{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 620, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0016129032258064516, "grad_norm": 8.676656723022461, "learning_rate": 1.6129032258064516e-06, "loss": 1.2418, "step": 1 }, { "epoch": 0.008064516129032258, "grad_norm": 4.749965190887451, "learning_rate": 8.064516129032258e-06, "loss": 1.1522, "step": 5 }, { "epoch": 0.016129032258064516, "grad_norm": 4.172841548919678, "learning_rate": 1.6129032258064517e-05, "loss": 0.9234, "step": 10 }, { "epoch": 0.024193548387096774, "grad_norm": 3.307924747467041, "learning_rate": 2.4193548387096777e-05, "loss": 0.5665, "step": 15 }, { "epoch": 0.03225806451612903, "grad_norm": 3.2205514907836914, "learning_rate": 3.2258064516129034e-05, "loss": 0.205, "step": 20 }, { "epoch": 0.04032258064516129, "grad_norm": 1.3202248811721802, "learning_rate": 4.032258064516129e-05, "loss": 0.0567, "step": 25 }, { "epoch": 0.04838709677419355, "grad_norm": 2.270284652709961, "learning_rate": 4.8387096774193554e-05, "loss": 0.0348, "step": 30 }, { "epoch": 0.056451612903225805, "grad_norm": 1.9001686573028564, "learning_rate": 5.645161290322582e-05, "loss": 0.0256, "step": 35 }, { "epoch": 0.06451612903225806, "grad_norm": 0.8090413212776184, "learning_rate": 6.451612903225807e-05, "loss": 0.0234, "step": 40 }, { "epoch": 0.07258064516129033, "grad_norm": 0.7908226251602173, "learning_rate": 7.258064516129033e-05, "loss": 0.0177, "step": 45 }, { "epoch": 0.08064516129032258, "grad_norm": 3.9440441131591797, "learning_rate": 8.064516129032258e-05, "loss": 0.0148, "step": 50 }, { "epoch": 0.08870967741935484, "grad_norm": 0.7870569825172424, "learning_rate": 8.870967741935484e-05, "loss": 0.0186, "step": 55 }, { "epoch": 0.0967741935483871, "grad_norm": 1.7015715837478638, "learning_rate": 9.677419354838711e-05, "loss": 0.0197, "step": 60 }, { "epoch": 0.10483870967741936, "grad_norm": 0.7983545660972595, "learning_rate": 9.99928681279855e-05, "loss": 0.0148, "step": 65 }, { "epoch": 0.11290322580645161, "grad_norm": 1.1775078773498535, "learning_rate": 9.994929183335238e-05, "loss": 0.0139, "step": 70 }, { "epoch": 0.12096774193548387, "grad_norm": 4.544706344604492, "learning_rate": 9.986613588305434e-05, "loss": 0.0201, "step": 75 }, { "epoch": 0.12903225806451613, "grad_norm": 1.9559928178787231, "learning_rate": 9.974346616959476e-05, "loss": 0.0157, "step": 80 }, { "epoch": 0.13709677419354838, "grad_norm": 1.6153545379638672, "learning_rate": 9.95813798960538e-05, "loss": 0.0188, "step": 85 }, { "epoch": 0.14516129032258066, "grad_norm": 300.6769104003906, "learning_rate": 9.938000549906508e-05, "loss": 0.0761, "step": 90 }, { "epoch": 0.1532258064516129, "grad_norm": 0.5071149468421936, "learning_rate": 9.913950254704291e-05, "loss": 0.0236, "step": 95 }, { "epoch": 0.16129032258064516, "grad_norm": 1.064196228981018, "learning_rate": 9.88600616137407e-05, "loss": 1.1514, "step": 100 }, { "epoch": 0.1693548387096774, "grad_norm": 415.9707946777344, "learning_rate": 9.854190412724114e-05, "loss": 0.0835, "step": 105 }, { "epoch": 0.1774193548387097, "grad_norm": 1.221725583076477, "learning_rate": 9.818528219449705e-05, "loss": 0.0189, "step": 110 }, { "epoch": 0.18548387096774194, "grad_norm": 0.40028271079063416, "learning_rate": 9.779047840156288e-05, "loss": 0.0197, "step": 115 }, { "epoch": 0.1935483870967742, "grad_norm": 3.2106378078460693, "learning_rate": 9.735780558967434e-05, "loss": 1.0777, "step": 120 }, { "epoch": 0.20161290322580644, "grad_norm": 544.4739379882812, "learning_rate": 9.688760660735402e-05, "loss": 0.0779, "step": 125 }, { "epoch": 0.20967741935483872, "grad_norm": 0.48118099570274353, "learning_rate": 9.63802540387394e-05, "loss": 0.7549, "step": 130 }, { "epoch": 0.21774193548387097, "grad_norm": 0.43920445442199707, "learning_rate": 9.583614990834828e-05, "loss": 0.0193, "step": 135 }, { "epoch": 0.22580645161290322, "grad_norm": 1.4750373363494873, "learning_rate": 9.525572536251607e-05, "loss": 0.02, "step": 140 }, { "epoch": 0.23387096774193547, "grad_norm": 2.262516498565674, "learning_rate": 9.463944032775658e-05, "loss": 0.0216, "step": 145 }, { "epoch": 0.24193548387096775, "grad_norm": 1.0215312242507935, "learning_rate": 9.398778314631801e-05, "loss": 0.018, "step": 150 }, { "epoch": 0.25, "grad_norm": 0.8741445541381836, "learning_rate": 9.330127018922194e-05, "loss": 0.0178, "step": 155 }, { "epoch": 0.25806451612903225, "grad_norm": 0.5452761650085449, "learning_rate": 9.258044544709274e-05, "loss": 0.0323, "step": 160 }, { "epoch": 0.2661290322580645, "grad_norm": 0.43162572383880615, "learning_rate": 9.182588009910118e-05, "loss": 0.0201, "step": 165 }, { "epoch": 0.27419354838709675, "grad_norm": 0.25898757576942444, "learning_rate": 9.103817206036382e-05, "loss": 0.0198, "step": 170 }, { "epoch": 0.28225806451612906, "grad_norm": 2.178893566131592, "learning_rate": 9.021794550815713e-05, "loss": 0.0159, "step": 175 }, { "epoch": 0.2903225806451613, "grad_norm": 145.104736328125, "learning_rate": 8.936585038732141e-05, "loss": 0.5195, "step": 180 }, { "epoch": 0.29838709677419356, "grad_norm": 6.8297834396362305, "learning_rate": 8.848256189524662e-05, "loss": 0.021, "step": 185 }, { "epoch": 0.3064516129032258, "grad_norm": 1.4729770421981812, "learning_rate": 8.756877994684818e-05, "loss": 0.0159, "step": 190 }, { "epoch": 0.31451612903225806, "grad_norm": 2.1729257106781006, "learning_rate": 8.66252286199567e-05, "loss": 0.0152, "step": 195 }, { "epoch": 0.3225806451612903, "grad_norm": 0.6754555106163025, "learning_rate": 8.565265558156101e-05, "loss": 0.0117, "step": 200 }, { "epoch": 0.33064516129032256, "grad_norm": 0.3731600046157837, "learning_rate": 8.46518314953594e-05, "loss": 0.0101, "step": 205 }, { "epoch": 0.3387096774193548, "grad_norm": 0.3129900395870209, "learning_rate": 8.362354941108802e-05, "loss": 0.0161, "step": 210 }, { "epoch": 0.3467741935483871, "grad_norm": 0.3701813519001007, "learning_rate": 8.256862413611113e-05, "loss": 0.0156, "step": 215 }, { "epoch": 0.3548387096774194, "grad_norm": 10.514623641967773, "learning_rate": 8.148789158977013e-05, "loss": 0.0072, "step": 220 }, { "epoch": 0.3629032258064516, "grad_norm": 0.25649479031562805, "learning_rate": 8.038220814100403e-05, "loss": 0.0089, "step": 225 }, { "epoch": 0.3709677419354839, "grad_norm": 0.36153924465179443, "learning_rate": 7.925244992976538e-05, "loss": 0.0091, "step": 230 }, { "epoch": 0.3790322580645161, "grad_norm": 0.15555064380168915, "learning_rate": 7.809951217276985e-05, "loss": 0.1414, "step": 235 }, { "epoch": 0.3870967741935484, "grad_norm": 0.5916360020637512, "learning_rate": 7.692430845412945e-05, "loss": 0.0153, "step": 240 }, { "epoch": 0.3951612903225806, "grad_norm": 0.47686517238616943, "learning_rate": 7.572777000143145e-05, "loss": 0.0154, "step": 245 }, { "epoch": 0.4032258064516129, "grad_norm": 0.5245814919471741, "learning_rate": 7.451084494783668e-05, "loss": 0.02, "step": 250 }, { "epoch": 0.4112903225806452, "grad_norm": 3.145771026611328, "learning_rate": 7.327449758078193e-05, "loss": 0.0121, "step": 255 }, { "epoch": 0.41935483870967744, "grad_norm": 0.47086361050605774, "learning_rate": 7.201970757788172e-05, "loss": 0.01, "step": 260 }, { "epoch": 0.4274193548387097, "grad_norm": 0.8778440356254578, "learning_rate": 7.074746923063497e-05, "loss": 0.0083, "step": 265 }, { "epoch": 0.43548387096774194, "grad_norm": 0.1912572830915451, "learning_rate": 6.945879065655165e-05, "loss": 0.018, "step": 270 }, { "epoch": 0.4435483870967742, "grad_norm": 0.3541414737701416, "learning_rate": 6.815469300032374e-05, "loss": 0.0103, "step": 275 }, { "epoch": 0.45161290322580644, "grad_norm": 0.39805546402931213, "learning_rate": 6.683620962467357e-05, "loss": 0.0123, "step": 280 }, { "epoch": 0.4596774193548387, "grad_norm": 0.3862231373786926, "learning_rate": 6.550438529152056e-05, "loss": 0.0085, "step": 285 }, { "epoch": 0.46774193548387094, "grad_norm": 0.7873578071594238, "learning_rate": 6.416027533411519e-05, "loss": 0.0078, "step": 290 }, { "epoch": 0.47580645161290325, "grad_norm": 0.9767886996269226, "learning_rate": 6.28049448207966e-05, "loss": 0.0089, "step": 295 }, { "epoch": 0.4838709677419355, "grad_norm": 0.659842312335968, "learning_rate": 6.14394677110356e-05, "loss": 0.0085, "step": 300 }, { "epoch": 0.49193548387096775, "grad_norm": 0.15667948126792908, "learning_rate": 6.006492600443301e-05, "loss": 0.0106, "step": 305 }, { "epoch": 0.5, "grad_norm": 0.37710684537887573, "learning_rate": 5.868240888334653e-05, "loss": 0.0125, "step": 310 }, { "epoch": 0.5080645161290323, "grad_norm": 0.22124210000038147, "learning_rate": 5.7293011849826217e-05, "loss": 0.0121, "step": 315 }, { "epoch": 0.5161290322580645, "grad_norm": 0.894884467124939, "learning_rate": 5.5897835857542317e-05, "loss": 0.0059, "step": 320 }, { "epoch": 0.5241935483870968, "grad_norm": 0.2709536850452423, "learning_rate": 5.449798643939304e-05, "loss": 0.0074, "step": 325 }, { "epoch": 0.532258064516129, "grad_norm": 0.15670421719551086, "learning_rate": 5.30945728314841e-05, "loss": 0.0131, "step": 330 }, { "epoch": 0.5403225806451613, "grad_norm": 0.31018203496932983, "learning_rate": 5.168870709417342e-05, "loss": 0.0101, "step": 335 }, { "epoch": 0.5483870967741935, "grad_norm": 0.20672403275966644, "learning_rate": 5.02815032308783e-05, "loss": 0.0057, "step": 340 }, { "epoch": 0.5564516129032258, "grad_norm": 0.18754148483276367, "learning_rate": 4.887407630534271e-05, "loss": 0.0063, "step": 345 }, { "epoch": 0.5645161290322581, "grad_norm": 0.29782626032829285, "learning_rate": 4.746754155806437e-05, "loss": 0.0071, "step": 350 }, { "epoch": 0.5725806451612904, "grad_norm": 0.1842086762189865, "learning_rate": 4.6063013522581915e-05, "loss": 0.012, "step": 355 }, { "epoch": 0.5806451612903226, "grad_norm": 3.5033631324768066, "learning_rate": 4.466160514232206e-05, "loss": 0.0071, "step": 360 }, { "epoch": 0.5887096774193549, "grad_norm": 0.11994602531194687, "learning_rate": 4.3264426888706965e-05, "loss": 0.0056, "step": 365 }, { "epoch": 0.5967741935483871, "grad_norm": 0.21798093616962433, "learning_rate": 4.187258588122019e-05, "loss": 0.0096, "step": 370 }, { "epoch": 0.6048387096774194, "grad_norm": 1.366426944732666, "learning_rate": 4.048718501012895e-05, "loss": 0.0113, "step": 375 }, { "epoch": 0.6129032258064516, "grad_norm": 0.2101077437400818, "learning_rate": 3.9109322062557424e-05, "loss": 0.0105, "step": 380 }, { "epoch": 0.6209677419354839, "grad_norm": 0.2682858407497406, "learning_rate": 3.77400888526038e-05, "loss": 0.0079, "step": 385 }, { "epoch": 0.6290322580645161, "grad_norm": 0.3401815593242645, "learning_rate": 3.6380570356190344e-05, "loss": 0.0087, "step": 390 }, { "epoch": 0.6370967741935484, "grad_norm": 1.0749493837356567, "learning_rate": 3.5031843851332104e-05, "loss": 0.0077, "step": 395 }, { "epoch": 0.6451612903225806, "grad_norm": 0.600229024887085, "learning_rate": 3.3694978064505254e-05, "loss": 0.0088, "step": 400 }, { "epoch": 0.6532258064516129, "grad_norm": 11.734594345092773, "learning_rate": 3.2371032323791754e-05, "loss": 0.0107, "step": 405 }, { "epoch": 0.6612903225806451, "grad_norm": 0.9927507638931274, "learning_rate": 3.10610557194712e-05, "loss": 0.0146, "step": 410 }, { "epoch": 0.6693548387096774, "grad_norm": 0.46601602435112, "learning_rate": 2.9766086272725196e-05, "loss": 0.0123, "step": 415 }, { "epoch": 0.6774193548387096, "grad_norm": 11.156332969665527, "learning_rate": 2.848715011311271e-05, "loss": 0.0134, "step": 420 }, { "epoch": 0.6854838709677419, "grad_norm": 0.34076985716819763, "learning_rate": 2.72252606654683e-05, "loss": 0.011, "step": 425 }, { "epoch": 0.6935483870967742, "grad_norm": 0.14691153168678284, "learning_rate": 2.598141784686775e-05, "loss": 0.0113, "step": 430 }, { "epoch": 0.7016129032258065, "grad_norm": 0.19039344787597656, "learning_rate": 2.4756607274296846e-05, "loss": 0.0066, "step": 435 }, { "epoch": 0.7096774193548387, "grad_norm": 0.4377165734767914, "learning_rate": 2.3551799483651894e-05, "loss": 0.0056, "step": 440 }, { "epoch": 0.717741935483871, "grad_norm": 0.16886833310127258, "learning_rate": 2.2367949160690072e-05, "loss": 0.0075, "step": 445 }, { "epoch": 0.7258064516129032, "grad_norm": 5.967025279998779, "learning_rate": 2.120599438453968e-05, "loss": 0.0083, "step": 450 }, { "epoch": 0.7338709677419355, "grad_norm": 0.15985387563705444, "learning_rate": 2.0066855884369244e-05, "loss": 0.004, "step": 455 }, { "epoch": 0.7419354838709677, "grad_norm": 0.18566212058067322, "learning_rate": 1.8951436309804765e-05, "loss": 0.0049, "step": 460 }, { "epoch": 0.75, "grad_norm": 0.2376997023820877, "learning_rate": 1.7860619515673033e-05, "loss": 0.0071, "step": 465 }, { "epoch": 0.7580645161290323, "grad_norm": 0.24823206663131714, "learning_rate": 1.6795269861638042e-05, "loss": 0.0068, "step": 470 }, { "epoch": 0.7661290322580645, "grad_norm": 0.15533004701137543, "learning_rate": 1.575623152728518e-05, "loss": 0.0054, "step": 475 }, { "epoch": 0.7741935483870968, "grad_norm": 0.1936763972043991, "learning_rate": 1.4744327843196043e-05, "loss": 0.0053, "step": 480 }, { "epoch": 0.782258064516129, "grad_norm": 0.24305252730846405, "learning_rate": 1.3760360638544012e-05, "loss": 0.0042, "step": 485 }, { "epoch": 0.7903225806451613, "grad_norm": 0.23395785689353943, "learning_rate": 1.280510960572745e-05, "loss": 0.006, "step": 490 }, { "epoch": 0.7983870967741935, "grad_norm": 0.160566508769989, "learning_rate": 1.187933168254397e-05, "loss": 0.0036, "step": 495 }, { "epoch": 0.8064516129032258, "grad_norm": 0.173016756772995, "learning_rate": 1.0983760452395413e-05, "loss": 0.004, "step": 500 }, { "epoch": 0.8145161290322581, "grad_norm": 0.4186556935310364, "learning_rate": 1.0119105562998698e-05, "loss": 0.0062, "step": 505 }, { "epoch": 0.8225806451612904, "grad_norm": 0.29756516218185425, "learning_rate": 9.28605216406337e-06, "loss": 0.0046, "step": 510 }, { "epoch": 0.8306451612903226, "grad_norm": 0.1282072514295578, "learning_rate": 8.485260364381187e-06, "loss": 0.0033, "step": 515 }, { "epoch": 0.8387096774193549, "grad_norm": 0.1337880641222, "learning_rate": 7.717364708758024e-06, "loss": 0.0049, "step": 520 }, { "epoch": 0.8467741935483871, "grad_norm": 0.2575572431087494, "learning_rate": 6.982973675202676e-06, "loss": 0.0058, "step": 525 }, { "epoch": 0.8548387096774194, "grad_norm": 0.1366008073091507, "learning_rate": 6.282669192770896e-06, "loss": 0.005, "step": 530 }, { "epoch": 0.8629032258064516, "grad_norm": 2.0787034034729004, "learning_rate": 5.617006180446688e-06, "loss": 0.0075, "step": 535 }, { "epoch": 0.8709677419354839, "grad_norm": 0.17773574590682983, "learning_rate": 4.986512107426283e-06, "loss": 0.0054, "step": 540 }, { "epoch": 0.8790322580645161, "grad_norm": 0.0983528196811676, "learning_rate": 4.391686575153331e-06, "loss": 0.0043, "step": 545 }, { "epoch": 0.8870967741935484, "grad_norm": 0.13671350479125977, "learning_rate": 3.833000921436319e-06, "loss": 0.0031, "step": 550 }, { "epoch": 0.8951612903225806, "grad_norm": 0.6432064175605774, "learning_rate": 3.310897846962041e-06, "loss": 0.0048, "step": 555 }, { "epoch": 0.9032258064516129, "grad_norm": 0.5728097558021545, "learning_rate": 2.825791064500993e-06, "loss": 0.0053, "step": 560 }, { "epoch": 0.9112903225806451, "grad_norm": 0.6017826199531555, "learning_rate": 2.3780649710827554e-06, "loss": 0.0058, "step": 565 }, { "epoch": 0.9193548387096774, "grad_norm": 0.3495374321937561, "learning_rate": 1.9680743434010384e-06, "loss": 0.007, "step": 570 }, { "epoch": 0.9274193548387096, "grad_norm": 0.34849557280540466, "learning_rate": 1.5961440566897912e-06, "loss": 0.003, "step": 575 }, { "epoch": 0.9354838709677419, "grad_norm": 0.24747046828269958, "learning_rate": 1.2625688272930926e-06, "loss": 0.0042, "step": 580 }, { "epoch": 0.9435483870967742, "grad_norm": 0.5230942964553833, "learning_rate": 9.67612979132948e-07, "loss": 0.0045, "step": 585 }, { "epoch": 0.9516129032258065, "grad_norm": 0.14619432389736176, "learning_rate": 7.115102342598101e-07, "loss": 0.0034, "step": 590 }, { "epoch": 0.9596774193548387, "grad_norm": 0.47683650255203247, "learning_rate": 4.944635276520393e-07, "loss": 0.0036, "step": 595 }, { "epoch": 0.967741935483871, "grad_norm": 0.20121371746063232, "learning_rate": 3.1664484641086287e-07, "loss": 0.0054, "step": 600 }, { "epoch": 0.9758064516129032, "grad_norm": 0.12017033249139786, "learning_rate": 1.781950934783505e-07, "loss": 0.0059, "step": 605 }, { "epoch": 0.9838709677419355, "grad_norm": 0.39845508337020874, "learning_rate": 7.92239759864255e-08, "loss": 0.0043, "step": 610 }, { "epoch": 0.9919354838709677, "grad_norm": 0.21712489426136017, "learning_rate": 1.980991832524759e-08, "loss": 0.0056, "step": 615 }, { "epoch": 1.0, "grad_norm": 0.13640886545181274, "learning_rate": 0.0, "loss": 0.0046, "step": 620 }, { "epoch": 1.0, "eval_loss": 0.0026028414722532034, "eval_runtime": 3.8766, "eval_samples_per_second": 1.548, "eval_steps_per_second": 1.548, "step": 620 }, { "epoch": 1.0, "step": 620, "total_flos": 5.590502947474637e+16, "train_loss": 0.06471517855121244, "train_runtime": 1309.8316, "train_samples_per_second": 0.473, "train_steps_per_second": 0.473 } ], "logging_steps": 5, "max_steps": 620, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 5.590502947474637e+16, "train_batch_size": 1, "trial_name": null, "trial_params": null }