| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.983219390926041, | |
| "eval_steps": 500, | |
| "global_step": 300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 2.6769, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 2.6837, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.6307, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 2.5928, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9999417253661235e-05, | |
| "loss": 2.6261, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9994755690455154e-05, | |
| "loss": 2.6851, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.998543473718677e-05, | |
| "loss": 2.6421, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9971458739130598e-05, | |
| "loss": 2.7906, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.995283421166614e-05, | |
| "loss": 2.4095, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9929569837240567e-05, | |
| "loss": 2.2509, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.990167646132107e-05, | |
| "loss": 2.3942, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9869167087338908e-05, | |
| "loss": 2.4104, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.983205687062742e-05, | |
| "loss": 2.5293, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9790363111356838e-05, | |
| "loss": 2.6231, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9744105246469264e-05, | |
| "loss": 2.5915, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9693304840617456e-05, | |
| "loss": 2.4235, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.963798557611178e-05, | |
| "loss": 2.5622, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.957817324187987e-05, | |
| "loss": 2.3644, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9513895721444286e-05, | |
| "loss": 2.3714, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9445182979923657e-05, | |
| "loss": 2.5419, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.937206705006344e-05, | |
| "loss": 2.4557, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.9294582017302797e-05, | |
| "loss": 2.3226, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.921276400388451e-05, | |
| "loss": 2.3402, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9126651152015404e-05, | |
| "loss": 2.3476, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9036283606085057e-05, | |
| "loss": 2.3808, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.8941703493951163e-05, | |
| "loss": 2.4262, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.8842954907300236e-05, | |
| "loss": 2.3338, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.874008388109276e-05, | |
| "loss": 2.4052, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.863313837210247e-05, | |
| "loss": 2.2928, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.8522168236559693e-05, | |
| "loss": 2.4616, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.840722520690921e-05, | |
| "loss": 2.2975, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8288362867693414e-05, | |
| "loss": 2.379, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.816563663057211e-05, | |
| "loss": 2.3342, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8039103708490503e-05, | |
| "loss": 2.2533, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.790882308900746e-05, | |
| "loss": 2.2486, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7774855506796497e-05, | |
| "loss": 2.2496, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7637263415332272e-05, | |
| "loss": 2.3153, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.749611095777581e-05, | |
| "loss": 2.2758, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.7351463937072008e-05, | |
| "loss": 2.5391, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.7203389785273402e-05, | |
| "loss": 2.3089, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.705195753210446e-05, | |
| "loss": 2.2669, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6897237772781046e-05, | |
| "loss": 2.3467, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.673930263510011e-05, | |
| "loss": 2.3333, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.6578225745814907e-05, | |
| "loss": 2.4031, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.6414082196311402e-05, | |
| "loss": 2.3823, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.6246948507601915e-05, | |
| "loss": 2.2399, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.607690259465229e-05, | |
| "loss": 2.1444, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.5904023730059227e-05, | |
| "loss": 2.2844, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.57283925070947e-05, | |
| "loss": 2.2505, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.55500908021347e-05, | |
| "loss": 2.0644, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.536920173648984e-05, | |
| "loss": 2.2254, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.5185809637655548e-05, | |
| "loss": 2.3203, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 2.2265, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.4811859444908053e-05, | |
| "loss": 2.3855, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.4621475680399771e-05, | |
| "loss": 2.358, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.4428937460242417e-05, | |
| "loss": 2.3075, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.4234334542574906e-05, | |
| "loss": 2.267, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.4037757648064019e-05, | |
| "loss": 2.2067, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.3839298417611964e-05, | |
| "loss": 2.1989, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.3639049369634878e-05, | |
| "loss": 2.1818, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.3437103856932266e-05, | |
| "loss": 2.1537, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.3233556023167487e-05, | |
| "loss": 2.158, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.3028500758979507e-05, | |
| "loss": 2.3302, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.2822033657746478e-05, | |
| "loss": 2.3653, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.2614250971021658e-05, | |
| "loss": 2.1906, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.2405249563662539e-05, | |
| "loss": 2.3488, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.2195126868674052e-05, | |
| "loss": 2.3102, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.1983980841786899e-05, | |
| "loss": 2.4081, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.177190991579223e-05, | |
| "loss": 2.2191, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.1559012954653865e-05, | |
| "loss": 2.2446, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.1345389207419588e-05, | |
| "loss": 2.3967, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.1131138261952845e-05, | |
| "loss": 2.0867, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.0916359998506549e-05, | |
| "loss": 2.0844, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.070115454316054e-05, | |
| "loss": 2.1256, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0485622221144485e-05, | |
| "loss": 2.2274, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.0269863510067872e-05, | |
| "loss": 2.1116, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.0053978993079046e-05, | |
| "loss": 2.4296, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.838069311974986e-06, | |
| "loss": 2.3066, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.622235120283769e-06, | |
| "loss": 2.1659, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.406577036341548e-06, | |
| "loss": 2.3285, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.19119559638596e-06, | |
| "loss": 2.2788, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.976191207687775e-06, | |
| "loss": 2.2487, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.7616641017427e-06, | |
| "loss": 2.1421, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.5477142875451e-06, | |
| "loss": 2.3432, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.334441504965456e-06, | |
| "loss": 1.9564, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.1219451782533e-06, | |
| "loss": 2.1576, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.91032436968725e-06, | |
| "loss": 2.2133, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.699677733393827e-06, | |
| "loss": 2.2797, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.490103469356513e-06, | |
| "loss": 2.3495, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.2816992776365714e-06, | |
| "loss": 2.2715, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.0745623128268605e-06, | |
| "loss": 2.3838, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.868789138759977e-06, | |
| "loss": 2.3654, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.664475683491797e-06, | |
| "loss": 2.2189, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.461717194581394e-06, | |
| "loss": 2.3374, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.260608194688207e-06, | |
| "loss": 2.3332, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.061242437507131e-06, | |
| "loss": 2.2279, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.863712864062089e-06, | |
| "loss": 2.102, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.6681115593784705e-06, | |
| "loss": 2.1584, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.4745297095546125e-06, | |
| "loss": 2.1382, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.2830575592523415e-06, | |
| "loss": 2.0084, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.093784369626397e-06, | |
| "loss": 1.9967, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.9067983767123736e-06, | |
| "loss": 2.3797, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.722186750292511e-06, | |
| "loss": 2.3558, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 4.54003555325862e-06, | |
| "loss": 2.1337, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.360429701490935e-06, | |
| "loss": 2.2661, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 4.183452924271776e-06, | |
| "loss": 2.1616, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 4.009187725252309e-06, | |
| "loss": 2.355, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.837715343990727e-06, | |
| "loss": 2.224, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.669115718079702e-06, | |
| "loss": 2.2079, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.5034674458807893e-06, | |
| "loss": 2.086, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.3408477498831917e-06, | |
| "loss": 2.126, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.1813324407038826e-06, | |
| "loss": 2.1389, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.024995881745972e-06, | |
| "loss": 2.1352, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.8719109545317102e-06, | |
| "loss": 2.2183, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.722149024726307e-06, | |
| "loss": 2.2478, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.5757799088684654e-06, | |
| "loss": 2.2587, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.432871841823047e-06, | |
| "loss": 2.3191, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.293491444971109e-06, | |
| "loss": 2.2398, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.157703695152109e-06, | |
| "loss": 2.3926, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.025571894372794e-06, | |
| "loss": 2.2304, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1.897157640296825e-06, | |
| "loss": 2.2469, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1.7725207975289883e-06, | |
| "loss": 2.1065, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.6517194697072903e-06, | |
| "loss": 2.1856, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.534809972415998e-06, | |
| "loss": 2.1236, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.4218468069322576e-06, | |
| "loss": 2.1961, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.3128826348184886e-06, | |
| "loss": 2.0173, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.207968253372438e-06, | |
| "loss": 2.2847, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.1071525719463094e-06, | |
| "loss": 2.3931, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.010482589146048e-06, | |
| "loss": 2.2776, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 9.180033709213454e-07, | |
| "loss": 2.3249, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 8.297580295566576e-07, | |
| "loss": 2.1791, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 7.457877035729588e-07, | |
| "loss": 2.2177, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.661315385496426e-07, | |
| "loss": 2.2903, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.908266688755049e-07, | |
| "loss": 2.2478, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.199082004372958e-07, | |
| "loss": 2.2076, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.534091942539476e-07, | |
| "loss": 2.0342, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.913606510640644e-07, | |
| "loss": 2.2668, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.3379149687388866e-07, | |
| "loss": 2.1174, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 2.807285694724804e-07, | |
| "loss": 2.2882, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 2.3219660592038285e-07, | |
| "loss": 2.1654, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.8821823101760949e-07, | |
| "loss": 2.2209, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.4881394675633543e-07, | |
| "loss": 2.1442, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.1400212276321377e-07, | |
| "loss": 2.3807, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.379898773574924e-08, | |
| "loss": 2.4242, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 5.821862187675775e-08, | |
| "loss": 2.2171, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.727295033040035e-08, | |
| "loss": 2.277, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.0971737622883515e-08, | |
| "loss": 2.0345, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.322583110392692e-09, | |
| "loss": 2.0743, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.330917436402791e-09, | |
| "loss": 2.0902, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0, | |
| "loss": 2.07, | |
| "step": 300 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 300, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 2.591436490764288e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |