| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 178, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 307.2925252001398, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 2.14, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 348.5579299962218, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 2.1592, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 295.8217735171242, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 2.1112, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 320.95278420676283, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 2.0727, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 273.6994356101882, | |
| "learning_rate": 5.555555555555555e-06, | |
| "loss": 1.8522, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 137.18652764811333, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.5844, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 8.105539597979915, | |
| "learning_rate": 7.777777777777777e-06, | |
| "loss": 1.4424, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 4.0951798632526275, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 1.4244, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 2.6078027918916717, | |
| "learning_rate": 9.999999999999999e-06, | |
| "loss": 1.3763, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 1.9798839986563381, | |
| "learning_rate": 1.111111111111111e-05, | |
| "loss": 1.3583, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 2.065159602417454, | |
| "learning_rate": 1.2222222222222222e-05, | |
| "loss": 1.3335, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 1.3002391180584008, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 1.309, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 2.2825778137111357, | |
| "learning_rate": 1.4444444444444444e-05, | |
| "loss": 1.2932, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.0205928394685457, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 1.2689, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.3373431587246087, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 1.2489, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.8043774075647928, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 1.2302, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.8741531962350515, | |
| "learning_rate": 1.888888888888889e-05, | |
| "loss": 1.2195, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.7028376256413013, | |
| "learning_rate": 1.9999999999999998e-05, | |
| "loss": 1.2027, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.6648050831010063, | |
| "learning_rate": 2.111111111111111e-05, | |
| "loss": 1.1887, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.7354468182477213, | |
| "learning_rate": 2.222222222222222e-05, | |
| "loss": 1.1704, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.7839624362161648, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 1.1654, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.7752033041911668, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 1.1544, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.7396236066452132, | |
| "learning_rate": 2.5555555555555557e-05, | |
| "loss": 1.138, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.748517600591021, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 1.133, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.7065074963471533, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 1.1119, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.7289352652301433, | |
| "learning_rate": 2.8888888888888888e-05, | |
| "loss": 1.1132, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.7324972388581718, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1008, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.7020306466156484, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 1.1067, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.7080095311543003, | |
| "learning_rate": 3.222222222222223e-05, | |
| "loss": 1.1028, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.7149252394051553, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 1.0953, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.7285982439983161, | |
| "learning_rate": 3.444444444444445e-05, | |
| "loss": 1.0859, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.7057181936211316, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 1.0829, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.718290068719298, | |
| "learning_rate": 3.666666666666667e-05, | |
| "loss": 1.0658, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.6998775916382852, | |
| "learning_rate": 3.777777777777778e-05, | |
| "loss": 1.0687, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.6851057008807085, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 1.0701, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.6724489509428345, | |
| "learning_rate": 3.9999999999999996e-05, | |
| "loss": 1.0675, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.6480552435697329, | |
| "learning_rate": 4.1111111111111116e-05, | |
| "loss": 1.0537, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.6383559954105056, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 1.0463, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.6029067517028285, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 1.0422, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.5709974683191396, | |
| "learning_rate": 4.444444444444444e-05, | |
| "loss": 1.037, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.5363586512263623, | |
| "learning_rate": 4.555555555555556e-05, | |
| "loss": 1.0249, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.5064701225673895, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 1.0269, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.4492710010511058, | |
| "learning_rate": 4.777777777777778e-05, | |
| "loss": 1.014, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.4154607764986574, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 1.0209, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.3647804068160132, | |
| "learning_rate": 5e-05, | |
| "loss": 1.019, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.3247880218369176, | |
| "learning_rate": 5.1111111111111115e-05, | |
| "loss": 1.0142, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.3054048294238843, | |
| "learning_rate": 5.222222222222222e-05, | |
| "loss": 1.005, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.2911566588563454, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 1.0182, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.30175247842711633, | |
| "learning_rate": 5.4444444444444446e-05, | |
| "loss": 1.003, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.2957438299690764, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 1.0018, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.222811420448816, | |
| "learning_rate": 5.6666666666666664e-05, | |
| "loss": 0.9982, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.19282721337866324, | |
| "learning_rate": 5.7777777777777776e-05, | |
| "loss": 1.003, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.2271139316205582, | |
| "learning_rate": 5.888888888888889e-05, | |
| "loss": 0.9884, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.22400303611020278, | |
| "learning_rate": 6e-05, | |
| "loss": 0.9933, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.2640000653332148, | |
| "learning_rate": 5.999935744992388e-05, | |
| "loss": 0.9898, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.29683962789566454, | |
| "learning_rate": 5.999742982722021e-05, | |
| "loss": 0.9894, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.23953312072588218, | |
| "learning_rate": 5.999421721446195e-05, | |
| "loss": 0.9891, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.22760431232110737, | |
| "learning_rate": 5.9989719749266715e-05, | |
| "loss": 0.9794, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.28069530624064654, | |
| "learning_rate": 5.998393762429097e-05, | |
| "loss": 0.9827, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.33248489828390104, | |
| "learning_rate": 5.997687108722169e-05, | |
| "loss": 0.9829, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.37460157299926583, | |
| "learning_rate": 5.9968520440765807e-05, | |
| "loss": 0.9865, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.40204606775328766, | |
| "learning_rate": 5.9958886042637214e-05, | |
| "loss": 0.9872, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.2760468657673376, | |
| "learning_rate": 5.994796830554148e-05, | |
| "loss": 0.9825, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.25200297636338487, | |
| "learning_rate": 5.9935767697158103e-05, | |
| "loss": 0.9761, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.3387334098621161, | |
| "learning_rate": 5.992228474012056e-05, | |
| "loss": 0.9724, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.3352356569193807, | |
| "learning_rate": 5.990752001199384e-05, | |
| "loss": 0.9694, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.25307126239627026, | |
| "learning_rate": 5.989147414524976e-05, | |
| "loss": 0.9751, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.28603448445681706, | |
| "learning_rate": 5.987414782723985e-05, | |
| "loss": 0.9675, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.2512510471162242, | |
| "learning_rate": 5.985554180016591e-05, | |
| "loss": 0.9713, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.23308838498745504, | |
| "learning_rate": 5.98356568610482e-05, | |
| "loss": 0.9675, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.2507676754613168, | |
| "learning_rate": 5.981449386169134e-05, | |
| "loss": 0.9768, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.17834215895890057, | |
| "learning_rate": 5.979205370864779e-05, | |
| "loss": 0.9736, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.22157097251518248, | |
| "learning_rate": 5.976833736317901e-05, | |
| "loss": 0.9761, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.19478083716793773, | |
| "learning_rate": 5.9743345841214316e-05, | |
| "loss": 0.9578, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.23929984172471444, | |
| "learning_rate": 5.9717080213307314e-05, | |
| "loss": 0.9637, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.1792504123152509, | |
| "learning_rate": 5.968954160459011e-05, | |
| "loss": 0.9694, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.1942820411854134, | |
| "learning_rate": 5.966073119472502e-05, | |
| "loss": 0.9654, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.20225817833116058, | |
| "learning_rate": 5.963065021785414e-05, | |
| "loss": 0.9568, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.17871245685588816, | |
| "learning_rate": 5.9599299962546375e-05, | |
| "loss": 0.9672, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.19357967707631799, | |
| "learning_rate": 5.956668177174234e-05, | |
| "loss": 0.9581, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.17703904963665285, | |
| "learning_rate": 5.953279704269675e-05, | |
| "loss": 0.9399, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.18397738069475075, | |
| "learning_rate": 5.949764722691864e-05, | |
| "loss": 0.9582, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.1667458999382299, | |
| "learning_rate": 5.9461233830109117e-05, | |
| "loss": 0.9574, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.15246629979305598, | |
| "learning_rate": 5.9423558412096914e-05, | |
| "loss": 0.9624, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.147474418280348, | |
| "learning_rate": 5.938462258677154e-05, | |
| "loss": 0.9574, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.18908685471550463, | |
| "learning_rate": 5.934442802201417e-05, | |
| "loss": 0.9559, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.15962484297812873, | |
| "learning_rate": 5.930297643962617e-05, | |
| "loss": 0.9565, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.15626218745541218, | |
| "learning_rate": 5.926026961525538e-05, | |
| "loss": 0.9669, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.16363056001381104, | |
| "learning_rate": 5.921630937832001e-05, | |
| "loss": 0.9575, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.17266359267085402, | |
| "learning_rate": 5.91710976119303e-05, | |
| "loss": 0.9482, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.16154285700465842, | |
| "learning_rate": 5.9124636252807844e-05, | |
| "loss": 0.9486, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.1505238744935054, | |
| "learning_rate": 5.907692729120263e-05, | |
| "loss": 0.9465, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.1665515957602505, | |
| "learning_rate": 5.9027972770807796e-05, | |
| "loss": 0.9458, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.15288744931626277, | |
| "learning_rate": 5.897777478867205e-05, | |
| "loss": 0.9513, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.13893588088386338, | |
| "learning_rate": 5.892633549510988e-05, | |
| "loss": 0.9517, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.15518080707211024, | |
| "learning_rate": 5.887365709360941e-05, | |
| "loss": 0.956, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.16735904200015367, | |
| "learning_rate": 5.881974184073806e-05, | |
| "loss": 0.9644, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.16211582659946666, | |
| "learning_rate": 5.876459204604579e-05, | |
| "loss": 0.947, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.16067404496452142, | |
| "learning_rate": 5.8708210071966266e-05, | |
| "loss": 0.9493, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.16466229984590008, | |
| "learning_rate": 5.8650598333715604e-05, | |
| "loss": 0.9525, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.15635651077742227, | |
| "learning_rate": 5.8591759299188915e-05, | |
| "loss": 0.9462, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.1445739463712597, | |
| "learning_rate": 5.853169548885461e-05, | |
| "loss": 0.9557, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.16706069442335733, | |
| "learning_rate": 5.847040947564642e-05, | |
| "loss": 0.9571, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.14383289999902457, | |
| "learning_rate": 5.8407903884853173e-05, | |
| "loss": 0.9452, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.1409356084289515, | |
| "learning_rate": 5.8344181394006345e-05, | |
| "loss": 0.9452, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.1697312533943109, | |
| "learning_rate": 5.827924473276536e-05, | |
| "loss": 0.9567, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.17531082504359882, | |
| "learning_rate": 5.821309668280065e-05, | |
| "loss": 0.9462, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.1639474813042821, | |
| "learning_rate": 5.814574007767453e-05, | |
| "loss": 0.9485, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.17850681384081132, | |
| "learning_rate": 5.807717780271977e-05, | |
| "loss": 0.9366, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.16125613154173873, | |
| "learning_rate": 5.800741279491605e-05, | |
| "loss": 0.9451, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.197900179672024, | |
| "learning_rate": 5.7936448042764106e-05, | |
| "loss": 0.9495, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.21076815721637063, | |
| "learning_rate": 5.7864286586157726e-05, | |
| "loss": 0.9435, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.19782059920889028, | |
| "learning_rate": 5.7790931516253545e-05, | |
| "loss": 0.9416, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.1836244550876009, | |
| "learning_rate": 5.7716385975338605e-05, | |
| "loss": 0.9466, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.1613256822257701, | |
| "learning_rate": 5.764065315669578e-05, | |
| "loss": 0.9513, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.16741350820768655, | |
| "learning_rate": 5.756373630446695e-05, | |
| "loss": 0.9418, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.18843157688751017, | |
| "learning_rate": 5.748563871351408e-05, | |
| "loss": 0.945, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.15069085597597218, | |
| "learning_rate": 5.7406363729278026e-05, | |
| "loss": 0.9466, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.16662142102625724, | |
| "learning_rate": 5.7325914747635275e-05, | |
| "loss": 0.9486, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.19590155354881233, | |
| "learning_rate": 5.724429521475244e-05, | |
| "loss": 0.9435, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.15881882311241918, | |
| "learning_rate": 5.716150862693866e-05, | |
| "loss": 0.9466, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.14283264252478434, | |
| "learning_rate": 5.707755853049582e-05, | |
| "loss": 0.9412, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.16782472517291772, | |
| "learning_rate": 5.699244852156665e-05, | |
| "loss": 0.9382, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.16548141172284359, | |
| "learning_rate": 5.690618224598065e-05, | |
| "loss": 0.9479, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.13809707490800946, | |
| "learning_rate": 5.681876339909797e-05, | |
| "loss": 0.9429, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.18656335425800988, | |
| "learning_rate": 5.673019572565103e-05, | |
| "loss": 0.9381, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.1732296780937527, | |
| "learning_rate": 5.664048301958422e-05, | |
| "loss": 0.9431, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.14221356678314365, | |
| "learning_rate": 5.654962912389126e-05, | |
| "loss": 0.9523, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.15119656927035072, | |
| "learning_rate": 5.645763793045065e-05, | |
| "loss": 0.9392, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.14464956279442504, | |
| "learning_rate": 5.636451337985896e-05, | |
| "loss": 0.9384, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.13140580588464784, | |
| "learning_rate": 5.627025946126199e-05, | |
| "loss": 0.9372, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.14493562139149813, | |
| "learning_rate": 5.617488021218392e-05, | |
| "loss": 0.9358, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.14893216335321577, | |
| "learning_rate": 5.6078379718354315e-05, | |
| "loss": 0.9419, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.14722396099871643, | |
| "learning_rate": 5.5980762113533166e-05, | |
| "loss": 0.944, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.1484388680331261, | |
| "learning_rate": 5.588203157933376e-05, | |
| "loss": 0.946, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.14453486799975523, | |
| "learning_rate": 5.578219234504359e-05, | |
| "loss": 0.9502, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.13849264435281988, | |
| "learning_rate": 5.568124868744315e-05, | |
| "loss": 0.9339, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.16199659629172095, | |
| "learning_rate": 5.557920493062277e-05, | |
| "loss": 0.9238, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.16192446153254933, | |
| "learning_rate": 5.547606544579737e-05, | |
| "loss": 0.9336, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.14648464155352814, | |
| "learning_rate": 5.5371834651119204e-05, | |
| "loss": 0.9305, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.1383162638141149, | |
| "learning_rate": 5.5266517011488596e-05, | |
| "loss": 0.9391, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.1543998232523172, | |
| "learning_rate": 5.5160117038362726e-05, | |
| "loss": 0.9366, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.15730672921508043, | |
| "learning_rate": 5.5052639289562294e-05, | |
| "loss": 0.9346, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.13879299352222021, | |
| "learning_rate": 5.494408836907636e-05, | |
| "loss": 0.9364, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.15487093688296455, | |
| "learning_rate": 5.483446892686507e-05, | |
| "loss": 0.9246, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.14401066674945762, | |
| "learning_rate": 5.472378565866047e-05, | |
| "loss": 0.9361, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.14277478926628612, | |
| "learning_rate": 5.461204330576541e-05, | |
| "loss": 0.9389, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.16131149816932222, | |
| "learning_rate": 5.4499246654850374e-05, | |
| "loss": 0.9371, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.17407874387998404, | |
| "learning_rate": 5.4385400537748465e-05, | |
| "loss": 0.9372, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.17636478629943686, | |
| "learning_rate": 5.427050983124843e-05, | |
| "loss": 0.9343, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.16977563161518314, | |
| "learning_rate": 5.4154579456885744e-05, | |
| "loss": 0.9281, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.14318477872489785, | |
| "learning_rate": 5.403761438073182e-05, | |
| "loss": 0.9365, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.1397942734256781, | |
| "learning_rate": 5.3919619613181215e-05, | |
| "loss": 0.9469, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.137963649770748, | |
| "learning_rate": 5.3800600208737054e-05, | |
| "loss": 0.9359, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.15171364959388206, | |
| "learning_rate": 5.3680561265794496e-05, | |
| "loss": 0.9269, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.14709787438935637, | |
| "learning_rate": 5.3559507926422344e-05, | |
| "loss": 0.9383, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.12890500807018168, | |
| "learning_rate": 5.343744537614276e-05, | |
| "loss": 0.924, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.11549683107184264, | |
| "learning_rate": 5.331437884370913e-05, | |
| "loss": 0.9283, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.15022761689747757, | |
| "learning_rate": 5.319031360088211e-05, | |
| "loss": 0.9307, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.15714839886122223, | |
| "learning_rate": 5.306525496220379e-05, | |
| "loss": 0.935, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.16859132275494298, | |
| "learning_rate": 5.293920828477001e-05, | |
| "loss": 0.9239, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.13633581062447336, | |
| "learning_rate": 5.281217896800093e-05, | |
| "loss": 0.9414, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.1513556776118746, | |
| "learning_rate": 5.268417245340968e-05, | |
| "loss": 0.9338, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.1757338104436492, | |
| "learning_rate": 5.255519422436932e-05, | |
| "loss": 0.9351, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.19112425290854476, | |
| "learning_rate": 5.242524980587791e-05, | |
| "loss": 0.9333, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.17979027309488244, | |
| "learning_rate": 5.2294344764321825e-05, | |
| "loss": 0.9179, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.16529910589891425, | |
| "learning_rate": 5.2162484707237387e-05, | |
| "loss": 0.9356, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.20692577816446356, | |
| "learning_rate": 5.202967528307057e-05, | |
| "loss": 0.9276, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.197154695709334, | |
| "learning_rate": 5.1895922180935066e-05, | |
| "loss": 0.9303, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.14150974456692325, | |
| "learning_rate": 5.176123113036863e-05, | |
| "loss": 0.9364, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.15101923044037124, | |
| "learning_rate": 5.162560790108756e-05, | |
| "loss": 0.9219, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.19585100899813354, | |
| "learning_rate": 5.148905830273964e-05, | |
| "loss": 0.9282, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.16466476745138203, | |
| "learning_rate": 5.135158818465514e-05, | |
| "loss": 0.9267, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.13928826936678446, | |
| "learning_rate": 5.1213203435596425e-05, | |
| "loss": 0.9204, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.16777686357665667, | |
| "learning_rate": 5.107390998350555e-05, | |
| "loss": 0.9209, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.13941558577566035, | |
| "learning_rate": 5.093371379525041e-05, | |
| "loss": 0.933, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.1312376870568112, | |
| "learning_rate": 5.079262087636908e-05, | |
| "loss": 0.9273, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.13304976321730042, | |
| "learning_rate": 5.0650637270812615e-05, | |
| "loss": 0.9325, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.9060415029525757, | |
| "eval_runtime": 367.2491, | |
| "eval_samples_per_second": 35.657, | |
| "eval_steps_per_second": 0.054, | |
| "step": 178 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 534, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1405957714214912.0, | |
| "train_batch_size": 42, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |