| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.9904153354632586, |
| "eval_steps": 500, |
| "global_step": 78, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.038338658146964855, |
| "grad_norm": 6.569391250610352, |
| "learning_rate": 1.25e-06, |
| "loss": 1.0687, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.07667731629392971, |
| "grad_norm": 6.976002216339111, |
| "learning_rate": 2.5e-06, |
| "loss": 1.07, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.11501597444089456, |
| "grad_norm": 7.1917877197265625, |
| "learning_rate": 3.7500000000000005e-06, |
| "loss": 1.137, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.15335463258785942, |
| "grad_norm": 6.239739418029785, |
| "learning_rate": 5e-06, |
| "loss": 1.1053, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.19169329073482427, |
| "grad_norm": 4.6187615394592285, |
| "learning_rate": 6.25e-06, |
| "loss": 0.9952, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.23003194888178913, |
| "grad_norm": 2.926978826522827, |
| "learning_rate": 7.500000000000001e-06, |
| "loss": 1.0048, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.268370607028754, |
| "grad_norm": 4.541296005249023, |
| "learning_rate": 8.750000000000001e-06, |
| "loss": 1.0243, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.30670926517571884, |
| "grad_norm": 4.317328453063965, |
| "learning_rate": 1e-05, |
| "loss": 0.9556, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.3450479233226837, |
| "grad_norm": 4.892552375793457, |
| "learning_rate": 9.994965332706574e-06, |
| "loss": 0.9424, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.38338658146964855, |
| "grad_norm": 4.437831878662109, |
| "learning_rate": 9.979871469976197e-06, |
| "loss": 0.867, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.4217252396166134, |
| "grad_norm": 3.369062900543213, |
| "learning_rate": 9.954748808839675e-06, |
| "loss": 0.9222, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.46006389776357826, |
| "grad_norm": 2.2566771507263184, |
| "learning_rate": 9.91964794299315e-06, |
| "loss": 0.9229, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.4984025559105431, |
| "grad_norm": 2.014096736907959, |
| "learning_rate": 9.874639560909118e-06, |
| "loss": 0.8628, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.536741214057508, |
| "grad_norm": 1.8734163045883179, |
| "learning_rate": 9.819814303479268e-06, |
| "loss": 0.836, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.5750798722044729, |
| "grad_norm": 1.5196428298950195, |
| "learning_rate": 9.755282581475769e-06, |
| "loss": 0.8587, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.6134185303514377, |
| "grad_norm": 1.5723515748977661, |
| "learning_rate": 9.681174353198687e-06, |
| "loss": 0.8516, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.6517571884984026, |
| "grad_norm": 1.4177026748657227, |
| "learning_rate": 9.597638862757255e-06, |
| "loss": 0.7702, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.6900958466453674, |
| "grad_norm": 0.9884472489356995, |
| "learning_rate": 9.504844339512096e-06, |
| "loss": 0.85, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.7284345047923323, |
| "grad_norm": 1.1919715404510498, |
| "learning_rate": 9.40297765928369e-06, |
| "loss": 0.8992, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.7667731629392971, |
| "grad_norm": 1.0880603790283203, |
| "learning_rate": 9.292243968009332e-06, |
| "loss": 0.8034, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.805111821086262, |
| "grad_norm": 0.8762597441673279, |
| "learning_rate": 9.172866268606514e-06, |
| "loss": 0.8485, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.8434504792332268, |
| "grad_norm": 0.7751829028129578, |
| "learning_rate": 9.045084971874738e-06, |
| "loss": 0.7495, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.8817891373801917, |
| "grad_norm": 1.0985928773880005, |
| "learning_rate": 8.90915741234015e-06, |
| "loss": 0.8174, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.9201277955271565, |
| "grad_norm": 0.9542496204376221, |
| "learning_rate": 8.765357330018056e-06, |
| "loss": 0.8876, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.9584664536741214, |
| "grad_norm": 0.8492473363876343, |
| "learning_rate": 8.613974319136959e-06, |
| "loss": 0.8334, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.9968051118210862, |
| "grad_norm": 1.098617672920227, |
| "learning_rate": 8.455313244934324e-06, |
| "loss": 0.8659, |
| "step": 26 |
| }, |
| { |
| "epoch": 1.035143769968051, |
| "grad_norm": 2.166583299636841, |
| "learning_rate": 8.289693629698564e-06, |
| "loss": 1.3682, |
| "step": 27 |
| }, |
| { |
| "epoch": 1.073482428115016, |
| "grad_norm": 0.9129027724266052, |
| "learning_rate": 8.117449009293668e-06, |
| "loss": 0.7633, |
| "step": 28 |
| }, |
| { |
| "epoch": 1.1118210862619808, |
| "grad_norm": 0.5984949469566345, |
| "learning_rate": 7.938926261462366e-06, |
| "loss": 0.7596, |
| "step": 29 |
| }, |
| { |
| "epoch": 1.1501597444089458, |
| "grad_norm": 0.7058618664741516, |
| "learning_rate": 7.754484907260513e-06, |
| "loss": 0.8484, |
| "step": 30 |
| }, |
| { |
| "epoch": 1.1884984025559104, |
| "grad_norm": 0.7317571640014648, |
| "learning_rate": 7.564496387029532e-06, |
| "loss": 0.7711, |
| "step": 31 |
| }, |
| { |
| "epoch": 1.2268370607028753, |
| "grad_norm": 0.5173776745796204, |
| "learning_rate": 7.369343312364994e-06, |
| "loss": 0.7035, |
| "step": 32 |
| }, |
| { |
| "epoch": 1.2651757188498403, |
| "grad_norm": 0.6413378715515137, |
| "learning_rate": 7.169418695587791e-06, |
| "loss": 0.8469, |
| "step": 33 |
| }, |
| { |
| "epoch": 1.3035143769968052, |
| "grad_norm": 0.6376237869262695, |
| "learning_rate": 6.965125158269619e-06, |
| "loss": 0.7636, |
| "step": 34 |
| }, |
| { |
| "epoch": 1.34185303514377, |
| "grad_norm": 0.7505595088005066, |
| "learning_rate": 6.7568741204067145e-06, |
| "loss": 0.729, |
| "step": 35 |
| }, |
| { |
| "epoch": 1.3801916932907348, |
| "grad_norm": 0.6152018904685974, |
| "learning_rate": 6.545084971874738e-06, |
| "loss": 0.7501, |
| "step": 36 |
| }, |
| { |
| "epoch": 1.4185303514376997, |
| "grad_norm": 0.6682824492454529, |
| "learning_rate": 6.330184227833376e-06, |
| "loss": 0.7599, |
| "step": 37 |
| }, |
| { |
| "epoch": 1.4568690095846646, |
| "grad_norm": 0.5879465341567993, |
| "learning_rate": 6.112604669781572e-06, |
| "loss": 0.8, |
| "step": 38 |
| }, |
| { |
| "epoch": 1.4952076677316293, |
| "grad_norm": 0.6658037304878235, |
| "learning_rate": 5.892784473993184e-06, |
| "loss": 0.8194, |
| "step": 39 |
| }, |
| { |
| "epoch": 1.5335463258785942, |
| "grad_norm": 0.5833703279495239, |
| "learning_rate": 5.671166329088278e-06, |
| "loss": 0.8215, |
| "step": 40 |
| }, |
| { |
| "epoch": 1.571884984025559, |
| "grad_norm": 0.5278577208518982, |
| "learning_rate": 5.448196544517168e-06, |
| "loss": 0.7256, |
| "step": 41 |
| }, |
| { |
| "epoch": 1.610223642172524, |
| "grad_norm": 0.5681585073471069, |
| "learning_rate": 5.224324151752575e-06, |
| "loss": 0.7466, |
| "step": 42 |
| }, |
| { |
| "epoch": 1.648562300319489, |
| "grad_norm": 0.536124587059021, |
| "learning_rate": 5e-06, |
| "loss": 0.7235, |
| "step": 43 |
| }, |
| { |
| "epoch": 1.6869009584664538, |
| "grad_norm": 0.5382006168365479, |
| "learning_rate": 4.775675848247427e-06, |
| "loss": 0.821, |
| "step": 44 |
| }, |
| { |
| "epoch": 1.7252396166134185, |
| "grad_norm": 0.5785599946975708, |
| "learning_rate": 4.551803455482833e-06, |
| "loss": 0.7045, |
| "step": 45 |
| }, |
| { |
| "epoch": 1.7635782747603834, |
| "grad_norm": 0.5749883651733398, |
| "learning_rate": 4.3288336709117246e-06, |
| "loss": 0.821, |
| "step": 46 |
| }, |
| { |
| "epoch": 1.8019169329073481, |
| "grad_norm": 0.5753567218780518, |
| "learning_rate": 4.107215526006818e-06, |
| "loss": 0.7941, |
| "step": 47 |
| }, |
| { |
| "epoch": 1.840255591054313, |
| "grad_norm": 0.5819541811943054, |
| "learning_rate": 3.887395330218429e-06, |
| "loss": 0.7763, |
| "step": 48 |
| }, |
| { |
| "epoch": 1.878594249201278, |
| "grad_norm": 0.5433679819107056, |
| "learning_rate": 3.669815772166625e-06, |
| "loss": 0.7556, |
| "step": 49 |
| }, |
| { |
| "epoch": 1.9169329073482428, |
| "grad_norm": 0.5302234292030334, |
| "learning_rate": 3.4549150281252635e-06, |
| "loss": 0.6809, |
| "step": 50 |
| }, |
| { |
| "epoch": 1.9552715654952078, |
| "grad_norm": 0.4911814332008362, |
| "learning_rate": 3.2431258795932863e-06, |
| "loss": 0.7219, |
| "step": 51 |
| }, |
| { |
| "epoch": 1.9936102236421727, |
| "grad_norm": 0.4501954913139343, |
| "learning_rate": 3.0348748417303826e-06, |
| "loss": 0.775, |
| "step": 52 |
| }, |
| { |
| "epoch": 2.0319488817891376, |
| "grad_norm": 1.8575425148010254, |
| "learning_rate": 2.83058130441221e-06, |
| "loss": 1.1184, |
| "step": 53 |
| }, |
| { |
| "epoch": 2.070287539936102, |
| "grad_norm": 0.5785319209098816, |
| "learning_rate": 2.6306566876350072e-06, |
| "loss": 0.725, |
| "step": 54 |
| }, |
| { |
| "epoch": 2.108626198083067, |
| "grad_norm": 0.46213895082473755, |
| "learning_rate": 2.43550361297047e-06, |
| "loss": 0.687, |
| "step": 55 |
| }, |
| { |
| "epoch": 2.146964856230032, |
| "grad_norm": 0.4825771749019623, |
| "learning_rate": 2.245515092739488e-06, |
| "loss": 0.7641, |
| "step": 56 |
| }, |
| { |
| "epoch": 2.1853035143769968, |
| "grad_norm": 0.5464252829551697, |
| "learning_rate": 2.061073738537635e-06, |
| "loss": 0.8842, |
| "step": 57 |
| }, |
| { |
| "epoch": 2.2236421725239617, |
| "grad_norm": 0.38878318667411804, |
| "learning_rate": 1.8825509907063328e-06, |
| "loss": 0.6678, |
| "step": 58 |
| }, |
| { |
| "epoch": 2.2619808306709266, |
| "grad_norm": 0.37567567825317383, |
| "learning_rate": 1.7103063703014372e-06, |
| "loss": 0.7137, |
| "step": 59 |
| }, |
| { |
| "epoch": 2.3003194888178915, |
| "grad_norm": 0.45787665247917175, |
| "learning_rate": 1.544686755065677e-06, |
| "loss": 0.8036, |
| "step": 60 |
| }, |
| { |
| "epoch": 2.3386581469648564, |
| "grad_norm": 0.38240814208984375, |
| "learning_rate": 1.3860256808630429e-06, |
| "loss": 0.7092, |
| "step": 61 |
| }, |
| { |
| "epoch": 2.376996805111821, |
| "grad_norm": 0.4268067181110382, |
| "learning_rate": 1.234642669981946e-06, |
| "loss": 0.6451, |
| "step": 62 |
| }, |
| { |
| "epoch": 2.415335463258786, |
| "grad_norm": 0.4034426510334015, |
| "learning_rate": 1.0908425876598512e-06, |
| "loss": 0.6204, |
| "step": 63 |
| }, |
| { |
| "epoch": 2.4536741214057507, |
| "grad_norm": 0.43444254994392395, |
| "learning_rate": 9.549150281252633e-07, |
| "loss": 0.7801, |
| "step": 64 |
| }, |
| { |
| "epoch": 2.4920127795527156, |
| "grad_norm": 0.4256804883480072, |
| "learning_rate": 8.271337313934869e-07, |
| "loss": 0.734, |
| "step": 65 |
| }, |
| { |
| "epoch": 2.5303514376996805, |
| "grad_norm": 0.40886035561561584, |
| "learning_rate": 7.077560319906696e-07, |
| "loss": 0.7094, |
| "step": 66 |
| }, |
| { |
| "epoch": 2.5686900958466454, |
| "grad_norm": 0.41242608428001404, |
| "learning_rate": 5.9702234071631e-07, |
| "loss": 0.6316, |
| "step": 67 |
| }, |
| { |
| "epoch": 2.6070287539936103, |
| "grad_norm": 0.4434635639190674, |
| "learning_rate": 4.951556604879049e-07, |
| "loss": 0.7929, |
| "step": 68 |
| }, |
| { |
| "epoch": 2.6453674121405752, |
| "grad_norm": 0.4307124614715576, |
| "learning_rate": 4.0236113724274716e-07, |
| "loss": 0.7719, |
| "step": 69 |
| }, |
| { |
| "epoch": 2.68370607028754, |
| "grad_norm": 0.4264853894710541, |
| "learning_rate": 3.18825646801314e-07, |
| "loss": 0.8123, |
| "step": 70 |
| }, |
| { |
| "epoch": 2.722044728434505, |
| "grad_norm": 0.48485955595970154, |
| "learning_rate": 2.447174185242324e-07, |
| "loss": 0.7276, |
| "step": 71 |
| }, |
| { |
| "epoch": 2.7603833865814695, |
| "grad_norm": 0.4042198061943054, |
| "learning_rate": 1.801856965207338e-07, |
| "loss": 0.6724, |
| "step": 72 |
| }, |
| { |
| "epoch": 2.7987220447284344, |
| "grad_norm": 0.4416380524635315, |
| "learning_rate": 1.253604390908819e-07, |
| "loss": 0.7595, |
| "step": 73 |
| }, |
| { |
| "epoch": 2.8370607028753994, |
| "grad_norm": 0.39589574933052063, |
| "learning_rate": 8.035205700685167e-08, |
| "loss": 0.7724, |
| "step": 74 |
| }, |
| { |
| "epoch": 2.8753993610223643, |
| "grad_norm": 0.4274565279483795, |
| "learning_rate": 4.52511911603265e-08, |
| "loss": 0.7977, |
| "step": 75 |
| }, |
| { |
| "epoch": 2.913738019169329, |
| "grad_norm": 0.3615323007106781, |
| "learning_rate": 2.012853002380466e-08, |
| "loss": 0.6038, |
| "step": 76 |
| }, |
| { |
| "epoch": 2.952076677316294, |
| "grad_norm": 0.440108060836792, |
| "learning_rate": 5.034667293427053e-09, |
| "loss": 0.7618, |
| "step": 77 |
| }, |
| { |
| "epoch": 2.9904153354632586, |
| "grad_norm": 0.3972029685974121, |
| "learning_rate": 0.0, |
| "loss": 0.6367, |
| "step": 78 |
| }, |
| { |
| "epoch": 2.9904153354632586, |
| "step": 78, |
| "total_flos": 77393718411264.0, |
| "train_loss": 0.815421788356243, |
| "train_runtime": 7664.7103, |
| "train_samples_per_second": 0.979, |
| "train_steps_per_second": 0.01 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 78, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 77393718411264.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
