| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9961389961389961, | |
| "global_step": 129, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.285, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2933, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2853, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.2804, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.2834, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2e-05, | |
| "loss": 0.2805, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9996738360808566e-05, | |
| "loss": 0.2805, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9986955570892302e-05, | |
| "loss": 0.2906, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9970658011837404e-05, | |
| "loss": 0.2822, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.994785631499535e-05, | |
| "loss": 0.2994, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9918565354547738e-05, | |
| "loss": 0.3013, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9882804237803487e-05, | |
| "loss": 0.2954, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9840596292734573e-05, | |
| "loss": 0.2929, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9791969052758563e-05, | |
| "loss": 0.2956, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9736954238777793e-05, | |
| "loss": 0.3002, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9675587738486935e-05, | |
| "loss": 0.2918, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9607909582962478e-05, | |
| "loss": 0.2791, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9533963920549307e-05, | |
| "loss": 0.2777, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9453798988061535e-05, | |
| "loss": 0.2885, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.936746707931628e-05, | |
| "loss": 0.2957, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.927502451102095e-05, | |
| "loss": 0.2913, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9176531586036282e-05, | |
| "loss": 0.2945, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9072052554039123e-05, | |
| "loss": 0.2802, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8961655569610557e-05, | |
| "loss": 0.2885, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8845412647776795e-05, | |
| "loss": 0.2888, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8723399617031754e-05, | |
| "loss": 0.2772, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8595696069872013e-05, | |
| "loss": 0.2705, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8462385310876444e-05, | |
| "loss": 0.2831, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8323554302364273e-05, | |
| "loss": 0.28, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8179293607667177e-05, | |
| "loss": 0.28, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8029697332052277e-05, | |
| "loss": 0.2819, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7874863061334658e-05, | |
| "loss": 0.2847, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7714891798219432e-05, | |
| "loss": 0.2809, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7549887896414853e-05, | |
| "loss": 0.2666, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.7379958992559494e-05, | |
| "loss": 0.2837, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.720521593600787e-05, | |
| "loss": 0.2822, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7025772716520324e-05, | |
| "loss": 0.2764, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6841746389904306e-05, | |
| "loss": 0.2795, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.6653257001655652e-05, | |
| "loss": 0.2686, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.6460427508649546e-05, | |
| "loss": 0.2655, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6263383698932307e-05, | |
| "loss": 0.2715, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6062254109666383e-05, | |
| "loss": 0.276, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5857169943281948e-05, | |
| "loss": 0.274, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5648264981889936e-05, | |
| "loss": 0.2831, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.5435675500012212e-05, | |
| "loss": 0.2685, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.5219540175685938e-05, | |
| "loss": 0.2746, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.2624, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.477719818512263e-05, | |
| "loss": 0.2709, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.4551280070880089e-05, | |
| "loss": 0.2801, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.432239302994747e-05, | |
| "loss": 0.2724, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.4090686371713403e-05, | |
| "loss": 0.2661, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.385631124488136e-05, | |
| "loss": 0.262, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.361942053887118e-05, | |
| "loss": 0.2653, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3380168784085028e-05, | |
| "loss": 0.2618, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.3138712051102908e-05, | |
| "loss": 0.2619, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2895207848873488e-05, | |
| "loss": 0.2597, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.264981502196662e-05, | |
| "loss": 0.2592, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.2402693646954607e-05, | |
| "loss": 0.2658, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.2154004927989815e-05, | |
| "loss": 0.2535, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1903911091646684e-05, | |
| "loss": 0.2711, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.165257528109685e-05, | |
| "loss": 0.2601, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.1400161449686293e-05, | |
| "loss": 0.2532, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.1146834253984008e-05, | |
| "loss": 0.2621, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0892758946371943e-05, | |
| "loss": 0.2612, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0638101267246283e-05, | |
| "loss": 0.2521, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.0383027336900356e-05, | |
| "loss": 0.2611, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.012770354715974e-05, | |
| "loss": 0.256, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.872296452840266e-06, | |
| "loss": 0.2584, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.616972663099648e-06, | |
| "loss": 0.2502, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.361898732753715e-06, | |
| "loss": 0.2384, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.107241053628058e-06, | |
| "loss": 0.258, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.853165746015997e-06, | |
| "loss": 0.2536, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.599838550313714e-06, | |
| "loss": 0.263, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.347424718903152e-06, | |
| "loss": 0.247, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.096088908353316e-06, | |
| "loss": 0.2523, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.845995072010188e-06, | |
| "loss": 0.2454, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.597306353045393e-06, | |
| "loss": 0.2618, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.350184978033386e-06, | |
| "loss": 0.2536, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.104792151126515e-06, | |
| "loss": 0.2454, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.861287948897091e-06, | |
| "loss": 0.2636, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.619831215914974e-06, | |
| "loss": 0.2415, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.38057946112882e-06, | |
| "loss": 0.2415, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.1436887551186466e-06, | |
| "loss": 0.248, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.9093136282866014e-06, | |
| "loss": 0.247, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.67760697005253e-06, | |
| "loss": 0.245, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.448719929119916e-06, | |
| "loss": 0.2298, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.22280181487737e-06, | |
| "loss": 0.2465, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.2403, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.7804598243140664e-06, | |
| "loss": 0.242, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.56432449998779e-06, | |
| "loss": 0.2402, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.351735018110066e-06, | |
| "loss": 0.2423, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.142830056718052e-06, | |
| "loss": 0.2453, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.937745890333623e-06, | |
| "loss": 0.2467, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.7366163010676937e-06, | |
| "loss": 0.2354, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.5395724913504546e-06, | |
| "loss": 0.2497, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.3467429983443477e-06, | |
| "loss": 0.2362, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.1582536100956973e-06, | |
| "loss": 0.2444, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.9742272834796813e-06, | |
| "loss": 0.2495, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.7947840639921308e-06, | |
| "loss": 0.2242, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.620041007440508e-06, | |
| "loss": 0.2356, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.4501121035851494e-06, | |
| "loss": 0.2369, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2851082017805704e-06, | |
| "loss": 0.2424, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1251369386653454e-06, | |
| "loss": 0.248, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.9703026679477253e-06, | |
| "loss": 0.2354, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.820706392332824e-06, | |
| "loss": 0.2518, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.6764456976357279e-06, | |
| "loss": 0.2402, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.53761468912356e-06, | |
| "loss": 0.2509, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4043039301279904e-06, | |
| "loss": 0.2446, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2766003829682504e-06, | |
| "loss": 0.2343, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.1545873522232055e-06, | |
| "loss": 0.2363, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.0383444303894453e-06, | |
| "loss": 0.2332, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.279474459608806e-07, | |
| "loss": 0.2388, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.234684139637205e-07, | |
| "loss": 0.2414, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.249754889790539e-07, | |
| "loss": 0.2407, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.325329206837217e-07, | |
| "loss": 0.2363, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.462010119384665e-07, | |
| "loss": 0.2463, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.660360794506946e-07, | |
| "loss": 0.2328, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.920904170375239e-07, | |
| "loss": 0.2308, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.2441226151306403e-07, | |
| "loss": 0.2474, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.6304576122221035e-07, | |
| "loss": 0.2353, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.0803094724143879e-07, | |
| "loss": 0.2331, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.5940370726542864e-07, | |
| "loss": 0.2374, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1719576219651585e-07, | |
| "loss": 0.2279, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.143464545226298e-08, | |
| "loss": 0.2498, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.214368500465305e-08, | |
| "loss": 0.2452, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.9341988162595593e-08, | |
| "loss": 0.2374, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.3044429107700319e-08, | |
| "loss": 0.2405, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.2616391914364056e-09, | |
| "loss": 0.246, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2311, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 129, | |
| "total_flos": 1.0254156470912287e+18, | |
| "train_loss": 0.2606376674055129, | |
| "train_runtime": 20090.4454, | |
| "train_samples_per_second": 4.947, | |
| "train_steps_per_second": 0.006 | |
| } | |
| ], | |
| "max_steps": 129, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.0254156470912287e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |