| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 2122, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.5460187196731567, | |
| "learning_rate": 3.5211267605633804e-06, | |
| "loss": 2.7305, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1.3390659093856812, | |
| "learning_rate": 7.042253521126761e-06, | |
| "loss": 2.7192, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 1.4170739650726318, | |
| "learning_rate": 1.056338028169014e-05, | |
| "loss": 2.6137, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.9075584411621094, | |
| "learning_rate": 1.4084507042253522e-05, | |
| "loss": 2.3702, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.8087872862815857, | |
| "learning_rate": 1.7605633802816902e-05, | |
| "loss": 2.0872, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.6687940359115601, | |
| "learning_rate": 2.112676056338028e-05, | |
| "loss": 1.9849, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.48965755105018616, | |
| "learning_rate": 2.464788732394366e-05, | |
| "loss": 1.8513, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.44189536571502686, | |
| "learning_rate": 2.8169014084507043e-05, | |
| "loss": 1.7901, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.4670378863811493, | |
| "learning_rate": 2.9827134625458355e-05, | |
| "loss": 1.7426, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.5341612100601196, | |
| "learning_rate": 2.9434258774227347e-05, | |
| "loss": 1.663, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.32935911417007446, | |
| "learning_rate": 2.9041382922996336e-05, | |
| "loss": 1.6262, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.49150750041007996, | |
| "learning_rate": 2.864850707176532e-05, | |
| "loss": 1.5616, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.4343280792236328, | |
| "learning_rate": 2.825563122053431e-05, | |
| "loss": 1.554, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.46223270893096924, | |
| "learning_rate": 2.78627553693033e-05, | |
| "loss": 1.4944, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.5165560245513916, | |
| "learning_rate": 2.7469879518072288e-05, | |
| "loss": 1.4857, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.4836352467536926, | |
| "learning_rate": 2.707700366684128e-05, | |
| "loss": 1.4545, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.5089160799980164, | |
| "learning_rate": 2.668412781561027e-05, | |
| "loss": 1.3965, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.4400924742221832, | |
| "learning_rate": 2.6291251964379258e-05, | |
| "loss": 1.3703, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.520064115524292, | |
| "learning_rate": 2.5898376113148247e-05, | |
| "loss": 1.3869, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.4993634819984436, | |
| "learning_rate": 2.5505500261917232e-05, | |
| "loss": 1.3827, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.4747565686702728, | |
| "learning_rate": 2.511262441068622e-05, | |
| "loss": 1.3494, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.45222511887550354, | |
| "learning_rate": 2.4719748559455214e-05, | |
| "loss": 1.3295, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.5314898490905762, | |
| "learning_rate": 2.4326872708224202e-05, | |
| "loss": 1.2971, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.4976591467857361, | |
| "learning_rate": 2.393399685699319e-05, | |
| "loss": 1.2882, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.5847007632255554, | |
| "learning_rate": 2.354112100576218e-05, | |
| "loss": 1.3001, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.4775733947753906, | |
| "learning_rate": 2.314824515453117e-05, | |
| "loss": 1.3185, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.4997495710849762, | |
| "learning_rate": 2.2755369303300158e-05, | |
| "loss": 1.31, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.46225643157958984, | |
| "learning_rate": 2.2362493452069147e-05, | |
| "loss": 1.2577, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.5184119939804077, | |
| "learning_rate": 2.1969617600838136e-05, | |
| "loss": 1.2846, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.5487437844276428, | |
| "learning_rate": 2.1576741749607125e-05, | |
| "loss": 1.234, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.44502514600753784, | |
| "learning_rate": 2.1183865898376113e-05, | |
| "loss": 1.2652, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.5340321660041809, | |
| "learning_rate": 2.0790990047145102e-05, | |
| "loss": 1.2302, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.46381640434265137, | |
| "learning_rate": 2.039811419591409e-05, | |
| "loss": 1.2402, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.5519339442253113, | |
| "learning_rate": 2.0005238344683083e-05, | |
| "loss": 1.2924, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.47584566473960876, | |
| "learning_rate": 1.9612362493452072e-05, | |
| "loss": 1.2526, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.5393884778022766, | |
| "learning_rate": 1.9219486642221058e-05, | |
| "loss": 1.188, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.5071331858634949, | |
| "learning_rate": 1.8826610790990047e-05, | |
| "loss": 1.228, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.5538953542709351, | |
| "learning_rate": 1.8433734939759036e-05, | |
| "loss": 1.238, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.4920578598976135, | |
| "learning_rate": 1.8040859088528024e-05, | |
| "loss": 1.1884, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.5614542961120605, | |
| "learning_rate": 1.7647983237297017e-05, | |
| "loss": 1.1734, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.514637291431427, | |
| "learning_rate": 1.7255107386066006e-05, | |
| "loss": 1.1938, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.6818568110466003, | |
| "learning_rate": 1.6862231534834994e-05, | |
| "loss": 1.2151, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.5652730464935303, | |
| "learning_rate": 1.646935568360398e-05, | |
| "loss": 1.1932, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.46096867322921753, | |
| "learning_rate": 1.607647983237297e-05, | |
| "loss": 1.1895, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 0.5606017112731934, | |
| "learning_rate": 1.5683603981141958e-05, | |
| "loss": 1.1812, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.6677452325820923, | |
| "learning_rate": 1.5290728129910947e-05, | |
| "loss": 1.2344, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 0.4959929585456848, | |
| "learning_rate": 1.4897852278679937e-05, | |
| "loss": 1.1922, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 0.5288898348808289, | |
| "learning_rate": 1.4504976427448928e-05, | |
| "loss": 1.1824, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 0.49359196424484253, | |
| "learning_rate": 1.4112100576217915e-05, | |
| "loss": 1.2163, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 0.5195351839065552, | |
| "learning_rate": 1.3719224724986904e-05, | |
| "loss": 1.1688, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.5838152766227722, | |
| "learning_rate": 1.3326348873755894e-05, | |
| "loss": 1.1566, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 0.5872088074684143, | |
| "learning_rate": 1.2933473022524883e-05, | |
| "loss": 1.1685, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.5522224307060242, | |
| "learning_rate": 1.254059717129387e-05, | |
| "loss": 1.1745, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 0.56575608253479, | |
| "learning_rate": 1.2147721320062861e-05, | |
| "loss": 1.1895, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 0.5466055870056152, | |
| "learning_rate": 1.175484546883185e-05, | |
| "loss": 1.1445, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 0.5417299866676331, | |
| "learning_rate": 1.1361969617600839e-05, | |
| "loss": 1.2091, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 0.6165823936462402, | |
| "learning_rate": 1.0969093766369827e-05, | |
| "loss": 1.1533, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 0.6337828636169434, | |
| "learning_rate": 1.0576217915138816e-05, | |
| "loss": 1.1721, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 0.5324290990829468, | |
| "learning_rate": 1.0183342063907805e-05, | |
| "loss": 1.2054, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 0.5414568185806274, | |
| "learning_rate": 9.790466212676796e-06, | |
| "loss": 1.164, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.5330114364624023, | |
| "learning_rate": 9.397590361445783e-06, | |
| "loss": 1.1632, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 0.4752555787563324, | |
| "learning_rate": 9.004714510214772e-06, | |
| "loss": 1.123, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.5737998485565186, | |
| "learning_rate": 8.61183865898376e-06, | |
| "loss": 1.168, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 0.5172936320304871, | |
| "learning_rate": 8.218962807752751e-06, | |
| "loss": 1.1856, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 0.6017385721206665, | |
| "learning_rate": 7.826086956521738e-06, | |
| "loss": 1.1571, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 0.6177796125411987, | |
| "learning_rate": 7.433211105290728e-06, | |
| "loss": 1.166, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 0.5164004564285278, | |
| "learning_rate": 7.040335254059718e-06, | |
| "loss": 1.1979, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.6410110592842102, | |
| "learning_rate": 6.647459402828706e-06, | |
| "loss": 1.156, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 0.5372936129570007, | |
| "learning_rate": 6.254583551597696e-06, | |
| "loss": 1.1532, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 0.6036843657493591, | |
| "learning_rate": 5.8617077003666845e-06, | |
| "loss": 1.1793, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 0.6881660223007202, | |
| "learning_rate": 5.468831849135673e-06, | |
| "loss": 1.1709, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 0.5717695951461792, | |
| "learning_rate": 5.075955997904662e-06, | |
| "loss": 1.1441, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 0.5951740145683289, | |
| "learning_rate": 4.683080146673651e-06, | |
| "loss": 1.1855, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 0.628299355506897, | |
| "learning_rate": 4.29020429544264e-06, | |
| "loss": 1.1888, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 0.5050541758537292, | |
| "learning_rate": 3.89732844421163e-06, | |
| "loss": 1.1396, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "grad_norm": 0.5191190838813782, | |
| "learning_rate": 3.504452592980618e-06, | |
| "loss": 1.1955, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 0.5869964957237244, | |
| "learning_rate": 3.111576741749607e-06, | |
| "loss": 1.1314, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 0.5007169842720032, | |
| "learning_rate": 2.718700890518596e-06, | |
| "loss": 1.1516, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 0.6014929413795471, | |
| "learning_rate": 2.325825039287585e-06, | |
| "loss": 1.1901, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 0.8301869034767151, | |
| "learning_rate": 1.932949188056574e-06, | |
| "loss": 1.1197, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 0.5901632905006409, | |
| "learning_rate": 1.5400733368255633e-06, | |
| "loss": 1.1843, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "grad_norm": 0.580962598323822, | |
| "learning_rate": 1.1471974855945522e-06, | |
| "loss": 1.1456, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 0.5481582283973694, | |
| "learning_rate": 7.543216343635412e-07, | |
| "loss": 1.1644, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 0.5194025039672852, | |
| "learning_rate": 3.6144578313253016e-07, | |
| "loss": 1.1581, | |
| "step": 2100 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 2122, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "total_flos": 4558731463360512.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |