| { | |
| "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": 30.484978650801835, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 1.8871, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 30.606371885605782, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 1.8307, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 21.244108931810736, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.7922, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 27.90080784000707, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 1.7422, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 22.016657678494965, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 1.757, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 18.98537323764419, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 1.7216, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 15.294538652658218, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 1.6025, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 14.265933581855554, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 1.5677, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 12.840032703203521, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 1.4663, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 11.560868761621572, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 1.4061, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 9.965629256574427, | |
| "learning_rate": 2e-05, | |
| "loss": 1.4483, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 12.353598844828388, | |
| "learning_rate": 1.999867521457224e-05, | |
| "loss": 1.3607, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 11.182206078770232, | |
| "learning_rate": 1.9994701209300245e-05, | |
| "loss": 1.3203, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 9.055478658005105, | |
| "learning_rate": 1.9988079037124866e-05, | |
| "loss": 1.3627, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 8.94634827859796, | |
| "learning_rate": 1.9978810452637544e-05, | |
| "loss": 1.2695, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 8.397214663882789, | |
| "learning_rate": 1.9966897911615417e-05, | |
| "loss": 1.1791, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 8.293556598630499, | |
| "learning_rate": 1.995234457037063e-05, | |
| "loss": 1.1773, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 7.906819078071131, | |
| "learning_rate": 1.9935154284914063e-05, | |
| "loss": 1.1327, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 7.292321689794095, | |
| "learning_rate": 1.991533160993366e-05, | |
| "loss": 1.088, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 7.894048510877663, | |
| "learning_rate": 1.98928817975876e-05, | |
| "loss": 1.0844, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 7.102421640395548, | |
| "learning_rate": 1.9867810796112742e-05, | |
| "loss": 1.0804, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 7.598223086984867, | |
| "learning_rate": 1.9840125248248564e-05, | |
| "loss": 1.014, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 6.851825287149768, | |
| "learning_rate": 1.9809832489477144e-05, | |
| "loss": 0.9518, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 8.875055017435104, | |
| "learning_rate": 1.9776940546079552e-05, | |
| "loss": 1.0026, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 6.923610220037225, | |
| "learning_rate": 1.9741458133009258e-05, | |
| "loss": 0.9639, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 5.862173810721529, | |
| "learning_rate": 1.970339465158301e-05, | |
| "loss": 1.0026, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 6.940850076199128, | |
| "learning_rate": 1.9662760186989914e-05, | |
| "loss": 0.9844, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 7.868648086627184, | |
| "learning_rate": 1.9619565505619288e-05, | |
| "loss": 0.9796, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 5.6636206724227, | |
| "learning_rate": 1.9573822052208013e-05, | |
| "loss": 0.9808, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 5.903449792443762, | |
| "learning_rate": 1.9525541946808187e-05, | |
| "loss": 0.9122, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 5.4898704535759215, | |
| "learning_rate": 1.9474737981575833e-05, | |
| "loss": 0.9137, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 5.228428437706942, | |
| "learning_rate": 1.942142361738151e-05, | |
| "loss": 0.8407, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 5.1702666302810965, | |
| "learning_rate": 1.936561298024377e-05, | |
| "loss": 0.887, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 4.900231624500507, | |
| "learning_rate": 1.9307320857586377e-05, | |
| "loss": 0.8204, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 4.922333033003047, | |
| "learning_rate": 1.9246562694320258e-05, | |
| "loss": 0.8486, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 5.201888753755247, | |
| "learning_rate": 1.9183354588751274e-05, | |
| "loss": 0.907, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 4.860640284428521, | |
| "learning_rate": 1.9117713288314864e-05, | |
| "loss": 0.8574, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 4.580591572366175, | |
| "learning_rate": 1.904965618513868e-05, | |
| "loss": 0.811, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 5.297298448229, | |
| "learning_rate": 1.8979201311434434e-05, | |
| "loss": 0.8555, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 4.721873279220298, | |
| "learning_rate": 1.8906367334720125e-05, | |
| "loss": 0.8341, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 5.095801192042166, | |
| "learning_rate": 1.8831173552873946e-05, | |
| "loss": 0.8716, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 5.3441220873403905, | |
| "learning_rate": 1.8753639889021197e-05, | |
| "loss": 0.8781, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 4.532321987285937, | |
| "learning_rate": 1.8673786886255478e-05, | |
| "loss": 0.7815, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 4.679574395088194, | |
| "learning_rate": 1.8591635702195672e-05, | |
| "loss": 0.7853, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 5.075491918015324, | |
| "learning_rate": 1.8507208103380093e-05, | |
| "loss": 0.8402, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 4.317197369879349, | |
| "learning_rate": 1.8420526459499252e-05, | |
| "loss": 0.7753, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 4.589606044376589, | |
| "learning_rate": 1.8331613737468888e-05, | |
| "loss": 0.8122, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 4.6900469154988285, | |
| "learning_rate": 1.8240493495344695e-05, | |
| "loss": 0.7951, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 5.020594146645111, | |
| "learning_rate": 1.8147189876080463e-05, | |
| "loss": 0.8168, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 4.662493315581984, | |
| "learning_rate": 1.8051727601131228e-05, | |
| "loss": 0.7373, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 4.974273107131929, | |
| "learning_rate": 1.7954131963903134e-05, | |
| "loss": 0.7787, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 4.365204444458351, | |
| "learning_rate": 1.785442882305179e-05, | |
| "loss": 0.4912, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 3.7095298265573566, | |
| "learning_rate": 1.775264459563081e-05, | |
| "loss": 0.5061, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 3.9264808145353345, | |
| "learning_rate": 1.764880625009245e-05, | |
| "loss": 0.4638, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 4.517466872346605, | |
| "learning_rate": 1.7542941299142113e-05, | |
| "loss": 0.5132, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 4.393239176372648, | |
| "learning_rate": 1.7435077792448666e-05, | |
| "loss": 0.5032, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 4.102692831599629, | |
| "learning_rate": 1.7325244309212476e-05, | |
| "loss": 0.4896, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 4.306815339780854, | |
| "learning_rate": 1.7213469950593156e-05, | |
| "loss": 0.5276, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 3.669633026693288, | |
| "learning_rate": 1.709978433199901e-05, | |
| "loss": 0.4701, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 3.6349873257380647, | |
| "learning_rate": 1.6984217575240212e-05, | |
| "loss": 0.523, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 3.8514129423929644, | |
| "learning_rate": 1.6866800300547814e-05, | |
| "loss": 0.5191, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 4.0727927472839935, | |
| "learning_rate": 1.674756361846071e-05, | |
| "loss": 0.5099, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 3.8533512936957814, | |
| "learning_rate": 1.6626539121582687e-05, | |
| "loss": 0.4699, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 3.848655733525694, | |
| "learning_rate": 1.650375887621171e-05, | |
| "loss": 0.5236, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 3.7786610465222195, | |
| "learning_rate": 1.637925541384375e-05, | |
| "loss": 0.5004, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 4.129110657857101, | |
| "learning_rate": 1.6253061722553353e-05, | |
| "loss": 0.5111, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 3.563345959257114, | |
| "learning_rate": 1.612521123825317e-05, | |
| "loss": 0.5283, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 3.4507319240659045, | |
| "learning_rate": 1.5995737835834905e-05, | |
| "loss": 0.51, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 3.2981965367762176, | |
| "learning_rate": 1.586467582019392e-05, | |
| "loss": 0.479, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 3.495703648801503, | |
| "learning_rate": 1.5732059917139912e-05, | |
| "loss": 0.5205, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 3.362132799029953, | |
| "learning_rate": 1.5597925264196048e-05, | |
| "loss": 0.4693, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 3.6338693701653986, | |
| "learning_rate": 1.546230740128904e-05, | |
| "loss": 0.4955, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 3.454874967529182, | |
| "learning_rate": 1.53252422613326e-05, | |
| "loss": 0.4963, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 3.6257898193731557, | |
| "learning_rate": 1.5186766160706738e-05, | |
| "loss": 0.5488, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 3.355629895995887, | |
| "learning_rate": 1.504691578963549e-05, | |
| "loss": 0.454, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 3.429585448812729, | |
| "learning_rate": 1.4905728202465596e-05, | |
| "loss": 0.4872, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 3.472655975376114, | |
| "learning_rate": 1.4763240807848667e-05, | |
| "loss": 0.5173, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 3.280114985753271, | |
| "learning_rate": 1.4619491358829502e-05, | |
| "loss": 0.4988, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 3.932103045994452, | |
| "learning_rate": 1.4474517942843173e-05, | |
| "loss": 0.5476, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 3.5660509849664757, | |
| "learning_rate": 1.4328358971623455e-05, | |
| "loss": 0.4795, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 3.296496618171067, | |
| "learning_rate": 1.4181053171025392e-05, | |
| "loss": 0.4794, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 3.416440506559902, | |
| "learning_rate": 1.4032639570764595e-05, | |
| "loss": 0.4533, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 3.5490178428581873, | |
| "learning_rate": 1.3883157494076048e-05, | |
| "loss": 0.5145, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 3.7320219790842826, | |
| "learning_rate": 1.3732646547295128e-05, | |
| "loss": 0.5526, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 3.290676418674348, | |
| "learning_rate": 1.358114660936364e-05, | |
| "loss": 0.5242, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 3.26994133386278, | |
| "learning_rate": 1.34286978212636e-05, | |
| "loss": 0.4824, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 3.2021199952514676, | |
| "learning_rate": 1.32753405753816e-05, | |
| "loss": 0.5134, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 3.2338319774097224, | |
| "learning_rate": 1.3121115504806554e-05, | |
| "loss": 0.4774, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 3.375967522858161, | |
| "learning_rate": 1.2966063472563686e-05, | |
| "loss": 0.4944, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 3.416860311339495, | |
| "learning_rate": 1.2810225560787561e-05, | |
| "loss": 0.4879, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 3.347354692833765, | |
| "learning_rate": 1.2653643059837109e-05, | |
| "loss": 0.4956, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 3.1658491953383554, | |
| "learning_rate": 1.2496357457355423e-05, | |
| "loss": 0.5082, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 3.02531591610065, | |
| "learning_rate": 1.2338410427277342e-05, | |
| "loss": 0.4626, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 3.3426392038214097, | |
| "learning_rate": 1.2179843818787625e-05, | |
| "loss": 0.4599, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 3.229004903043567, | |
| "learning_rate": 1.202069964523272e-05, | |
| "loss": 0.4996, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 3.0090353325455195, | |
| "learning_rate": 1.186102007298904e-05, | |
| "loss": 0.4486, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 3.3271548867618006, | |
| "learning_rate": 1.1700847410290667e-05, | |
| "loss": 0.4955, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 3.3091390581467475, | |
| "learning_rate": 1.1540224096019495e-05, | |
| "loss": 0.4966, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 3.0321015410518535, | |
| "learning_rate": 1.137919268846074e-05, | |
| "loss": 0.4773, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 3.0009675055285556, | |
| "learning_rate": 1.121779585402684e-05, | |
| "loss": 0.4861, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 2.8711578569586167, | |
| "learning_rate": 1.105607635595266e-05, | |
| "loss": 0.4584, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 2.9151252669937917, | |
| "learning_rate": 1.0894077042965084e-05, | |
| "loss": 0.3349, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 3.3487435959729472, | |
| "learning_rate": 1.0731840837929946e-05, | |
| "loss": 0.2424, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 2.7928749335478122, | |
| "learning_rate": 1.0569410726479301e-05, | |
| "loss": 0.2299, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 2.863305397504332, | |
| "learning_rate": 1.0406829745622085e-05, | |
| "loss": 0.2422, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 2.995041564245414, | |
| "learning_rate": 1.0244140972341155e-05, | |
| "loss": 0.2295, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 3.213047049288632, | |
| "learning_rate": 1.008138751217973e-05, | |
| "loss": 0.2094, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 3.786514778538547, | |
| "learning_rate": 9.918612487820274e-06, | |
| "loss": 0.2269, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 3.2097534075464207, | |
| "learning_rate": 9.755859027658848e-06, | |
| "loss": 0.2115, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 3.240129447282013, | |
| "learning_rate": 9.593170254377915e-06, | |
| "loss": 0.2285, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 3.0328446745701236, | |
| "learning_rate": 9.430589273520704e-06, | |
| "loss": 0.2126, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 2.873994070791607, | |
| "learning_rate": 9.268159162070058e-06, | |
| "loss": 0.2212, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 2.8733178898454583, | |
| "learning_rate": 9.105922957034921e-06, | |
| "loss": 0.227, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 2.5368057315952943, | |
| "learning_rate": 8.943923644047343e-06, | |
| "loss": 0.2119, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 2.6178041557223652, | |
| "learning_rate": 8.782204145973162e-06, | |
| "loss": 0.2388, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 2.295055992590324, | |
| "learning_rate": 8.620807311539258e-06, | |
| "loss": 0.2242, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 2.417078301022775, | |
| "learning_rate": 8.45977590398051e-06, | |
| "loss": 0.2157, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 2.3674932603427945, | |
| "learning_rate": 8.299152589709336e-06, | |
| "loss": 0.219, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "grad_norm": 2.334842999589727, | |
| "learning_rate": 8.138979927010964e-06, | |
| "loss": 0.21, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 2.545817898850758, | |
| "learning_rate": 7.979300354767282e-06, | |
| "loss": 0.2218, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 2.6654355266890315, | |
| "learning_rate": 7.82015618121238e-06, | |
| "loss": 0.2333, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 2.5879654322914267, | |
| "learning_rate": 7.66158957272266e-06, | |
| "loss": 0.2343, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "grad_norm": 2.5602322289813277, | |
| "learning_rate": 7.503642542644581e-06, | |
| "loss": 0.2342, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 2.4852393224287463, | |
| "learning_rate": 7.346356940162895e-06, | |
| "loss": 0.2232, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 2.5558719980428672, | |
| "learning_rate": 7.189774439212442e-06, | |
| "loss": 0.2209, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 2.5873633196570007, | |
| "learning_rate": 7.033936527436318e-06, | |
| "loss": 0.2344, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 3.612159744659012, | |
| "learning_rate": 6.878884495193448e-06, | |
| "loss": 0.2367, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 2.334109200048589, | |
| "learning_rate": 6.724659424618401e-06, | |
| "loss": 0.2148, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 2.6257760853420087, | |
| "learning_rate": 6.571302178736404e-06, | |
| "loss": 0.2238, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 2.278029376248316, | |
| "learning_rate": 6.418853390636363e-06, | |
| "loss": 0.2148, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 2.3491752394670553, | |
| "learning_rate": 6.267353452704876e-06, | |
| "loss": 0.2205, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 2.1958711938169793, | |
| "learning_rate": 6.116842505923955e-06, | |
| "loss": 0.2112, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 2.2803769465934107, | |
| "learning_rate": 5.967360429235407e-06, | |
| "loss": 0.2097, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "grad_norm": 2.224556430792831, | |
| "learning_rate": 5.8189468289746075e-06, | |
| "loss": 0.2175, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "grad_norm": 2.304319604508385, | |
| "learning_rate": 5.671641028376547e-06, | |
| "loss": 0.2049, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 2.2487334819396962, | |
| "learning_rate": 5.525482057156833e-06, | |
| "loss": 0.2028, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 2.419261087836657, | |
| "learning_rate": 5.380508641170499e-06, | |
| "loss": 0.2127, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "grad_norm": 2.362959406351894, | |
| "learning_rate": 5.236759192151336e-06, | |
| "loss": 0.206, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 2.3417742730482467, | |
| "learning_rate": 5.094271797534404e-06, | |
| "loss": 0.2012, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 2.3759103335405785, | |
| "learning_rate": 4.953084210364508e-06, | |
| "loss": 0.2053, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 2.340796580729461, | |
| "learning_rate": 4.813233839293265e-06, | |
| "loss": 0.2098, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "grad_norm": 2.344815024947747, | |
| "learning_rate": 4.674757738667405e-06, | |
| "loss": 0.2099, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 2.4519121609958336, | |
| "learning_rate": 4.537692598710962e-06, | |
| "loss": 0.2139, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 2.334903041452081, | |
| "learning_rate": 4.402074735803955e-06, | |
| "loss": 0.2069, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 2.253294228641448, | |
| "learning_rate": 4.267940082860088e-06, | |
| "loss": 0.1931, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 2.439705169093854, | |
| "learning_rate": 4.135324179806079e-06, | |
| "loss": 0.2017, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 2.4105348954237655, | |
| "learning_rate": 4.004262164165098e-06, | |
| "loss": 0.216, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 2.240742500235594, | |
| "learning_rate": 3.874788761746836e-06, | |
| "loss": 0.1963, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 2.2879704809752166, | |
| "learning_rate": 3.74693827744665e-06, | |
| "loss": 0.1936, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 2.534216096649578, | |
| "learning_rate": 3.6207445861562497e-06, | |
| "loss": 0.2132, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 2.2676487795840052, | |
| "learning_rate": 3.4962411237882945e-06, | |
| "loss": 0.1917, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 2.1966856873556586, | |
| "learning_rate": 3.373460878417315e-06, | |
| "loss": 0.2021, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 1.9288689734502165, | |
| "learning_rate": 3.252436381539291e-06, | |
| "loss": 0.1441, | |
| "step": 153 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 204, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 14230609723392.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |