| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.08982708286548394, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00044913541432741973, | |
| "grad_norm": 1.8347047443967313e-05, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 46.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00044913541432741973, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 24.1819, | |
| "eval_samples_per_second": 155.074, | |
| "eval_steps_per_second": 77.537, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008982708286548395, | |
| "grad_norm": 1.178028014692245e-05, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 46.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001347406242982259, | |
| "grad_norm": 1.920527938636951e-05, | |
| "learning_rate": 1e-05, | |
| "loss": 46.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.001796541657309679, | |
| "grad_norm": 1.3640981705975719e-05, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0022456770716370988, | |
| "grad_norm": 1.654174593568314e-05, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002694812485964518, | |
| "grad_norm": 2.7698526537278667e-05, | |
| "learning_rate": 2e-05, | |
| "loss": 46.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.003143947900291938, | |
| "grad_norm": 2.164217221434228e-05, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 46.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003593083314619358, | |
| "grad_norm": 1.6897010937100276e-05, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.004042218728946777, | |
| "grad_norm": 1.6789213987067342e-05, | |
| "learning_rate": 3e-05, | |
| "loss": 46.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0044913541432741975, | |
| "grad_norm": 1.6697740647941828e-05, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 46.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004940489557601617, | |
| "grad_norm": 1.7767843019100837e-05, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 46.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005389624971929036, | |
| "grad_norm": 1.6536820112378336e-05, | |
| "learning_rate": 4e-05, | |
| "loss": 46.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.005838760386256457, | |
| "grad_norm": 1.60211438924307e-05, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.006287895800583876, | |
| "grad_norm": 1.2699316357611679e-05, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0067370312149112955, | |
| "grad_norm": 1.592025364516303e-05, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.007186166629238716, | |
| "grad_norm": 1.6692121789674275e-05, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007635302043566135, | |
| "grad_norm": 1.1570506103453226e-05, | |
| "learning_rate": 5.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.008084437457893555, | |
| "grad_norm": 1.7659904187894426e-05, | |
| "learning_rate": 6e-05, | |
| "loss": 46.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.008533572872220974, | |
| "grad_norm": 2.8337542971712537e-05, | |
| "learning_rate": 6.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.008982708286548395, | |
| "grad_norm": 2.210317506978754e-05, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.009431843700875815, | |
| "grad_norm": 1.847420571721159e-05, | |
| "learning_rate": 7e-05, | |
| "loss": 46.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009880979115203234, | |
| "grad_norm": 1.4571529391105287e-05, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 46.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.010330114529530653, | |
| "grad_norm": 2.3006352421361953e-05, | |
| "learning_rate": 7.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010779249943858073, | |
| "grad_norm": 1.4708183698530775e-05, | |
| "learning_rate": 8e-05, | |
| "loss": 46.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.011228385358185492, | |
| "grad_norm": 1.1890616406162735e-05, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.011677520772512913, | |
| "grad_norm": 1.3593901712738443e-05, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.012126656186840333, | |
| "grad_norm": 2.0232280803611502e-05, | |
| "learning_rate": 9e-05, | |
| "loss": 46.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.012575791601167752, | |
| "grad_norm": 2.0324196157162078e-05, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 46.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.013024927015495171, | |
| "grad_norm": 1.1772674952226225e-05, | |
| "learning_rate": 9.666666666666667e-05, | |
| "loss": 46.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.013474062429822591, | |
| "grad_norm": 2.784149364742916e-05, | |
| "learning_rate": 0.0001, | |
| "loss": 46.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013923197844150012, | |
| "grad_norm": 1.5066696505527943e-05, | |
| "learning_rate": 9.999146252290264e-05, | |
| "loss": 46.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.014372333258477431, | |
| "grad_norm": 2.315385063411668e-05, | |
| "learning_rate": 9.996585300715116e-05, | |
| "loss": 46.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01482146867280485, | |
| "grad_norm": 1.594394416315481e-05, | |
| "learning_rate": 9.99231801983717e-05, | |
| "loss": 46.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01527060408713227, | |
| "grad_norm": 3.582621138775721e-05, | |
| "learning_rate": 9.986345866928941e-05, | |
| "loss": 46.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01571973950145969, | |
| "grad_norm": 1.7413909517927095e-05, | |
| "learning_rate": 9.978670881475172e-05, | |
| "loss": 46.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01616887491578711, | |
| "grad_norm": 1.7645959815126844e-05, | |
| "learning_rate": 9.96929568447637e-05, | |
| "loss": 46.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01661801033011453, | |
| "grad_norm": 1.7726713849697262e-05, | |
| "learning_rate": 9.958223477553714e-05, | |
| "loss": 46.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.017067145744441948, | |
| "grad_norm": 1.669113407842815e-05, | |
| "learning_rate": 9.94545804185573e-05, | |
| "loss": 46.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.017516281158769367, | |
| "grad_norm": 3.179280611220747e-05, | |
| "learning_rate": 9.931003736767013e-05, | |
| "loss": 46.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.01796541657309679, | |
| "grad_norm": 1.825378421926871e-05, | |
| "learning_rate": 9.91486549841951e-05, | |
| "loss": 46.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01841455198742421, | |
| "grad_norm": 1.339137634204235e-05, | |
| "learning_rate": 9.89704883800683e-05, | |
| "loss": 46.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.01886368740175163, | |
| "grad_norm": 2.6022458769148216e-05, | |
| "learning_rate": 9.877559839902184e-05, | |
| "loss": 46.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01931282281607905, | |
| "grad_norm": 2.146918086509686e-05, | |
| "learning_rate": 9.85640515958057e-05, | |
| "loss": 46.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.019761958230406468, | |
| "grad_norm": 2.5427123546251096e-05, | |
| "learning_rate": 9.833592021345937e-05, | |
| "loss": 46.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.020211093644733887, | |
| "grad_norm": 3.769697286770679e-05, | |
| "learning_rate": 9.809128215864097e-05, | |
| "loss": 46.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.020660229059061307, | |
| "grad_norm": 2.2639473172603175e-05, | |
| "learning_rate": 9.783022097502204e-05, | |
| "loss": 46.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.021109364473388726, | |
| "grad_norm": 5.1706858357647434e-05, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 46.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.021558499887716145, | |
| "grad_norm": 2.7394435164751485e-05, | |
| "learning_rate": 9.725919140804099e-05, | |
| "loss": 46.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.022007635302043565, | |
| "grad_norm": 3.286890569142997e-05, | |
| "learning_rate": 9.694941803075283e-05, | |
| "loss": 46.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.022456770716370984, | |
| "grad_norm": 4.607827577274293e-05, | |
| "learning_rate": 9.662361147021779e-05, | |
| "loss": 46.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022456770716370984, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 24.1846, | |
| "eval_samples_per_second": 155.057, | |
| "eval_steps_per_second": 77.529, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022905906130698407, | |
| "grad_norm": 1.5578716556774452e-05, | |
| "learning_rate": 9.628188298907782e-05, | |
| "loss": 46.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.023355041545025827, | |
| "grad_norm": 1.1132679901493248e-05, | |
| "learning_rate": 9.592434928729616e-05, | |
| "loss": 46.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.023804176959353246, | |
| "grad_norm": 2.1629448383464478e-05, | |
| "learning_rate": 9.555113246230442e-05, | |
| "loss": 46.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.024253312373680665, | |
| "grad_norm": 2.475133806001395e-05, | |
| "learning_rate": 9.516235996730645e-05, | |
| "loss": 46.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.024702447788008085, | |
| "grad_norm": 3.32804411300458e-05, | |
| "learning_rate": 9.475816456775313e-05, | |
| "loss": 46.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.025151583202335504, | |
| "grad_norm": 3.513158662826754e-05, | |
| "learning_rate": 9.43386842960031e-05, | |
| "loss": 46.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.025600718616662924, | |
| "grad_norm": 2.9858412744943053e-05, | |
| "learning_rate": 9.39040624041849e-05, | |
| "loss": 46.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.026049854030990343, | |
| "grad_norm": 3.1021958420751616e-05, | |
| "learning_rate": 9.345444731527642e-05, | |
| "loss": 46.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.026498989445317762, | |
| "grad_norm": 3.1924246286507696e-05, | |
| "learning_rate": 9.298999257241863e-05, | |
| "loss": 46.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.026948124859645182, | |
| "grad_norm": 2.357485755055677e-05, | |
| "learning_rate": 9.251085678648072e-05, | |
| "loss": 46.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0273972602739726, | |
| "grad_norm": 4.4487918785307556e-05, | |
| "learning_rate": 9.201720358189464e-05, | |
| "loss": 46.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.027846395688300024, | |
| "grad_norm": 6.167806714074686e-05, | |
| "learning_rate": 9.150920154077754e-05, | |
| "loss": 46.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.028295531102627444, | |
| "grad_norm": 2.354049138375558e-05, | |
| "learning_rate": 9.098702414536107e-05, | |
| "loss": 46.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.028744666516954863, | |
| "grad_norm": 2.1689204004360363e-05, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 46.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.029193801931282282, | |
| "grad_norm": 3.892782478942536e-05, | |
| "learning_rate": 8.9900861364012e-05, | |
| "loss": 46.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0296429373456097, | |
| "grad_norm": 3.200348874088377e-05, | |
| "learning_rate": 8.933724690167417e-05, | |
| "loss": 46.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03009207275993712, | |
| "grad_norm": 3.3991811505984515e-05, | |
| "learning_rate": 8.876019880555649e-05, | |
| "loss": 46.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03054120817426454, | |
| "grad_norm": 3.4045824577333406e-05, | |
| "learning_rate": 8.816991413705516e-05, | |
| "loss": 46.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03099034358859196, | |
| "grad_norm": 3.314235800644383e-05, | |
| "learning_rate": 8.756659447784368e-05, | |
| "loss": 46.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03143947900291938, | |
| "grad_norm": 4.097645796719007e-05, | |
| "learning_rate": 8.695044586103296e-05, | |
| "loss": 46.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0318886144172468, | |
| "grad_norm": 3.730815660674125e-05, | |
| "learning_rate": 8.632167870081121e-05, | |
| "loss": 46.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03233774983157422, | |
| "grad_norm": 4.687355612986721e-05, | |
| "learning_rate": 8.568050772058762e-05, | |
| "loss": 46.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03278688524590164, | |
| "grad_norm": 5.8023280871566385e-05, | |
| "learning_rate": 8.502715187966455e-05, | |
| "loss": 46.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03323602066022906, | |
| "grad_norm": 3.261071105953306e-05, | |
| "learning_rate": 8.436183429846313e-05, | |
| "loss": 46.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03368515607455648, | |
| "grad_norm": 6.231103907339275e-05, | |
| "learning_rate": 8.368478218232787e-05, | |
| "loss": 46.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.034134291488883896, | |
| "grad_norm": 5.714733197237365e-05, | |
| "learning_rate": 8.299622674393614e-05, | |
| "loss": 46.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03458342690321132, | |
| "grad_norm": 4.1154456994263455e-05, | |
| "learning_rate": 8.229640312433937e-05, | |
| "loss": 46.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.035032562317538735, | |
| "grad_norm": 9.290780144510791e-05, | |
| "learning_rate": 8.158555031266254e-05, | |
| "loss": 46.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03548169773186616, | |
| "grad_norm": 2.987302104884293e-05, | |
| "learning_rate": 8.086391106448965e-05, | |
| "loss": 46.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03593083314619358, | |
| "grad_norm": 3.9257214666577056e-05, | |
| "learning_rate": 8.013173181896283e-05, | |
| "loss": 46.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.036379968560520996, | |
| "grad_norm": 5.808036439702846e-05, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 46.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.03682910397484842, | |
| "grad_norm": 6.756263610441238e-05, | |
| "learning_rate": 7.863675700402526e-05, | |
| "loss": 46.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.037278239389175835, | |
| "grad_norm": 2.79947435046779e-05, | |
| "learning_rate": 7.787447196714427e-05, | |
| "loss": 46.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03772737480350326, | |
| "grad_norm": 6.227039557415992e-05, | |
| "learning_rate": 7.710266782362247e-05, | |
| "loss": 46.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.038176510217830674, | |
| "grad_norm": 6.594270962523296e-05, | |
| "learning_rate": 7.63216081438678e-05, | |
| "loss": 46.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0386256456321581, | |
| "grad_norm": 8.637751307105646e-05, | |
| "learning_rate": 7.553155965904535e-05, | |
| "loss": 46.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03907478104648551, | |
| "grad_norm": 0.00010642538836691529, | |
| "learning_rate": 7.473279216998895e-05, | |
| "loss": 46.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.039523916460812936, | |
| "grad_norm": 0.00010907182877417654, | |
| "learning_rate": 7.392557845506432e-05, | |
| "loss": 46.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03997305187514035, | |
| "grad_norm": 8.498036186210811e-05, | |
| "learning_rate": 7.311019417701566e-05, | |
| "loss": 46.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.040422187289467774, | |
| "grad_norm": 6.34290263406001e-05, | |
| "learning_rate": 7.228691778882693e-05, | |
| "loss": 46.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0408713227037952, | |
| "grad_norm": 4.713748785434291e-05, | |
| "learning_rate": 7.145603043863045e-05, | |
| "loss": 46.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04132045811812261, | |
| "grad_norm": 8.825240365695208e-05, | |
| "learning_rate": 7.061781587369519e-05, | |
| "loss": 46.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.041769593532450036, | |
| "grad_norm": 6.958351150387898e-05, | |
| "learning_rate": 6.977256034352712e-05, | |
| "loss": 46.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04221872894677745, | |
| "grad_norm": 8.253315172623843e-05, | |
| "learning_rate": 6.892055250211552e-05, | |
| "loss": 46.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.042667864361104875, | |
| "grad_norm": 6.820957059971988e-05, | |
| "learning_rate": 6.806208330935766e-05, | |
| "loss": 46.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04311699977543229, | |
| "grad_norm": 8.223598706535995e-05, | |
| "learning_rate": 6.719744593169641e-05, | |
| "loss": 46.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.043566135189759714, | |
| "grad_norm": 7.749118231004104e-05, | |
| "learning_rate": 6.632693564200416e-05, | |
| "loss": 46.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.04401527060408713, | |
| "grad_norm": 5.018940646550618e-05, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 46.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04446440601841455, | |
| "grad_norm": 0.00011348233965691179, | |
| "learning_rate": 6.456948734446624e-05, | |
| "loss": 46.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.04491354143274197, | |
| "grad_norm": 6.80896409903653e-05, | |
| "learning_rate": 6.368314950360415e-05, | |
| "loss": 46.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04491354143274197, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 24.1705, | |
| "eval_samples_per_second": 155.148, | |
| "eval_steps_per_second": 77.574, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04536267684706939, | |
| "grad_norm": 7.611372711835429e-05, | |
| "learning_rate": 6.279213887972179e-05, | |
| "loss": 46.0, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.045811812261396814, | |
| "grad_norm": 9.4139642897062e-05, | |
| "learning_rate": 6.189675975213094e-05, | |
| "loss": 46.0, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04626094767572423, | |
| "grad_norm": 5.4080392146715894e-05, | |
| "learning_rate": 6.099731789198344e-05, | |
| "loss": 46.0, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04671008309005165, | |
| "grad_norm": 5.4170694056665525e-05, | |
| "learning_rate": 6.009412045785051e-05, | |
| "loss": 46.0, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.04715921850437907, | |
| "grad_norm": 0.00016534763562958688, | |
| "learning_rate": 5.918747589082853e-05, | |
| "loss": 46.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04760835391870649, | |
| "grad_norm": 4.377401637611911e-05, | |
| "learning_rate": 5.82776938092065e-05, | |
| "loss": 46.0, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04805748933303391, | |
| "grad_norm": 6.235337787074968e-05, | |
| "learning_rate": 5.736508490273188e-05, | |
| "loss": 46.0, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04850662474736133, | |
| "grad_norm": 6.442901940317824e-05, | |
| "learning_rate": 5.644996082651017e-05, | |
| "loss": 46.0, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04895576016168875, | |
| "grad_norm": 0.000264510337729007, | |
| "learning_rate": 5.553263409457504e-05, | |
| "loss": 46.0, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.04940489557601617, | |
| "grad_norm": 0.00013142352690920234, | |
| "learning_rate": 5.4613417973165106e-05, | |
| "loss": 46.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.049854030990343585, | |
| "grad_norm": 4.1707517084432766e-05, | |
| "learning_rate": 5.3692626373743706e-05, | |
| "loss": 46.0, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.05030316640467101, | |
| "grad_norm": 9.263215906685218e-05, | |
| "learning_rate": 5.27705737457985e-05, | |
| "loss": 46.0, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.05075230181899843, | |
| "grad_norm": 4.6955046855146065e-05, | |
| "learning_rate": 5.184757496945726e-05, | |
| "loss": 46.0, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.05120143723332585, | |
| "grad_norm": 8.900297689251602e-05, | |
| "learning_rate": 5.092394524795649e-05, | |
| "loss": 46.0, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05165057264765327, | |
| "grad_norm": 4.0383063605986536e-05, | |
| "learning_rate": 5e-05, | |
| "loss": 46.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.052099708061980686, | |
| "grad_norm": 0.00014041118265595287, | |
| "learning_rate": 4.907605475204352e-05, | |
| "loss": 46.0, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.05254884347630811, | |
| "grad_norm": 4.6739267418161035e-05, | |
| "learning_rate": 4.8152425030542766e-05, | |
| "loss": 46.0, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.052997978890635525, | |
| "grad_norm": 4.048235496156849e-05, | |
| "learning_rate": 4.72294262542015e-05, | |
| "loss": 46.0, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.05344711430496295, | |
| "grad_norm": 0.0001663568546064198, | |
| "learning_rate": 4.6307373626256306e-05, | |
| "loss": 46.0, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.053896249719290364, | |
| "grad_norm": 5.4521020501852036e-05, | |
| "learning_rate": 4.5386582026834906e-05, | |
| "loss": 46.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.054345385133617786, | |
| "grad_norm": 7.633578206878155e-05, | |
| "learning_rate": 4.446736590542497e-05, | |
| "loss": 46.0, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0547945205479452, | |
| "grad_norm": 0.00010656577796908095, | |
| "learning_rate": 4.3550039173489845e-05, | |
| "loss": 46.0, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.055243655962272625, | |
| "grad_norm": 9.359400428365916e-05, | |
| "learning_rate": 4.2634915097268115e-05, | |
| "loss": 46.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05569279137660005, | |
| "grad_norm": 9.334505739388987e-05, | |
| "learning_rate": 4.1722306190793495e-05, | |
| "loss": 46.0, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.056141926790927464, | |
| "grad_norm": 0.00011861397797474638, | |
| "learning_rate": 4.0812524109171476e-05, | |
| "loss": 46.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05659106220525489, | |
| "grad_norm": 0.00010963834210997447, | |
| "learning_rate": 3.99058795421495e-05, | |
| "loss": 46.0, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0570401976195823, | |
| "grad_norm": 9.579626930644736e-05, | |
| "learning_rate": 3.9002682108016585e-05, | |
| "loss": 46.0, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.057489333033909726, | |
| "grad_norm": 9.241644147550687e-05, | |
| "learning_rate": 3.8103240247869075e-05, | |
| "loss": 46.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05793846844823714, | |
| "grad_norm": 0.0001144165507867001, | |
| "learning_rate": 3.720786112027822e-05, | |
| "loss": 46.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.058387603862564565, | |
| "grad_norm": 0.00014389632269740105, | |
| "learning_rate": 3.631685049639586e-05, | |
| "loss": 46.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05883673927689198, | |
| "grad_norm": 0.00012475732364691794, | |
| "learning_rate": 3.543051265553377e-05, | |
| "loss": 46.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0592858746912194, | |
| "grad_norm": 0.00019938830519095063, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 46.0, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.05973501010554682, | |
| "grad_norm": 0.00015618746692780405, | |
| "learning_rate": 3.367306435799584e-05, | |
| "loss": 46.0, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.06018414551987424, | |
| "grad_norm": 9.118753223447129e-05, | |
| "learning_rate": 3.2802554068303596e-05, | |
| "loss": 46.0, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.060633280934201665, | |
| "grad_norm": 0.0001693732338026166, | |
| "learning_rate": 3.1937916690642356e-05, | |
| "loss": 46.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.06108241634852908, | |
| "grad_norm": 0.00010582958930172026, | |
| "learning_rate": 3.107944749788449e-05, | |
| "loss": 46.0, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.061531551762856504, | |
| "grad_norm": 0.0001773022231645882, | |
| "learning_rate": 3.0227439656472877e-05, | |
| "loss": 46.0, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.06198068717718392, | |
| "grad_norm": 0.00011500522668939084, | |
| "learning_rate": 2.9382184126304834e-05, | |
| "loss": 46.0, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.06242982259151134, | |
| "grad_norm": 0.0001146023059845902, | |
| "learning_rate": 2.8543969561369556e-05, | |
| "loss": 46.0, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06287895800583876, | |
| "grad_norm": 9.070700616575778e-05, | |
| "learning_rate": 2.771308221117309e-05, | |
| "loss": 46.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06332809342016618, | |
| "grad_norm": 0.00012379302643239498, | |
| "learning_rate": 2.688980582298435e-05, | |
| "loss": 46.0, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0637772288344936, | |
| "grad_norm": 0.00010337288404116407, | |
| "learning_rate": 2.607442154493568e-05, | |
| "loss": 46.0, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06422636424882101, | |
| "grad_norm": 7.891904533607885e-05, | |
| "learning_rate": 2.5267207830011068e-05, | |
| "loss": 46.0, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06467549966314844, | |
| "grad_norm": 0.0001713550736894831, | |
| "learning_rate": 2.446844034095466e-05, | |
| "loss": 46.0, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06512463507747586, | |
| "grad_norm": 9.049165237229317e-05, | |
| "learning_rate": 2.3678391856132204e-05, | |
| "loss": 46.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06557377049180328, | |
| "grad_norm": 0.0001013107830658555, | |
| "learning_rate": 2.2897332176377528e-05, | |
| "loss": 46.0, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0660229059061307, | |
| "grad_norm": 7.251360511872917e-05, | |
| "learning_rate": 2.2125528032855724e-05, | |
| "loss": 46.0, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06647204132045811, | |
| "grad_norm": 0.00011842833191622049, | |
| "learning_rate": 2.136324299597474e-05, | |
| "loss": 46.0, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06692117673478554, | |
| "grad_norm": 6.4388572354801e-05, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 46.0, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06737031214911296, | |
| "grad_norm": 0.0001277444389415905, | |
| "learning_rate": 1.9868268181037185e-05, | |
| "loss": 46.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06737031214911296, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 24.1729, | |
| "eval_samples_per_second": 155.132, | |
| "eval_steps_per_second": 77.566, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06781944756344038, | |
| "grad_norm": 0.00017438446229789406, | |
| "learning_rate": 1.9136088935510362e-05, | |
| "loss": 46.0, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06826858297776779, | |
| "grad_norm": 9.030107321450487e-05, | |
| "learning_rate": 1.8414449687337464e-05, | |
| "loss": 46.0, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06871771839209521, | |
| "grad_norm": 0.0001165097564808093, | |
| "learning_rate": 1.7703596875660645e-05, | |
| "loss": 46.0, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.06916685380642264, | |
| "grad_norm": 7.154911145335063e-05, | |
| "learning_rate": 1.700377325606388e-05, | |
| "loss": 46.0, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.06961598922075006, | |
| "grad_norm": 8.026312571018934e-05, | |
| "learning_rate": 1.631521781767214e-05, | |
| "loss": 46.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.07006512463507747, | |
| "grad_norm": 0.0001564437843626365, | |
| "learning_rate": 1.5638165701536868e-05, | |
| "loss": 46.0, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.07051426004940489, | |
| "grad_norm": 9.968695667339489e-05, | |
| "learning_rate": 1.4972848120335453e-05, | |
| "loss": 46.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.07096339546373231, | |
| "grad_norm": 0.0001090860168915242, | |
| "learning_rate": 1.4319492279412388e-05, | |
| "loss": 46.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.07141253087805974, | |
| "grad_norm": 0.0001526081468909979, | |
| "learning_rate": 1.3678321299188801e-05, | |
| "loss": 46.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.07186166629238716, | |
| "grad_norm": 0.0001304527249885723, | |
| "learning_rate": 1.3049554138967051e-05, | |
| "loss": 46.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07231080170671457, | |
| "grad_norm": 6.977524026297033e-05, | |
| "learning_rate": 1.2433405522156332e-05, | |
| "loss": 46.0, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07275993712104199, | |
| "grad_norm": 7.938971975818276e-05, | |
| "learning_rate": 1.183008586294485e-05, | |
| "loss": 46.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07320907253536942, | |
| "grad_norm": 7.140285015339032e-05, | |
| "learning_rate": 1.1239801194443506e-05, | |
| "loss": 46.0, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.07365820794969684, | |
| "grad_norm": 0.00010763693717308342, | |
| "learning_rate": 1.066275309832584e-05, | |
| "loss": 46.0, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.07410734336402425, | |
| "grad_norm": 0.00019754729873966426, | |
| "learning_rate": 1.0099138635988026e-05, | |
| "loss": 46.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07455647877835167, | |
| "grad_norm": 0.00010536264016991481, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 46.0, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.0750056141926791, | |
| "grad_norm": 0.00015691340377088636, | |
| "learning_rate": 9.012975854638949e-06, | |
| "loss": 46.0, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07545474960700652, | |
| "grad_norm": 9.983302152249962e-05, | |
| "learning_rate": 8.490798459222476e-06, | |
| "loss": 46.0, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07590388502133394, | |
| "grad_norm": 6.743606354575604e-05, | |
| "learning_rate": 7.982796418105371e-06, | |
| "loss": 46.0, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.07635302043566135, | |
| "grad_norm": 0.0001256086106877774, | |
| "learning_rate": 7.489143213519301e-06, | |
| "loss": 46.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07680215584998877, | |
| "grad_norm": 9.930363012244925e-05, | |
| "learning_rate": 7.010007427581378e-06, | |
| "loss": 46.0, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0772512912643162, | |
| "grad_norm": 0.000161067713634111, | |
| "learning_rate": 6.5455526847235825e-06, | |
| "loss": 46.0, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07770042667864362, | |
| "grad_norm": 9.258305362891406e-05, | |
| "learning_rate": 6.0959375958151045e-06, | |
| "loss": 46.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07814956209297103, | |
| "grad_norm": 0.00010014686995418742, | |
| "learning_rate": 5.6613157039969055e-06, | |
| "loss": 46.0, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07859869750729845, | |
| "grad_norm": 0.00010816480062203482, | |
| "learning_rate": 5.241835432246889e-06, | |
| "loss": 46.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07904783292162587, | |
| "grad_norm": 6.613504228880629e-05, | |
| "learning_rate": 4.837640032693558e-06, | |
| "loss": 46.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.0794969683359533, | |
| "grad_norm": 0.00016118783969432116, | |
| "learning_rate": 4.448867537695578e-06, | |
| "loss": 46.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0799461037502807, | |
| "grad_norm": 0.00010963629756588489, | |
| "learning_rate": 4.075650712703849e-06, | |
| "loss": 46.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.08039523916460813, | |
| "grad_norm": 8.793177403276786e-05, | |
| "learning_rate": 3.71811701092219e-06, | |
| "loss": 46.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.08084437457893555, | |
| "grad_norm": 7.429590914398432e-05, | |
| "learning_rate": 3.376388529782215e-06, | |
| "loss": 46.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08129350999326297, | |
| "grad_norm": 0.00013769243378192186, | |
| "learning_rate": 3.0505819692471792e-06, | |
| "loss": 46.0, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.0817426454075904, | |
| "grad_norm": 7.244458538480103e-05, | |
| "learning_rate": 2.7408085919590264e-06, | |
| "loss": 46.0, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.0821917808219178, | |
| "grad_norm": 0.00016487624088767916, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 46.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.08264091623624523, | |
| "grad_norm": 0.000193040439626202, | |
| "learning_rate": 2.1697790249779636e-06, | |
| "loss": 46.0, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08309005165057265, | |
| "grad_norm": 0.00019107495609205216, | |
| "learning_rate": 1.908717841359048e-06, | |
| "loss": 46.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08353918706490007, | |
| "grad_norm": 0.0001634787186048925, | |
| "learning_rate": 1.6640797865406288e-06, | |
| "loss": 46.0, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08398832247922748, | |
| "grad_norm": 5.865022467332892e-05, | |
| "learning_rate": 1.4359484041943038e-06, | |
| "loss": 46.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0844374578935549, | |
| "grad_norm": 0.0001328428043052554, | |
| "learning_rate": 1.2244016009781701e-06, | |
| "loss": 46.0, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08488659330788233, | |
| "grad_norm": 0.00016936052998062223, | |
| "learning_rate": 1.0295116199317057e-06, | |
| "loss": 46.0, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.08533572872220975, | |
| "grad_norm": 0.00014457838551606983, | |
| "learning_rate": 8.513450158049108e-07, | |
| "loss": 46.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.08578486413653717, | |
| "grad_norm": 8.662261825520545e-05, | |
| "learning_rate": 6.899626323298713e-07, | |
| "loss": 46.0, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08623399955086458, | |
| "grad_norm": 4.997271753381938e-05, | |
| "learning_rate": 5.454195814427021e-07, | |
| "loss": 46.0, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.086683134965192, | |
| "grad_norm": 0.00014428974827751517, | |
| "learning_rate": 4.177652244628627e-07, | |
| "loss": 46.0, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.08713227037951943, | |
| "grad_norm": 0.0001020684139803052, | |
| "learning_rate": 3.0704315523631953e-07, | |
| "loss": 46.0, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08758140579384685, | |
| "grad_norm": 0.00013089347339700907, | |
| "learning_rate": 2.1329118524827662e-07, | |
| "loss": 46.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.08803054120817426, | |
| "grad_norm": 0.00010549910075496882, | |
| "learning_rate": 1.3654133071059893e-07, | |
| "loss": 46.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08847967662250168, | |
| "grad_norm": 0.00019823636102955788, | |
| "learning_rate": 7.681980162830282e-08, | |
| "loss": 46.0, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.0889288120368291, | |
| "grad_norm": 0.0002306640089955181, | |
| "learning_rate": 3.4146992848854695e-08, | |
| "loss": 46.0, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.08937794745115653, | |
| "grad_norm": 9.864150342764333e-05, | |
| "learning_rate": 8.537477097364522e-09, | |
| "loss": 46.0, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.08982708286548394, | |
| "grad_norm": 0.00015147426165640354, | |
| "learning_rate": 0.0, | |
| "loss": 46.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08982708286548394, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 24.1793, | |
| "eval_samples_per_second": 155.091, | |
| "eval_steps_per_second": 77.546, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 32399651241984.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |