873118552e-06, | |
| "loss": 0.9674, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0393402542819231e-06, | |
| "loss": 1.0506, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0190161034520797e-06, | |
| "loss": 0.9507, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.988819648238379e-07, | |
| "loss": 0.7685, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.789382643785894e-07, | |
| "loss": 0.9887, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.591854240685882e-07, | |
| "loss": 1.0764, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.396238618080322e-07, | |
| "loss": 1.0055, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.202539914642184e-07, | |
| "loss": 0.7657, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.010762228487813e-07, | |
| "loss": 1.1411, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.82090961709029e-07, | |
| "loss": 0.8572, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.632986097193574e-07, | |
| "loss": 1.1252, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.446995644727473e-07, | |
| "loss": 0.914, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.2629421947236e-07, | |
| "loss": 0.9381, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 8.080829641232013e-07, | |
| "loss": 0.9887, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.900661837238977e-07, | |
| "loss": 0.8681, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.722442594585312e-07, | |
| "loss": 0.9718, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.546175683885815e-07, | |
| "loss": 0.9835, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.371864834449405e-07, | |
| "loss": 1.0341, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.19951373420037e-07, | |
| "loss": 1.038, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.029126029600197e-07, | |
| "loss": 0.8693, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.860705325570493e-07, | |
| "loss": 1.09, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.694255185416687e-07, | |
| "loss": 0.9367, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.529779130752678e-07, | |
| "loss": 1.004, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.367280641426277e-07, | |
| "loss": 0.9833, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.206763155445628e-07, | |
| "loss": 0.928, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.048230068906447e-07, | |
| "loss": 0.8501, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.891684735920166e-07, | |
| "loss": 1.0349, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.737130468542973e-07, | |
| "loss": 1.0553, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.58457053670578e-07, | |
| "loss": 0.8809, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.434008168144944e-07, | |
| "loss": 0.778, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.285446548334072e-07, | |
| "loss": 0.9346, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.138888820416587e-07, | |
| "loss": 0.9802, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.994338085139261e-07, | |
| "loss": 1.079, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.851797400786506e-07, | |
| "loss": 1.0603, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.711269783115813e-07, | |
| "loss": 0.9383, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.572758205293848e-07, | |
| "loss": 0.9539, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.4362655978336e-07, | |
| "loss": 0.9954, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.301794848532326e-07, | |
| "loss": 1.0169, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.169348802410522e-07, | |
| "loss": 0.8573, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.0389302616516747e-07, | |
| "loss": 0.9394, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.9105419855430146e-07, | |
| "loss": 0.8933, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.7841866904170797e-07, | |
| "loss": 0.924, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.659867049594312e-07, | |
| "loss": 1.0655, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.5375856933264843e-07, | |
| "loss": 1.0421, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4173452087410186e-07, | |
| "loss": 0.8971, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.2991481397862567e-07, | |
| "loss": 0.987, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.1829969871776556e-07, | |
| "loss": 0.8905, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.068894208344897e-07, | |
| "loss": 0.8928, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.9568422173798293e-07, | |
| "loss": 1.0383, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.846843384985476e-07, | |
| "loss": 1.012, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.738900038425796e-07, | |
| "loss": 0.8459, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.633014461476524e-07, | |
| "loss": 0.9434, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.5291888943767993e-07, | |
| "loss": 0.9878, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.4274255337817465e-07, | |
| "loss": 0.9085, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.3277265327160903e-07, | |
| "loss": 0.8787, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.2300940005285377e-07, | |
| "loss": 1.0057, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.134530002847146e-07, | |
| "loss": 0.8496, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.0410365615356364e-07, | |
| "loss": 1.0935, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.9496156546506273e-07, | |
| "loss": 1.078, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.860269216399768e-07, | |
| "loss": 0.9329, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7729991371008504e-07, | |
| "loss": 1.0333, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.6878072631417386e-07, | |
| "loss": 0.7911, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.6046953969413913e-07, | |
| "loss": 1.0209, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.5236652969116805e-07, | |
| "loss": 0.9866, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.444718677420176e-07, | |
| "loss": 0.919, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.367857208753931e-07, | |
| "loss": 0.9626, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.2930825170840877e-07, | |
| "loss": 1.0431, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.2203961844315048e-07, | |
| "loss": 0.8534, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1497997486332512e-07, | |
| "loss": 1.0252, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.0812947033101207e-07, | |
| "loss": 1.0337, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0148824978349791e-07, | |
| "loss": 0.9547, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.505645373021455e-08, | |
| "loss": 1.042, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.883421824976479e-08, | |
| "loss": 1.0198, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.282167498703919e-08, | |
| "loss": 0.9271, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.701895115043823e-08, | |
| "loss": 1.0045, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.142616950917447e-08, | |
| "loss": 0.9935, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.604344839068022e-08, | |
| "loss": 1.0396, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.08709016780984e-08, | |
| "loss": 0.9228, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.590863880788111e-08, | |
| "loss": 0.9778, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.115676476746489e-08, | |
| "loss": 1.1956, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.6615380093055774e-08, | |
| "loss": 0.9367, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.228458086750098e-08, | |
| "loss": 1.044, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.8164458718255025e-08, | |
| "loss": 1.0658, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.425510081544237e-08, | |
| "loss": 1.0386, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.05565898700122e-08, | |
| "loss": 0.8835, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.7069004131987652e-08, | |
| "loss": 0.7686, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.379241738881377e-08, | |
| "loss": 0.9699, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.0726898963793208e-08, | |
| "loss": 1.0765, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7872513714617403e-08, | |
| "loss": 1.0337, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.5229322032002113e-08, | |
| "loss": 1.078, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.2797379838402901e-08, | |
| "loss": 0.9252, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0576738586831614e-08, | |
| "loss": 1.0517, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.567445259775042e-09, | |
| "loss": 1.0319, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.7695423681901625e-09, | |
| "loss": 1.1047, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.183067950617071e-09, | |
| "loss": 0.9721, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.808055572362967e-09, | |
| "loss": 0.9197, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.6445343247982758e-09, | |
| "loss": 1.0672, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.6925288247393589e-09, | |
| "loss": 0.9606, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.52059213927825e-10, | |
| "loss": 1.0111, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.2314115860642157e-10, | |
| "loss": 0.8451, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0578584918374824e-10, | |
| "loss": 0.9923, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.0306, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 705, | |
| "total_flos": 262405283512320.0, | |
| "train_loss": 1.0077643413070245, | |
| "train_runtime": 9660.1894, | |
| "train_samples_per_second": 4.665, | |
| "train_steps_per_second": 0.073 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 705, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 400, | |
| "total_flos": 262405283512320.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |