| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.2953367875647668, | |
| "eval_steps": 10000000, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.012953367875647668, | |
| "grad_norm": 25.608994557453922, | |
| "learning_rate": 6.476683937823834e-09, | |
| "loss": 2.9207, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.025906735751295335, | |
| "grad_norm": 25.006509435749546, | |
| "learning_rate": 1.2953367875647667e-08, | |
| "loss": 2.8035, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.038860103626943004, | |
| "grad_norm": 26.07092571978953, | |
| "learning_rate": 1.9430051813471502e-08, | |
| "loss": 2.8884, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05181347150259067, | |
| "grad_norm": 24.9115069323374, | |
| "learning_rate": 2.5906735751295334e-08, | |
| "loss": 2.8786, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06476683937823834, | |
| "grad_norm": 26.35515887725646, | |
| "learning_rate": 3.238341968911917e-08, | |
| "loss": 2.8856, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07772020725388601, | |
| "grad_norm": 26.802778574551475, | |
| "learning_rate": 3.8860103626943005e-08, | |
| "loss": 2.8532, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09067357512953368, | |
| "grad_norm": 26.534962747282748, | |
| "learning_rate": 4.533678756476684e-08, | |
| "loss": 2.8952, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.10362694300518134, | |
| "grad_norm": 23.051599328506548, | |
| "learning_rate": 5.181347150259067e-08, | |
| "loss": 2.8437, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11658031088082901, | |
| "grad_norm": 24.471329875902512, | |
| "learning_rate": 5.8290155440414504e-08, | |
| "loss": 2.8237, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12953367875647667, | |
| "grad_norm": 24.33959816120034, | |
| "learning_rate": 6.476683937823834e-08, | |
| "loss": 2.8199, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14248704663212436, | |
| "grad_norm": 25.214396561848275, | |
| "learning_rate": 7.124352331606218e-08, | |
| "loss": 2.8259, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15544041450777202, | |
| "grad_norm": 20.508728927761148, | |
| "learning_rate": 7.772020725388601e-08, | |
| "loss": 2.7907, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16839378238341968, | |
| "grad_norm": 21.133956193530185, | |
| "learning_rate": 8.419689119170984e-08, | |
| "loss": 2.7776, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.18134715025906736, | |
| "grad_norm": 20.216670897828962, | |
| "learning_rate": 9.067357512953368e-08, | |
| "loss": 2.6967, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.19430051813471502, | |
| "grad_norm": 19.577892832244164, | |
| "learning_rate": 9.715025906735751e-08, | |
| "loss": 2.6823, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.20725388601036268, | |
| "grad_norm": 14.26088592523892, | |
| "learning_rate": 1.0362694300518134e-07, | |
| "loss": 2.6147, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.22020725388601037, | |
| "grad_norm": 12.389614421651768, | |
| "learning_rate": 1.1010362694300518e-07, | |
| "loss": 2.5705, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.23316062176165803, | |
| "grad_norm": 10.537108203822184, | |
| "learning_rate": 1.1658031088082901e-07, | |
| "loss": 2.5779, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.24611398963730569, | |
| "grad_norm": 8.566097835293071, | |
| "learning_rate": 1.2305699481865284e-07, | |
| "loss": 2.5464, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25906735751295334, | |
| "grad_norm": 7.5406588821241884, | |
| "learning_rate": 1.2953367875647668e-07, | |
| "loss": 2.5462, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.27202072538860106, | |
| "grad_norm": 6.406776125251227, | |
| "learning_rate": 1.3601036269430052e-07, | |
| "loss": 2.4201, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2849740932642487, | |
| "grad_norm": 5.0333340242378135, | |
| "learning_rate": 1.4248704663212436e-07, | |
| "loss": 2.4404, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2979274611398964, | |
| "grad_norm": 4.96460074385302, | |
| "learning_rate": 1.4896373056994818e-07, | |
| "loss": 2.4234, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.31088082901554404, | |
| "grad_norm": 4.718132167499178, | |
| "learning_rate": 1.5544041450777202e-07, | |
| "loss": 2.447, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.3238341968911917, | |
| "grad_norm": 4.473414832247284, | |
| "learning_rate": 1.6191709844559583e-07, | |
| "loss": 2.3905, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.33678756476683935, | |
| "grad_norm": 4.231641138749174, | |
| "learning_rate": 1.6839378238341968e-07, | |
| "loss": 2.3678, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.34974093264248707, | |
| "grad_norm": 3.950230234878744, | |
| "learning_rate": 1.7487046632124352e-07, | |
| "loss": 2.3378, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3626943005181347, | |
| "grad_norm": 3.789360959290258, | |
| "learning_rate": 1.8134715025906736e-07, | |
| "loss": 2.3507, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.3756476683937824, | |
| "grad_norm": 3.752329492676831, | |
| "learning_rate": 1.8782383419689118e-07, | |
| "loss": 2.3752, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.38860103626943004, | |
| "grad_norm": 3.8201206496833313, | |
| "learning_rate": 1.9430051813471502e-07, | |
| "loss": 2.3437, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4015544041450777, | |
| "grad_norm": 3.8776779124718175, | |
| "learning_rate": 2.0077720207253883e-07, | |
| "loss": 2.3889, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.41450777202072536, | |
| "grad_norm": 3.989802925727435, | |
| "learning_rate": 2.0725388601036267e-07, | |
| "loss": 2.3129, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.4274611398963731, | |
| "grad_norm": 3.651464736452137, | |
| "learning_rate": 2.1373056994818652e-07, | |
| "loss": 2.3276, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.44041450777202074, | |
| "grad_norm": 3.784690196455203, | |
| "learning_rate": 2.2020725388601036e-07, | |
| "loss": 2.31, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.4533678756476684, | |
| "grad_norm": 3.5900123693564683, | |
| "learning_rate": 2.2668393782383417e-07, | |
| "loss": 2.3023, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.46632124352331605, | |
| "grad_norm": 3.616764531573499, | |
| "learning_rate": 2.3316062176165802e-07, | |
| "loss": 2.2977, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4792746113989637, | |
| "grad_norm": 3.4970049253402076, | |
| "learning_rate": 2.3963730569948183e-07, | |
| "loss": 2.32, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.49222797927461137, | |
| "grad_norm": 3.7715349943050733, | |
| "learning_rate": 2.4611398963730567e-07, | |
| "loss": 2.274, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5051813471502591, | |
| "grad_norm": 3.7123935945294897, | |
| "learning_rate": 2.525906735751295e-07, | |
| "loss": 2.3086, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5181347150259067, | |
| "grad_norm": 3.372291772174901, | |
| "learning_rate": 2.5906735751295336e-07, | |
| "loss": 2.2496, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5310880829015544, | |
| "grad_norm": 3.784294595936631, | |
| "learning_rate": 2.655440414507772e-07, | |
| "loss": 2.2858, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.5440414507772021, | |
| "grad_norm": 3.4351612072380675, | |
| "learning_rate": 2.7202072538860104e-07, | |
| "loss": 2.28, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5569948186528497, | |
| "grad_norm": 3.3411063107189753, | |
| "learning_rate": 2.7849740932642483e-07, | |
| "loss": 2.2764, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.5699481865284974, | |
| "grad_norm": 3.3765768734179993, | |
| "learning_rate": 2.849740932642487e-07, | |
| "loss": 2.3278, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.582901554404145, | |
| "grad_norm": 3.501479583740947, | |
| "learning_rate": 2.914507772020725e-07, | |
| "loss": 2.275, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5958549222797928, | |
| "grad_norm": 3.4545990517473895, | |
| "learning_rate": 2.9792746113989635e-07, | |
| "loss": 2.292, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.6088082901554405, | |
| "grad_norm": 3.4921131143644137, | |
| "learning_rate": 3.044041450777202e-07, | |
| "loss": 2.306, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6217616580310881, | |
| "grad_norm": 3.3569942808113202, | |
| "learning_rate": 3.1088082901554404e-07, | |
| "loss": 2.2818, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6347150259067358, | |
| "grad_norm": 3.473735761418456, | |
| "learning_rate": 3.173575129533679e-07, | |
| "loss": 2.292, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6476683937823834, | |
| "grad_norm": 3.4098418845884506, | |
| "learning_rate": 3.2383419689119167e-07, | |
| "loss": 2.2388, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6606217616580311, | |
| "grad_norm": 3.2884572160731103, | |
| "learning_rate": 3.303108808290155e-07, | |
| "loss": 2.2485, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.6735751295336787, | |
| "grad_norm": 3.49059998854233, | |
| "learning_rate": 3.3678756476683935e-07, | |
| "loss": 2.2546, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6865284974093264, | |
| "grad_norm": 3.228603206545892, | |
| "learning_rate": 3.432642487046632e-07, | |
| "loss": 2.2611, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6994818652849741, | |
| "grad_norm": 3.201214883425356, | |
| "learning_rate": 3.4974093264248704e-07, | |
| "loss": 2.2423, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7124352331606217, | |
| "grad_norm": 3.174572440904334, | |
| "learning_rate": 3.562176165803109e-07, | |
| "loss": 2.2575, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7253886010362695, | |
| "grad_norm": 3.6637540799374086, | |
| "learning_rate": 3.626943005181347e-07, | |
| "loss": 2.2703, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7383419689119171, | |
| "grad_norm": 3.2144805242394456, | |
| "learning_rate": 3.691709844559585e-07, | |
| "loss": 2.2536, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.7512953367875648, | |
| "grad_norm": 3.2832998982931647, | |
| "learning_rate": 3.7564766839378235e-07, | |
| "loss": 2.2923, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7642487046632125, | |
| "grad_norm": 3.339591881411431, | |
| "learning_rate": 3.8212435233160625e-07, | |
| "loss": 2.2582, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.7772020725388601, | |
| "grad_norm": 3.1793659057987114, | |
| "learning_rate": 3.8860103626943004e-07, | |
| "loss": 2.2725, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7901554404145078, | |
| "grad_norm": 3.105051267095931, | |
| "learning_rate": 3.950777202072539e-07, | |
| "loss": 2.2609, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.8031088082901554, | |
| "grad_norm": 3.1058601287467837, | |
| "learning_rate": 4.0155440414507767e-07, | |
| "loss": 2.2564, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.8160621761658031, | |
| "grad_norm": 3.2145397711050374, | |
| "learning_rate": 4.0803108808290156e-07, | |
| "loss": 2.2643, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8290155440414507, | |
| "grad_norm": 2.9574287184508896, | |
| "learning_rate": 4.1450777202072535e-07, | |
| "loss": 2.2198, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8419689119170984, | |
| "grad_norm": 3.385438951564467, | |
| "learning_rate": 4.209844559585492e-07, | |
| "loss": 2.2482, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.8549222797927462, | |
| "grad_norm": 3.232841492939655, | |
| "learning_rate": 4.2746113989637303e-07, | |
| "loss": 2.2553, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.8678756476683938, | |
| "grad_norm": 3.2498023529603226, | |
| "learning_rate": 4.339378238341969e-07, | |
| "loss": 2.2091, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.8808290155440415, | |
| "grad_norm": 3.387272764637681, | |
| "learning_rate": 4.404145077720207e-07, | |
| "loss": 2.2803, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.8937823834196891, | |
| "grad_norm": 3.205923523135513, | |
| "learning_rate": 4.468911917098445e-07, | |
| "loss": 2.2416, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.9067357512953368, | |
| "grad_norm": 3.328638392908686, | |
| "learning_rate": 4.5336787564766835e-07, | |
| "loss": 2.2512, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9196891191709845, | |
| "grad_norm": 3.223397811767207, | |
| "learning_rate": 4.5984455958549224e-07, | |
| "loss": 2.2233, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9326424870466321, | |
| "grad_norm": 2.903434123937875, | |
| "learning_rate": 4.6632124352331603e-07, | |
| "loss": 2.2221, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9455958549222798, | |
| "grad_norm": 3.167214093551616, | |
| "learning_rate": 4.7279792746113987e-07, | |
| "loss": 2.1797, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.9585492227979274, | |
| "grad_norm": 2.8228184362789936, | |
| "learning_rate": 4.792746113989637e-07, | |
| "loss": 2.2113, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9715025906735751, | |
| "grad_norm": 3.2547987666473506, | |
| "learning_rate": 4.857512953367875e-07, | |
| "loss": 2.2603, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.9844559585492227, | |
| "grad_norm": 3.2819941887944317, | |
| "learning_rate": 4.922279792746113e-07, | |
| "loss": 2.2197, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.9974093264248705, | |
| "grad_norm": 3.1793621366904765, | |
| "learning_rate": 4.987046632124352e-07, | |
| "loss": 2.1802, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.0103626943005182, | |
| "grad_norm": 3.129077964136437, | |
| "learning_rate": 5.05181347150259e-07, | |
| "loss": 2.2115, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.0233160621761659, | |
| "grad_norm": 3.624077927910293, | |
| "learning_rate": 5.116580310880829e-07, | |
| "loss": 2.2505, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.0362694300518134, | |
| "grad_norm": 3.001991950399265, | |
| "learning_rate": 5.181347150259067e-07, | |
| "loss": 2.2186, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.049222797927461, | |
| "grad_norm": 3.2056710194420077, | |
| "learning_rate": 5.246113989637306e-07, | |
| "loss": 2.2441, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.0621761658031088, | |
| "grad_norm": 3.3095386913147307, | |
| "learning_rate": 5.310880829015544e-07, | |
| "loss": 2.2022, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.0751295336787565, | |
| "grad_norm": 3.2748460342108276, | |
| "learning_rate": 5.375647668393782e-07, | |
| "loss": 2.2102, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.0880829015544042, | |
| "grad_norm": 3.1482065886570108, | |
| "learning_rate": 5.440414507772021e-07, | |
| "loss": 2.186, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.1010362694300517, | |
| "grad_norm": 3.2541206943048016, | |
| "learning_rate": 5.505181347150258e-07, | |
| "loss": 2.2447, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.1139896373056994, | |
| "grad_norm": 3.35168061407981, | |
| "learning_rate": 5.569948186528497e-07, | |
| "loss": 2.287, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.1269430051813472, | |
| "grad_norm": 3.324702715900074, | |
| "learning_rate": 5.634715025906735e-07, | |
| "loss": 2.2381, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.1398963730569949, | |
| "grad_norm": 2.981057376298506, | |
| "learning_rate": 5.699481865284974e-07, | |
| "loss": 2.253, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.1528497409326426, | |
| "grad_norm": 3.2319048665014734, | |
| "learning_rate": 5.764248704663213e-07, | |
| "loss": 2.234, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.16580310880829, | |
| "grad_norm": 3.197206460323895, | |
| "learning_rate": 5.82901554404145e-07, | |
| "loss": 2.2401, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.1787564766839378, | |
| "grad_norm": 3.128653204841994, | |
| "learning_rate": 5.893782383419689e-07, | |
| "loss": 2.2044, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.1917098445595855, | |
| "grad_norm": 3.194111726652451, | |
| "learning_rate": 5.958549222797927e-07, | |
| "loss": 2.1694, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.2046632124352332, | |
| "grad_norm": 3.07095427002542, | |
| "learning_rate": 6.023316062176166e-07, | |
| "loss": 2.2416, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.2176165803108807, | |
| "grad_norm": 3.0432092192539777, | |
| "learning_rate": 6.088082901554404e-07, | |
| "loss": 2.2327, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.2305699481865284, | |
| "grad_norm": 3.045795113156715, | |
| "learning_rate": 6.152849740932642e-07, | |
| "loss": 2.1957, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.2435233160621761, | |
| "grad_norm": 3.363436323137103, | |
| "learning_rate": 6.217616580310881e-07, | |
| "loss": 2.2182, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.2564766839378239, | |
| "grad_norm": 3.0067901045334633, | |
| "learning_rate": 6.282383419689119e-07, | |
| "loss": 2.1986, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.2694300518134716, | |
| "grad_norm": 3.047870744487349, | |
| "learning_rate": 6.347150259067358e-07, | |
| "loss": 2.2217, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.2823834196891193, | |
| "grad_norm": 3.2325772640891395, | |
| "learning_rate": 6.411917098445595e-07, | |
| "loss": 2.2146, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.2953367875647668, | |
| "grad_norm": 3.2882251454495006, | |
| "learning_rate": 6.476683937823833e-07, | |
| "loss": 2.203, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 15440, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 15311042248704.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |