| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 153, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.385727144574295, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 1.7526, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 7.407346233892389, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 1.7086, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.453240159782491, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.6948, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 5.231881985607868, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 1.6844, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 4.290775811530979, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 1.6481, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 4.865595819070934, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 1.559, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 4.787864348488922, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 1.5407, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 5.343516383725201, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 1.5289, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 5.876818436372016, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 1.4746, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 5.090544083750136, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 1.3829, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 4.736310405894691, | |
| "learning_rate": 2e-05, | |
| "loss": 1.4213, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 4.585247241293968, | |
| "learning_rate": 1.999867521457224e-05, | |
| "loss": 1.3315, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 4.247578099135077, | |
| "learning_rate": 1.9994701209300245e-05, | |
| "loss": 1.269, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 4.100439299332423, | |
| "learning_rate": 1.9988079037124866e-05, | |
| "loss": 1.237, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 3.900696491747024, | |
| "learning_rate": 1.9978810452637544e-05, | |
| "loss": 1.206, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 3.7254601213601424, | |
| "learning_rate": 1.9966897911615417e-05, | |
| "loss": 1.1384, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 3.4594232764317203, | |
| "learning_rate": 1.995234457037063e-05, | |
| "loss": 1.1137, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 3.479977282965165, | |
| "learning_rate": 1.9935154284914063e-05, | |
| "loss": 1.0582, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 3.089440166288491, | |
| "learning_rate": 1.991533160993366e-05, | |
| "loss": 1.0306, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 3.296297140236444, | |
| "learning_rate": 1.98928817975876e-05, | |
| "loss": 1.0031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 3.092912502793922, | |
| "learning_rate": 1.9867810796112742e-05, | |
| "loss": 1.0027, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.8301743306090144, | |
| "learning_rate": 1.9840125248248564e-05, | |
| "loss": 0.915, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 2.700584849840672, | |
| "learning_rate": 1.9809832489477144e-05, | |
| "loss": 0.8695, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.9985312999308578, | |
| "learning_rate": 1.9776940546079552e-05, | |
| "loss": 0.8793, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 2.865876357974003, | |
| "learning_rate": 1.9741458133009258e-05, | |
| "loss": 0.8702, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 2.7813509804899423, | |
| "learning_rate": 1.970339465158301e-05, | |
| "loss": 0.8633, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 2.778002439526667, | |
| "learning_rate": 1.9662760186989914e-05, | |
| "loss": 0.8495, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 2.65981269219971, | |
| "learning_rate": 1.9619565505619288e-05, | |
| "loss": 0.8463, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.7575466470931684, | |
| "learning_rate": 1.9573822052208013e-05, | |
| "loss": 0.8125, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 2.359003505133353, | |
| "learning_rate": 1.9525541946808187e-05, | |
| "loss": 0.7691, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.362088089814385, | |
| "learning_rate": 1.9474737981575833e-05, | |
| "loss": 0.7924, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 2.446279901118075, | |
| "learning_rate": 1.942142361738151e-05, | |
| "loss": 0.6905, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 2.575618103062542, | |
| "learning_rate": 1.936561298024377e-05, | |
| "loss": 0.7403, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.4765301530813417, | |
| "learning_rate": 1.9307320857586377e-05, | |
| "loss": 0.6781, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 2.514368537213703, | |
| "learning_rate": 1.9246562694320258e-05, | |
| "loss": 0.6946, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 2.4362145971600135, | |
| "learning_rate": 1.9183354588751274e-05, | |
| "loss": 0.7167, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 2.497750987289733, | |
| "learning_rate": 1.9117713288314864e-05, | |
| "loss": 0.7087, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 2.4766895730904745, | |
| "learning_rate": 1.904965618513868e-05, | |
| "loss": 0.6525, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 2.738681388493496, | |
| "learning_rate": 1.8979201311434434e-05, | |
| "loss": 0.6883, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 2.3966146991738966, | |
| "learning_rate": 1.8906367334720125e-05, | |
| "loss": 0.65, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.3250167887091786, | |
| "learning_rate": 1.8831173552873946e-05, | |
| "loss": 0.6635, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.4981823035208532, | |
| "learning_rate": 1.8753639889021197e-05, | |
| "loss": 0.6694, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.193667735760868, | |
| "learning_rate": 1.8673786886255478e-05, | |
| "loss": 0.5838, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 2.2455740949475835, | |
| "learning_rate": 1.8591635702195672e-05, | |
| "loss": 0.5871, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 2.2660780881135194, | |
| "learning_rate": 1.8507208103380093e-05, | |
| "loss": 0.6389, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 2.1782462437953147, | |
| "learning_rate": 1.8420526459499252e-05, | |
| "loss": 0.5775, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.300815653260401, | |
| "learning_rate": 1.8331613737468888e-05, | |
| "loss": 0.6236, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.0552955568974483, | |
| "learning_rate": 1.8240493495344695e-05, | |
| "loss": 0.5946, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 2.3014809235970017, | |
| "learning_rate": 1.8147189876080463e-05, | |
| "loss": 0.6149, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.0483445577828654, | |
| "learning_rate": 1.8051727601131228e-05, | |
| "loss": 0.5565, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.1431784888417904, | |
| "learning_rate": 1.7954131963903134e-05, | |
| "loss": 0.5689, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 1.870213438898714, | |
| "learning_rate": 1.785442882305179e-05, | |
| "loss": 0.3835, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 1.7742745864796516, | |
| "learning_rate": 1.775264459563081e-05, | |
| "loss": 0.381, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 1.85913693482777, | |
| "learning_rate": 1.764880625009245e-05, | |
| "loss": 0.3634, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 1.8223835701505764, | |
| "learning_rate": 1.7542941299142113e-05, | |
| "loss": 0.3964, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 2.0376834810173623, | |
| "learning_rate": 1.7435077792448666e-05, | |
| "loss": 0.3912, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 1.9367444811085355, | |
| "learning_rate": 1.7325244309212476e-05, | |
| "loss": 0.3652, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 2.0044934340131806, | |
| "learning_rate": 1.7213469950593156e-05, | |
| "loss": 0.406, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 1.7651282767994878, | |
| "learning_rate": 1.709978433199901e-05, | |
| "loss": 0.3607, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 1.9042056017789768, | |
| "learning_rate": 1.6984217575240212e-05, | |
| "loss": 0.3984, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 1.8589445625686343, | |
| "learning_rate": 1.6866800300547814e-05, | |
| "loss": 0.3906, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 2.0540269145149384, | |
| "learning_rate": 1.674756361846071e-05, | |
| "loss": 0.4072, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 1.8054523911306564, | |
| "learning_rate": 1.6626539121582687e-05, | |
| "loss": 0.3643, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.8676585477146084, | |
| "learning_rate": 1.650375887621171e-05, | |
| "loss": 0.4015, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 1.819143157098772, | |
| "learning_rate": 1.637925541384375e-05, | |
| "loss": 0.3769, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 1.7391959194884734, | |
| "learning_rate": 1.6253061722553353e-05, | |
| "loss": 0.3937, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.8992984530610373, | |
| "learning_rate": 1.612521123825317e-05, | |
| "loss": 0.3879, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 1.8399116519781489, | |
| "learning_rate": 1.5995737835834905e-05, | |
| "loss": 0.3796, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 1.744045618081832, | |
| "learning_rate": 1.586467582019392e-05, | |
| "loss": 0.3615, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.732449536710179, | |
| "learning_rate": 1.5732059917139912e-05, | |
| "loss": 0.3807, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 1.8491352122059705, | |
| "learning_rate": 1.5597925264196048e-05, | |
| "loss": 0.3554, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 1.7969522210818722, | |
| "learning_rate": 1.546230740128904e-05, | |
| "loss": 0.369, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 1.8026036373190188, | |
| "learning_rate": 1.53252422613326e-05, | |
| "loss": 0.3757, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 1.7840199265885361, | |
| "learning_rate": 1.5186766160706738e-05, | |
| "loss": 0.4167, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 1.7787153715407427, | |
| "learning_rate": 1.504691578963549e-05, | |
| "loss": 0.3474, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 1.8175150450617676, | |
| "learning_rate": 1.4905728202465596e-05, | |
| "loss": 0.3764, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 1.8592965886869042, | |
| "learning_rate": 1.4763240807848667e-05, | |
| "loss": 0.3993, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 1.861633852080652, | |
| "learning_rate": 1.4619491358829502e-05, | |
| "loss": 0.3844, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 1.7028160864977242, | |
| "learning_rate": 1.4474517942843173e-05, | |
| "loss": 0.3896, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 1.7419533336971356, | |
| "learning_rate": 1.4328358971623455e-05, | |
| "loss": 0.3558, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 1.663271147203201, | |
| "learning_rate": 1.4181053171025392e-05, | |
| "loss": 0.3551, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 1.6863853163892444, | |
| "learning_rate": 1.4032639570764595e-05, | |
| "loss": 0.3337, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 1.793591094205731, | |
| "learning_rate": 1.3883157494076048e-05, | |
| "loss": 0.3781, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 1.6945142844490944, | |
| "learning_rate": 1.3732646547295128e-05, | |
| "loss": 0.4093, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 1.8592456163542173, | |
| "learning_rate": 1.358114660936364e-05, | |
| "loss": 0.3814, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 1.6574433544024902, | |
| "learning_rate": 1.34286978212636e-05, | |
| "loss": 0.3578, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 1.7633364539067733, | |
| "learning_rate": 1.32753405753816e-05, | |
| "loss": 0.3856, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 1.8239540708598452, | |
| "learning_rate": 1.3121115504806554e-05, | |
| "loss": 0.3644, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 1.7596504605594738, | |
| "learning_rate": 1.2966063472563686e-05, | |
| "loss": 0.3591, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 1.5895053962743115, | |
| "learning_rate": 1.2810225560787561e-05, | |
| "loss": 0.3588, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 1.6063701391748393, | |
| "learning_rate": 1.2653643059837109e-05, | |
| "loss": 0.3653, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 1.635228537006481, | |
| "learning_rate": 1.2496357457355423e-05, | |
| "loss": 0.3827, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 1.5414929249094127, | |
| "learning_rate": 1.2338410427277342e-05, | |
| "loss": 0.3378, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 1.4758404777924012, | |
| "learning_rate": 1.2179843818787625e-05, | |
| "loss": 0.3325, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 1.6415892900669684, | |
| "learning_rate": 1.202069964523272e-05, | |
| "loss": 0.3811, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 1.523419893603189, | |
| "learning_rate": 1.186102007298904e-05, | |
| "loss": 0.3241, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 1.6484777075035715, | |
| "learning_rate": 1.1700847410290667e-05, | |
| "loss": 0.3702, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 1.5525179346249605, | |
| "learning_rate": 1.1540224096019495e-05, | |
| "loss": 0.3619, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 1.516882070329218, | |
| "learning_rate": 1.137919268846074e-05, | |
| "loss": 0.3581, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 1.56335517369179, | |
| "learning_rate": 1.121779585402684e-05, | |
| "loss": 0.3549, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 1.521782470020805, | |
| "learning_rate": 1.105607635595266e-05, | |
| "loss": 0.3477, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.3640551624778041, | |
| "learning_rate": 1.0894077042965084e-05, | |
| "loss": 0.2394, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 1.444257443908557, | |
| "learning_rate": 1.0731840837929946e-05, | |
| "loss": 0.1767, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 1.482114013548794, | |
| "learning_rate": 1.0569410726479301e-05, | |
| "loss": 0.183, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 1.3296971608762402, | |
| "learning_rate": 1.0406829745622085e-05, | |
| "loss": 0.1774, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 1.2009794251635362, | |
| "learning_rate": 1.0244140972341155e-05, | |
| "loss": 0.1712, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 1.14191410951918, | |
| "learning_rate": 1.008138751217973e-05, | |
| "loss": 0.1463, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 1.2767421227177214, | |
| "learning_rate": 9.918612487820274e-06, | |
| "loss": 0.1701, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 1.1705901668353922, | |
| "learning_rate": 9.755859027658848e-06, | |
| "loss": 0.1435, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 1.3281334820644817, | |
| "learning_rate": 9.593170254377915e-06, | |
| "loss": 0.1743, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 1.280366732691976, | |
| "learning_rate": 9.430589273520704e-06, | |
| "loss": 0.1508, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 1.3111725861250203, | |
| "learning_rate": 9.268159162070058e-06, | |
| "loss": 0.1544, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 1.3647916018519441, | |
| "learning_rate": 9.105922957034921e-06, | |
| "loss": 0.1679, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 1.2719254204880934, | |
| "learning_rate": 8.943923644047343e-06, | |
| "loss": 0.1553, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 1.3747641621255784, | |
| "learning_rate": 8.782204145973162e-06, | |
| "loss": 0.1753, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 1.2489287077269695, | |
| "learning_rate": 8.620807311539258e-06, | |
| "loss": 0.164, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 1.1953887915106376, | |
| "learning_rate": 8.45977590398051e-06, | |
| "loss": 0.1537, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 1.2483013054915837, | |
| "learning_rate": 8.299152589709336e-06, | |
| "loss": 0.1627, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "grad_norm": 1.1082607637515602, | |
| "learning_rate": 8.138979927010964e-06, | |
| "loss": 0.1505, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 1.1877785406198442, | |
| "learning_rate": 7.979300354767282e-06, | |
| "loss": 0.1718, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 1.2007862495511872, | |
| "learning_rate": 7.82015618121238e-06, | |
| "loss": 0.1796, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 1.268623285589921, | |
| "learning_rate": 7.66158957272266e-06, | |
| "loss": 0.1713, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "grad_norm": 1.2625050648502694, | |
| "learning_rate": 7.503642542644581e-06, | |
| "loss": 0.1719, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 1.1282996425547969, | |
| "learning_rate": 7.346356940162895e-06, | |
| "loss": 0.1598, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 1.138917768830711, | |
| "learning_rate": 7.189774439212442e-06, | |
| "loss": 0.1576, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 1.1962501488409478, | |
| "learning_rate": 7.033936527436318e-06, | |
| "loss": 0.1762, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 1.19112885794854, | |
| "learning_rate": 6.878884495193448e-06, | |
| "loss": 0.1749, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 1.173014938321161, | |
| "learning_rate": 6.724659424618401e-06, | |
| "loss": 0.1596, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 1.177107256530893, | |
| "learning_rate": 6.571302178736404e-06, | |
| "loss": 0.1646, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 1.050951011586738, | |
| "learning_rate": 6.418853390636363e-06, | |
| "loss": 0.1592, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 1.1776390226869722, | |
| "learning_rate": 6.267353452704876e-06, | |
| "loss": 0.1567, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 1.0758437748665028, | |
| "learning_rate": 6.116842505923955e-06, | |
| "loss": 0.154, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 1.0784529587978813, | |
| "learning_rate": 5.967360429235407e-06, | |
| "loss": 0.1561, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "grad_norm": 1.1696386233204266, | |
| "learning_rate": 5.8189468289746075e-06, | |
| "loss": 0.159, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "grad_norm": 1.1371629509054215, | |
| "learning_rate": 5.671641028376547e-06, | |
| "loss": 0.1574, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 1.0887944902862232, | |
| "learning_rate": 5.525482057156833e-06, | |
| "loss": 0.1547, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 1.207419126341501, | |
| "learning_rate": 5.380508641170499e-06, | |
| "loss": 0.162, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "grad_norm": 1.0198211842292424, | |
| "learning_rate": 5.236759192151336e-06, | |
| "loss": 0.1537, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 1.1458982044698929, | |
| "learning_rate": 5.094271797534404e-06, | |
| "loss": 0.1521, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 1.099882752498822, | |
| "learning_rate": 4.953084210364508e-06, | |
| "loss": 0.15, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 1.139969519826751, | |
| "learning_rate": 4.813233839293265e-06, | |
| "loss": 0.1593, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "grad_norm": 1.0693934037605295, | |
| "learning_rate": 4.674757738667405e-06, | |
| "loss": 0.154, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 1.1635281149070174, | |
| "learning_rate": 4.537692598710962e-06, | |
| "loss": 0.1692, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 1.04490476219924, | |
| "learning_rate": 4.402074735803955e-06, | |
| "loss": 0.147, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 1.0932888148614304, | |
| "learning_rate": 4.267940082860088e-06, | |
| "loss": 0.1498, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 1.1168758784687194, | |
| "learning_rate": 4.135324179806079e-06, | |
| "loss": 0.1499, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 1.1217893774994945, | |
| "learning_rate": 4.004262164165098e-06, | |
| "loss": 0.1596, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 1.0376730211600298, | |
| "learning_rate": 3.874788761746836e-06, | |
| "loss": 0.1524, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 1.1181081966120519, | |
| "learning_rate": 3.74693827744665e-06, | |
| "loss": 0.1426, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 1.158024328625209, | |
| "learning_rate": 3.6207445861562497e-06, | |
| "loss": 0.1673, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 1.1391490636515222, | |
| "learning_rate": 3.4962411237882945e-06, | |
| "loss": 0.1486, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 1.0483715121883341, | |
| "learning_rate": 3.373460878417315e-06, | |
| "loss": 0.1563, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.8483422663230202, | |
| "learning_rate": 3.252436381539291e-06, | |
| "loss": 0.1084, | |
| "step": 153 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 204, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 46868594688000.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |