| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.873015873015873, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.015873015873015872, | |
| "grad_norm": 0.4978693127632141, | |
| "learning_rate": 5e-05, | |
| "loss": 0.9589, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.031746031746031744, | |
| "grad_norm": 0.4809919595718384, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9265, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.047619047619047616, | |
| "grad_norm": 0.5111315250396729, | |
| "learning_rate": 9.999975227016531e-05, | |
| "loss": 0.9665, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.06349206349206349, | |
| "grad_norm": 0.670375406742096, | |
| "learning_rate": 9.999900908311602e-05, | |
| "loss": 0.9922, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07936507936507936, | |
| "grad_norm": 0.6541376113891602, | |
| "learning_rate": 9.999777044621652e-05, | |
| "loss": 0.7833, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09523809523809523, | |
| "grad_norm": 0.8511355519294739, | |
| "learning_rate": 9.999603637174071e-05, | |
| "loss": 0.8339, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.933815598487854, | |
| "learning_rate": 9.999380687687188e-05, | |
| "loss": 0.7231, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.12698412698412698, | |
| "grad_norm": 1.2876204252243042, | |
| "learning_rate": 9.999108198370249e-05, | |
| "loss": 0.9078, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.14285714285714285, | |
| "grad_norm": 1.2767138481140137, | |
| "learning_rate": 9.998786171923407e-05, | |
| "loss": 0.8635, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.15873015873015872, | |
| "grad_norm": 1.2097059488296509, | |
| "learning_rate": 9.998414611537681e-05, | |
| "loss": 0.8626, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.1746031746031746, | |
| "grad_norm": 1.4272280931472778, | |
| "learning_rate": 9.997993520894937e-05, | |
| "loss": 0.9185, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 0.8439553380012512, | |
| "learning_rate": 9.997522904167844e-05, | |
| "loss": 0.6789, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.20634920634920634, | |
| "grad_norm": 0.8309609293937683, | |
| "learning_rate": 9.997002766019832e-05, | |
| "loss": 0.8529, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 0.9561108350753784, | |
| "learning_rate": 9.996433111605052e-05, | |
| "loss": 0.8475, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.23809523809523808, | |
| "grad_norm": 0.6515036821365356, | |
| "learning_rate": 9.99581394656832e-05, | |
| "loss": 0.8345, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.25396825396825395, | |
| "grad_norm": 0.45641326904296875, | |
| "learning_rate": 9.995145277045061e-05, | |
| "loss": 0.6654, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2698412698412698, | |
| "grad_norm": 0.2796855568885803, | |
| "learning_rate": 9.994427109661253e-05, | |
| "loss": 0.6643, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 0.2947935461997986, | |
| "learning_rate": 9.993659451533353e-05, | |
| "loss": 0.7327, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.30158730158730157, | |
| "grad_norm": 0.3294975161552429, | |
| "learning_rate": 9.992842310268233e-05, | |
| "loss": 0.7466, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.31746031746031744, | |
| "grad_norm": 0.25266706943511963, | |
| "learning_rate": 9.991975693963107e-05, | |
| "loss": 0.6628, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.26995259523391724, | |
| "learning_rate": 9.99105961120544e-05, | |
| "loss": 0.7145, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.3492063492063492, | |
| "grad_norm": 0.29081106185913086, | |
| "learning_rate": 9.990094071072877e-05, | |
| "loss": 0.6947, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.36507936507936506, | |
| "grad_norm": 0.277067095041275, | |
| "learning_rate": 9.989079083133139e-05, | |
| "loss": 0.7225, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 0.27529773116111755, | |
| "learning_rate": 9.988014657443941e-05, | |
| "loss": 0.7122, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.3968253968253968, | |
| "grad_norm": 0.30684614181518555, | |
| "learning_rate": 9.986900804552878e-05, | |
| "loss": 0.7015, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.4126984126984127, | |
| "grad_norm": 0.30238044261932373, | |
| "learning_rate": 9.985737535497337e-05, | |
| "loss": 0.5781, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.42857142857142855, | |
| "grad_norm": 0.30560147762298584, | |
| "learning_rate": 9.984524861804376e-05, | |
| "loss": 0.5947, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.280203253030777, | |
| "learning_rate": 9.983262795490613e-05, | |
| "loss": 0.7072, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.4603174603174603, | |
| "grad_norm": 0.28849631547927856, | |
| "learning_rate": 9.981951349062106e-05, | |
| "loss": 0.7074, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.47619047619047616, | |
| "grad_norm": 0.2815149426460266, | |
| "learning_rate": 9.980590535514233e-05, | |
| "loss": 0.5274, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.49206349206349204, | |
| "grad_norm": 0.26764699816703796, | |
| "learning_rate": 9.979180368331558e-05, | |
| "loss": 0.6645, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.5079365079365079, | |
| "grad_norm": 0.29958057403564453, | |
| "learning_rate": 9.9777208614877e-05, | |
| "loss": 0.7361, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.5238095238095238, | |
| "grad_norm": 0.26811736822128296, | |
| "learning_rate": 9.976212029445194e-05, | |
| "loss": 0.6962, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.5396825396825397, | |
| "grad_norm": 0.2567647695541382, | |
| "learning_rate": 9.97465388715535e-05, | |
| "loss": 0.6077, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 0.25592276453971863, | |
| "learning_rate": 9.9730464500581e-05, | |
| "loss": 0.6288, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.24128927290439606, | |
| "learning_rate": 9.971389734081848e-05, | |
| "loss": 0.5665, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.5873015873015873, | |
| "grad_norm": 0.2471931427717209, | |
| "learning_rate": 9.969683755643317e-05, | |
| "loss": 0.7, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.6031746031746031, | |
| "grad_norm": 0.24910229444503784, | |
| "learning_rate": 9.967928531647374e-05, | |
| "loss": 0.5286, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.6190476190476191, | |
| "grad_norm": 0.29654461145401, | |
| "learning_rate": 9.966124079486872e-05, | |
| "loss": 0.6379, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.6349206349206349, | |
| "grad_norm": 0.23167571425437927, | |
| "learning_rate": 9.96427041704248e-05, | |
| "loss": 0.5028, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.6507936507936508, | |
| "grad_norm": 0.3802570402622223, | |
| "learning_rate": 9.962367562682496e-05, | |
| "loss": 0.7501, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.2911546230316162, | |
| "learning_rate": 9.960415535262671e-05, | |
| "loss": 0.7529, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.6825396825396826, | |
| "grad_norm": 0.27725136280059814, | |
| "learning_rate": 9.958414354126022e-05, | |
| "loss": 0.6338, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6984126984126984, | |
| "grad_norm": 0.29778677225112915, | |
| "learning_rate": 9.956364039102642e-05, | |
| "loss": 0.6084, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 0.3038597106933594, | |
| "learning_rate": 9.954264610509497e-05, | |
| "loss": 0.7813, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.7301587301587301, | |
| "grad_norm": 0.24961970746517181, | |
| "learning_rate": 9.952116089150232e-05, | |
| "loss": 0.5784, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.746031746031746, | |
| "grad_norm": 0.41124090552330017, | |
| "learning_rate": 9.94991849631496e-05, | |
| "loss": 0.8362, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 0.2612743079662323, | |
| "learning_rate": 9.947671853780054e-05, | |
| "loss": 0.5879, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.7777777777777778, | |
| "grad_norm": 0.3509594798088074, | |
| "learning_rate": 9.94537618380793e-05, | |
| "loss": 0.6429, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.7936507936507936, | |
| "grad_norm": 0.4222470223903656, | |
| "learning_rate": 9.943031509146825e-05, | |
| "loss": 0.8086, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.8095238095238095, | |
| "grad_norm": 0.34031662344932556, | |
| "learning_rate": 9.940637853030572e-05, | |
| "loss": 0.7058, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.8253968253968254, | |
| "grad_norm": 0.25386595726013184, | |
| "learning_rate": 9.938195239178374e-05, | |
| "loss": 0.5537, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.8412698412698413, | |
| "grad_norm": 0.27435001730918884, | |
| "learning_rate": 9.935703691794565e-05, | |
| "loss": 0.5793, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.8571428571428571, | |
| "grad_norm": 0.360727995634079, | |
| "learning_rate": 9.933163235568367e-05, | |
| "loss": 0.6103, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.873015873015873, | |
| "grad_norm": 0.29674389958381653, | |
| "learning_rate": 9.930573895673657e-05, | |
| "loss": 0.7375, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 0.3319956958293915, | |
| "learning_rate": 9.927935697768698e-05, | |
| "loss": 0.5953, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.9047619047619048, | |
| "grad_norm": 0.3237013518810272, | |
| "learning_rate": 9.925248667995907e-05, | |
| "loss": 0.6891, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.9206349206349206, | |
| "grad_norm": 0.2946189343929291, | |
| "learning_rate": 9.922512832981584e-05, | |
| "loss": 0.5815, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.9365079365079365, | |
| "grad_norm": 0.31961193680763245, | |
| "learning_rate": 9.919728219835643e-05, | |
| "loss": 0.6767, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.30548524856567383, | |
| "learning_rate": 9.916894856151357e-05, | |
| "loss": 0.6222, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.9682539682539683, | |
| "grad_norm": 0.2908201515674591, | |
| "learning_rate": 9.914012770005072e-05, | |
| "loss": 0.6102, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.9841269841269841, | |
| "grad_norm": 0.3024301826953888, | |
| "learning_rate": 9.91108198995594e-05, | |
| "loss": 0.6281, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.37488242983818054, | |
| "learning_rate": 9.908102545045625e-05, | |
| "loss": 0.5405, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.0158730158730158, | |
| "grad_norm": 0.3462425172328949, | |
| "learning_rate": 9.905074464798024e-05, | |
| "loss": 0.5831, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.0317460317460316, | |
| "grad_norm": 0.32379499077796936, | |
| "learning_rate": 9.901997779218967e-05, | |
| "loss": 0.6897, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.0476190476190477, | |
| "grad_norm": 0.3253431022167206, | |
| "learning_rate": 9.898872518795932e-05, | |
| "loss": 0.5935, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.0634920634920635, | |
| "grad_norm": 0.31801578402519226, | |
| "learning_rate": 9.895698714497724e-05, | |
| "loss": 0.5721, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.0793650793650793, | |
| "grad_norm": 0.29547229409217834, | |
| "learning_rate": 9.892476397774186e-05, | |
| "loss": 0.5041, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.0952380952380953, | |
| "grad_norm": 0.30208516120910645, | |
| "learning_rate": 9.889205600555877e-05, | |
| "loss": 0.5027, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 0.37307029962539673, | |
| "learning_rate": 9.885886355253758e-05, | |
| "loss": 0.6963, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.126984126984127, | |
| "grad_norm": 0.31057053804397583, | |
| "learning_rate": 9.882518694758875e-05, | |
| "loss": 0.4872, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 0.35556697845458984, | |
| "learning_rate": 9.879102652442024e-05, | |
| "loss": 0.6017, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.1587301587301586, | |
| "grad_norm": 0.37607231736183167, | |
| "learning_rate": 9.875638262153431e-05, | |
| "loss": 0.6837, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.1746031746031746, | |
| "grad_norm": 0.34590160846710205, | |
| "learning_rate": 9.872125558222409e-05, | |
| "loss": 0.5724, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.1904761904761905, | |
| "grad_norm": 0.3449731469154358, | |
| "learning_rate": 9.868564575457023e-05, | |
| "loss": 0.6157, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.2063492063492063, | |
| "grad_norm": 0.4771505892276764, | |
| "learning_rate": 9.864955349143734e-05, | |
| "loss": 0.5829, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.2222222222222223, | |
| "grad_norm": 0.374600887298584, | |
| "learning_rate": 9.861297915047069e-05, | |
| "loss": 0.6213, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.2380952380952381, | |
| "grad_norm": 0.40953242778778076, | |
| "learning_rate": 9.857592309409247e-05, | |
| "loss": 0.5805, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.253968253968254, | |
| "grad_norm": 0.3891858756542206, | |
| "learning_rate": 9.853838568949831e-05, | |
| "loss": 0.5201, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.2698412698412698, | |
| "grad_norm": 0.4599400758743286, | |
| "learning_rate": 9.850036730865364e-05, | |
| "loss": 0.6509, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.2857142857142856, | |
| "grad_norm": 0.47590476274490356, | |
| "learning_rate": 9.846186832828989e-05, | |
| "loss": 0.5522, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.3015873015873016, | |
| "grad_norm": 0.42077696323394775, | |
| "learning_rate": 9.842288912990096e-05, | |
| "loss": 0.6272, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.3174603174603174, | |
| "grad_norm": 0.4116186201572418, | |
| "learning_rate": 9.838343009973925e-05, | |
| "loss": 0.5974, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.4247848689556122, | |
| "learning_rate": 9.83434916288119e-05, | |
| "loss": 0.4948, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.3492063492063493, | |
| "grad_norm": 0.3873782455921173, | |
| "learning_rate": 9.830307411287695e-05, | |
| "loss": 0.496, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.3650793650793651, | |
| "grad_norm": 0.4587806463241577, | |
| "learning_rate": 9.82621779524394e-05, | |
| "loss": 0.6617, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.380952380952381, | |
| "grad_norm": 0.4379841089248657, | |
| "learning_rate": 9.822080355274719e-05, | |
| "loss": 0.5294, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.3968253968253967, | |
| "grad_norm": 0.464910626411438, | |
| "learning_rate": 9.817895132378725e-05, | |
| "loss": 0.6855, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.4126984126984126, | |
| "grad_norm": 0.4157741963863373, | |
| "learning_rate": 9.813662168028144e-05, | |
| "loss": 0.5563, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.4285714285714286, | |
| "grad_norm": 0.4436641037464142, | |
| "learning_rate": 9.809381504168234e-05, | |
| "loss": 0.5291, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.4444444444444444, | |
| "grad_norm": 0.4180889427661896, | |
| "learning_rate": 9.805053183216923e-05, | |
| "loss": 0.5158, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.4603174603174602, | |
| "grad_norm": 0.4933961033821106, | |
| "learning_rate": 9.800677248064382e-05, | |
| "loss": 0.6885, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.4761904761904763, | |
| "grad_norm": 0.4813699722290039, | |
| "learning_rate": 9.796253742072596e-05, | |
| "loss": 0.6305, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.492063492063492, | |
| "grad_norm": 0.4272967278957367, | |
| "learning_rate": 9.791782709074944e-05, | |
| "loss": 0.5119, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.507936507936508, | |
| "grad_norm": 0.4510858356952667, | |
| "learning_rate": 9.787264193375753e-05, | |
| "loss": 0.5693, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.5238095238095237, | |
| "grad_norm": 0.5349589586257935, | |
| "learning_rate": 9.782698239749873e-05, | |
| "loss": 0.6708, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.5396825396825395, | |
| "grad_norm": 0.5285341739654541, | |
| "learning_rate": 9.778084893442218e-05, | |
| "loss": 0.6712, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 0.4625120460987091, | |
| "learning_rate": 9.77342420016733e-05, | |
| "loss": 0.5257, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.5714285714285714, | |
| "grad_norm": 0.4635828733444214, | |
| "learning_rate": 9.768716206108921e-05, | |
| "loss": 0.482, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.5873015873015874, | |
| "grad_norm": 0.47050222754478455, | |
| "learning_rate": 9.763960957919413e-05, | |
| "loss": 0.4347, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.6031746031746033, | |
| "grad_norm": 0.42742452025413513, | |
| "learning_rate": 9.759158502719481e-05, | |
| "loss": 0.4208, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.619047619047619, | |
| "grad_norm": 0.48628243803977966, | |
| "learning_rate": 9.754308888097583e-05, | |
| "loss": 0.5814, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.6349206349206349, | |
| "grad_norm": 0.4874871075153351, | |
| "learning_rate": 9.749412162109485e-05, | |
| "loss": 0.5278, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.6507936507936507, | |
| "grad_norm": 0.5010098814964294, | |
| "learning_rate": 9.744468373277797e-05, | |
| "loss": 0.5341, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.4798610508441925, | |
| "learning_rate": 9.739477570591473e-05, | |
| "loss": 0.5088, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.6825396825396826, | |
| "grad_norm": 0.5140134692192078, | |
| "learning_rate": 9.734439803505345e-05, | |
| "loss": 0.5922, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.6984126984126984, | |
| "grad_norm": 0.49391329288482666, | |
| "learning_rate": 9.729355121939621e-05, | |
| "loss": 0.5445, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.7142857142857144, | |
| "grad_norm": 0.5012408494949341, | |
| "learning_rate": 9.724223576279395e-05, | |
| "loss": 0.5175, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.7301587301587302, | |
| "grad_norm": 0.5038516521453857, | |
| "learning_rate": 9.719045217374143e-05, | |
| "loss": 0.4399, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.746031746031746, | |
| "grad_norm": 0.49503833055496216, | |
| "learning_rate": 9.713820096537225e-05, | |
| "loss": 0.483, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.7619047619047619, | |
| "grad_norm": 0.5000967979431152, | |
| "learning_rate": 9.708548265545375e-05, | |
| "loss": 0.6131, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.504001796245575, | |
| "learning_rate": 9.703229776638185e-05, | |
| "loss": 0.5121, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.7936507936507935, | |
| "grad_norm": 0.5135077238082886, | |
| "learning_rate": 9.697864682517592e-05, | |
| "loss": 0.4606, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.8095238095238095, | |
| "grad_norm": 0.5064616799354553, | |
| "learning_rate": 9.692453036347351e-05, | |
| "loss": 0.4862, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.8253968253968254, | |
| "grad_norm": 0.5660854578018188, | |
| "learning_rate": 9.686994891752508e-05, | |
| "loss": 0.5925, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.8412698412698414, | |
| "grad_norm": 0.5516797304153442, | |
| "learning_rate": 9.681490302818874e-05, | |
| "loss": 0.5986, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.8571428571428572, | |
| "grad_norm": 0.5815762281417847, | |
| "learning_rate": 9.675939324092486e-05, | |
| "loss": 0.6187, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.873015873015873, | |
| "grad_norm": 0.5087511539459229, | |
| "learning_rate": 9.670342010579065e-05, | |
| "loss": 0.499, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.8888888888888888, | |
| "grad_norm": 0.45885273814201355, | |
| "learning_rate": 9.664698417743475e-05, | |
| "loss": 0.4405, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 0.537526547908783, | |
| "learning_rate": 9.659008601509168e-05, | |
| "loss": 0.5208, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.9206349206349205, | |
| "grad_norm": 0.4978830814361572, | |
| "learning_rate": 9.653272618257631e-05, | |
| "loss": 0.5475, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.9365079365079365, | |
| "grad_norm": 0.5565654635429382, | |
| "learning_rate": 9.647490524827834e-05, | |
| "loss": 0.5459, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.9523809523809523, | |
| "grad_norm": 0.5845757126808167, | |
| "learning_rate": 9.641662378515659e-05, | |
| "loss": 0.6169, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.9682539682539684, | |
| "grad_norm": 0.5273924469947815, | |
| "learning_rate": 9.635788237073334e-05, | |
| "loss": 0.519, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.9841269841269842, | |
| "grad_norm": 0.5515849590301514, | |
| "learning_rate": 9.629868158708861e-05, | |
| "loss": 0.52, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.7463253736495972, | |
| "learning_rate": 9.623902202085444e-05, | |
| "loss": 0.5024, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.015873015873016, | |
| "grad_norm": 0.5206636190414429, | |
| "learning_rate": 9.617890426320899e-05, | |
| "loss": 0.4819, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.0317460317460316, | |
| "grad_norm": 0.4978935122489929, | |
| "learning_rate": 9.611832890987076e-05, | |
| "loss": 0.4031, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.0476190476190474, | |
| "grad_norm": 0.5565934181213379, | |
| "learning_rate": 9.605729656109265e-05, | |
| "loss": 0.5879, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.0634920634920633, | |
| "grad_norm": 0.5003566741943359, | |
| "learning_rate": 9.599580782165598e-05, | |
| "loss": 0.3628, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.0793650793650795, | |
| "grad_norm": 0.4868488609790802, | |
| "learning_rate": 9.593386330086458e-05, | |
| "loss": 0.3807, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.0952380952380953, | |
| "grad_norm": 0.5097118616104126, | |
| "learning_rate": 9.587146361253868e-05, | |
| "loss": 0.4166, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.111111111111111, | |
| "grad_norm": 0.5274227857589722, | |
| "learning_rate": 9.580860937500884e-05, | |
| "loss": 0.385, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.126984126984127, | |
| "grad_norm": 0.5781636238098145, | |
| "learning_rate": 9.57453012111099e-05, | |
| "loss": 0.3981, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.142857142857143, | |
| "grad_norm": 0.6308386921882629, | |
| "learning_rate": 9.568153974817464e-05, | |
| "loss": 0.4357, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.1587301587301586, | |
| "grad_norm": 0.6387614011764526, | |
| "learning_rate": 9.561732561802778e-05, | |
| "loss": 0.4168, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.1746031746031744, | |
| "grad_norm": 0.6377487182617188, | |
| "learning_rate": 9.555265945697953e-05, | |
| "loss": 0.3831, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.1904761904761907, | |
| "grad_norm": 0.7271438241004944, | |
| "learning_rate": 9.548754190581939e-05, | |
| "loss": 0.3844, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.2063492063492065, | |
| "grad_norm": 0.8928720951080322, | |
| "learning_rate": 9.542197360980978e-05, | |
| "loss": 0.5863, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.8302777409553528, | |
| "learning_rate": 9.53559552186796e-05, | |
| "loss": 0.4477, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.238095238095238, | |
| "grad_norm": 0.7997470498085022, | |
| "learning_rate": 9.528948738661784e-05, | |
| "loss": 0.3644, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.253968253968254, | |
| "grad_norm": 0.8765047192573547, | |
| "learning_rate": 9.522257077226717e-05, | |
| "loss": 0.3806, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.2698412698412698, | |
| "grad_norm": 0.7953476309776306, | |
| "learning_rate": 9.51552060387172e-05, | |
| "loss": 0.3829, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 0.8067965507507324, | |
| "learning_rate": 9.508739385349812e-05, | |
| "loss": 0.4414, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.3015873015873014, | |
| "grad_norm": 0.7154417037963867, | |
| "learning_rate": 9.501913488857399e-05, | |
| "loss": 0.3377, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.317460317460317, | |
| "grad_norm": 0.8233152627944946, | |
| "learning_rate": 9.49504298203361e-05, | |
| "loss": 0.4463, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 0.8649589419364929, | |
| "learning_rate": 9.488127932959625e-05, | |
| "loss": 0.3966, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.3492063492063493, | |
| "grad_norm": 0.834513247013092, | |
| "learning_rate": 9.481168410158003e-05, | |
| "loss": 0.5009, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.365079365079365, | |
| "grad_norm": 0.7996335625648499, | |
| "learning_rate": 9.474164482592002e-05, | |
| "loss": 0.4546, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 0.9039611220359802, | |
| "learning_rate": 9.467116219664894e-05, | |
| "loss": 0.4492, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.3968253968253967, | |
| "grad_norm": 0.8271594643592834, | |
| "learning_rate": 9.460023691219277e-05, | |
| "loss": 0.3569, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.4126984126984126, | |
| "grad_norm": 0.9009270071983337, | |
| "learning_rate": 9.45288696753639e-05, | |
| "loss": 0.4727, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.4285714285714284, | |
| "grad_norm": 0.7487375736236572, | |
| "learning_rate": 9.445706119335407e-05, | |
| "loss": 0.3298, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.8869822025299072, | |
| "learning_rate": 9.438481217772744e-05, | |
| "loss": 0.4476, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.4603174603174605, | |
| "grad_norm": 0.8800178170204163, | |
| "learning_rate": 9.431212334441343e-05, | |
| "loss": 0.4377, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.4761904761904763, | |
| "grad_norm": 0.8610995411872864, | |
| "learning_rate": 9.423899541369978e-05, | |
| "loss": 0.409, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.492063492063492, | |
| "grad_norm": 0.8344472050666809, | |
| "learning_rate": 9.41654291102253e-05, | |
| "loss": 0.427, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.507936507936508, | |
| "grad_norm": 0.9956201314926147, | |
| "learning_rate": 9.409142516297269e-05, | |
| "loss": 0.5661, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.5238095238095237, | |
| "grad_norm": 0.8969646692276001, | |
| "learning_rate": 9.401698430526142e-05, | |
| "loss": 0.4215, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.5396825396825395, | |
| "grad_norm": 0.918438732624054, | |
| "learning_rate": 9.394210727474028e-05, | |
| "loss": 0.4774, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.5555555555555554, | |
| "grad_norm": 0.8604788780212402, | |
| "learning_rate": 9.386679481338033e-05, | |
| "loss": 0.3978, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.571428571428571, | |
| "grad_norm": 0.7847458124160767, | |
| "learning_rate": 9.379104766746722e-05, | |
| "loss": 0.3602, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.5873015873015874, | |
| "grad_norm": 0.8306839466094971, | |
| "learning_rate": 9.371486658759416e-05, | |
| "loss": 0.466, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.6031746031746033, | |
| "grad_norm": 0.8458616137504578, | |
| "learning_rate": 9.363825232865413e-05, | |
| "loss": 0.4077, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.619047619047619, | |
| "grad_norm": 0.933336615562439, | |
| "learning_rate": 9.356120564983266e-05, | |
| "loss": 0.4652, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.634920634920635, | |
| "grad_norm": 0.9182778596878052, | |
| "learning_rate": 9.348372731460023e-05, | |
| "loss": 0.3775, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.6507936507936507, | |
| "grad_norm": 0.9331458806991577, | |
| "learning_rate": 9.340581809070459e-05, | |
| "loss": 0.4362, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.8755380511283875, | |
| "learning_rate": 9.332747875016332e-05, | |
| "loss": 0.363, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.682539682539683, | |
| "grad_norm": 0.8975720405578613, | |
| "learning_rate": 9.324871006925613e-05, | |
| "loss": 0.4007, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.6984126984126986, | |
| "grad_norm": 1.1305972337722778, | |
| "learning_rate": 9.316951282851707e-05, | |
| "loss": 0.5013, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.7142857142857144, | |
| "grad_norm": 0.8970773220062256, | |
| "learning_rate": 9.308988781272694e-05, | |
| "loss": 0.4052, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.7301587301587302, | |
| "grad_norm": 1.0294140577316284, | |
| "learning_rate": 9.300983581090541e-05, | |
| "loss": 0.4707, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.746031746031746, | |
| "grad_norm": 0.9334731698036194, | |
| "learning_rate": 9.292935761630326e-05, | |
| "loss": 0.3639, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.761904761904762, | |
| "grad_norm": 0.9174486398696899, | |
| "learning_rate": 9.284845402639446e-05, | |
| "loss": 0.3959, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.9317827224731445, | |
| "learning_rate": 9.276712584286833e-05, | |
| "loss": 0.3916, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.7936507936507935, | |
| "grad_norm": 0.9498136639595032, | |
| "learning_rate": 9.26853738716216e-05, | |
| "loss": 0.4551, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.8095238095238093, | |
| "grad_norm": 0.8333742022514343, | |
| "learning_rate": 9.260319892275034e-05, | |
| "loss": 0.3518, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.825396825396825, | |
| "grad_norm": 0.8575045466423035, | |
| "learning_rate": 9.2520601810542e-05, | |
| "loss": 0.3623, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.8412698412698414, | |
| "grad_norm": 1.100193977355957, | |
| "learning_rate": 9.243758335346735e-05, | |
| "loss": 0.5737, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.9462725520133972, | |
| "learning_rate": 9.235414437417234e-05, | |
| "loss": 0.4491, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.873015873015873, | |
| "grad_norm": 0.8208152651786804, | |
| "learning_rate": 9.227028569946996e-05, | |
| "loss": 0.3799, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.888888888888889, | |
| "grad_norm": 0.8733758330345154, | |
| "learning_rate": 9.2186008160332e-05, | |
| "loss": 0.4313, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.9047619047619047, | |
| "grad_norm": 0.8397769927978516, | |
| "learning_rate": 9.210131259188095e-05, | |
| "loss": 0.3718, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.9206349206349205, | |
| "grad_norm": 1.0263302326202393, | |
| "learning_rate": 9.201619983338153e-05, | |
| "loss": 0.5163, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.9365079365079367, | |
| "grad_norm": 0.7651734948158264, | |
| "learning_rate": 9.193067072823251e-05, | |
| "loss": 0.3483, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.9523809523809526, | |
| "grad_norm": 0.92905592918396, | |
| "learning_rate": 9.18447261239584e-05, | |
| "loss": 0.5041, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.9682539682539684, | |
| "grad_norm": 0.8523809909820557, | |
| "learning_rate": 9.175836687220084e-05, | |
| "loss": 0.381, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.984126984126984, | |
| "grad_norm": 0.8607370257377625, | |
| "learning_rate": 9.167159382871039e-05, | |
| "loss": 0.3953, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 1.321708083152771, | |
| "learning_rate": 9.15844078533379e-05, | |
| "loss": 0.4583, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 3.015873015873016, | |
| "grad_norm": 0.7019425630569458, | |
| "learning_rate": 9.149680981002609e-05, | |
| "loss": 0.2773, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 3.0317460317460316, | |
| "grad_norm": 0.6896389126777649, | |
| "learning_rate": 9.140880056680088e-05, | |
| "loss": 0.2746, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 3.0476190476190474, | |
| "grad_norm": 0.779511570930481, | |
| "learning_rate": 9.13203809957629e-05, | |
| "loss": 0.3052, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 3.0634920634920633, | |
| "grad_norm": 0.8268155455589294, | |
| "learning_rate": 9.123155197307876e-05, | |
| "loss": 0.3045, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 3.0793650793650795, | |
| "grad_norm": 0.7496017813682556, | |
| "learning_rate": 9.114231437897244e-05, | |
| "loss": 0.2231, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 3.0952380952380953, | |
| "grad_norm": 0.8415669798851013, | |
| "learning_rate": 9.105266909771653e-05, | |
| "loss": 0.2298, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 3.111111111111111, | |
| "grad_norm": 1.05263090133667, | |
| "learning_rate": 9.096261701762342e-05, | |
| "loss": 0.2488, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 3.126984126984127, | |
| "grad_norm": 1.238415241241455, | |
| "learning_rate": 9.087215903103662e-05, | |
| "loss": 0.2806, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 3.142857142857143, | |
| "grad_norm": 1.1588196754455566, | |
| "learning_rate": 9.078129603432181e-05, | |
| "loss": 0.245, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 3.1587301587301586, | |
| "grad_norm": 1.584652304649353, | |
| "learning_rate": 9.069002892785797e-05, | |
| "loss": 0.295, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 3.1746031746031744, | |
| "grad_norm": 1.3894325494766235, | |
| "learning_rate": 9.059835861602853e-05, | |
| "loss": 0.2349, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.1904761904761907, | |
| "grad_norm": 1.66408109664917, | |
| "learning_rate": 9.050628600721234e-05, | |
| "loss": 0.2627, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 3.2063492063492065, | |
| "grad_norm": 1.2087987661361694, | |
| "learning_rate": 9.041381201377468e-05, | |
| "loss": 0.2159, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 3.2222222222222223, | |
| "grad_norm": 1.369932770729065, | |
| "learning_rate": 9.032093755205822e-05, | |
| "loss": 0.2341, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 3.238095238095238, | |
| "grad_norm": 1.6366993188858032, | |
| "learning_rate": 9.0227663542374e-05, | |
| "loss": 0.2893, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 3.253968253968254, | |
| "grad_norm": 1.529963731765747, | |
| "learning_rate": 9.013399090899217e-05, | |
| "loss": 0.2395, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 3.2698412698412698, | |
| "grad_norm": 1.7285979986190796, | |
| "learning_rate": 9.003992058013302e-05, | |
| "loss": 0.3451, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 3.2857142857142856, | |
| "grad_norm": 1.3240851163864136, | |
| "learning_rate": 8.99454534879576e-05, | |
| "loss": 0.2469, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 3.3015873015873014, | |
| "grad_norm": 1.3964006900787354, | |
| "learning_rate": 8.985059056855858e-05, | |
| "loss": 0.2456, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 3.317460317460317, | |
| "grad_norm": 1.405621886253357, | |
| "learning_rate": 8.975533276195102e-05, | |
| "loss": 0.2347, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 1.3338896036148071, | |
| "learning_rate": 8.965968101206291e-05, | |
| "loss": 0.2988, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.3492063492063493, | |
| "grad_norm": 1.329379677772522, | |
| "learning_rate": 8.956363626672595e-05, | |
| "loss": 0.2651, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 3.365079365079365, | |
| "grad_norm": 1.3324720859527588, | |
| "learning_rate": 8.94671994776661e-05, | |
| "loss": 0.2527, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 3.380952380952381, | |
| "grad_norm": 1.2702524662017822, | |
| "learning_rate": 8.937037160049416e-05, | |
| "loss": 0.2763, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 3.3968253968253967, | |
| "grad_norm": 1.270229458808899, | |
| "learning_rate": 8.927315359469626e-05, | |
| "loss": 0.236, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 3.4126984126984126, | |
| "grad_norm": 1.3164818286895752, | |
| "learning_rate": 8.917554642362443e-05, | |
| "loss": 0.2476, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 3.4285714285714284, | |
| "grad_norm": 1.2434004545211792, | |
| "learning_rate": 8.907755105448704e-05, | |
| "loss": 0.2387, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 3.4444444444444446, | |
| "grad_norm": 1.0932611227035522, | |
| "learning_rate": 8.89791684583391e-05, | |
| "loss": 0.2195, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 3.4603174603174605, | |
| "grad_norm": 1.334930181503296, | |
| "learning_rate": 8.888039961007282e-05, | |
| "loss": 0.2725, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 3.4761904761904763, | |
| "grad_norm": 1.1716219186782837, | |
| "learning_rate": 8.87812454884078e-05, | |
| "loss": 0.2515, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 3.492063492063492, | |
| "grad_norm": 1.1771153211593628, | |
| "learning_rate": 8.868170707588142e-05, | |
| "loss": 0.2286, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 3.507936507936508, | |
| "grad_norm": 1.2309902906417847, | |
| "learning_rate": 8.858178535883905e-05, | |
| "loss": 0.2365, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 3.5238095238095237, | |
| "grad_norm": 0.9976351261138916, | |
| "learning_rate": 8.848148132742431e-05, | |
| "loss": 0.22, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 3.5396825396825395, | |
| "grad_norm": 1.1791083812713623, | |
| "learning_rate": 8.838079597556925e-05, | |
| "loss": 0.2683, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 3.5555555555555554, | |
| "grad_norm": 1.1750749349594116, | |
| "learning_rate": 8.827973030098448e-05, | |
| "loss": 0.2396, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 3.571428571428571, | |
| "grad_norm": 1.054264783859253, | |
| "learning_rate": 8.81782853051493e-05, | |
| "loss": 0.2396, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 3.5873015873015874, | |
| "grad_norm": 1.1976933479309082, | |
| "learning_rate": 8.807646199330187e-05, | |
| "loss": 0.2393, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 3.6031746031746033, | |
| "grad_norm": 1.4662325382232666, | |
| "learning_rate": 8.797426137442897e-05, | |
| "loss": 0.3188, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 3.619047619047619, | |
| "grad_norm": 1.5771795511245728, | |
| "learning_rate": 8.787168446125638e-05, | |
| "loss": 0.3204, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 3.634920634920635, | |
| "grad_norm": 1.3994357585906982, | |
| "learning_rate": 8.776873227023852e-05, | |
| "loss": 0.3045, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 3.6507936507936507, | |
| "grad_norm": 1.9753646850585938, | |
| "learning_rate": 8.766540582154859e-05, | |
| "loss": 0.2306, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.6666666666666665, | |
| "grad_norm": 1.4474598169326782, | |
| "learning_rate": 8.756170613906833e-05, | |
| "loss": 0.2581, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 3.682539682539683, | |
| "grad_norm": 1.1273548603057861, | |
| "learning_rate": 8.745763425037797e-05, | |
| "loss": 0.2213, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 3.6984126984126986, | |
| "grad_norm": 1.0989768505096436, | |
| "learning_rate": 8.735319118674596e-05, | |
| "loss": 0.2063, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 3.7142857142857144, | |
| "grad_norm": 1.243393063545227, | |
| "learning_rate": 8.724837798311882e-05, | |
| "loss": 0.2539, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 3.7301587301587302, | |
| "grad_norm": 1.1233344078063965, | |
| "learning_rate": 8.714319567811088e-05, | |
| "loss": 0.2225, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 3.746031746031746, | |
| "grad_norm": 1.2728500366210938, | |
| "learning_rate": 8.703764531399392e-05, | |
| "loss": 0.246, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 3.761904761904762, | |
| "grad_norm": 1.2673249244689941, | |
| "learning_rate": 8.69317279366869e-05, | |
| "loss": 0.2881, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 3.7777777777777777, | |
| "grad_norm": 1.4421532154083252, | |
| "learning_rate": 8.682544459574562e-05, | |
| "loss": 0.3309, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 3.7936507936507935, | |
| "grad_norm": 1.217529296875, | |
| "learning_rate": 8.671879634435224e-05, | |
| "loss": 0.2815, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 3.8095238095238093, | |
| "grad_norm": 1.1456962823867798, | |
| "learning_rate": 8.661178423930491e-05, | |
| "loss": 0.2557, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.825396825396825, | |
| "grad_norm": 1.0717531442642212, | |
| "learning_rate": 8.650440934100728e-05, | |
| "loss": 0.2471, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 3.8412698412698414, | |
| "grad_norm": 1.217034935951233, | |
| "learning_rate": 8.6396672713458e-05, | |
| "loss": 0.2883, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 3.857142857142857, | |
| "grad_norm": 1.237244725227356, | |
| "learning_rate": 8.628857542424009e-05, | |
| "loss": 0.2953, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 3.873015873015873, | |
| "grad_norm": 1.2947179079055786, | |
| "learning_rate": 8.618011854451056e-05, | |
| "loss": 0.3134, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "grad_norm": 1.2005493640899658, | |
| "learning_rate": 8.607130314898956e-05, | |
| "loss": 0.2655, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 3.9047619047619047, | |
| "grad_norm": 1.387406826019287, | |
| "learning_rate": 8.596213031594991e-05, | |
| "loss": 0.3133, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 3.9206349206349205, | |
| "grad_norm": 1.297012209892273, | |
| "learning_rate": 8.585260112720631e-05, | |
| "loss": 0.2747, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 3.9365079365079367, | |
| "grad_norm": 1.12217378616333, | |
| "learning_rate": 8.57427166681047e-05, | |
| "loss": 0.2444, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 3.9523809523809526, | |
| "grad_norm": 1.2482068538665771, | |
| "learning_rate": 8.56324780275114e-05, | |
| "loss": 0.2887, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 3.9682539682539684, | |
| "grad_norm": 1.2814184427261353, | |
| "learning_rate": 8.552188629780244e-05, | |
| "loss": 0.284, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.984126984126984, | |
| "grad_norm": 1.1486774682998657, | |
| "learning_rate": 8.541094257485265e-05, | |
| "loss": 0.2636, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 1.6360046863555908, | |
| "learning_rate": 8.529964795802485e-05, | |
| "loss": 0.2305, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 4.015873015873016, | |
| "grad_norm": 0.7815824151039124, | |
| "learning_rate": 8.518800355015892e-05, | |
| "loss": 0.1427, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 4.031746031746032, | |
| "grad_norm": 0.9590736031532288, | |
| "learning_rate": 8.507601045756085e-05, | |
| "loss": 0.1609, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 4.0476190476190474, | |
| "grad_norm": 0.9721108078956604, | |
| "learning_rate": 8.49636697899919e-05, | |
| "loss": 0.1429, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 4.063492063492063, | |
| "grad_norm": 1.0513888597488403, | |
| "learning_rate": 8.485098266065744e-05, | |
| "loss": 0.1344, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 4.079365079365079, | |
| "grad_norm": 1.1911511421203613, | |
| "learning_rate": 8.473795018619604e-05, | |
| "loss": 0.135, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 4.095238095238095, | |
| "grad_norm": 1.052157998085022, | |
| "learning_rate": 8.462457348666835e-05, | |
| "loss": 0.1146, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 4.111111111111111, | |
| "grad_norm": 1.4159713983535767, | |
| "learning_rate": 8.4510853685546e-05, | |
| "loss": 0.1359, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 4.1269841269841265, | |
| "grad_norm": 1.6234732866287231, | |
| "learning_rate": 8.439679190970052e-05, | |
| "loss": 0.1634, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 4.142857142857143, | |
| "grad_norm": 1.2149155139923096, | |
| "learning_rate": 8.428238928939207e-05, | |
| "loss": 0.1051, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 4.158730158730159, | |
| "grad_norm": 1.527443528175354, | |
| "learning_rate": 8.416764695825834e-05, | |
| "loss": 0.1519, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 4.174603174603175, | |
| "grad_norm": 1.3665393590927124, | |
| "learning_rate": 8.405256605330331e-05, | |
| "loss": 0.1366, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 4.190476190476191, | |
| "grad_norm": 1.2650479078292847, | |
| "learning_rate": 8.39371477148859e-05, | |
| "loss": 0.1314, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 4.2063492063492065, | |
| "grad_norm": 0.9967718124389648, | |
| "learning_rate": 8.382139308670875e-05, | |
| "loss": 0.1173, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 4.222222222222222, | |
| "grad_norm": 1.1094558238983154, | |
| "learning_rate": 8.370530331580686e-05, | |
| "loss": 0.1126, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 4.238095238095238, | |
| "grad_norm": 1.0152033567428589, | |
| "learning_rate": 8.35888795525362e-05, | |
| "loss": 0.089, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 4.253968253968254, | |
| "grad_norm": 1.2841627597808838, | |
| "learning_rate": 8.347212295056239e-05, | |
| "loss": 0.1292, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 4.26984126984127, | |
| "grad_norm": 1.416364073753357, | |
| "learning_rate": 8.335503466684915e-05, | |
| "loss": 0.1444, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 4.285714285714286, | |
| "grad_norm": 1.2542331218719482, | |
| "learning_rate": 8.323761586164695e-05, | |
| "loss": 0.1313, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 4.301587301587301, | |
| "grad_norm": 1.3430452346801758, | |
| "learning_rate": 8.311986769848141e-05, | |
| "loss": 0.1405, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 4.317460317460317, | |
| "grad_norm": 1.3169519901275635, | |
| "learning_rate": 8.300179134414188e-05, | |
| "loss": 0.1429, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 4.333333333333333, | |
| "grad_norm": 1.2539156675338745, | |
| "learning_rate": 8.288338796866976e-05, | |
| "loss": 0.1382, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 4.349206349206349, | |
| "grad_norm": 1.365218997001648, | |
| "learning_rate": 8.276465874534702e-05, | |
| "loss": 0.1236, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 4.365079365079365, | |
| "grad_norm": 1.4856258630752563, | |
| "learning_rate": 8.264560485068446e-05, | |
| "loss": 0.1516, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 4.380952380952381, | |
| "grad_norm": 1.139467477798462, | |
| "learning_rate": 8.252622746441021e-05, | |
| "loss": 0.1187, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 4.396825396825397, | |
| "grad_norm": 1.1698997020721436, | |
| "learning_rate": 8.240652776945781e-05, | |
| "loss": 0.133, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 4.412698412698413, | |
| "grad_norm": 1.284920334815979, | |
| "learning_rate": 8.228650695195472e-05, | |
| "loss": 0.1564, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 4.428571428571429, | |
| "grad_norm": 1.2975406646728516, | |
| "learning_rate": 8.216616620121043e-05, | |
| "loss": 0.1476, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "grad_norm": 1.28453528881073, | |
| "learning_rate": 8.204550670970469e-05, | |
| "loss": 0.1444, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 4.4603174603174605, | |
| "grad_norm": 1.2703144550323486, | |
| "learning_rate": 8.192452967307576e-05, | |
| "loss": 0.1627, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 4.476190476190476, | |
| "grad_norm": 1.2940740585327148, | |
| "learning_rate": 8.180323629010848e-05, | |
| "loss": 0.1384, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 4.492063492063492, | |
| "grad_norm": 1.2578924894332886, | |
| "learning_rate": 8.168162776272244e-05, | |
| "loss": 0.1301, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 4.507936507936508, | |
| "grad_norm": 1.2214442491531372, | |
| "learning_rate": 8.155970529596006e-05, | |
| "loss": 0.139, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 4.523809523809524, | |
| "grad_norm": 1.436343789100647, | |
| "learning_rate": 8.143747009797464e-05, | |
| "loss": 0.1522, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 4.5396825396825395, | |
| "grad_norm": 1.179060459136963, | |
| "learning_rate": 8.131492338001839e-05, | |
| "loss": 0.1236, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 4.555555555555555, | |
| "grad_norm": 1.3683005571365356, | |
| "learning_rate": 8.119206635643045e-05, | |
| "loss": 0.1489, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 4.571428571428571, | |
| "grad_norm": 1.2832778692245483, | |
| "learning_rate": 8.106890024462481e-05, | |
| "loss": 0.1388, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 4.587301587301587, | |
| "grad_norm": 1.0831190347671509, | |
| "learning_rate": 8.094542626507828e-05, | |
| "loss": 0.1219, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 4.603174603174603, | |
| "grad_norm": 1.212108850479126, | |
| "learning_rate": 8.082164564131845e-05, | |
| "loss": 0.1331, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 4.619047619047619, | |
| "grad_norm": 1.157487154006958, | |
| "learning_rate": 8.069755959991142e-05, | |
| "loss": 0.1306, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 4.634920634920634, | |
| "grad_norm": 1.194389820098877, | |
| "learning_rate": 8.057316937044977e-05, | |
| "loss": 0.1361, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 4.650793650793651, | |
| "grad_norm": 1.2109564542770386, | |
| "learning_rate": 8.044847618554034e-05, | |
| "loss": 0.138, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 4.666666666666667, | |
| "grad_norm": 1.0707926750183105, | |
| "learning_rate": 8.032348128079203e-05, | |
| "loss": 0.1078, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 4.682539682539683, | |
| "grad_norm": 1.179071307182312, | |
| "learning_rate": 8.019818589480352e-05, | |
| "loss": 0.1397, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 4.698412698412699, | |
| "grad_norm": 1.3288228511810303, | |
| "learning_rate": 8.0072591269151e-05, | |
| "loss": 0.1613, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 4.714285714285714, | |
| "grad_norm": 1.24984872341156, | |
| "learning_rate": 7.994669864837594e-05, | |
| "loss": 0.1457, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 4.73015873015873, | |
| "grad_norm": 1.2009999752044678, | |
| "learning_rate": 7.982050927997264e-05, | |
| "loss": 0.1257, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 4.746031746031746, | |
| "grad_norm": 1.207233190536499, | |
| "learning_rate": 7.969402441437594e-05, | |
| "loss": 0.1567, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 4.761904761904762, | |
| "grad_norm": 1.1672086715698242, | |
| "learning_rate": 7.956724530494887e-05, | |
| "loss": 0.1274, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.777777777777778, | |
| "grad_norm": 1.506867527961731, | |
| "learning_rate": 7.944017320797013e-05, | |
| "loss": 0.139, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 4.7936507936507935, | |
| "grad_norm": 1.4278178215026855, | |
| "learning_rate": 7.931280938262169e-05, | |
| "loss": 0.1357, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 4.809523809523809, | |
| "grad_norm": 1.599716067314148, | |
| "learning_rate": 7.918515509097634e-05, | |
| "loss": 0.1704, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 4.825396825396825, | |
| "grad_norm": 1.3049015998840332, | |
| "learning_rate": 7.905721159798513e-05, | |
| "loss": 0.1379, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 4.841269841269841, | |
| "grad_norm": 1.3524868488311768, | |
| "learning_rate": 7.89289801714649e-05, | |
| "loss": 0.1545, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 4.857142857142857, | |
| "grad_norm": 1.2142527103424072, | |
| "learning_rate": 7.880046208208563e-05, | |
| "loss": 0.1453, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 4.8730158730158735, | |
| "grad_norm": 1.1084891557693481, | |
| "learning_rate": 7.867165860335792e-05, | |
| "loss": 0.1427, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 4.888888888888889, | |
| "grad_norm": 1.266802191734314, | |
| "learning_rate": 7.854257101162037e-05, | |
| "loss": 0.1396, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 4.904761904761905, | |
| "grad_norm": 1.1826775074005127, | |
| "learning_rate": 7.841320058602688e-05, | |
| "loss": 0.1514, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 4.920634920634921, | |
| "grad_norm": 1.4232659339904785, | |
| "learning_rate": 7.828354860853399e-05, | |
| "loss": 0.1472, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 4.936507936507937, | |
| "grad_norm": 1.1436830759048462, | |
| "learning_rate": 7.815361636388827e-05, | |
| "loss": 0.1249, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 4.9523809523809526, | |
| "grad_norm": 1.3001309633255005, | |
| "learning_rate": 7.802340513961342e-05, | |
| "loss": 0.1663, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 4.968253968253968, | |
| "grad_norm": 1.4213690757751465, | |
| "learning_rate": 7.789291622599767e-05, | |
| "loss": 0.1538, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 4.984126984126984, | |
| "grad_norm": 1.5043220520019531, | |
| "learning_rate": 7.776215091608085e-05, | |
| "loss": 0.151, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.6825261116027832, | |
| "learning_rate": 7.763111050564178e-05, | |
| "loss": 0.1485, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 5.015873015873016, | |
| "grad_norm": 0.8601322174072266, | |
| "learning_rate": 7.749979629318516e-05, | |
| "loss": 0.0703, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 5.031746031746032, | |
| "grad_norm": 0.7637147903442383, | |
| "learning_rate": 7.736820957992895e-05, | |
| "loss": 0.0633, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 5.0476190476190474, | |
| "grad_norm": 0.8896054625511169, | |
| "learning_rate": 7.723635166979133e-05, | |
| "loss": 0.0652, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 5.063492063492063, | |
| "grad_norm": 0.9216472506523132, | |
| "learning_rate": 7.710422386937784e-05, | |
| "loss": 0.0585, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 5.079365079365079, | |
| "grad_norm": 0.7166661024093628, | |
| "learning_rate": 7.697182748796841e-05, | |
| "loss": 0.0531, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 5.095238095238095, | |
| "grad_norm": 0.9962891936302185, | |
| "learning_rate": 7.683916383750436e-05, | |
| "loss": 0.072, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 5.111111111111111, | |
| "grad_norm": 0.7969011068344116, | |
| "learning_rate": 7.670623423257548e-05, | |
| "loss": 0.0554, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 5.1269841269841265, | |
| "grad_norm": 0.8427059650421143, | |
| "learning_rate": 7.657303999040693e-05, | |
| "loss": 0.0534, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 5.142857142857143, | |
| "grad_norm": 0.9813700914382935, | |
| "learning_rate": 7.64395824308462e-05, | |
| "loss": 0.0696, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 5.158730158730159, | |
| "grad_norm": 0.8625731468200684, | |
| "learning_rate": 7.630586287635008e-05, | |
| "loss": 0.0562, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 5.174603174603175, | |
| "grad_norm": 0.9820646047592163, | |
| "learning_rate": 7.617188265197148e-05, | |
| "loss": 0.063, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 5.190476190476191, | |
| "grad_norm": 1.0742745399475098, | |
| "learning_rate": 7.603764308534636e-05, | |
| "loss": 0.0689, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 5.2063492063492065, | |
| "grad_norm": 0.9532903432846069, | |
| "learning_rate": 7.590314550668054e-05, | |
| "loss": 0.0667, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 5.222222222222222, | |
| "grad_norm": 0.8958349227905273, | |
| "learning_rate": 7.576839124873653e-05, | |
| "loss": 0.0538, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 5.238095238095238, | |
| "grad_norm": 0.9804512858390808, | |
| "learning_rate": 7.563338164682036e-05, | |
| "loss": 0.0689, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 5.253968253968254, | |
| "grad_norm": 1.0487632751464844, | |
| "learning_rate": 7.549811803876825e-05, | |
| "loss": 0.0671, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 5.26984126984127, | |
| "grad_norm": 0.9195834994316101, | |
| "learning_rate": 7.536260176493348e-05, | |
| "loss": 0.0669, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 5.285714285714286, | |
| "grad_norm": 0.9964186549186707, | |
| "learning_rate": 7.5226834168173e-05, | |
| "loss": 0.0688, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 5.301587301587301, | |
| "grad_norm": 0.8904131054878235, | |
| "learning_rate": 7.509081659383417e-05, | |
| "loss": 0.0636, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 5.317460317460317, | |
| "grad_norm": 0.965900182723999, | |
| "learning_rate": 7.495455038974146e-05, | |
| "loss": 0.0769, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 5.333333333333333, | |
| "grad_norm": 0.9203529357910156, | |
| "learning_rate": 7.481803690618303e-05, | |
| "loss": 0.0554, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 5.349206349206349, | |
| "grad_norm": 1.2281473875045776, | |
| "learning_rate": 7.46812774958974e-05, | |
| "loss": 0.0735, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 5.365079365079365, | |
| "grad_norm": 1.0948208570480347, | |
| "learning_rate": 7.454427351405999e-05, | |
| "loss": 0.0705, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 5.380952380952381, | |
| "grad_norm": 1.0401225090026855, | |
| "learning_rate": 7.440702631826977e-05, | |
| "loss": 0.07, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 5.396825396825397, | |
| "grad_norm": 0.9042516350746155, | |
| "learning_rate": 7.426953726853574e-05, | |
| "loss": 0.0628, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 5.412698412698413, | |
| "grad_norm": 0.9594908356666565, | |
| "learning_rate": 7.413180772726348e-05, | |
| "loss": 0.0606, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 5.428571428571429, | |
| "grad_norm": 1.0593825578689575, | |
| "learning_rate": 7.399383905924165e-05, | |
| "loss": 0.0652, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 5.444444444444445, | |
| "grad_norm": 1.0469237565994263, | |
| "learning_rate": 7.385563263162847e-05, | |
| "loss": 0.0636, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 5.4603174603174605, | |
| "grad_norm": 0.9159653782844543, | |
| "learning_rate": 7.371718981393815e-05, | |
| "loss": 0.0566, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 5.476190476190476, | |
| "grad_norm": 0.9596768021583557, | |
| "learning_rate": 7.357851197802735e-05, | |
| "loss": 0.0659, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 5.492063492063492, | |
| "grad_norm": 0.8929640054702759, | |
| "learning_rate": 7.343960049808156e-05, | |
| "loss": 0.0586, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 5.507936507936508, | |
| "grad_norm": 0.859683632850647, | |
| "learning_rate": 7.330045675060149e-05, | |
| "loss": 0.0522, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 5.523809523809524, | |
| "grad_norm": 1.026452898979187, | |
| "learning_rate": 7.316108211438945e-05, | |
| "loss": 0.0679, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 5.5396825396825395, | |
| "grad_norm": 0.9891062378883362, | |
| "learning_rate": 7.302147797053569e-05, | |
| "loss": 0.072, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 0.9392737150192261, | |
| "learning_rate": 7.288164570240463e-05, | |
| "loss": 0.062, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 5.571428571428571, | |
| "grad_norm": 1.1346358060836792, | |
| "learning_rate": 7.274158669562126e-05, | |
| "loss": 0.0666, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 5.587301587301587, | |
| "grad_norm": 0.8670554757118225, | |
| "learning_rate": 7.26013023380574e-05, | |
| "loss": 0.0572, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 5.603174603174603, | |
| "grad_norm": 1.020330786705017, | |
| "learning_rate": 7.246079401981784e-05, | |
| "loss": 0.0617, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 5.619047619047619, | |
| "grad_norm": 1.0136491060256958, | |
| "learning_rate": 7.232006313322667e-05, | |
| "loss": 0.0853, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 5.634920634920634, | |
| "grad_norm": 1.010423183441162, | |
| "learning_rate": 7.217911107281352e-05, | |
| "loss": 0.0705, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 5.650793650793651, | |
| "grad_norm": 0.9768037796020508, | |
| "learning_rate": 7.203793923529956e-05, | |
| "loss": 0.0853, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 5.666666666666667, | |
| "grad_norm": 0.9990655183792114, | |
| "learning_rate": 7.189654901958385e-05, | |
| "loss": 0.0715, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 5.682539682539683, | |
| "grad_norm": 1.0247498750686646, | |
| "learning_rate": 7.175494182672939e-05, | |
| "loss": 0.0712, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 5.698412698412699, | |
| "grad_norm": 1.0099873542785645, | |
| "learning_rate": 7.161311905994922e-05, | |
| "loss": 0.0712, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 5.714285714285714, | |
| "grad_norm": 1.0355095863342285, | |
| "learning_rate": 7.147108212459257e-05, | |
| "loss": 0.0722, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 5.73015873015873, | |
| "grad_norm": 1.1409605741500854, | |
| "learning_rate": 7.13288324281309e-05, | |
| "loss": 0.0688, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 5.746031746031746, | |
| "grad_norm": 1.1082065105438232, | |
| "learning_rate": 7.118637138014396e-05, | |
| "loss": 0.0781, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 5.761904761904762, | |
| "grad_norm": 1.1074239015579224, | |
| "learning_rate": 7.104370039230583e-05, | |
| "loss": 0.0705, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 5.777777777777778, | |
| "grad_norm": 0.9265062212944031, | |
| "learning_rate": 7.090082087837091e-05, | |
| "loss": 0.0593, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 5.7936507936507935, | |
| "grad_norm": 0.911005437374115, | |
| "learning_rate": 7.075773425415994e-05, | |
| "loss": 0.0678, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 5.809523809523809, | |
| "grad_norm": 1.0349949598312378, | |
| "learning_rate": 7.061444193754596e-05, | |
| "loss": 0.078, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 5.825396825396825, | |
| "grad_norm": 1.0515737533569336, | |
| "learning_rate": 7.047094534844023e-05, | |
| "loss": 0.0666, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 5.841269841269841, | |
| "grad_norm": 1.179187297821045, | |
| "learning_rate": 7.032724590877821e-05, | |
| "loss": 0.0774, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 5.857142857142857, | |
| "grad_norm": 1.1190379858016968, | |
| "learning_rate": 7.018334504250545e-05, | |
| "loss": 0.0891, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 5.8730158730158735, | |
| "grad_norm": 0.9958922863006592, | |
| "learning_rate": 7.003924417556343e-05, | |
| "loss": 0.0711, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 5.888888888888889, | |
| "grad_norm": 1.053802728652954, | |
| "learning_rate": 6.989494473587554e-05, | |
| "loss": 0.0759, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 5.904761904761905, | |
| "grad_norm": 0.9447202682495117, | |
| "learning_rate": 6.975044815333282e-05, | |
| "loss": 0.0713, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 5.920634920634921, | |
| "grad_norm": 0.9191451668739319, | |
| "learning_rate": 6.960575585977984e-05, | |
| "loss": 0.0655, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 5.936507936507937, | |
| "grad_norm": 1.1037213802337646, | |
| "learning_rate": 6.946086928900054e-05, | |
| "loss": 0.0831, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 5.9523809523809526, | |
| "grad_norm": 0.9468006491661072, | |
| "learning_rate": 6.931578987670396e-05, | |
| "loss": 0.059, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 5.968253968253968, | |
| "grad_norm": 1.1110552549362183, | |
| "learning_rate": 6.917051906051006e-05, | |
| "loss": 0.0709, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 5.984126984126984, | |
| "grad_norm": 1.1933718919754028, | |
| "learning_rate": 6.902505827993541e-05, | |
| "loss": 0.1004, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 1.4565590620040894, | |
| "learning_rate": 6.887940897637908e-05, | |
| "loss": 0.0915, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 6.015873015873016, | |
| "grad_norm": 0.6238571405410767, | |
| "learning_rate": 6.873357259310815e-05, | |
| "loss": 0.0431, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 6.031746031746032, | |
| "grad_norm": 0.4840649366378784, | |
| "learning_rate": 6.858755057524354e-05, | |
| "loss": 0.0358, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 6.0476190476190474, | |
| "grad_norm": 0.48597481846809387, | |
| "learning_rate": 6.844134436974567e-05, | |
| "loss": 0.0222, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 6.063492063492063, | |
| "grad_norm": 0.6410611867904663, | |
| "learning_rate": 6.829495542540013e-05, | |
| "loss": 0.0404, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 6.079365079365079, | |
| "grad_norm": 0.5220045447349548, | |
| "learning_rate": 6.814838519280324e-05, | |
| "loss": 0.0303, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 6.095238095238095, | |
| "grad_norm": 0.6196178793907166, | |
| "learning_rate": 6.80016351243478e-05, | |
| "loss": 0.0391, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 6.111111111111111, | |
| "grad_norm": 0.64337158203125, | |
| "learning_rate": 6.785470667420862e-05, | |
| "loss": 0.0338, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 6.1269841269841265, | |
| "grad_norm": 0.8072399497032166, | |
| "learning_rate": 6.77076012983281e-05, | |
| "loss": 0.0413, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 6.142857142857143, | |
| "grad_norm": 0.6252787709236145, | |
| "learning_rate": 6.75603204544019e-05, | |
| "loss": 0.0332, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 6.158730158730159, | |
| "grad_norm": 0.7571528553962708, | |
| "learning_rate": 6.741286560186437e-05, | |
| "loss": 0.0375, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 6.174603174603175, | |
| "grad_norm": 0.5972614884376526, | |
| "learning_rate": 6.726523820187413e-05, | |
| "loss": 0.0333, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 6.190476190476191, | |
| "grad_norm": 0.6365858316421509, | |
| "learning_rate": 6.711743971729967e-05, | |
| "loss": 0.0264, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 6.2063492063492065, | |
| "grad_norm": 0.7397788763046265, | |
| "learning_rate": 6.696947161270476e-05, | |
| "loss": 0.0319, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 6.222222222222222, | |
| "grad_norm": 0.6979987025260925, | |
| "learning_rate": 6.682133535433393e-05, | |
| "loss": 0.0415, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 6.238095238095238, | |
| "grad_norm": 0.6048802733421326, | |
| "learning_rate": 6.667303241009803e-05, | |
| "loss": 0.031, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 6.253968253968254, | |
| "grad_norm": 0.7918148040771484, | |
| "learning_rate": 6.652456424955963e-05, | |
| "loss": 0.0342, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 6.26984126984127, | |
| "grad_norm": 0.5297304391860962, | |
| "learning_rate": 6.637593234391843e-05, | |
| "loss": 0.0283, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 6.285714285714286, | |
| "grad_norm": 0.6882847547531128, | |
| "learning_rate": 6.622713816599673e-05, | |
| "loss": 0.0327, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 6.301587301587301, | |
| "grad_norm": 0.5969606637954712, | |
| "learning_rate": 6.60781831902248e-05, | |
| "loss": 0.0344, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 6.317460317460317, | |
| "grad_norm": 0.5623995065689087, | |
| "learning_rate": 6.592906889262632e-05, | |
| "loss": 0.0292, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 6.333333333333333, | |
| "grad_norm": 0.7312327027320862, | |
| "learning_rate": 6.577979675080369e-05, | |
| "loss": 0.0358, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 6.349206349206349, | |
| "grad_norm": 0.5290599465370178, | |
| "learning_rate": 6.563036824392344e-05, | |
| "loss": 0.0265, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 6.365079365079365, | |
| "grad_norm": 0.604269802570343, | |
| "learning_rate": 6.548078485270152e-05, | |
| "loss": 0.0311, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 6.380952380952381, | |
| "grad_norm": 0.6508985161781311, | |
| "learning_rate": 6.533104805938873e-05, | |
| "loss": 0.0325, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 6.396825396825397, | |
| "grad_norm": 0.7835598587989807, | |
| "learning_rate": 6.518115934775585e-05, | |
| "loss": 0.0311, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 6.412698412698413, | |
| "grad_norm": 0.6879574656486511, | |
| "learning_rate": 6.503112020307916e-05, | |
| "loss": 0.039, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 6.428571428571429, | |
| "grad_norm": 0.8170531392097473, | |
| "learning_rate": 6.488093211212555e-05, | |
| "loss": 0.0476, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 6.444444444444445, | |
| "grad_norm": 0.635261058807373, | |
| "learning_rate": 6.473059656313782e-05, | |
| "loss": 0.0315, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 6.4603174603174605, | |
| "grad_norm": 0.6152068972587585, | |
| "learning_rate": 6.458011504582005e-05, | |
| "loss": 0.0303, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 6.476190476190476, | |
| "grad_norm": 0.6500536799430847, | |
| "learning_rate": 6.442948905132266e-05, | |
| "loss": 0.0227, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 6.492063492063492, | |
| "grad_norm": 0.792615532875061, | |
| "learning_rate": 6.427872007222777e-05, | |
| "loss": 0.0254, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 6.507936507936508, | |
| "grad_norm": 0.7331106066703796, | |
| "learning_rate": 6.412780960253436e-05, | |
| "loss": 0.0307, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 6.523809523809524, | |
| "grad_norm": 0.7086438536643982, | |
| "learning_rate": 6.397675913764347e-05, | |
| "loss": 0.0275, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 6.5396825396825395, | |
| "grad_norm": 0.8358487486839294, | |
| "learning_rate": 6.382557017434332e-05, | |
| "loss": 0.0466, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 6.555555555555555, | |
| "grad_norm": 0.6510606408119202, | |
| "learning_rate": 6.367424421079463e-05, | |
| "loss": 0.037, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 6.571428571428571, | |
| "grad_norm": 0.8983582854270935, | |
| "learning_rate": 6.352278274651561e-05, | |
| "loss": 0.0379, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 6.587301587301587, | |
| "grad_norm": 0.7613969445228577, | |
| "learning_rate": 6.337118728236721e-05, | |
| "loss": 0.0358, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 6.603174603174603, | |
| "grad_norm": 0.8371831774711609, | |
| "learning_rate": 6.321945932053822e-05, | |
| "loss": 0.046, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 6.619047619047619, | |
| "grad_norm": 0.7133164405822754, | |
| "learning_rate": 6.306760036453035e-05, | |
| "loss": 0.0276, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 6.634920634920634, | |
| "grad_norm": 0.6740472316741943, | |
| "learning_rate": 6.291561191914333e-05, | |
| "loss": 0.0383, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 6.650793650793651, | |
| "grad_norm": 0.6885079741477966, | |
| "learning_rate": 6.276349549046007e-05, | |
| "loss": 0.0368, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 0.8201141953468323, | |
| "learning_rate": 6.261125258583171e-05, | |
| "loss": 0.0487, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 6.682539682539683, | |
| "grad_norm": 0.6679426431655884, | |
| "learning_rate": 6.245888471386263e-05, | |
| "loss": 0.0318, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 6.698412698412699, | |
| "grad_norm": 0.8221629858016968, | |
| "learning_rate": 6.230639338439549e-05, | |
| "loss": 0.0392, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 6.714285714285714, | |
| "grad_norm": 0.7618691921234131, | |
| "learning_rate": 6.215378010849641e-05, | |
| "loss": 0.0373, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 6.73015873015873, | |
| "grad_norm": 0.7761756181716919, | |
| "learning_rate": 6.200104639843985e-05, | |
| "loss": 0.0366, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 6.746031746031746, | |
| "grad_norm": 0.8383869528770447, | |
| "learning_rate": 6.184819376769364e-05, | |
| "loss": 0.0375, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 6.761904761904762, | |
| "grad_norm": 0.78884357213974, | |
| "learning_rate": 6.169522373090412e-05, | |
| "loss": 0.0487, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 6.777777777777778, | |
| "grad_norm": 0.7803629040718079, | |
| "learning_rate": 6.154213780388092e-05, | |
| "loss": 0.0373, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 6.7936507936507935, | |
| "grad_norm": 0.5684940218925476, | |
| "learning_rate": 6.138893750358212e-05, | |
| "loss": 0.0297, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 6.809523809523809, | |
| "grad_norm": 0.7369560599327087, | |
| "learning_rate": 6.123562434809912e-05, | |
| "loss": 0.0372, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 6.825396825396825, | |
| "grad_norm": 0.47202688455581665, | |
| "learning_rate": 6.108219985664161e-05, | |
| "loss": 0.0243, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 6.841269841269841, | |
| "grad_norm": 0.6708411574363708, | |
| "learning_rate": 6.0928665549522554e-05, | |
| "loss": 0.0348, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 6.857142857142857, | |
| "grad_norm": 0.8175257444381714, | |
| "learning_rate": 6.0775022948143115e-05, | |
| "loss": 0.05, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 6.8730158730158735, | |
| "grad_norm": 0.7456179261207581, | |
| "learning_rate": 6.06212735749775e-05, | |
| "loss": 0.0356, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 6.888888888888889, | |
| "grad_norm": 0.615135908126831, | |
| "learning_rate": 6.046741895355802e-05, | |
| "loss": 0.0292, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 6.904761904761905, | |
| "grad_norm": 0.6926703453063965, | |
| "learning_rate": 6.031346060845986e-05, | |
| "loss": 0.035, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 6.920634920634921, | |
| "grad_norm": 0.9521751403808594, | |
| "learning_rate": 6.015940006528602e-05, | |
| "loss": 0.0478, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 6.936507936507937, | |
| "grad_norm": 0.6635673642158508, | |
| "learning_rate": 6.0005238850652234e-05, | |
| "loss": 0.0405, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 6.9523809523809526, | |
| "grad_norm": 0.6299306154251099, | |
| "learning_rate": 5.9850978492171794e-05, | |
| "loss": 0.0328, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 6.968253968253968, | |
| "grad_norm": 0.7513844966888428, | |
| "learning_rate": 5.96966205184404e-05, | |
| "loss": 0.0335, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 6.984126984126984, | |
| "grad_norm": 0.9874755144119263, | |
| "learning_rate": 5.954216645902109e-05, | |
| "loss": 0.0416, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 0.8250815272331238, | |
| "learning_rate": 5.9387617844429e-05, | |
| "loss": 0.0368, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 7.015873015873016, | |
| "grad_norm": 0.4338611364364624, | |
| "learning_rate": 5.923297620611623e-05, | |
| "loss": 0.0189, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 7.031746031746032, | |
| "grad_norm": 0.5719791054725647, | |
| "learning_rate": 5.907824307645669e-05, | |
| "loss": 0.0169, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 7.0476190476190474, | |
| "grad_norm": 0.38255706429481506, | |
| "learning_rate": 5.892341998873089e-05, | |
| "loss": 0.0186, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 7.063492063492063, | |
| "grad_norm": 0.3592822253704071, | |
| "learning_rate": 5.876850847711073e-05, | |
| "loss": 0.0166, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 7.079365079365079, | |
| "grad_norm": 0.6182012557983398, | |
| "learning_rate": 5.861351007664434e-05, | |
| "loss": 0.0236, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 7.095238095238095, | |
| "grad_norm": 0.5176107883453369, | |
| "learning_rate": 5.845842632324088e-05, | |
| "loss": 0.0253, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 7.111111111111111, | |
| "grad_norm": 0.4049137830734253, | |
| "learning_rate": 5.83032587536552e-05, | |
| "loss": 0.0221, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 7.1269841269841265, | |
| "grad_norm": 0.4034527540206909, | |
| "learning_rate": 5.814800890547278e-05, | |
| "loss": 0.0182, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 7.142857142857143, | |
| "grad_norm": 0.4478590488433838, | |
| "learning_rate": 5.799267831709442e-05, | |
| "loss": 0.0208, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 7.158730158730159, | |
| "grad_norm": 0.4524051547050476, | |
| "learning_rate": 5.78372685277209e-05, | |
| "loss": 0.0147, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 7.174603174603175, | |
| "grad_norm": 0.4985044300556183, | |
| "learning_rate": 5.7681781077337905e-05, | |
| "loss": 0.0198, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 7.190476190476191, | |
| "grad_norm": 0.4616793692111969, | |
| "learning_rate": 5.752621750670068e-05, | |
| "loss": 0.0171, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 7.2063492063492065, | |
| "grad_norm": 0.4235040247440338, | |
| "learning_rate": 5.737057935731868e-05, | |
| "loss": 0.0159, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 7.222222222222222, | |
| "grad_norm": 0.42039763927459717, | |
| "learning_rate": 5.721486817144044e-05, | |
| "loss": 0.0168, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 7.238095238095238, | |
| "grad_norm": 0.40982750058174133, | |
| "learning_rate": 5.705908549203823e-05, | |
| "loss": 0.0153, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 7.253968253968254, | |
| "grad_norm": 0.44600027799606323, | |
| "learning_rate": 5.690323286279274e-05, | |
| "loss": 0.0167, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 7.26984126984127, | |
| "grad_norm": 0.5298761129379272, | |
| "learning_rate": 5.674731182807781e-05, | |
| "loss": 0.0158, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 7.285714285714286, | |
| "grad_norm": 0.3657887279987335, | |
| "learning_rate": 5.659132393294514e-05, | |
| "loss": 0.0188, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 7.301587301587301, | |
| "grad_norm": 0.4426786005496979, | |
| "learning_rate": 5.643527072310891e-05, | |
| "loss": 0.0197, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 7.317460317460317, | |
| "grad_norm": 0.5749462842941284, | |
| "learning_rate": 5.627915374493061e-05, | |
| "loss": 0.0181, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 7.333333333333333, | |
| "grad_norm": 0.5059666633605957, | |
| "learning_rate": 5.612297454540352e-05, | |
| "loss": 0.0206, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 7.349206349206349, | |
| "grad_norm": 0.5599040389060974, | |
| "learning_rate": 5.596673467213756e-05, | |
| "loss": 0.0148, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 7.365079365079365, | |
| "grad_norm": 0.5010665059089661, | |
| "learning_rate": 5.581043567334383e-05, | |
| "loss": 0.0186, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 7.380952380952381, | |
| "grad_norm": 0.49025240540504456, | |
| "learning_rate": 5.5654079097819345e-05, | |
| "loss": 0.0237, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 7.396825396825397, | |
| "grad_norm": 0.4369467794895172, | |
| "learning_rate": 5.5497666494931654e-05, | |
| "loss": 0.017, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 7.412698412698413, | |
| "grad_norm": 0.4754543602466583, | |
| "learning_rate": 5.5341199414603493e-05, | |
| "loss": 0.0202, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 7.428571428571429, | |
| "grad_norm": 0.4779890179634094, | |
| "learning_rate": 5.518467940729739e-05, | |
| "loss": 0.0221, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 7.444444444444445, | |
| "grad_norm": 0.5082346796989441, | |
| "learning_rate": 5.502810802400039e-05, | |
| "loss": 0.0191, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 7.4603174603174605, | |
| "grad_norm": 0.4045872688293457, | |
| "learning_rate": 5.487148681620862e-05, | |
| "loss": 0.0181, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 7.476190476190476, | |
| "grad_norm": 0.306020587682724, | |
| "learning_rate": 5.4714817335911894e-05, | |
| "loss": 0.011, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 7.492063492063492, | |
| "grad_norm": 0.4682234823703766, | |
| "learning_rate": 5.455810113557839e-05, | |
| "loss": 0.0126, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 7.507936507936508, | |
| "grad_norm": 0.46444806456565857, | |
| "learning_rate": 5.440133976813926e-05, | |
| "loss": 0.0205, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 7.523809523809524, | |
| "grad_norm": 1.0911283493041992, | |
| "learning_rate": 5.4244534786973214e-05, | |
| "loss": 0.0209, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 7.5396825396825395, | |
| "grad_norm": 0.4805389642715454, | |
| "learning_rate": 5.40876877458911e-05, | |
| "loss": 0.0186, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 7.555555555555555, | |
| "grad_norm": 0.5102893114089966, | |
| "learning_rate": 5.3930800199120616e-05, | |
| "loss": 0.02, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 7.571428571428571, | |
| "grad_norm": 0.44652751088142395, | |
| "learning_rate": 5.377387370129079e-05, | |
| "loss": 0.0176, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 7.587301587301587, | |
| "grad_norm": 0.5319653153419495, | |
| "learning_rate": 5.361690980741663e-05, | |
| "loss": 0.0276, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 7.603174603174603, | |
| "grad_norm": 0.42663267254829407, | |
| "learning_rate": 5.345991007288371e-05, | |
| "loss": 0.0165, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 7.619047619047619, | |
| "grad_norm": 0.5141676068305969, | |
| "learning_rate": 5.330287605343279e-05, | |
| "loss": 0.0206, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 7.634920634920634, | |
| "grad_norm": 0.37202200293540955, | |
| "learning_rate": 5.314580930514431e-05, | |
| "loss": 0.014, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 7.650793650793651, | |
| "grad_norm": 0.5131287574768066, | |
| "learning_rate": 5.298871138442307e-05, | |
| "loss": 0.018, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 7.666666666666667, | |
| "grad_norm": 0.5241144895553589, | |
| "learning_rate": 5.283158384798275e-05, | |
| "loss": 0.0174, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 7.682539682539683, | |
| "grad_norm": 0.4443790316581726, | |
| "learning_rate": 5.267442825283048e-05, | |
| "loss": 0.0194, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 7.698412698412699, | |
| "grad_norm": 0.46092358231544495, | |
| "learning_rate": 5.2517246156251455e-05, | |
| "loss": 0.0138, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 7.714285714285714, | |
| "grad_norm": 0.5907039046287537, | |
| "learning_rate": 5.236003911579345e-05, | |
| "loss": 0.028, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 7.73015873015873, | |
| "grad_norm": 0.5472407341003418, | |
| "learning_rate": 5.220280868925145e-05, | |
| "loss": 0.0201, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 7.746031746031746, | |
| "grad_norm": 0.522294282913208, | |
| "learning_rate": 5.204555643465215e-05, | |
| "loss": 0.021, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 7.761904761904762, | |
| "grad_norm": 0.5975657105445862, | |
| "learning_rate": 5.1888283910238555e-05, | |
| "loss": 0.0198, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 7.777777777777778, | |
| "grad_norm": 0.6385313868522644, | |
| "learning_rate": 5.173099267445451e-05, | |
| "loss": 0.0222, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 7.7936507936507935, | |
| "grad_norm": 0.5334087014198303, | |
| "learning_rate": 5.157368428592933e-05, | |
| "loss": 0.0183, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 7.809523809523809, | |
| "grad_norm": 0.6203488111495972, | |
| "learning_rate": 5.1416360303462206e-05, | |
| "loss": 0.0329, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 7.825396825396825, | |
| "grad_norm": 0.5505366325378418, | |
| "learning_rate": 5.125902228600693e-05, | |
| "loss": 0.0169, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 7.841269841269841, | |
| "grad_norm": 0.4648919999599457, | |
| "learning_rate": 5.110167179265636e-05, | |
| "loss": 0.0182, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 7.857142857142857, | |
| "grad_norm": 0.3623007833957672, | |
| "learning_rate": 5.094431038262693e-05, | |
| "loss": 0.0155, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 7.8730158730158735, | |
| "grad_norm": 0.4798755347728729, | |
| "learning_rate": 5.078693961524329e-05, | |
| "loss": 0.02, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 7.888888888888889, | |
| "grad_norm": 0.5778583288192749, | |
| "learning_rate": 5.062956104992285e-05, | |
| "loss": 0.0318, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 7.904761904761905, | |
| "grad_norm": 0.37309491634368896, | |
| "learning_rate": 5.0472176246160184e-05, | |
| "loss": 0.0116, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 7.920634920634921, | |
| "grad_norm": 0.6432266235351562, | |
| "learning_rate": 5.031478676351179e-05, | |
| "loss": 0.0188, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 7.936507936507937, | |
| "grad_norm": 0.43156516551971436, | |
| "learning_rate": 5.01573941615805e-05, | |
| "loss": 0.0179, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.9523809523809526, | |
| "grad_norm": 0.553710401058197, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0192, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 7.968253968253968, | |
| "grad_norm": 0.39197760820388794, | |
| "learning_rate": 4.984260583841953e-05, | |
| "loss": 0.0177, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 7.984126984126984, | |
| "grad_norm": 0.5970882773399353, | |
| "learning_rate": 4.9685213236488216e-05, | |
| "loss": 0.025, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 0.44673952460289, | |
| "learning_rate": 4.9527823753839834e-05, | |
| "loss": 0.0121, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 8.015873015873016, | |
| "grad_norm": 0.3288459777832031, | |
| "learning_rate": 4.937043895007717e-05, | |
| "loss": 0.0167, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 8.031746031746032, | |
| "grad_norm": 0.410833477973938, | |
| "learning_rate": 4.9213060384756716e-05, | |
| "loss": 0.0147, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 8.047619047619047, | |
| "grad_norm": 0.34271591901779175, | |
| "learning_rate": 4.9055689617373084e-05, | |
| "loss": 0.0108, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 8.063492063492063, | |
| "grad_norm": 0.22280845046043396, | |
| "learning_rate": 4.8898328207343666e-05, | |
| "loss": 0.0076, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 8.079365079365079, | |
| "grad_norm": 0.404482364654541, | |
| "learning_rate": 4.874097771399308e-05, | |
| "loss": 0.0124, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 8.095238095238095, | |
| "grad_norm": 0.3690173327922821, | |
| "learning_rate": 4.858363969653781e-05, | |
| "loss": 0.0167, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 8.11111111111111, | |
| "grad_norm": 0.31355366110801697, | |
| "learning_rate": 4.8426315714070684e-05, | |
| "loss": 0.0143, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 8.126984126984127, | |
| "grad_norm": 0.24391916394233704, | |
| "learning_rate": 4.8269007325545506e-05, | |
| "loss": 0.0111, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 8.142857142857142, | |
| "grad_norm": 0.39755526185035706, | |
| "learning_rate": 4.8111716089761456e-05, | |
| "loss": 0.0145, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 8.158730158730158, | |
| "grad_norm": 0.27595722675323486, | |
| "learning_rate": 4.7954443565347865e-05, | |
| "loss": 0.01, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 8.174603174603174, | |
| "grad_norm": 0.304116815328598, | |
| "learning_rate": 4.779719131074857e-05, | |
| "loss": 0.0105, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 8.19047619047619, | |
| "grad_norm": 0.2722436487674713, | |
| "learning_rate": 4.7639960884206576e-05, | |
| "loss": 0.0089, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 8.206349206349206, | |
| "grad_norm": 0.2728959321975708, | |
| "learning_rate": 4.7482753843748564e-05, | |
| "loss": 0.0108, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 8.222222222222221, | |
| "grad_norm": 0.2411596029996872, | |
| "learning_rate": 4.7325571747169545e-05, | |
| "loss": 0.0085, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 8.238095238095237, | |
| "grad_norm": 0.23578131198883057, | |
| "learning_rate": 4.716841615201726e-05, | |
| "loss": 0.008, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 8.253968253968253, | |
| "grad_norm": 0.3611275255680084, | |
| "learning_rate": 4.7011288615576934e-05, | |
| "loss": 0.0141, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 8.26984126984127, | |
| "grad_norm": 0.3158744275569916, | |
| "learning_rate": 4.6854190694855694e-05, | |
| "loss": 0.0115, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 8.285714285714286, | |
| "grad_norm": 0.40253180265426636, | |
| "learning_rate": 4.6697123946567227e-05, | |
| "loss": 0.013, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 8.301587301587302, | |
| "grad_norm": 0.290996789932251, | |
| "learning_rate": 4.65400899271163e-05, | |
| "loss": 0.0103, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 8.317460317460318, | |
| "grad_norm": 0.37486013770103455, | |
| "learning_rate": 4.63830901925834e-05, | |
| "loss": 0.0155, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 0.42451635003089905, | |
| "learning_rate": 4.6226126298709224e-05, | |
| "loss": 0.0175, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 8.34920634920635, | |
| "grad_norm": 0.4372078776359558, | |
| "learning_rate": 4.60691998008794e-05, | |
| "loss": 0.0203, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 8.365079365079366, | |
| "grad_norm": 0.3044324517250061, | |
| "learning_rate": 4.5912312254108905e-05, | |
| "loss": 0.0139, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 8.380952380952381, | |
| "grad_norm": 0.39817896485328674, | |
| "learning_rate": 4.575546521302681e-05, | |
| "loss": 0.0135, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 8.396825396825397, | |
| "grad_norm": 0.3401551842689514, | |
| "learning_rate": 4.5598660231860746e-05, | |
| "loss": 0.0107, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 8.412698412698413, | |
| "grad_norm": 0.3589102625846863, | |
| "learning_rate": 4.544189886442162e-05, | |
| "loss": 0.0131, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 8.428571428571429, | |
| "grad_norm": 0.4164977967739105, | |
| "learning_rate": 4.528518266408811e-05, | |
| "loss": 0.015, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 8.444444444444445, | |
| "grad_norm": 0.5136562585830688, | |
| "learning_rate": 4.5128513183791386e-05, | |
| "loss": 0.016, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 8.46031746031746, | |
| "grad_norm": 0.36152708530426025, | |
| "learning_rate": 4.49718919759996e-05, | |
| "loss": 0.015, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 8.476190476190476, | |
| "grad_norm": 0.2721676230430603, | |
| "learning_rate": 4.481532059270262e-05, | |
| "loss": 0.0083, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 8.492063492063492, | |
| "grad_norm": 0.2820744216442108, | |
| "learning_rate": 4.465880058539652e-05, | |
| "loss": 0.01, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 8.507936507936508, | |
| "grad_norm": 0.3638380467891693, | |
| "learning_rate": 4.450233350506836e-05, | |
| "loss": 0.0101, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 8.523809523809524, | |
| "grad_norm": 0.3278939723968506, | |
| "learning_rate": 4.4345920902180647e-05, | |
| "loss": 0.0104, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 8.53968253968254, | |
| "grad_norm": 0.3926644027233124, | |
| "learning_rate": 4.418956432665618e-05, | |
| "loss": 0.0125, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 8.555555555555555, | |
| "grad_norm": 0.3797055780887604, | |
| "learning_rate": 4.403326532786245e-05, | |
| "loss": 0.0111, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 8.571428571428571, | |
| "grad_norm": 0.26904818415641785, | |
| "learning_rate": 4.387702545459649e-05, | |
| "loss": 0.009, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 8.587301587301587, | |
| "grad_norm": 0.32789549231529236, | |
| "learning_rate": 4.3720846255069406e-05, | |
| "loss": 0.0075, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 8.603174603174603, | |
| "grad_norm": 0.19732752442359924, | |
| "learning_rate": 4.356472927689109e-05, | |
| "loss": 0.008, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 8.619047619047619, | |
| "grad_norm": 0.23964589834213257, | |
| "learning_rate": 4.3408676067054866e-05, | |
| "loss": 0.0102, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 8.634920634920634, | |
| "grad_norm": 0.4041917026042938, | |
| "learning_rate": 4.32526881719222e-05, | |
| "loss": 0.0188, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 8.65079365079365, | |
| "grad_norm": 0.4420047998428345, | |
| "learning_rate": 4.3096767137207256e-05, | |
| "loss": 0.0138, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 8.666666666666666, | |
| "grad_norm": 0.43801549077033997, | |
| "learning_rate": 4.2940914507961775e-05, | |
| "loss": 0.012, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 8.682539682539682, | |
| "grad_norm": 0.24375741183757782, | |
| "learning_rate": 4.278513182855956e-05, | |
| "loss": 0.0078, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 8.698412698412698, | |
| "grad_norm": 0.48987898230552673, | |
| "learning_rate": 4.262942064268134e-05, | |
| "loss": 0.0184, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 8.714285714285714, | |
| "grad_norm": 0.38676026463508606, | |
| "learning_rate": 4.247378249329933e-05, | |
| "loss": 0.0122, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 8.73015873015873, | |
| "grad_norm": 0.20567281544208527, | |
| "learning_rate": 4.23182189226621e-05, | |
| "loss": 0.0076, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 8.746031746031747, | |
| "grad_norm": 0.28698331117630005, | |
| "learning_rate": 4.21627314722791e-05, | |
| "loss": 0.0084, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 8.761904761904763, | |
| "grad_norm": 0.3160061836242676, | |
| "learning_rate": 4.20073216829056e-05, | |
| "loss": 0.0111, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 8.777777777777779, | |
| "grad_norm": 0.2930062711238861, | |
| "learning_rate": 4.185199109452721e-05, | |
| "loss": 0.0107, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 8.793650793650794, | |
| "grad_norm": 0.3634200692176819, | |
| "learning_rate": 4.169674124634481e-05, | |
| "loss": 0.0101, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 8.80952380952381, | |
| "grad_norm": 0.37438124418258667, | |
| "learning_rate": 4.1541573676759126e-05, | |
| "loss": 0.014, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 8.825396825396826, | |
| "grad_norm": 0.3476526141166687, | |
| "learning_rate": 4.138648992335566e-05, | |
| "loss": 0.0129, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 8.841269841269842, | |
| "grad_norm": 0.18964612483978271, | |
| "learning_rate": 4.12314915228893e-05, | |
| "loss": 0.0062, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 8.857142857142858, | |
| "grad_norm": 0.35653162002563477, | |
| "learning_rate": 4.107658001126913e-05, | |
| "loss": 0.0131, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 8.873015873015873, | |
| "grad_norm": 0.38258370757102966, | |
| "learning_rate": 4.092175692354333e-05, | |
| "loss": 0.0119, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 8.88888888888889, | |
| "grad_norm": 0.2177157700061798, | |
| "learning_rate": 4.0767023793883785e-05, | |
| "loss": 0.0062, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 8.904761904761905, | |
| "grad_norm": 0.3157006502151489, | |
| "learning_rate": 4.0612382155571026e-05, | |
| "loss": 0.0116, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 8.920634920634921, | |
| "grad_norm": 0.5421932935714722, | |
| "learning_rate": 4.045783354097893e-05, | |
| "loss": 0.0251, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 8.936507936507937, | |
| "grad_norm": 0.4682704210281372, | |
| "learning_rate": 4.0303379481559623e-05, | |
| "loss": 0.0193, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 8.952380952380953, | |
| "grad_norm": 0.36263760924339294, | |
| "learning_rate": 4.0149021507828224e-05, | |
| "loss": 0.0155, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 8.968253968253968, | |
| "grad_norm": 0.3147249221801758, | |
| "learning_rate": 3.9994761149347784e-05, | |
| "loss": 0.0114, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 8.984126984126984, | |
| "grad_norm": 0.41839832067489624, | |
| "learning_rate": 3.984059993471399e-05, | |
| "loss": 0.0154, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.37561434507369995, | |
| "learning_rate": 3.968653939154017e-05, | |
| "loss": 0.0103, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 9.015873015873016, | |
| "grad_norm": 0.31883716583251953, | |
| "learning_rate": 3.9532581046442e-05, | |
| "loss": 0.0082, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 9.031746031746032, | |
| "grad_norm": 0.23053289949893951, | |
| "learning_rate": 3.937872642502252e-05, | |
| "loss": 0.0073, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 9.047619047619047, | |
| "grad_norm": 0.25523173809051514, | |
| "learning_rate": 3.9224977051856904e-05, | |
| "loss": 0.008, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 9.063492063492063, | |
| "grad_norm": 0.20138682425022125, | |
| "learning_rate": 3.907133445047747e-05, | |
| "loss": 0.007, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 9.079365079365079, | |
| "grad_norm": 0.2522388696670532, | |
| "learning_rate": 3.8917800143358404e-05, | |
| "loss": 0.0064, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 9.095238095238095, | |
| "grad_norm": 0.32254767417907715, | |
| "learning_rate": 3.8764375651900906e-05, | |
| "loss": 0.0121, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 9.11111111111111, | |
| "grad_norm": 0.2257680743932724, | |
| "learning_rate": 3.861106249641789e-05, | |
| "loss": 0.0069, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 9.126984126984127, | |
| "grad_norm": 0.20319634675979614, | |
| "learning_rate": 3.84578621961191e-05, | |
| "loss": 0.0083, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 9.142857142857142, | |
| "grad_norm": 0.21617092192173004, | |
| "learning_rate": 3.830477626909589e-05, | |
| "loss": 0.0081, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 9.158730158730158, | |
| "grad_norm": 0.3438735902309418, | |
| "learning_rate": 3.8151806232306374e-05, | |
| "loss": 0.0113, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 9.174603174603174, | |
| "grad_norm": 0.29311296343803406, | |
| "learning_rate": 3.7998953601560175e-05, | |
| "loss": 0.0097, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 9.19047619047619, | |
| "grad_norm": 0.16206145286560059, | |
| "learning_rate": 3.784621989150361e-05, | |
| "loss": 0.0059, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 9.206349206349206, | |
| "grad_norm": 0.22121606767177582, | |
| "learning_rate": 3.769360661560453e-05, | |
| "loss": 0.0084, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 9.222222222222221, | |
| "grad_norm": 0.25994566082954407, | |
| "learning_rate": 3.75411152861374e-05, | |
| "loss": 0.0104, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 9.238095238095237, | |
| "grad_norm": 0.18151433765888214, | |
| "learning_rate": 3.73887474141683e-05, | |
| "loss": 0.0056, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 9.253968253968253, | |
| "grad_norm": 0.18867704272270203, | |
| "learning_rate": 3.723650450953994e-05, | |
| "loss": 0.006, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 9.26984126984127, | |
| "grad_norm": 0.3016846776008606, | |
| "learning_rate": 3.708438808085668e-05, | |
| "loss": 0.0136, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 9.285714285714286, | |
| "grad_norm": 0.41189849376678467, | |
| "learning_rate": 3.693239963546967e-05, | |
| "loss": 0.0168, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 9.301587301587302, | |
| "grad_norm": 0.2735559940338135, | |
| "learning_rate": 3.6780540679461784e-05, | |
| "loss": 0.0097, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 9.317460317460318, | |
| "grad_norm": 0.23788434267044067, | |
| "learning_rate": 3.662881271763279e-05, | |
| "loss": 0.0068, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 9.333333333333334, | |
| "grad_norm": 0.14663733541965485, | |
| "learning_rate": 3.64772172534844e-05, | |
| "loss": 0.0046, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 9.34920634920635, | |
| "grad_norm": 0.3166827857494354, | |
| "learning_rate": 3.63257557892054e-05, | |
| "loss": 0.0082, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 9.365079365079366, | |
| "grad_norm": 0.24929101765155792, | |
| "learning_rate": 3.6174429825656685e-05, | |
| "loss": 0.0104, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 9.380952380952381, | |
| "grad_norm": 0.27766042947769165, | |
| "learning_rate": 3.602324086235655e-05, | |
| "loss": 0.0079, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 9.396825396825397, | |
| "grad_norm": 0.25808480381965637, | |
| "learning_rate": 3.587219039746564e-05, | |
| "loss": 0.0076, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 9.412698412698413, | |
| "grad_norm": 0.2501043677330017, | |
| "learning_rate": 3.572127992777223e-05, | |
| "loss": 0.0103, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 9.428571428571429, | |
| "grad_norm": 0.2836500108242035, | |
| "learning_rate": 3.557051094867735e-05, | |
| "loss": 0.0082, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 9.444444444444445, | |
| "grad_norm": 0.3479957580566406, | |
| "learning_rate": 3.541988495417997e-05, | |
| "loss": 0.0126, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 9.46031746031746, | |
| "grad_norm": 0.2896635830402374, | |
| "learning_rate": 3.5269403436862175e-05, | |
| "loss": 0.0072, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 9.476190476190476, | |
| "grad_norm": 0.2840765416622162, | |
| "learning_rate": 3.511906788787447e-05, | |
| "loss": 0.0101, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 9.492063492063492, | |
| "grad_norm": 0.3210354745388031, | |
| "learning_rate": 3.496887979692084e-05, | |
| "loss": 0.0085, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 9.507936507936508, | |
| "grad_norm": 0.27587252855300903, | |
| "learning_rate": 3.481884065224415e-05, | |
| "loss": 0.0087, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 9.523809523809524, | |
| "grad_norm": 0.3219284117221832, | |
| "learning_rate": 3.466895194061128e-05, | |
| "loss": 0.009, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 9.53968253968254, | |
| "grad_norm": 0.17630243301391602, | |
| "learning_rate": 3.451921514729848e-05, | |
| "loss": 0.0059, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 9.555555555555555, | |
| "grad_norm": 0.25327348709106445, | |
| "learning_rate": 3.436963175607656e-05, | |
| "loss": 0.0081, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 9.571428571428571, | |
| "grad_norm": 0.3768535554409027, | |
| "learning_rate": 3.422020324919632e-05, | |
| "loss": 0.0113, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 9.587301587301587, | |
| "grad_norm": 0.1651473492383957, | |
| "learning_rate": 3.4070931107373675e-05, | |
| "loss": 0.0049, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 9.603174603174603, | |
| "grad_norm": 0.23368506133556366, | |
| "learning_rate": 3.39218168097752e-05, | |
| "loss": 0.008, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 9.619047619047619, | |
| "grad_norm": 0.1572844684123993, | |
| "learning_rate": 3.377286183400328e-05, | |
| "loss": 0.0048, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 9.634920634920634, | |
| "grad_norm": 0.2425893396139145, | |
| "learning_rate": 3.362406765608158e-05, | |
| "loss": 0.0084, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 9.65079365079365, | |
| "grad_norm": 0.280091255903244, | |
| "learning_rate": 3.3475435750440356e-05, | |
| "loss": 0.0114, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 9.666666666666666, | |
| "grad_norm": 0.34356409311294556, | |
| "learning_rate": 3.332696758990197e-05, | |
| "loss": 0.0101, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 9.682539682539682, | |
| "grad_norm": 0.26575177907943726, | |
| "learning_rate": 3.3178664645666066e-05, | |
| "loss": 0.0076, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 9.698412698412698, | |
| "grad_norm": 0.38795173168182373, | |
| "learning_rate": 3.303052838729525e-05, | |
| "loss": 0.0141, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 9.714285714285714, | |
| "grad_norm": 0.17991788685321808, | |
| "learning_rate": 3.2882560282700336e-05, | |
| "loss": 0.0071, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 9.73015873015873, | |
| "grad_norm": 0.26826414465904236, | |
| "learning_rate": 3.273476179812588e-05, | |
| "loss": 0.0084, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 9.746031746031747, | |
| "grad_norm": 0.4353213906288147, | |
| "learning_rate": 3.258713439813566e-05, | |
| "loss": 0.0138, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 9.761904761904763, | |
| "grad_norm": 0.27039167284965515, | |
| "learning_rate": 3.243967954559811e-05, | |
| "loss": 0.0075, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 9.777777777777779, | |
| "grad_norm": 0.1729506552219391, | |
| "learning_rate": 3.229239870167191e-05, | |
| "loss": 0.0066, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 9.793650793650794, | |
| "grad_norm": 0.31375908851623535, | |
| "learning_rate": 3.2145293325791395e-05, | |
| "loss": 0.0091, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 9.80952380952381, | |
| "grad_norm": 0.2373589277267456, | |
| "learning_rate": 3.199836487565222e-05, | |
| "loss": 0.0077, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 9.825396825396826, | |
| "grad_norm": 0.3218036890029907, | |
| "learning_rate": 3.1851614807196774e-05, | |
| "loss": 0.0142, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 9.841269841269842, | |
| "grad_norm": 0.2621251940727234, | |
| "learning_rate": 3.170504457459989e-05, | |
| "loss": 0.0085, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 9.857142857142858, | |
| "grad_norm": 0.2235831618309021, | |
| "learning_rate": 3.155865563025433e-05, | |
| "loss": 0.0085, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 9.873015873015873, | |
| "grad_norm": 0.3102441728115082, | |
| "learning_rate": 3.1412449424756474e-05, | |
| "loss": 0.0091, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 9.88888888888889, | |
| "grad_norm": 0.3454819321632385, | |
| "learning_rate": 3.1266427406891856e-05, | |
| "loss": 0.0078, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 9.904761904761905, | |
| "grad_norm": 0.1699669510126114, | |
| "learning_rate": 3.112059102362093e-05, | |
| "loss": 0.005, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 9.920634920634921, | |
| "grad_norm": 0.21184861660003662, | |
| "learning_rate": 3.0974941720064585e-05, | |
| "loss": 0.0059, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 9.936507936507937, | |
| "grad_norm": 0.21373149752616882, | |
| "learning_rate": 3.082948093948997e-05, | |
| "loss": 0.0067, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 9.952380952380953, | |
| "grad_norm": 0.17170457541942596, | |
| "learning_rate": 3.0684210123296055e-05, | |
| "loss": 0.0061, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 9.968253968253968, | |
| "grad_norm": 0.33514630794525146, | |
| "learning_rate": 3.053913071099947e-05, | |
| "loss": 0.0136, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 9.984126984126984, | |
| "grad_norm": 0.34444811940193176, | |
| "learning_rate": 3.0394244140220163e-05, | |
| "loss": 0.0129, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 0.2810363173484802, | |
| "learning_rate": 3.0249551846667207e-05, | |
| "loss": 0.0072, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 10.015873015873016, | |
| "grad_norm": 0.16898448765277863, | |
| "learning_rate": 3.010505526412447e-05, | |
| "loss": 0.0057, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 10.031746031746032, | |
| "grad_norm": 0.27064862847328186, | |
| "learning_rate": 2.996075582443658e-05, | |
| "loss": 0.0081, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 10.047619047619047, | |
| "grad_norm": 0.11674167960882187, | |
| "learning_rate": 2.981665495749457e-05, | |
| "loss": 0.0044, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 10.063492063492063, | |
| "grad_norm": 0.18693989515304565, | |
| "learning_rate": 2.9672754091221805e-05, | |
| "loss": 0.0071, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 10.079365079365079, | |
| "grad_norm": 0.19624684751033783, | |
| "learning_rate": 2.9529054651559772e-05, | |
| "loss": 0.0065, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 10.095238095238095, | |
| "grad_norm": 0.13836269080638885, | |
| "learning_rate": 2.938555806245406e-05, | |
| "loss": 0.0045, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 10.11111111111111, | |
| "grad_norm": 0.2417069971561432, | |
| "learning_rate": 2.9242265745840063e-05, | |
| "loss": 0.0091, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 10.126984126984127, | |
| "grad_norm": 0.18066619336605072, | |
| "learning_rate": 2.9099179121629117e-05, | |
| "loss": 0.006, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 10.142857142857142, | |
| "grad_norm": 0.2307615429162979, | |
| "learning_rate": 2.895629960769417e-05, | |
| "loss": 0.0078, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 10.158730158730158, | |
| "grad_norm": 0.1858942061662674, | |
| "learning_rate": 2.881362861985606e-05, | |
| "loss": 0.007, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 10.174603174603174, | |
| "grad_norm": 0.20081129670143127, | |
| "learning_rate": 2.867116757186911e-05, | |
| "loss": 0.0073, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 10.19047619047619, | |
| "grad_norm": 0.2889654338359833, | |
| "learning_rate": 2.8528917875407433e-05, | |
| "loss": 0.0088, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 10.206349206349206, | |
| "grad_norm": 0.22024375200271606, | |
| "learning_rate": 2.838688094005078e-05, | |
| "loss": 0.0061, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 10.222222222222221, | |
| "grad_norm": 0.2205890566110611, | |
| "learning_rate": 2.8245058173270622e-05, | |
| "loss": 0.0072, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 10.238095238095237, | |
| "grad_norm": 0.21441209316253662, | |
| "learning_rate": 2.8103450980416136e-05, | |
| "loss": 0.0054, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 10.253968253968253, | |
| "grad_norm": 0.18930909037590027, | |
| "learning_rate": 2.796206076470044e-05, | |
| "loss": 0.0066, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 10.26984126984127, | |
| "grad_norm": 0.16868965327739716, | |
| "learning_rate": 2.7820888927186483e-05, | |
| "loss": 0.0048, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 10.285714285714286, | |
| "grad_norm": 0.3065090775489807, | |
| "learning_rate": 2.7679936866773315e-05, | |
| "loss": 0.0088, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 10.301587301587302, | |
| "grad_norm": 0.21105839312076569, | |
| "learning_rate": 2.753920598018217e-05, | |
| "loss": 0.0057, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 10.317460317460318, | |
| "grad_norm": 0.07848194986581802, | |
| "learning_rate": 2.739869766194263e-05, | |
| "loss": 0.0031, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 10.333333333333334, | |
| "grad_norm": 0.23540142178535461, | |
| "learning_rate": 2.7258413304378734e-05, | |
| "loss": 0.0078, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 10.34920634920635, | |
| "grad_norm": 0.2934277057647705, | |
| "learning_rate": 2.7118354297595396e-05, | |
| "loss": 0.0065, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 10.365079365079366, | |
| "grad_norm": 0.2042340338230133, | |
| "learning_rate": 2.6978522029464325e-05, | |
| "loss": 0.005, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 10.380952380952381, | |
| "grad_norm": 0.2258983999490738, | |
| "learning_rate": 2.683891788561055e-05, | |
| "loss": 0.0074, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 10.396825396825397, | |
| "grad_norm": 0.18975599110126495, | |
| "learning_rate": 2.669954324939852e-05, | |
| "loss": 0.0071, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 10.412698412698413, | |
| "grad_norm": 0.16135640442371368, | |
| "learning_rate": 2.6560399501918465e-05, | |
| "loss": 0.0058, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 10.428571428571429, | |
| "grad_norm": 0.30178365111351013, | |
| "learning_rate": 2.6421488021972673e-05, | |
| "loss": 0.0086, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 10.444444444444445, | |
| "grad_norm": 0.3351801037788391, | |
| "learning_rate": 2.6282810186061862e-05, | |
| "loss": 0.0132, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 10.46031746031746, | |
| "grad_norm": 0.25116395950317383, | |
| "learning_rate": 2.6144367368371535e-05, | |
| "loss": 0.0081, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 10.476190476190476, | |
| "grad_norm": 0.2531328797340393, | |
| "learning_rate": 2.600616094075835e-05, | |
| "loss": 0.0082, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 10.492063492063492, | |
| "grad_norm": 0.22533273696899414, | |
| "learning_rate": 2.5868192272736514e-05, | |
| "loss": 0.0065, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 10.507936507936508, | |
| "grad_norm": 0.18789933621883392, | |
| "learning_rate": 2.5730462731464273e-05, | |
| "loss": 0.0048, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 10.523809523809524, | |
| "grad_norm": 0.2593654990196228, | |
| "learning_rate": 2.5592973681730236e-05, | |
| "loss": 0.008, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 10.53968253968254, | |
| "grad_norm": 0.2563331425189972, | |
| "learning_rate": 2.5455726485940012e-05, | |
| "loss": 0.0099, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 10.555555555555555, | |
| "grad_norm": 0.2012241631746292, | |
| "learning_rate": 2.5318722504102604e-05, | |
| "loss": 0.0051, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 10.571428571428571, | |
| "grad_norm": 0.3327932059764862, | |
| "learning_rate": 2.5181963093816962e-05, | |
| "loss": 0.0077, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 10.587301587301587, | |
| "grad_norm": 0.2965086102485657, | |
| "learning_rate": 2.504544961025853e-05, | |
| "loss": 0.0089, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 10.603174603174603, | |
| "grad_norm": 0.2296365350484848, | |
| "learning_rate": 2.4909183406165836e-05, | |
| "loss": 0.0068, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 10.619047619047619, | |
| "grad_norm": 0.3457624018192291, | |
| "learning_rate": 2.4773165831827018e-05, | |
| "loss": 0.0083, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 10.634920634920634, | |
| "grad_norm": 0.20112329721450806, | |
| "learning_rate": 2.4637398235066527e-05, | |
| "loss": 0.0061, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 10.65079365079365, | |
| "grad_norm": 0.19829870760440826, | |
| "learning_rate": 2.450188196123177e-05, | |
| "loss": 0.0063, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 10.666666666666666, | |
| "grad_norm": 0.17704661190509796, | |
| "learning_rate": 2.4366618353179644e-05, | |
| "loss": 0.0045, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 10.682539682539682, | |
| "grad_norm": 0.27905184030532837, | |
| "learning_rate": 2.423160875126348e-05, | |
| "loss": 0.009, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 10.698412698412698, | |
| "grad_norm": 0.18189361691474915, | |
| "learning_rate": 2.4096854493319477e-05, | |
| "loss": 0.0069, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 10.714285714285714, | |
| "grad_norm": 0.2877546548843384, | |
| "learning_rate": 2.3962356914653657e-05, | |
| "loss": 0.0064, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 10.73015873015873, | |
| "grad_norm": 0.27436089515686035, | |
| "learning_rate": 2.3828117348028528e-05, | |
| "loss": 0.009, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 10.746031746031747, | |
| "grad_norm": 0.11570344120264053, | |
| "learning_rate": 2.3694137123649946e-05, | |
| "loss": 0.0038, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 10.761904761904763, | |
| "grad_norm": 0.29015523195266724, | |
| "learning_rate": 2.3560417569153796e-05, | |
| "loss": 0.0079, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 10.777777777777779, | |
| "grad_norm": 0.23264740407466888, | |
| "learning_rate": 2.342696000959309e-05, | |
| "loss": 0.0087, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 10.793650793650794, | |
| "grad_norm": 0.23853233456611633, | |
| "learning_rate": 2.3293765767424537e-05, | |
| "loss": 0.0068, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 10.80952380952381, | |
| "grad_norm": 0.11449386179447174, | |
| "learning_rate": 2.3160836162495653e-05, | |
| "loss": 0.0033, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 10.825396825396826, | |
| "grad_norm": 0.15624088048934937, | |
| "learning_rate": 2.3028172512031604e-05, | |
| "loss": 0.005, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 10.841269841269842, | |
| "grad_norm": 0.17482654750347137, | |
| "learning_rate": 2.289577613062218e-05, | |
| "loss": 0.0053, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 10.857142857142858, | |
| "grad_norm": 0.1657302975654602, | |
| "learning_rate": 2.2763648330208688e-05, | |
| "loss": 0.0044, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 10.873015873015873, | |
| "grad_norm": 0.3183576762676239, | |
| "learning_rate": 2.2631790420071064e-05, | |
| "loss": 0.0087, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 10.88888888888889, | |
| "grad_norm": 0.2113347351551056, | |
| "learning_rate": 2.2500203706814856e-05, | |
| "loss": 0.0057, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 10.904761904761905, | |
| "grad_norm": 0.20787814259529114, | |
| "learning_rate": 2.2368889494358235e-05, | |
| "loss": 0.0066, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 10.920634920634921, | |
| "grad_norm": 0.19461645185947418, | |
| "learning_rate": 2.2237849083919142e-05, | |
| "loss": 0.0056, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 10.936507936507937, | |
| "grad_norm": 0.3162117302417755, | |
| "learning_rate": 2.2107083774002364e-05, | |
| "loss": 0.0102, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 10.952380952380953, | |
| "grad_norm": 0.1498049944639206, | |
| "learning_rate": 2.1976594860386597e-05, | |
| "loss": 0.0054, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 10.968253968253968, | |
| "grad_norm": 0.25862017273902893, | |
| "learning_rate": 2.1846383636111743e-05, | |
| "loss": 0.0063, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 10.984126984126984, | |
| "grad_norm": 0.2787252962589264, | |
| "learning_rate": 2.1716451391466008e-05, | |
| "loss": 0.004, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 0.5165538787841797, | |
| "learning_rate": 2.1586799413973135e-05, | |
| "loss": 0.0117, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 11.015873015873016, | |
| "grad_norm": 0.16975046694278717, | |
| "learning_rate": 2.1457428988379635e-05, | |
| "loss": 0.0053, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 11.031746031746032, | |
| "grad_norm": 0.09435385465621948, | |
| "learning_rate": 2.1328341396642093e-05, | |
| "loss": 0.0032, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 11.047619047619047, | |
| "grad_norm": 0.0928262248635292, | |
| "learning_rate": 2.1199537917914386e-05, | |
| "loss": 0.0031, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 11.063492063492063, | |
| "grad_norm": 0.1879938691854477, | |
| "learning_rate": 2.107101982853511e-05, | |
| "loss": 0.0052, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 11.079365079365079, | |
| "grad_norm": 0.13509397208690643, | |
| "learning_rate": 2.0942788402014867e-05, | |
| "loss": 0.005, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 11.095238095238095, | |
| "grad_norm": 0.10293649882078171, | |
| "learning_rate": 2.0814844909023663e-05, | |
| "loss": 0.0038, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "grad_norm": 0.26907050609588623, | |
| "learning_rate": 2.068719061737831e-05, | |
| "loss": 0.0086, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 11.126984126984127, | |
| "grad_norm": 0.1459931880235672, | |
| "learning_rate": 2.0559826792029884e-05, | |
| "loss": 0.0045, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 11.142857142857142, | |
| "grad_norm": 0.10803816467523575, | |
| "learning_rate": 2.0432754695051136e-05, | |
| "loss": 0.0034, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 11.158730158730158, | |
| "grad_norm": 0.07795245200395584, | |
| "learning_rate": 2.0305975585624058e-05, | |
| "loss": 0.0031, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 11.174603174603174, | |
| "grad_norm": 0.14636225998401642, | |
| "learning_rate": 2.0179490720027372e-05, | |
| "loss": 0.0055, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 11.19047619047619, | |
| "grad_norm": 0.0945882797241211, | |
| "learning_rate": 2.005330135162408e-05, | |
| "loss": 0.0036, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 11.206349206349206, | |
| "grad_norm": 0.16662253439426422, | |
| "learning_rate": 1.992740873084899e-05, | |
| "loss": 0.0042, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 11.222222222222221, | |
| "grad_norm": 0.2733784019947052, | |
| "learning_rate": 1.9801814105196497e-05, | |
| "loss": 0.0066, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 11.238095238095237, | |
| "grad_norm": 0.27156999707221985, | |
| "learning_rate": 1.9676518719207977e-05, | |
| "loss": 0.0069, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 11.253968253968253, | |
| "grad_norm": 0.23552264273166656, | |
| "learning_rate": 1.9551523814459665e-05, | |
| "loss": 0.0071, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 11.26984126984127, | |
| "grad_norm": 0.09834027290344238, | |
| "learning_rate": 1.9426830629550242e-05, | |
| "loss": 0.0035, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 11.285714285714286, | |
| "grad_norm": 0.1471029371023178, | |
| "learning_rate": 1.9302440400088606e-05, | |
| "loss": 0.0055, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 11.301587301587302, | |
| "grad_norm": 0.20986461639404297, | |
| "learning_rate": 1.917835435868155e-05, | |
| "loss": 0.0063, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 11.317460317460318, | |
| "grad_norm": 0.29454532265663147, | |
| "learning_rate": 1.9054573734921714e-05, | |
| "loss": 0.0098, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 11.333333333333334, | |
| "grad_norm": 0.1742410510778427, | |
| "learning_rate": 1.8931099755375203e-05, | |
| "loss": 0.0044, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 11.34920634920635, | |
| "grad_norm": 0.13173726201057434, | |
| "learning_rate": 1.880793364356956e-05, | |
| "loss": 0.0055, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 11.365079365079366, | |
| "grad_norm": 0.20177853107452393, | |
| "learning_rate": 1.8685076619981608e-05, | |
| "loss": 0.006, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 11.380952380952381, | |
| "grad_norm": 0.1103038340806961, | |
| "learning_rate": 1.8562529902025372e-05, | |
| "loss": 0.0037, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 11.396825396825397, | |
| "grad_norm": 0.22189675271511078, | |
| "learning_rate": 1.844029470403993e-05, | |
| "loss": 0.0066, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 11.412698412698413, | |
| "grad_norm": 0.21314705908298492, | |
| "learning_rate": 1.8318372237277565e-05, | |
| "loss": 0.0065, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 11.428571428571429, | |
| "grad_norm": 0.1456424593925476, | |
| "learning_rate": 1.8196763709891524e-05, | |
| "loss": 0.0049, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 11.444444444444445, | |
| "grad_norm": 0.1834188550710678, | |
| "learning_rate": 1.8075470326924243e-05, | |
| "loss": 0.0067, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 11.46031746031746, | |
| "grad_norm": 0.2855736017227173, | |
| "learning_rate": 1.795449329029531e-05, | |
| "loss": 0.009, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 11.476190476190476, | |
| "grad_norm": 0.15806177258491516, | |
| "learning_rate": 1.7833833798789595e-05, | |
| "loss": 0.0044, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 11.492063492063492, | |
| "grad_norm": 0.16890814900398254, | |
| "learning_rate": 1.7713493048045294e-05, | |
| "loss": 0.0056, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 11.507936507936508, | |
| "grad_norm": 0.24409544467926025, | |
| "learning_rate": 1.7593472230542202e-05, | |
| "loss": 0.0069, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 11.523809523809524, | |
| "grad_norm": 0.2861270010471344, | |
| "learning_rate": 1.747377253558982e-05, | |
| "loss": 0.0078, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 11.53968253968254, | |
| "grad_norm": 0.17466863989830017, | |
| "learning_rate": 1.7354395149315534e-05, | |
| "loss": 0.0044, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 11.555555555555555, | |
| "grad_norm": 0.2202078104019165, | |
| "learning_rate": 1.7235341254653005e-05, | |
| "loss": 0.0071, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 11.571428571428571, | |
| "grad_norm": 0.25968992710113525, | |
| "learning_rate": 1.711661203133026e-05, | |
| "loss": 0.0052, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 11.587301587301587, | |
| "grad_norm": 0.10932864248752594, | |
| "learning_rate": 1.6998208655858137e-05, | |
| "loss": 0.0033, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 11.603174603174603, | |
| "grad_norm": 0.1846671849489212, | |
| "learning_rate": 1.6880132301518598e-05, | |
| "loss": 0.0049, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 11.619047619047619, | |
| "grad_norm": 0.18320026993751526, | |
| "learning_rate": 1.6762384138353078e-05, | |
| "loss": 0.0048, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 11.634920634920634, | |
| "grad_norm": 0.18667708337306976, | |
| "learning_rate": 1.6644965333150847e-05, | |
| "loss": 0.0041, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 11.65079365079365, | |
| "grad_norm": 0.29703792929649353, | |
| "learning_rate": 1.6527877049437622e-05, | |
| "loss": 0.0098, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 11.666666666666666, | |
| "grad_norm": 0.1451849490404129, | |
| "learning_rate": 1.6411120447463807e-05, | |
| "loss": 0.0034, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 11.682539682539682, | |
| "grad_norm": 0.28783440589904785, | |
| "learning_rate": 1.6294696684193154e-05, | |
| "loss": 0.009, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 11.698412698412698, | |
| "grad_norm": 0.22581429779529572, | |
| "learning_rate": 1.617860691329126e-05, | |
| "loss": 0.0044, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 11.714285714285714, | |
| "grad_norm": 0.20482461154460907, | |
| "learning_rate": 1.6062852285114123e-05, | |
| "loss": 0.007, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 11.73015873015873, | |
| "grad_norm": 0.10219337791204453, | |
| "learning_rate": 1.5947433946696693e-05, | |
| "loss": 0.0031, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 11.746031746031747, | |
| "grad_norm": 0.2273254692554474, | |
| "learning_rate": 1.583235304174167e-05, | |
| "loss": 0.0069, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 11.761904761904763, | |
| "grad_norm": 0.3083495497703552, | |
| "learning_rate": 1.5717610710607948e-05, | |
| "loss": 0.0116, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 11.777777777777779, | |
| "grad_norm": 0.2324836254119873, | |
| "learning_rate": 1.5603208090299498e-05, | |
| "loss": 0.0065, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 11.793650793650794, | |
| "grad_norm": 0.14565986394882202, | |
| "learning_rate": 1.5489146314454002e-05, | |
| "loss": 0.0041, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 11.80952380952381, | |
| "grad_norm": 0.18284986913204193, | |
| "learning_rate": 1.537542651333167e-05, | |
| "loss": 0.0043, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 11.825396825396826, | |
| "grad_norm": 0.21167722344398499, | |
| "learning_rate": 1.5262049813803958e-05, | |
| "loss": 0.0066, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 11.841269841269842, | |
| "grad_norm": 0.16525444388389587, | |
| "learning_rate": 1.5149017339342574e-05, | |
| "loss": 0.0047, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 11.857142857142858, | |
| "grad_norm": 0.17935959994792938, | |
| "learning_rate": 1.503633021000812e-05, | |
| "loss": 0.0053, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 11.873015873015873, | |
| "grad_norm": 0.2582390010356903, | |
| "learning_rate": 1.4923989542439159e-05, | |
| "loss": 0.0052, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 11.88888888888889, | |
| "grad_norm": 0.06719334423542023, | |
| "learning_rate": 1.4811996449841098e-05, | |
| "loss": 0.0025, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 11.904761904761905, | |
| "grad_norm": 0.19448348879814148, | |
| "learning_rate": 1.4700352041975163e-05, | |
| "loss": 0.0059, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 11.920634920634921, | |
| "grad_norm": 0.30000415444374084, | |
| "learning_rate": 1.458905742514734e-05, | |
| "loss": 0.0089, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 11.936507936507937, | |
| "grad_norm": 0.19624555110931396, | |
| "learning_rate": 1.447811370219757e-05, | |
| "loss": 0.0067, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 11.952380952380953, | |
| "grad_norm": 0.16108612716197968, | |
| "learning_rate": 1.4367521972488612e-05, | |
| "loss": 0.0036, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 11.968253968253968, | |
| "grad_norm": 0.10793477296829224, | |
| "learning_rate": 1.4257283331895315e-05, | |
| "loss": 0.0032, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 11.984126984126984, | |
| "grad_norm": 0.19331948459148407, | |
| "learning_rate": 1.4147398872793693e-05, | |
| "loss": 0.0054, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "grad_norm": 0.3868754208087921, | |
| "learning_rate": 1.4037869684050115e-05, | |
| "loss": 0.0066, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 12.015873015873016, | |
| "grad_norm": 0.1854810267686844, | |
| "learning_rate": 1.3928696851010443e-05, | |
| "loss": 0.0052, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 12.031746031746032, | |
| "grad_norm": 0.1465175747871399, | |
| "learning_rate": 1.3819881455489458e-05, | |
| "loss": 0.0064, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 12.047619047619047, | |
| "grad_norm": 0.09918566048145294, | |
| "learning_rate": 1.3711424575759912e-05, | |
| "loss": 0.0033, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 12.063492063492063, | |
| "grad_norm": 0.1635628491640091, | |
| "learning_rate": 1.3603327286542023e-05, | |
| "loss": 0.0044, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 12.079365079365079, | |
| "grad_norm": 0.1613842099905014, | |
| "learning_rate": 1.3495590658992718e-05, | |
| "loss": 0.0048, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 12.095238095238095, | |
| "grad_norm": 0.13634873926639557, | |
| "learning_rate": 1.33882157606951e-05, | |
| "loss": 0.0034, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 12.11111111111111, | |
| "grad_norm": 0.15302757918834686, | |
| "learning_rate": 1.3281203655647756e-05, | |
| "loss": 0.0047, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 12.126984126984127, | |
| "grad_norm": 0.10601391643285751, | |
| "learning_rate": 1.317455540425439e-05, | |
| "loss": 0.0031, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 12.142857142857142, | |
| "grad_norm": 0.16901229321956635, | |
| "learning_rate": 1.3068272063313102e-05, | |
| "loss": 0.004, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 12.158730158730158, | |
| "grad_norm": 0.11270225793123245, | |
| "learning_rate": 1.2962354686006084e-05, | |
| "loss": 0.0036, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 12.174603174603174, | |
| "grad_norm": 0.17881913483142853, | |
| "learning_rate": 1.2856804321889115e-05, | |
| "loss": 0.0061, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 12.19047619047619, | |
| "grad_norm": 0.27680760622024536, | |
| "learning_rate": 1.2751622016881182e-05, | |
| "loss": 0.0087, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 12.206349206349206, | |
| "grad_norm": 0.14763417840003967, | |
| "learning_rate": 1.2646808813254042e-05, | |
| "loss": 0.0039, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 12.222222222222221, | |
| "grad_norm": 0.21186058223247528, | |
| "learning_rate": 1.2542365749622049e-05, | |
| "loss": 0.0065, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 12.238095238095237, | |
| "grad_norm": 0.13028453290462494, | |
| "learning_rate": 1.2438293860931682e-05, | |
| "loss": 0.0037, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 12.253968253968253, | |
| "grad_norm": 0.1220482587814331, | |
| "learning_rate": 1.2334594178451425e-05, | |
| "loss": 0.0034, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 12.26984126984127, | |
| "grad_norm": 0.10451938956975937, | |
| "learning_rate": 1.2231267729761487e-05, | |
| "loss": 0.0034, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 12.285714285714286, | |
| "grad_norm": 0.06596413254737854, | |
| "learning_rate": 1.2128315538743646e-05, | |
| "loss": 0.0025, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 12.301587301587302, | |
| "grad_norm": 0.18053588271141052, | |
| "learning_rate": 1.2025738625571026e-05, | |
| "loss": 0.0043, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 12.317460317460318, | |
| "grad_norm": 0.2295704185962677, | |
| "learning_rate": 1.1923538006698154e-05, | |
| "loss": 0.0076, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 12.333333333333334, | |
| "grad_norm": 0.21795432269573212, | |
| "learning_rate": 1.1821714694850689e-05, | |
| "loss": 0.0062, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 12.34920634920635, | |
| "grad_norm": 0.110650934278965, | |
| "learning_rate": 1.172026969901553e-05, | |
| "loss": 0.0033, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 12.365079365079366, | |
| "grad_norm": 0.14939086139202118, | |
| "learning_rate": 1.161920402443077e-05, | |
| "loss": 0.0053, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 12.380952380952381, | |
| "grad_norm": 0.14100809395313263, | |
| "learning_rate": 1.1518518672575701e-05, | |
| "loss": 0.0047, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 12.396825396825397, | |
| "grad_norm": 0.1589258462190628, | |
| "learning_rate": 1.1418214641160958e-05, | |
| "loss": 0.0041, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 12.412698412698413, | |
| "grad_norm": 0.22199559211730957, | |
| "learning_rate": 1.1318292924118584e-05, | |
| "loss": 0.0048, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 12.428571428571429, | |
| "grad_norm": 0.1654834747314453, | |
| "learning_rate": 1.1218754511592217e-05, | |
| "loss": 0.0052, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 12.444444444444445, | |
| "grad_norm": 0.18298682570457458, | |
| "learning_rate": 1.1119600389927182e-05, | |
| "loss": 0.0053, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 12.46031746031746, | |
| "grad_norm": 0.13524076342582703, | |
| "learning_rate": 1.1020831541660915e-05, | |
| "loss": 0.0038, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 12.476190476190476, | |
| "grad_norm": 0.17973224818706512, | |
| "learning_rate": 1.092244894551298e-05, | |
| "loss": 0.0047, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 12.492063492063492, | |
| "grad_norm": 0.06217047572135925, | |
| "learning_rate": 1.0824453576375576e-05, | |
| "loss": 0.0026, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 12.507936507936508, | |
| "grad_norm": 0.17186515033245087, | |
| "learning_rate": 1.0726846405303754e-05, | |
| "loss": 0.0043, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 12.523809523809524, | |
| "grad_norm": 0.22013287246227264, | |
| "learning_rate": 1.062962839950587e-05, | |
| "loss": 0.0057, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 12.53968253968254, | |
| "grad_norm": 0.1783435344696045, | |
| "learning_rate": 1.0532800522333897e-05, | |
| "loss": 0.0057, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 12.555555555555555, | |
| "grad_norm": 0.21852487325668335, | |
| "learning_rate": 1.0436363733274057e-05, | |
| "loss": 0.0053, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 12.571428571428571, | |
| "grad_norm": 0.22835583984851837, | |
| "learning_rate": 1.0340318987937097e-05, | |
| "loss": 0.0056, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 12.587301587301587, | |
| "grad_norm": 0.14611005783081055, | |
| "learning_rate": 1.0244667238048988e-05, | |
| "loss": 0.004, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 12.603174603174603, | |
| "grad_norm": 0.13122573494911194, | |
| "learning_rate": 1.014940943144142e-05, | |
| "loss": 0.0034, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 12.619047619047619, | |
| "grad_norm": 0.1692192703485489, | |
| "learning_rate": 1.0054546512042424e-05, | |
| "loss": 0.0036, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 12.634920634920634, | |
| "grad_norm": 0.10081874579191208, | |
| "learning_rate": 9.960079419866985e-06, | |
| "loss": 0.0028, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 12.65079365079365, | |
| "grad_norm": 0.16554361581802368, | |
| "learning_rate": 9.866009091007833e-06, | |
| "loss": 0.004, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 12.666666666666666, | |
| "grad_norm": 0.11980407685041428, | |
| "learning_rate": 9.772336457626014e-06, | |
| "loss": 0.0033, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 12.682539682539682, | |
| "grad_norm": 0.19346101582050323, | |
| "learning_rate": 9.679062447941778e-06, | |
| "loss": 0.0054, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 12.698412698412698, | |
| "grad_norm": 0.21870972216129303, | |
| "learning_rate": 9.586187986225325e-06, | |
| "loss": 0.0056, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 12.714285714285714, | |
| "grad_norm": 0.18945957720279694, | |
| "learning_rate": 9.493713992787672e-06, | |
| "loss": 0.0056, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 12.73015873015873, | |
| "grad_norm": 0.25288915634155273, | |
| "learning_rate": 9.401641383971477e-06, | |
| "loss": 0.0067, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 12.746031746031747, | |
| "grad_norm": 0.15972785651683807, | |
| "learning_rate": 9.309971072142038e-06, | |
| "loss": 0.0041, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 12.761904761904763, | |
| "grad_norm": 0.2357502579689026, | |
| "learning_rate": 9.218703965678204e-06, | |
| "loss": 0.0059, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 12.777777777777779, | |
| "grad_norm": 0.23380345106124878, | |
| "learning_rate": 9.127840968963381e-06, | |
| "loss": 0.0072, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 12.793650793650794, | |
| "grad_norm": 0.13809677958488464, | |
| "learning_rate": 9.03738298237658e-06, | |
| "loss": 0.0046, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 12.80952380952381, | |
| "grad_norm": 0.26843348145484924, | |
| "learning_rate": 8.94733090228349e-06, | |
| "loss": 0.007, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 12.825396825396826, | |
| "grad_norm": 0.30479297041893005, | |
| "learning_rate": 8.857685621027568e-06, | |
| "loss": 0.0072, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 12.841269841269842, | |
| "grad_norm": 0.09838364273309708, | |
| "learning_rate": 8.768448026921245e-06, | |
| "loss": 0.0032, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 12.857142857142858, | |
| "grad_norm": 0.13536061346530914, | |
| "learning_rate": 8.67961900423711e-06, | |
| "loss": 0.0031, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 12.873015873015873, | |
| "grad_norm": 0.12725569307804108, | |
| "learning_rate": 8.591199433199126e-06, | |
| "loss": 0.0034, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 12.88888888888889, | |
| "grad_norm": 0.1910911351442337, | |
| "learning_rate": 8.503190189973914e-06, | |
| "loss": 0.0048, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 12.904761904761905, | |
| "grad_norm": 0.08065954595804214, | |
| "learning_rate": 8.415592146662104e-06, | |
| "loss": 0.0027, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 12.920634920634921, | |
| "grad_norm": 0.20949719846248627, | |
| "learning_rate": 8.328406171289621e-06, | |
| "loss": 0.0056, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 12.936507936507937, | |
| "grad_norm": 0.11893566697835922, | |
| "learning_rate": 8.24163312779917e-06, | |
| "loss": 0.0036, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 12.952380952380953, | |
| "grad_norm": 0.28728553652763367, | |
| "learning_rate": 8.155273876041614e-06, | |
| "loss": 0.0098, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 12.968253968253968, | |
| "grad_norm": 0.2053646296262741, | |
| "learning_rate": 8.069329271767484e-06, | |
| "loss": 0.0057, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 12.984126984126984, | |
| "grad_norm": 0.186600461602211, | |
| "learning_rate": 7.983800166618482e-06, | |
| "loss": 0.0044, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 0.18637099862098694, | |
| "learning_rate": 7.898687408119065e-06, | |
| "loss": 0.0034, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 13.015873015873016, | |
| "grad_norm": 0.23288948833942413, | |
| "learning_rate": 7.813991839667995e-06, | |
| "loss": 0.006, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 13.031746031746032, | |
| "grad_norm": 0.11603759229183197, | |
| "learning_rate": 7.72971430053005e-06, | |
| "loss": 0.0032, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 13.047619047619047, | |
| "grad_norm": 0.11359909176826477, | |
| "learning_rate": 7.645855625827658e-06, | |
| "loss": 0.0036, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 13.063492063492063, | |
| "grad_norm": 0.1750001609325409, | |
| "learning_rate": 7.56241664653266e-06, | |
| "loss": 0.0047, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 13.079365079365079, | |
| "grad_norm": 0.08407314866781235, | |
| "learning_rate": 7.4793981894580034e-06, | |
| "loss": 0.003, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 13.095238095238095, | |
| "grad_norm": 0.15450453758239746, | |
| "learning_rate": 7.396801077249676e-06, | |
| "loss": 0.004, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 13.11111111111111, | |
| "grad_norm": 0.1506980061531067, | |
| "learning_rate": 7.3146261283784104e-06, | |
| "loss": 0.004, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 13.126984126984127, | |
| "grad_norm": 0.0932818278670311, | |
| "learning_rate": 7.2328741571316696e-06, | |
| "loss": 0.0028, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 13.142857142857142, | |
| "grad_norm": 0.1964637041091919, | |
| "learning_rate": 7.1515459736055505e-06, | |
| "loss": 0.0047, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 13.158730158730158, | |
| "grad_norm": 0.11378604173660278, | |
| "learning_rate": 7.070642383696763e-06, | |
| "loss": 0.0036, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 13.174603174603174, | |
| "grad_norm": 0.07380079478025436, | |
| "learning_rate": 6.990164189094589e-06, | |
| "loss": 0.0024, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 13.19047619047619, | |
| "grad_norm": 0.11589548736810684, | |
| "learning_rate": 6.910112187273066e-06, | |
| "loss": 0.0036, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 13.206349206349206, | |
| "grad_norm": 0.2268502563238144, | |
| "learning_rate": 6.830487171482935e-06, | |
| "loss": 0.0065, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 13.222222222222221, | |
| "grad_norm": 0.1941031664609909, | |
| "learning_rate": 6.751289930743882e-06, | |
| "loss": 0.0043, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 13.238095238095237, | |
| "grad_norm": 0.14726468920707703, | |
| "learning_rate": 6.6725212498366885e-06, | |
| "loss": 0.0044, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 13.253968253968253, | |
| "grad_norm": 0.09331656992435455, | |
| "learning_rate": 6.594181909295427e-06, | |
| "loss": 0.003, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 13.26984126984127, | |
| "grad_norm": 0.1862584948539734, | |
| "learning_rate": 6.516272685399788e-06, | |
| "loss": 0.005, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 13.285714285714286, | |
| "grad_norm": 0.14406166970729828, | |
| "learning_rate": 6.438794350167337e-06, | |
| "loss": 0.0039, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 13.301587301587302, | |
| "grad_norm": 0.09058280289173126, | |
| "learning_rate": 6.36174767134588e-06, | |
| "loss": 0.0033, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 13.317460317460318, | |
| "grad_norm": 0.1405523121356964, | |
| "learning_rate": 6.285133412405858e-06, | |
| "loss": 0.0035, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 13.333333333333334, | |
| "grad_norm": 0.14805886149406433, | |
| "learning_rate": 6.20895233253278e-06, | |
| "loss": 0.0046, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 13.34920634920635, | |
| "grad_norm": 0.2134266048669815, | |
| "learning_rate": 6.133205186619695e-06, | |
| "loss": 0.0064, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 13.365079365079366, | |
| "grad_norm": 0.09715571254491806, | |
| "learning_rate": 6.057892725259717e-06, | |
| "loss": 0.003, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 13.380952380952381, | |
| "grad_norm": 0.16215340793132782, | |
| "learning_rate": 5.983015694738597e-06, | |
| "loss": 0.0052, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 13.396825396825397, | |
| "grad_norm": 0.12831249833106995, | |
| "learning_rate": 5.908574837027309e-06, | |
| "loss": 0.0035, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 13.412698412698413, | |
| "grad_norm": 0.23066161572933197, | |
| "learning_rate": 5.83457088977471e-06, | |
| "loss": 0.006, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 13.428571428571429, | |
| "grad_norm": 0.16153094172477722, | |
| "learning_rate": 5.761004586300234e-06, | |
| "loss": 0.0032, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 13.444444444444445, | |
| "grad_norm": 0.1263124942779541, | |
| "learning_rate": 5.687876655586583e-06, | |
| "loss": 0.0033, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 13.46031746031746, | |
| "grad_norm": 0.22407254576683044, | |
| "learning_rate": 5.615187822272583e-06, | |
| "loss": 0.0069, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 13.476190476190476, | |
| "grad_norm": 0.2908068001270294, | |
| "learning_rate": 5.542938806645931e-06, | |
| "loss": 0.0087, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 13.492063492063492, | |
| "grad_norm": 0.18590912222862244, | |
| "learning_rate": 5.4711303246361144e-06, | |
| "loss": 0.0048, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 13.507936507936508, | |
| "grad_norm": 0.17562605440616608, | |
| "learning_rate": 5.399763087807236e-06, | |
| "loss": 0.0044, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 13.523809523809524, | |
| "grad_norm": 0.07766014337539673, | |
| "learning_rate": 5.328837803351083e-06, | |
| "loss": 0.0025, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 13.53968253968254, | |
| "grad_norm": 0.16575992107391357, | |
| "learning_rate": 5.258355174079993e-06, | |
| "loss": 0.0045, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 13.555555555555555, | |
| "grad_norm": 0.1963498741388321, | |
| "learning_rate": 5.188315898419971e-06, | |
| "loss": 0.0061, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 13.571428571428571, | |
| "grad_norm": 0.2301764041185379, | |
| "learning_rate": 5.118720670403748e-06, | |
| "loss": 0.0051, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 13.587301587301587, | |
| "grad_norm": 0.16544826328754425, | |
| "learning_rate": 5.04957017966391e-06, | |
| "loss": 0.0033, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 13.603174603174603, | |
| "grad_norm": 0.17942006886005402, | |
| "learning_rate": 4.980865111426003e-06, | |
| "loss": 0.004, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 13.619047619047619, | |
| "grad_norm": 0.14243295788764954, | |
| "learning_rate": 4.912606146501886e-06, | |
| "loss": 0.0035, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 13.634920634920634, | |
| "grad_norm": 0.14227573573589325, | |
| "learning_rate": 4.844793961282812e-06, | |
| "loss": 0.0034, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 13.65079365079365, | |
| "grad_norm": 0.14716386795043945, | |
| "learning_rate": 4.777429227732844e-06, | |
| "loss": 0.0033, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 13.666666666666666, | |
| "grad_norm": 0.2278168946504593, | |
| "learning_rate": 4.710512613382151e-06, | |
| "loss": 0.006, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 13.682539682539682, | |
| "grad_norm": 0.2408359944820404, | |
| "learning_rate": 4.644044781320422e-06, | |
| "loss": 0.0072, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 13.698412698412698, | |
| "grad_norm": 0.23890067636966705, | |
| "learning_rate": 4.578026390190232e-06, | |
| "loss": 0.0051, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 13.714285714285714, | |
| "grad_norm": 0.2770053446292877, | |
| "learning_rate": 4.5124580941806165e-06, | |
| "loss": 0.0078, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 13.73015873015873, | |
| "grad_norm": 0.16485559940338135, | |
| "learning_rate": 4.447340543020473e-06, | |
| "loss": 0.0033, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 13.746031746031747, | |
| "grad_norm": 0.1674467772245407, | |
| "learning_rate": 4.382674381972224e-06, | |
| "loss": 0.0041, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 13.761904761904763, | |
| "grad_norm": 0.09436249732971191, | |
| "learning_rate": 4.318460251825357e-06, | |
| "loss": 0.0031, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 13.777777777777779, | |
| "grad_norm": 0.0673573687672615, | |
| "learning_rate": 4.254698788890127e-06, | |
| "loss": 0.0025, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 13.793650793650794, | |
| "grad_norm": 0.18255870044231415, | |
| "learning_rate": 4.191390624991159e-06, | |
| "loss": 0.0047, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 13.80952380952381, | |
| "grad_norm": 0.13948306441307068, | |
| "learning_rate": 4.12853638746134e-06, | |
| "loss": 0.0032, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 13.825396825396826, | |
| "grad_norm": 0.24183286726474762, | |
| "learning_rate": 4.0661366991354365e-06, | |
| "loss": 0.0063, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 13.841269841269842, | |
| "grad_norm": 0.10236512869596481, | |
| "learning_rate": 4.004192178344029e-06, | |
| "loss": 0.003, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 13.857142857142858, | |
| "grad_norm": 0.10468772053718567, | |
| "learning_rate": 3.942703438907358e-06, | |
| "loss": 0.003, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 13.873015873015873, | |
| "grad_norm": 0.1839323341846466, | |
| "learning_rate": 3.881671090129247e-06, | |
| "loss": 0.0047, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 13.88888888888889, | |
| "grad_norm": 0.245498925447464, | |
| "learning_rate": 3.821095736791008e-06, | |
| "loss": 0.0063, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 13.904761904761905, | |
| "grad_norm": 0.08903949707746506, | |
| "learning_rate": 3.7609779791455744e-06, | |
| "loss": 0.0027, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 13.920634920634921, | |
| "grad_norm": 0.10096840560436249, | |
| "learning_rate": 3.7013184129113976e-06, | |
| "loss": 0.0025, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 13.936507936507937, | |
| "grad_norm": 0.16196174919605255, | |
| "learning_rate": 3.6421176292666783e-06, | |
| "loss": 0.0049, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 13.952380952380953, | |
| "grad_norm": 0.2010921686887741, | |
| "learning_rate": 3.58337621484342e-06, | |
| "loss": 0.0047, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 13.968253968253968, | |
| "grad_norm": 0.20379731059074402, | |
| "learning_rate": 3.525094751721669e-06, | |
| "loss": 0.0049, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 13.984126984126984, | |
| "grad_norm": 0.13787353038787842, | |
| "learning_rate": 3.4672738174236884e-06, | |
| "loss": 0.0038, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "grad_norm": 0.21144546568393707, | |
| "learning_rate": 3.4099139849083307e-06, | |
| "loss": 0.0058, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 14.015873015873016, | |
| "grad_norm": 0.16598111391067505, | |
| "learning_rate": 3.353015822565253e-06, | |
| "loss": 0.0046, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 14.031746031746032, | |
| "grad_norm": 0.11316211521625519, | |
| "learning_rate": 3.296579894209345e-06, | |
| "loss": 0.0033, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 14.047619047619047, | |
| "grad_norm": 0.1642863005399704, | |
| "learning_rate": 3.2406067590751433e-06, | |
| "loss": 0.0042, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 14.063492063492063, | |
| "grad_norm": 0.06231338158249855, | |
| "learning_rate": 3.1850969718112745e-06, | |
| "loss": 0.0023, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 14.079365079365079, | |
| "grad_norm": 0.07541368156671524, | |
| "learning_rate": 3.1300510824749273e-06, | |
| "loss": 0.0027, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 14.095238095238095, | |
| "grad_norm": 0.10008185356855392, | |
| "learning_rate": 3.0754696365265068e-06, | |
| "loss": 0.0028, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 14.11111111111111, | |
| "grad_norm": 0.10464094579219818, | |
| "learning_rate": 3.0213531748240764e-06, | |
| "loss": 0.0031, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 14.126984126984127, | |
| "grad_norm": 0.09949090331792831, | |
| "learning_rate": 2.9677022336181413e-06, | |
| "loss": 0.003, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 14.142857142857142, | |
| "grad_norm": 0.08555309474468231, | |
| "learning_rate": 2.914517344546258e-06, | |
| "loss": 0.0028, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 14.158730158730158, | |
| "grad_norm": 0.10682200640439987, | |
| "learning_rate": 2.8617990346277657e-06, | |
| "loss": 0.0032, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 14.174603174603174, | |
| "grad_norm": 0.09806779026985168, | |
| "learning_rate": 2.8095478262585907e-06, | |
| "loss": 0.0035, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 14.19047619047619, | |
| "grad_norm": 0.13682028651237488, | |
| "learning_rate": 2.7577642372060673e-06, | |
| "loss": 0.003, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 14.206349206349206, | |
| "grad_norm": 0.1651875525712967, | |
| "learning_rate": 2.7064487806037985e-06, | |
| "loss": 0.0043, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 14.222222222222221, | |
| "grad_norm": 0.14128713309764862, | |
| "learning_rate": 2.6556019649465525e-06, | |
| "loss": 0.0032, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 14.238095238095237, | |
| "grad_norm": 0.18472391366958618, | |
| "learning_rate": 2.6052242940852787e-06, | |
| "loss": 0.0055, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 14.253968253968253, | |
| "grad_norm": 0.12015866488218307, | |
| "learning_rate": 2.5553162672220465e-06, | |
| "loss": 0.0027, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 14.26984126984127, | |
| "grad_norm": 0.20532798767089844, | |
| "learning_rate": 2.5058783789051467e-06, | |
| "loss": 0.006, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 14.285714285714286, | |
| "grad_norm": 0.11659039556980133, | |
| "learning_rate": 2.45691111902418e-06, | |
| "loss": 0.0033, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 14.301587301587302, | |
| "grad_norm": 0.22555606067180634, | |
| "learning_rate": 2.4084149728051952e-06, | |
| "loss": 0.0057, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 14.317460317460318, | |
| "grad_norm": 0.10364361107349396, | |
| "learning_rate": 2.360390420805869e-06, | |
| "loss": 0.003, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 14.333333333333334, | |
| "grad_norm": 0.15920886397361755, | |
| "learning_rate": 2.3128379389108e-06, | |
| "loss": 0.0039, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 14.34920634920635, | |
| "grad_norm": 0.10381603240966797, | |
| "learning_rate": 2.2657579983267064e-06, | |
| "loss": 0.0028, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 14.365079365079366, | |
| "grad_norm": 0.17512689530849457, | |
| "learning_rate": 2.219151065577829e-06, | |
| "loss": 0.0046, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 14.380952380952381, | |
| "grad_norm": 0.22503690421581268, | |
| "learning_rate": 2.1730176025012816e-06, | |
| "loss": 0.0063, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 14.396825396825397, | |
| "grad_norm": 0.17018793523311615, | |
| "learning_rate": 2.1273580662424796e-06, | |
| "loss": 0.0048, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 14.412698412698413, | |
| "grad_norm": 0.22725212574005127, | |
| "learning_rate": 2.082172909250568e-06, | |
| "loss": 0.0069, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 14.428571428571429, | |
| "grad_norm": 0.12136708199977875, | |
| "learning_rate": 2.0374625792740464e-06, | |
| "loss": 0.003, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 14.444444444444445, | |
| "grad_norm": 0.06128573417663574, | |
| "learning_rate": 1.993227519356189e-06, | |
| "loss": 0.0022, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 14.46031746031746, | |
| "grad_norm": 0.0800539031624794, | |
| "learning_rate": 1.9494681678307703e-06, | |
| "loss": 0.0025, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 14.476190476190476, | |
| "grad_norm": 0.2250363528728485, | |
| "learning_rate": 1.906184958317664e-06, | |
| "loss": 0.0056, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 14.492063492063492, | |
| "grad_norm": 0.06574003398418427, | |
| "learning_rate": 1.8633783197185783e-06, | |
| "loss": 0.0025, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 14.507936507936508, | |
| "grad_norm": 0.1733701229095459, | |
| "learning_rate": 1.8210486762127499e-06, | |
| "loss": 0.0052, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 14.523809523809524, | |
| "grad_norm": 0.14052851498126984, | |
| "learning_rate": 1.7791964472528232e-06, | |
| "loss": 0.0035, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 14.53968253968254, | |
| "grad_norm": 0.20883136987686157, | |
| "learning_rate": 1.737822047560611e-06, | |
| "loss": 0.006, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 14.555555555555555, | |
| "grad_norm": 0.18126244843006134, | |
| "learning_rate": 1.696925887123052e-06, | |
| "loss": 0.0043, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 14.571428571428571, | |
| "grad_norm": 0.19093488156795502, | |
| "learning_rate": 1.656508371188109e-06, | |
| "loss": 0.0045, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 14.587301587301587, | |
| "grad_norm": 0.16476662456989288, | |
| "learning_rate": 1.6165699002607671e-06, | |
| "loss": 0.0037, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 14.603174603174603, | |
| "grad_norm": 0.12128468602895737, | |
| "learning_rate": 1.5771108700990412e-06, | |
| "loss": 0.0034, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 14.619047619047619, | |
| "grad_norm": 0.07109358161687851, | |
| "learning_rate": 1.538131671710108e-06, | |
| "loss": 0.0027, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 14.634920634920634, | |
| "grad_norm": 0.12868039309978485, | |
| "learning_rate": 1.4996326913463754e-06, | |
| "loss": 0.0042, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 14.65079365079365, | |
| "grad_norm": 0.10166194289922714, | |
| "learning_rate": 1.461614310501691e-06, | |
| "loss": 0.0027, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 14.666666666666666, | |
| "grad_norm": 0.1676546037197113, | |
| "learning_rate": 1.4240769059075342e-06, | |
| "loss": 0.0045, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 14.682539682539682, | |
| "grad_norm": 0.16010187566280365, | |
| "learning_rate": 1.387020849529319e-06, | |
| "loss": 0.0048, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 14.698412698412698, | |
| "grad_norm": 0.22581593692302704, | |
| "learning_rate": 1.3504465085626638e-06, | |
| "loss": 0.0076, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 14.714285714285714, | |
| "grad_norm": 0.1580013781785965, | |
| "learning_rate": 1.3143542454297885e-06, | |
| "loss": 0.004, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 14.73015873015873, | |
| "grad_norm": 0.2010050266981125, | |
| "learning_rate": 1.2787444177759068e-06, | |
| "loss": 0.0058, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 14.746031746031747, | |
| "grad_norm": 0.2182077020406723, | |
| "learning_rate": 1.243617378465689e-06, | |
| "loss": 0.0053, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 14.761904761904763, | |
| "grad_norm": 0.23424509167671204, | |
| "learning_rate": 1.208973475579761e-06, | |
| "loss": 0.0055, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 14.777777777777779, | |
| "grad_norm": 0.1593056619167328, | |
| "learning_rate": 1.1748130524112666e-06, | |
| "loss": 0.0038, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 14.793650793650794, | |
| "grad_norm": 0.1183331161737442, | |
| "learning_rate": 1.1411364474624264e-06, | |
| "loss": 0.0035, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 14.80952380952381, | |
| "grad_norm": 0.1267019659280777, | |
| "learning_rate": 1.1079439944412406e-06, | |
| "loss": 0.0037, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 14.825396825396826, | |
| "grad_norm": 0.1250416487455368, | |
| "learning_rate": 1.075236022258147e-06, | |
| "loss": 0.0033, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 14.841269841269842, | |
| "grad_norm": 0.19605623185634613, | |
| "learning_rate": 1.0430128550227625e-06, | |
| "loss": 0.0034, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 14.857142857142858, | |
| "grad_norm": 0.18377277255058289, | |
| "learning_rate": 1.0112748120406856e-06, | |
| "loss": 0.007, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 14.873015873015873, | |
| "grad_norm": 0.1912008672952652, | |
| "learning_rate": 9.800222078103271e-07, | |
| "loss": 0.0042, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 14.88888888888889, | |
| "grad_norm": 0.1927856057882309, | |
| "learning_rate": 9.492553520197733e-07, | |
| "loss": 0.0055, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 14.904761904761905, | |
| "grad_norm": 0.103274405002594, | |
| "learning_rate": 9.189745495437608e-07, | |
| "loss": 0.0034, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 14.920634920634921, | |
| "grad_norm": 0.1846938282251358, | |
| "learning_rate": 8.891801004406119e-07, | |
| "loss": 0.0047, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 14.936507936507937, | |
| "grad_norm": 0.12870021164417267, | |
| "learning_rate": 8.59872299949288e-07, | |
| "loss": 0.0028, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 14.952380952380953, | |
| "grad_norm": 0.09814100712537766, | |
| "learning_rate": 8.31051438486441e-07, | |
| "loss": 0.0027, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 14.968253968253968, | |
| "grad_norm": 0.12259647250175476, | |
| "learning_rate": 8.027178016435765e-07, | |
| "loss": 0.003, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 14.984126984126984, | |
| "grad_norm": 0.2572350800037384, | |
| "learning_rate": 7.748716701841685e-07, | |
| "loss": 0.006, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 0.26040682196617126, | |
| "learning_rate": 7.475133200409212e-07, | |
| "loss": 0.0048, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 15.015873015873016, | |
| "grad_norm": 0.1166323646903038, | |
| "learning_rate": 7.206430223130278e-07, | |
| "loss": 0.0028, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 15.031746031746032, | |
| "grad_norm": 0.11518598347902298, | |
| "learning_rate": 6.9426104326345e-07, | |
| "loss": 0.0031, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 15.047619047619047, | |
| "grad_norm": 0.18673783540725708, | |
| "learning_rate": 6.683676443163311e-07, | |
| "loss": 0.0048, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 15.063492063492063, | |
| "grad_norm": 0.1127839982509613, | |
| "learning_rate": 6.429630820543598e-07, | |
| "loss": 0.0031, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 15.079365079365079, | |
| "grad_norm": 0.18263711035251617, | |
| "learning_rate": 6.180476082162656e-07, | |
| "loss": 0.004, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 15.095238095238095, | |
| "grad_norm": 0.1486678421497345, | |
| "learning_rate": 5.936214696942887e-07, | |
| "loss": 0.0037, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 15.11111111111111, | |
| "grad_norm": 0.2178022712469101, | |
| "learning_rate": 5.696849085317646e-07, | |
| "loss": 0.0057, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 15.126984126984127, | |
| "grad_norm": 0.12073294818401337, | |
| "learning_rate": 5.462381619207091e-07, | |
| "loss": 0.0031, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 15.142857142857142, | |
| "grad_norm": 0.12311496585607529, | |
| "learning_rate": 5.232814621994598e-07, | |
| "loss": 0.0036, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 15.158730158730158, | |
| "grad_norm": 0.16713330149650574, | |
| "learning_rate": 5.008150368503994e-07, | |
| "loss": 0.0038, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 15.174603174603174, | |
| "grad_norm": 0.1170608177781105, | |
| "learning_rate": 4.788391084976862e-07, | |
| "loss": 0.0033, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 15.19047619047619, | |
| "grad_norm": 0.06233490630984306, | |
| "learning_rate": 4.573538949050327e-07, | |
| "loss": 0.0023, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 15.206349206349206, | |
| "grad_norm": 0.13149504363536835, | |
| "learning_rate": 4.363596089735911e-07, | |
| "loss": 0.0031, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 15.222222222222221, | |
| "grad_norm": 0.16984321177005768, | |
| "learning_rate": 4.1585645873978284e-07, | |
| "loss": 0.0046, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 15.238095238095237, | |
| "grad_norm": 0.14544299244880676, | |
| "learning_rate": 3.958446473733002e-07, | |
| "loss": 0.0033, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 15.253968253968253, | |
| "grad_norm": 0.17623476684093475, | |
| "learning_rate": 3.7632437317505207e-07, | |
| "loss": 0.0046, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 15.26984126984127, | |
| "grad_norm": 0.12246549874544144, | |
| "learning_rate": 3.572958295752049e-07, | |
| "loss": 0.0034, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 15.285714285714286, | |
| "grad_norm": 0.14989396929740906, | |
| "learning_rate": 3.387592051312782e-07, | |
| "loss": 0.0036, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 15.301587301587302, | |
| "grad_norm": 0.19900646805763245, | |
| "learning_rate": 3.207146835262742e-07, | |
| "loss": 0.0057, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 15.317460317460318, | |
| "grad_norm": 0.1741442084312439, | |
| "learning_rate": 3.0316244356683454e-07, | |
| "loss": 0.0047, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 15.333333333333334, | |
| "grad_norm": 0.15245862305164337, | |
| "learning_rate": 2.8610265918151414e-07, | |
| "loss": 0.0046, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 15.34920634920635, | |
| "grad_norm": 0.19708728790283203, | |
| "learning_rate": 2.695354994190047e-07, | |
| "loss": 0.0058, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 15.365079365079366, | |
| "grad_norm": 0.13684900104999542, | |
| "learning_rate": 2.534611284465083e-07, | |
| "loss": 0.0037, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 15.380952380952381, | |
| "grad_norm": 0.18838024139404297, | |
| "learning_rate": 2.3787970554806084e-07, | |
| "loss": 0.0043, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 15.396825396825397, | |
| "grad_norm": 0.18869999051094055, | |
| "learning_rate": 2.2279138512300567e-07, | |
| "loss": 0.0056, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 15.412698412698413, | |
| "grad_norm": 0.14952099323272705, | |
| "learning_rate": 2.0819631668442253e-07, | |
| "loss": 0.0038, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 15.428571428571429, | |
| "grad_norm": 0.20797456800937653, | |
| "learning_rate": 1.940946448576675e-07, | |
| "loss": 0.0056, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 15.444444444444445, | |
| "grad_norm": 0.17077018320560455, | |
| "learning_rate": 1.8048650937893542e-07, | |
| "loss": 0.0049, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 15.46031746031746, | |
| "grad_norm": 0.16229721903800964, | |
| "learning_rate": 1.6737204509387206e-07, | |
| "loss": 0.0038, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 15.476190476190476, | |
| "grad_norm": 0.06878882646560669, | |
| "learning_rate": 1.5475138195623629e-07, | |
| "loss": 0.0024, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 15.492063492063492, | |
| "grad_norm": 0.1002248004078865, | |
| "learning_rate": 1.4262464502663443e-07, | |
| "loss": 0.0028, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 15.507936507936508, | |
| "grad_norm": 0.1598724126815796, | |
| "learning_rate": 1.309919544712268e-07, | |
| "loss": 0.0051, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 15.523809523809524, | |
| "grad_norm": 0.20375491678714752, | |
| "learning_rate": 1.1985342556060652e-07, | |
| "loss": 0.0048, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 15.53968253968254, | |
| "grad_norm": 0.15220007300376892, | |
| "learning_rate": 1.0920916866861142e-07, | |
| "loss": 0.0037, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 15.555555555555555, | |
| "grad_norm": 0.13166747987270355, | |
| "learning_rate": 9.905928927123609e-08, | |
| "loss": 0.0041, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 15.571428571428571, | |
| "grad_norm": 0.16521938145160675, | |
| "learning_rate": 8.940388794559939e-08, | |
| "loss": 0.0043, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 15.587301587301587, | |
| "grad_norm": 0.22669538855552673, | |
| "learning_rate": 8.02430603689397e-08, | |
| "loss": 0.006, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 15.603174603174603, | |
| "grad_norm": 0.09708595275878906, | |
| "learning_rate": 7.157689731767669e-08, | |
| "loss": 0.0025, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 15.619047619047619, | |
| "grad_norm": 0.2131219506263733, | |
| "learning_rate": 6.340548466648443e-08, | |
| "loss": 0.0051, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 15.634920634920634, | |
| "grad_norm": 0.1999976485967636, | |
| "learning_rate": 5.572890338748082e-08, | |
| "loss": 0.0046, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 15.65079365079365, | |
| "grad_norm": 0.10222487151622772, | |
| "learning_rate": 4.8547229549383844e-08, | |
| "loss": 0.0037, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 15.666666666666666, | |
| "grad_norm": 0.25009259581565857, | |
| "learning_rate": 4.186053431680104e-08, | |
| "loss": 0.0068, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 15.682539682539682, | |
| "grad_norm": 0.06356369704008102, | |
| "learning_rate": 3.566888394948009e-08, | |
| "loss": 0.0022, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 15.698412698412698, | |
| "grad_norm": 0.13318653404712677, | |
| "learning_rate": 2.997233980168157e-08, | |
| "loss": 0.0038, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 15.714285714285714, | |
| "grad_norm": 0.05918239429593086, | |
| "learning_rate": 2.4770958321568283e-08, | |
| "loss": 0.0022, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 15.73015873015873, | |
| "grad_norm": 0.1082151010632515, | |
| "learning_rate": 2.0064791050633526e-08, | |
| "loss": 0.0031, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 15.746031746031747, | |
| "grad_norm": 0.22153517603874207, | |
| "learning_rate": 1.5853884623195925e-08, | |
| "loss": 0.0049, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 15.761904761904763, | |
| "grad_norm": 0.09333167225122452, | |
| "learning_rate": 1.2138280765944254e-08, | |
| "loss": 0.0028, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 15.777777777777779, | |
| "grad_norm": 0.14806319773197174, | |
| "learning_rate": 8.918016297515541e-09, | |
| "loss": 0.0031, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 15.793650793650794, | |
| "grad_norm": 0.15807633101940155, | |
| "learning_rate": 6.193123128134248e-09, | |
| "loss": 0.0041, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 15.80952380952381, | |
| "grad_norm": 0.1491064578294754, | |
| "learning_rate": 3.963628259290308e-09, | |
| "loss": 0.0039, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 15.825396825396826, | |
| "grad_norm": 0.1288636475801468, | |
| "learning_rate": 2.229553783478222e-09, | |
| "loss": 0.0035, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 15.841269841269842, | |
| "grad_norm": 0.17619061470031738, | |
| "learning_rate": 9.90916883986115e-10, | |
| "loss": 0.0056, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 15.857142857142858, | |
| "grad_norm": 0.1407734900712967, | |
| "learning_rate": 2.477298346958978e-10, | |
| "loss": 0.0038, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 15.873015873015873, | |
| "grad_norm": 0.172795370221138, | |
| "learning_rate": 0.0, | |
| "loss": 0.0048, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 15.873015873015873, | |
| "step": 1000, | |
| "total_flos": 1.610662192030679e+17, | |
| "train_loss": 0.14257763476669788, | |
| "train_runtime": 58489.782, | |
| "train_samples_per_second": 0.274, | |
| "train_steps_per_second": 0.017 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 16, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.610662192030679e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |