| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.08410959480202704, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 8.410959480202704e-05, | |
| "grad_norm": 0.27725711464881897, | |
| "learning_rate": 0.00019999950652018584, | |
| "loss": 0.8848, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00016821918960405408, | |
| "grad_norm": 0.2778919041156769, | |
| "learning_rate": 0.0001999980260856137, | |
| "loss": 0.5957, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0002523287844060811, | |
| "grad_norm": 0.5110663771629333, | |
| "learning_rate": 0.000199995558710895, | |
| "loss": 1.2037, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00033643837920810816, | |
| "grad_norm": 0.4022953510284424, | |
| "learning_rate": 0.00019999210442038162, | |
| "loss": 1.1604, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0004205479740101352, | |
| "grad_norm": 0.5099120736122131, | |
| "learning_rate": 0.00019998766324816607, | |
| "loss": 1.1557, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0005046575688121622, | |
| "grad_norm": 0.4215346872806549, | |
| "learning_rate": 0.0001999822352380809, | |
| "loss": 1.0749, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0005887671636141893, | |
| "grad_norm": 0.5019777417182922, | |
| "learning_rate": 0.00019997582044369843, | |
| "loss": 0.8256, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0006728767584162163, | |
| "grad_norm": 0.39011430740356445, | |
| "learning_rate": 0.00019996841892833, | |
| "loss": 0.7847, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0007569863532182434, | |
| "grad_norm": 0.49417048692703247, | |
| "learning_rate": 0.00019996003076502565, | |
| "loss": 1.0241, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0008410959480202704, | |
| "grad_norm": 0.5206520557403564, | |
| "learning_rate": 0.00019995065603657316, | |
| "loss": 0.8424, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0009252055428222974, | |
| "grad_norm": 0.4732626974582672, | |
| "learning_rate": 0.0001999402948354973, | |
| "loss": 0.7602, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0010093151376243245, | |
| "grad_norm": 0.7119062542915344, | |
| "learning_rate": 0.00019992894726405893, | |
| "loss": 0.9964, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0010934247324263515, | |
| "grad_norm": 0.6489611864089966, | |
| "learning_rate": 0.000199916613434254, | |
| "loss": 0.8901, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0011775343272283786, | |
| "grad_norm": 0.6794676184654236, | |
| "learning_rate": 0.0001999032934678125, | |
| "loss": 0.8788, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0012616439220304056, | |
| "grad_norm": 0.5384562015533447, | |
| "learning_rate": 0.00019988898749619702, | |
| "loss": 0.8936, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0013457535168324326, | |
| "grad_norm": 0.6276252865791321, | |
| "learning_rate": 0.00019987369566060176, | |
| "loss": 0.7279, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0014298631116344597, | |
| "grad_norm": 0.8388938903808594, | |
| "learning_rate": 0.00019985741811195097, | |
| "loss": 0.9929, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0015139727064364867, | |
| "grad_norm": 0.5140799283981323, | |
| "learning_rate": 0.00019984015501089752, | |
| "loss": 0.6327, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0015980823012385138, | |
| "grad_norm": 0.6707816123962402, | |
| "learning_rate": 0.0001998219065278212, | |
| "loss": 0.8234, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0016821918960405408, | |
| "grad_norm": 0.721817672252655, | |
| "learning_rate": 0.00019980267284282717, | |
| "loss": 0.8038, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0017663014908425678, | |
| "grad_norm": 0.6394887566566467, | |
| "learning_rate": 0.00019978245414574417, | |
| "loss": 0.7706, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0018504110856445949, | |
| "grad_norm": 0.523406982421875, | |
| "learning_rate": 0.00019976125063612252, | |
| "loss": 0.6581, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.001934520680446622, | |
| "grad_norm": 0.5513215065002441, | |
| "learning_rate": 0.00019973906252323238, | |
| "loss": 0.9306, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002018630275248649, | |
| "grad_norm": 0.5572922825813293, | |
| "learning_rate": 0.0001997158900260614, | |
| "loss": 0.7959, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.002102739870050676, | |
| "grad_norm": 0.567721426486969, | |
| "learning_rate": 0.0001996917333733128, | |
| "loss": 0.684, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.002186849464852703, | |
| "grad_norm": 0.8525415062904358, | |
| "learning_rate": 0.00019966659280340297, | |
| "loss": 0.6086, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0022709590596547303, | |
| "grad_norm": 0.6331901550292969, | |
| "learning_rate": 0.00019964046856445924, | |
| "loss": 0.8775, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.002355068654456757, | |
| "grad_norm": 0.5875001549720764, | |
| "learning_rate": 0.00019961336091431727, | |
| "loss": 0.7884, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0024391782492587844, | |
| "grad_norm": 0.4703105092048645, | |
| "learning_rate": 0.00019958527012051857, | |
| "loss": 0.8803, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.002523287844060811, | |
| "grad_norm": 0.46804890036582947, | |
| "learning_rate": 0.00019955619646030802, | |
| "loss": 0.7249, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0026073974388628384, | |
| "grad_norm": 0.7693579792976379, | |
| "learning_rate": 0.00019952614022063084, | |
| "loss": 1.0132, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0026915070336648653, | |
| "grad_norm": 0.7424771785736084, | |
| "learning_rate": 0.00019949510169813003, | |
| "loss": 0.9041, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0027756166284668925, | |
| "grad_norm": 0.6431573033332825, | |
| "learning_rate": 0.00019946308119914323, | |
| "loss": 0.8694, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0028597262232689193, | |
| "grad_norm": 0.5848168730735779, | |
| "learning_rate": 0.0001994300790396999, | |
| "loss": 0.7372, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0029438358180709466, | |
| "grad_norm": 0.5710355043411255, | |
| "learning_rate": 0.000199396095545518, | |
| "loss": 0.7431, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0030279454128729734, | |
| "grad_norm": 0.531509518623352, | |
| "learning_rate": 0.00019936113105200085, | |
| "loss": 0.7768, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0031120550076750007, | |
| "grad_norm": 0.4804302752017975, | |
| "learning_rate": 0.00019932518590423394, | |
| "loss": 0.7301, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0031961646024770275, | |
| "grad_norm": 0.5045036673545837, | |
| "learning_rate": 0.00019928826045698136, | |
| "loss": 0.6649, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0032802741972790548, | |
| "grad_norm": 0.41377341747283936, | |
| "learning_rate": 0.0001992503550746824, | |
| "loss": 0.8951, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0033643837920810816, | |
| "grad_norm": 0.534492552280426, | |
| "learning_rate": 0.0001992114701314478, | |
| "loss": 0.6731, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.003448493386883109, | |
| "grad_norm": 0.36469462513923645, | |
| "learning_rate": 0.0001991716060110563, | |
| "loss": 0.6212, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0035326029816851357, | |
| "grad_norm": 0.44110405445098877, | |
| "learning_rate": 0.00019913076310695068, | |
| "loss": 0.8254, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.003616712576487163, | |
| "grad_norm": 0.4397999942302704, | |
| "learning_rate": 0.00019908894182223388, | |
| "loss": 0.7209, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0037008221712891897, | |
| "grad_norm": 0.5739462971687317, | |
| "learning_rate": 0.00019904614256966512, | |
| "loss": 0.8095, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.003784931766091217, | |
| "grad_norm": 0.5561273097991943, | |
| "learning_rate": 0.00019900236577165576, | |
| "loss": 0.7944, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.003869041360893244, | |
| "grad_norm": 0.5051570534706116, | |
| "learning_rate": 0.0001989576118602651, | |
| "loss": 0.8223, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.003953150955695271, | |
| "grad_norm": 0.4156644642353058, | |
| "learning_rate": 0.00019891188127719618, | |
| "loss": 0.7488, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.004037260550497298, | |
| "grad_norm": 0.520811140537262, | |
| "learning_rate": 0.0001988651744737914, | |
| "loss": 0.8407, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.004121370145299325, | |
| "grad_norm": 0.7556874752044678, | |
| "learning_rate": 0.00019881749191102808, | |
| "loss": 0.7462, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.004205479740101352, | |
| "grad_norm": 0.692021369934082, | |
| "learning_rate": 0.00019876883405951377, | |
| "loss": 0.9717, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.004289589334903379, | |
| "grad_norm": 0.7206075191497803, | |
| "learning_rate": 0.00019871920139948192, | |
| "loss": 0.7717, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.004373698929705406, | |
| "grad_norm": 0.5266078114509583, | |
| "learning_rate": 0.0001986685944207868, | |
| "loss": 0.843, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.004457808524507433, | |
| "grad_norm": 0.4582868218421936, | |
| "learning_rate": 0.0001986170136228989, | |
| "loss": 0.709, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.004541918119309461, | |
| "grad_norm": 0.5274102687835693, | |
| "learning_rate": 0.00019856445951489982, | |
| "loss": 0.6895, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.004626027714111487, | |
| "grad_norm": 0.5545452237129211, | |
| "learning_rate": 0.0001985109326154774, | |
| "loss": 0.7461, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.004710137308913514, | |
| "grad_norm": 0.6029804944992065, | |
| "learning_rate": 0.00019845643345292054, | |
| "loss": 0.7165, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.004794246903715541, | |
| "grad_norm": 0.5293462872505188, | |
| "learning_rate": 0.00019840096256511398, | |
| "loss": 0.6772, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.004878356498517569, | |
| "grad_norm": 0.5571961402893066, | |
| "learning_rate": 0.00019834452049953297, | |
| "loss": 0.5852, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0049624660933195956, | |
| "grad_norm": 0.5451956391334534, | |
| "learning_rate": 0.00019828710781323792, | |
| "loss": 0.7897, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.005046575688121622, | |
| "grad_norm": 0.6682764887809753, | |
| "learning_rate": 0.0001982287250728689, | |
| "loss": 0.8288, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.005130685282923649, | |
| "grad_norm": 0.6017076373100281, | |
| "learning_rate": 0.0001981693728546399, | |
| "loss": 0.8949, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.005214794877725677, | |
| "grad_norm": 0.5544684529304504, | |
| "learning_rate": 0.0001981090517443334, | |
| "loss": 0.6256, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.005298904472527704, | |
| "grad_norm": 0.5648895502090454, | |
| "learning_rate": 0.00019804776233729444, | |
| "loss": 0.6528, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0053830140673297305, | |
| "grad_norm": 0.4875372350215912, | |
| "learning_rate": 0.0001979855052384247, | |
| "loss": 0.7416, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.005467123662131757, | |
| "grad_norm": 0.649088978767395, | |
| "learning_rate": 0.00019792228106217658, | |
| "loss": 0.7148, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.005551233256933785, | |
| "grad_norm": 0.39869123697280884, | |
| "learning_rate": 0.00019785809043254722, | |
| "loss": 0.5991, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.005635342851735812, | |
| "grad_norm": 0.6871259212493896, | |
| "learning_rate": 0.0001977929339830722, | |
| "loss": 0.6648, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.005719452446537839, | |
| "grad_norm": 0.5564553737640381, | |
| "learning_rate": 0.00019772681235681936, | |
| "loss": 0.842, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0058035620413398655, | |
| "grad_norm": 0.6598843336105347, | |
| "learning_rate": 0.00019765972620638248, | |
| "loss": 1.0846, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.005887671636141893, | |
| "grad_norm": 0.5316952466964722, | |
| "learning_rate": 0.00019759167619387476, | |
| "loss": 0.8352, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.00597178123094392, | |
| "grad_norm": 0.923975944519043, | |
| "learning_rate": 0.00019752266299092236, | |
| "loss": 0.761, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.006055890825745947, | |
| "grad_norm": 0.6253910064697266, | |
| "learning_rate": 0.00019745268727865774, | |
| "loss": 0.8125, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.006140000420547974, | |
| "grad_norm": 0.5049079060554504, | |
| "learning_rate": 0.0001973817497477129, | |
| "loss": 0.7948, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.006224110015350001, | |
| "grad_norm": 0.5132893919944763, | |
| "learning_rate": 0.00019730985109821266, | |
| "loss": 0.7674, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.006308219610152028, | |
| "grad_norm": 0.5373795032501221, | |
| "learning_rate": 0.00019723699203976766, | |
| "loss": 0.9345, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.006392329204954055, | |
| "grad_norm": 0.658785343170166, | |
| "learning_rate": 0.0001971631732914674, | |
| "loss": 0.7366, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.006476438799756082, | |
| "grad_norm": 0.42672979831695557, | |
| "learning_rate": 0.0001970883955818731, | |
| "loss": 0.8569, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0065605483945581095, | |
| "grad_norm": 0.6121531128883362, | |
| "learning_rate": 0.0001970126596490106, | |
| "loss": 0.7555, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.006644657989360136, | |
| "grad_norm": 0.5736799836158752, | |
| "learning_rate": 0.00019693596624036292, | |
| "loss": 0.7651, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.006728767584162163, | |
| "grad_norm": 0.5765815377235413, | |
| "learning_rate": 0.0001968583161128631, | |
| "loss": 0.7362, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00681287717896419, | |
| "grad_norm": 0.5843616127967834, | |
| "learning_rate": 0.00019677971003288655, | |
| "loss": 0.8952, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.006896986773766218, | |
| "grad_norm": 0.5047760009765625, | |
| "learning_rate": 0.00019670014877624353, | |
| "loss": 0.6147, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0069810963685682445, | |
| "grad_norm": 0.5199994444847107, | |
| "learning_rate": 0.00019661963312817148, | |
| "loss": 0.8876, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.007065205963370271, | |
| "grad_norm": 0.49524548649787903, | |
| "learning_rate": 0.0001965381638833274, | |
| "loss": 0.7281, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.007149315558172298, | |
| "grad_norm": 0.4999952018260956, | |
| "learning_rate": 0.00019645574184577982, | |
| "loss": 0.7967, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.007233425152974326, | |
| "grad_norm": 0.5383725762367249, | |
| "learning_rate": 0.000196372367829001, | |
| "loss": 0.7426, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.007317534747776353, | |
| "grad_norm": 0.4904189705848694, | |
| "learning_rate": 0.00019628804265585877, | |
| "loss": 0.8622, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0074016443425783795, | |
| "grad_norm": 0.5760535001754761, | |
| "learning_rate": 0.0001962027671586086, | |
| "loss": 0.7718, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.007485753937380406, | |
| "grad_norm": 0.5867496728897095, | |
| "learning_rate": 0.0001961165421788852, | |
| "loss": 0.7467, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.007569863532182434, | |
| "grad_norm": 0.5310566425323486, | |
| "learning_rate": 0.0001960293685676943, | |
| "loss": 0.7393, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.007653973126984461, | |
| "grad_norm": 0.4217927157878876, | |
| "learning_rate": 0.0001959412471854043, | |
| "loss": 0.696, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.007738082721786488, | |
| "grad_norm": 0.6704983711242676, | |
| "learning_rate": 0.0001958521789017376, | |
| "loss": 0.8934, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.007822192316588514, | |
| "grad_norm": 0.543559193611145, | |
| "learning_rate": 0.00019576216459576222, | |
| "loss": 0.547, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.007906301911390542, | |
| "grad_norm": 0.47226864099502563, | |
| "learning_rate": 0.00019567120515588308, | |
| "loss": 0.6723, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.007990411506192568, | |
| "grad_norm": 0.41364991664886475, | |
| "learning_rate": 0.00019557930147983302, | |
| "loss": 0.6751, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.008074521100994596, | |
| "grad_norm": 0.595898449420929, | |
| "learning_rate": 0.00019548645447466431, | |
| "loss": 0.8247, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.008158630695796623, | |
| "grad_norm": 0.6959226131439209, | |
| "learning_rate": 0.00019539266505673938, | |
| "loss": 0.6472, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.00824274029059865, | |
| "grad_norm": 0.4702865779399872, | |
| "learning_rate": 0.00019529793415172192, | |
| "loss": 0.6569, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.008326849885400677, | |
| "grad_norm": 0.5658578872680664, | |
| "learning_rate": 0.00019520226269456768, | |
| "loss": 0.8219, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.008410959480202705, | |
| "grad_norm": 0.5993213057518005, | |
| "learning_rate": 0.00019510565162951537, | |
| "loss": 0.7014, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.00849506907500473, | |
| "grad_norm": 0.5463736057281494, | |
| "learning_rate": 0.00019500810191007718, | |
| "loss": 0.657, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.008579178669806758, | |
| "grad_norm": 0.5861630439758301, | |
| "learning_rate": 0.00019490961449902946, | |
| "loss": 0.7181, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.008663288264608784, | |
| "grad_norm": 0.5364252328872681, | |
| "learning_rate": 0.0001948101903684032, | |
| "loss": 0.7727, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.008747397859410812, | |
| "grad_norm": 0.6941062808036804, | |
| "learning_rate": 0.00019470983049947444, | |
| "loss": 0.7413, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.00883150745421284, | |
| "grad_norm": 0.7105836272239685, | |
| "learning_rate": 0.00019460853588275454, | |
| "loss": 0.8972, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.008915617049014866, | |
| "grad_norm": 0.4372139871120453, | |
| "learning_rate": 0.00019450630751798048, | |
| "loss": 0.5836, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.008999726643816893, | |
| "grad_norm": 0.6489446759223938, | |
| "learning_rate": 0.000194403146414105, | |
| "loss": 0.877, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.009083836238618921, | |
| "grad_norm": 0.5020645260810852, | |
| "learning_rate": 0.00019429905358928646, | |
| "loss": 0.6035, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.009167945833420947, | |
| "grad_norm": 0.7856727242469788, | |
| "learning_rate": 0.00019419403007087907, | |
| "loss": 0.715, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.009252055428222975, | |
| "grad_norm": 0.6128354072570801, | |
| "learning_rate": 0.00019408807689542257, | |
| "loss": 0.8376, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.009336165023025, | |
| "grad_norm": 0.5398645401000977, | |
| "learning_rate": 0.00019398119510863197, | |
| "loss": 0.7696, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.009420274617827028, | |
| "grad_norm": 0.575010359287262, | |
| "learning_rate": 0.00019387338576538744, | |
| "loss": 0.6847, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.009504384212629056, | |
| "grad_norm": 0.5950291156768799, | |
| "learning_rate": 0.00019376464992972356, | |
| "loss": 0.8757, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.009588493807431082, | |
| "grad_norm": 0.5944623947143555, | |
| "learning_rate": 0.00019365498867481923, | |
| "loss": 0.7609, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.00967260340223311, | |
| "grad_norm": 0.4505338668823242, | |
| "learning_rate": 0.00019354440308298675, | |
| "loss": 0.8388, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.009756712997035137, | |
| "grad_norm": 0.4795793294906616, | |
| "learning_rate": 0.00019343289424566122, | |
| "loss": 0.8994, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.009840822591837163, | |
| "grad_norm": 0.4607990086078644, | |
| "learning_rate": 0.00019332046326338986, | |
| "loss": 0.6956, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.009924932186639191, | |
| "grad_norm": 0.5157178640365601, | |
| "learning_rate": 0.0001932071112458211, | |
| "loss": 0.5357, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.010009041781441217, | |
| "grad_norm": 0.5431911945343018, | |
| "learning_rate": 0.00019309283931169356, | |
| "loss": 0.7704, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.010093151376243245, | |
| "grad_norm": 0.5151223540306091, | |
| "learning_rate": 0.00019297764858882514, | |
| "loss": 0.6825, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.010177260971045272, | |
| "grad_norm": 0.5834304094314575, | |
| "learning_rate": 0.00019286154021410173, | |
| "loss": 0.5308, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.010261370565847298, | |
| "grad_norm": 0.6810638904571533, | |
| "learning_rate": 0.00019274451533346615, | |
| "loss": 0.7112, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.010345480160649326, | |
| "grad_norm": 0.48631688952445984, | |
| "learning_rate": 0.00019262657510190666, | |
| "loss": 0.7143, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.010429589755451354, | |
| "grad_norm": 0.6056327819824219, | |
| "learning_rate": 0.0001925077206834458, | |
| "loss": 0.6615, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.01051369935025338, | |
| "grad_norm": 0.4841483235359192, | |
| "learning_rate": 0.0001923879532511287, | |
| "loss": 0.7337, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.010597808945055407, | |
| "grad_norm": 0.5701360702514648, | |
| "learning_rate": 0.0001922672739870115, | |
| "loss": 0.5952, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.010681918539857433, | |
| "grad_norm": 0.6007951498031616, | |
| "learning_rate": 0.00019214568408214985, | |
| "loss": 0.7754, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.010766028134659461, | |
| "grad_norm": 0.6508734822273254, | |
| "learning_rate": 0.00019202318473658705, | |
| "loss": 0.9641, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.010850137729461489, | |
| "grad_norm": 0.608518123626709, | |
| "learning_rate": 0.00019189977715934213, | |
| "loss": 0.8748, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.010934247324263515, | |
| "grad_norm": 0.43734872341156006, | |
| "learning_rate": 0.00019177546256839812, | |
| "loss": 0.742, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.011018356919065542, | |
| "grad_norm": 0.5520758628845215, | |
| "learning_rate": 0.0001916502421906898, | |
| "loss": 0.5509, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.01110246651386757, | |
| "grad_norm": 0.3889513909816742, | |
| "learning_rate": 0.00019152411726209176, | |
| "loss": 0.4892, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.011186576108669596, | |
| "grad_norm": 0.5818157196044922, | |
| "learning_rate": 0.00019139708902740613, | |
| "loss": 0.6986, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.011270685703471624, | |
| "grad_norm": 0.5319333076477051, | |
| "learning_rate": 0.0001912691587403503, | |
| "loss": 0.6992, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.011354795298273651, | |
| "grad_norm": 0.54970782995224, | |
| "learning_rate": 0.00019114032766354453, | |
| "loss": 0.7275, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.011438904893075677, | |
| "grad_norm": 0.4689725637435913, | |
| "learning_rate": 0.00019101059706849957, | |
| "loss": 0.7424, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.011523014487877705, | |
| "grad_norm": 0.625424861907959, | |
| "learning_rate": 0.00019087996823560402, | |
| "loss": 0.6456, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.011607124082679731, | |
| "grad_norm": 0.4590742290019989, | |
| "learning_rate": 0.0001907484424541117, | |
| "loss": 0.792, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.011691233677481759, | |
| "grad_norm": 0.48178377747535706, | |
| "learning_rate": 0.00019061602102212898, | |
| "loss": 0.8383, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.011775343272283786, | |
| "grad_norm": 0.49192577600479126, | |
| "learning_rate": 0.00019048270524660196, | |
| "loss": 0.7532, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.011859452867085812, | |
| "grad_norm": 0.5243375301361084, | |
| "learning_rate": 0.0001903484964433035, | |
| "loss": 0.8344, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.01194356246188784, | |
| "grad_norm": 0.4813317060470581, | |
| "learning_rate": 0.00019021339593682028, | |
| "loss": 0.9669, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.012027672056689868, | |
| "grad_norm": 0.49398431181907654, | |
| "learning_rate": 0.00019007740506053983, | |
| "loss": 0.6194, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.012111781651491894, | |
| "grad_norm": 0.5070907473564148, | |
| "learning_rate": 0.0001899405251566371, | |
| "loss": 0.4834, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.012195891246293921, | |
| "grad_norm": 0.5289168953895569, | |
| "learning_rate": 0.00018980275757606157, | |
| "loss": 0.6171, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.012280000841095947, | |
| "grad_norm": 0.6366864442825317, | |
| "learning_rate": 0.00018966410367852362, | |
| "loss": 0.8194, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.012364110435897975, | |
| "grad_norm": 0.7337288856506348, | |
| "learning_rate": 0.00018952456483248119, | |
| "loss": 0.8086, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.012448220030700003, | |
| "grad_norm": 0.683866560459137, | |
| "learning_rate": 0.0001893841424151264, | |
| "loss": 0.609, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.012532329625502029, | |
| "grad_norm": 0.5067821145057678, | |
| "learning_rate": 0.0001892428378123718, | |
| "loss": 0.8969, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.012616439220304056, | |
| "grad_norm": 0.48310670256614685, | |
| "learning_rate": 0.0001891006524188368, | |
| "loss": 0.6375, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.012700548815106084, | |
| "grad_norm": 0.5184181928634644, | |
| "learning_rate": 0.00018895758763783383, | |
| "loss": 0.7395, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.01278465840990811, | |
| "grad_norm": 0.4921519458293915, | |
| "learning_rate": 0.00018881364488135448, | |
| "loss": 0.8635, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.012868768004710138, | |
| "grad_norm": 0.42886897921562195, | |
| "learning_rate": 0.00018866882557005567, | |
| "loss": 0.6536, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.012952877599512164, | |
| "grad_norm": 0.5397732257843018, | |
| "learning_rate": 0.00018852313113324552, | |
| "loss": 0.698, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.013036987194314191, | |
| "grad_norm": 0.6344212889671326, | |
| "learning_rate": 0.00018837656300886937, | |
| "loss": 0.7285, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.013121096789116219, | |
| "grad_norm": 0.5104556679725647, | |
| "learning_rate": 0.00018822912264349534, | |
| "loss": 0.5943, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.013205206383918245, | |
| "grad_norm": 0.5612802505493164, | |
| "learning_rate": 0.00018808081149230036, | |
| "loss": 0.7107, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.013289315978720273, | |
| "grad_norm": 0.47731515765190125, | |
| "learning_rate": 0.00018793163101905563, | |
| "loss": 0.7419, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.0133734255735223, | |
| "grad_norm": 0.47317004203796387, | |
| "learning_rate": 0.00018778158269611218, | |
| "loss": 0.6444, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.013457535168324326, | |
| "grad_norm": 0.5630394816398621, | |
| "learning_rate": 0.00018763066800438636, | |
| "loss": 0.66, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.013541644763126354, | |
| "grad_norm": 0.4079178273677826, | |
| "learning_rate": 0.0001874788884333453, | |
| "loss": 0.6648, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.01362575435792838, | |
| "grad_norm": 0.4921400547027588, | |
| "learning_rate": 0.00018732624548099204, | |
| "loss": 0.6975, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.013709863952730408, | |
| "grad_norm": 0.5111605525016785, | |
| "learning_rate": 0.0001871727406538509, | |
| "loss": 0.7385, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.013793973547532435, | |
| "grad_norm": 0.6458659172058105, | |
| "learning_rate": 0.0001870183754669526, | |
| "loss": 0.7494, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.013878083142334461, | |
| "grad_norm": 0.5011044144630432, | |
| "learning_rate": 0.00018686315144381913, | |
| "loss": 0.5897, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.013962192737136489, | |
| "grad_norm": 0.524375319480896, | |
| "learning_rate": 0.000186707070116449, | |
| "loss": 0.721, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.014046302331938517, | |
| "grad_norm": 0.6293932795524597, | |
| "learning_rate": 0.0001865501330253019, | |
| "loss": 0.8962, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.014130411926740543, | |
| "grad_norm": 0.6843417882919312, | |
| "learning_rate": 0.00018639234171928353, | |
| "loss": 0.9065, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.01421452152154257, | |
| "grad_norm": 0.5260113477706909, | |
| "learning_rate": 0.0001862336977557304, | |
| "loss": 0.6079, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.014298631116344596, | |
| "grad_norm": 0.4797818064689636, | |
| "learning_rate": 0.0001860742027003944, | |
| "loss": 0.839, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.014382740711146624, | |
| "grad_norm": 0.5858710408210754, | |
| "learning_rate": 0.00018591385812742725, | |
| "loss": 0.7166, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.014466850305948652, | |
| "grad_norm": 0.47610747814178467, | |
| "learning_rate": 0.00018575266561936523, | |
| "loss": 0.5801, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.014550959900750678, | |
| "grad_norm": 0.7322710156440735, | |
| "learning_rate": 0.00018559062676711332, | |
| "loss": 0.7788, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.014635069495552705, | |
| "grad_norm": 0.5088084936141968, | |
| "learning_rate": 0.0001854277431699295, | |
| "loss": 0.6265, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.014719179090354733, | |
| "grad_norm": 0.5596153736114502, | |
| "learning_rate": 0.00018526401643540922, | |
| "loss": 0.7236, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.014803288685156759, | |
| "grad_norm": 0.7297357320785522, | |
| "learning_rate": 0.00018509944817946922, | |
| "loss": 0.6347, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.014887398279958787, | |
| "grad_norm": 0.45782700181007385, | |
| "learning_rate": 0.00018493404002633166, | |
| "loss": 0.6243, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.014971507874760813, | |
| "grad_norm": 0.46937909722328186, | |
| "learning_rate": 0.00018476779360850832, | |
| "loss": 0.7372, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.01505561746956284, | |
| "grad_norm": 0.4801153540611267, | |
| "learning_rate": 0.00018460071056678422, | |
| "loss": 0.6242, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.015139727064364868, | |
| "grad_norm": 0.42098918557167053, | |
| "learning_rate": 0.00018443279255020152, | |
| "loss": 0.823, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.015223836659166894, | |
| "grad_norm": 0.5428348183631897, | |
| "learning_rate": 0.00018426404121604323, | |
| "loss": 0.9499, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.015307946253968922, | |
| "grad_norm": 0.5740225315093994, | |
| "learning_rate": 0.00018409445822981693, | |
| "loss": 0.8347, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.01539205584877095, | |
| "grad_norm": 0.5669785141944885, | |
| "learning_rate": 0.00018392404526523817, | |
| "loss": 0.7831, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.015476165443572975, | |
| "grad_norm": 0.5455140471458435, | |
| "learning_rate": 0.0001837528040042142, | |
| "loss": 0.6875, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.015560275038375003, | |
| "grad_norm": 0.5965580344200134, | |
| "learning_rate": 0.00018358073613682706, | |
| "loss": 0.8344, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.01564438463317703, | |
| "grad_norm": 0.4952075779438019, | |
| "learning_rate": 0.00018340784336131713, | |
| "loss": 0.4143, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.01572849422797906, | |
| "grad_norm": 0.584922730922699, | |
| "learning_rate": 0.00018323412738406635, | |
| "loss": 0.6031, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.015812603822781084, | |
| "grad_norm": 0.5585885047912598, | |
| "learning_rate": 0.00018305958991958127, | |
| "loss": 0.7149, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.01589671341758311, | |
| "grad_norm": 0.6081206202507019, | |
| "learning_rate": 0.0001828842326904762, | |
| "loss": 0.7388, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.015980823012385136, | |
| "grad_norm": 0.5838598608970642, | |
| "learning_rate": 0.00018270805742745617, | |
| "loss": 0.7166, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.016064932607187166, | |
| "grad_norm": 0.5591777563095093, | |
| "learning_rate": 0.00018253106586929997, | |
| "loss": 0.6833, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.01614904220198919, | |
| "grad_norm": 0.49327778816223145, | |
| "learning_rate": 0.00018235325976284275, | |
| "loss": 0.6518, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.016233151796791218, | |
| "grad_norm": 0.5115340948104858, | |
| "learning_rate": 0.00018217464086295904, | |
| "loss": 0.9741, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.016317261391593247, | |
| "grad_norm": 0.44589102268218994, | |
| "learning_rate": 0.00018199521093254523, | |
| "loss": 0.7523, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.016401370986395273, | |
| "grad_norm": 0.4889369606971741, | |
| "learning_rate": 0.00018181497174250236, | |
| "loss": 0.5771, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0164854805811973, | |
| "grad_norm": 0.5447149276733398, | |
| "learning_rate": 0.00018163392507171842, | |
| "loss": 0.7554, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.01656959017599933, | |
| "grad_norm": 0.4478766918182373, | |
| "learning_rate": 0.00018145207270705096, | |
| "loss": 0.7293, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.016653699770801354, | |
| "grad_norm": 0.4789084494113922, | |
| "learning_rate": 0.0001812694164433094, | |
| "loss": 0.7598, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.01673780936560338, | |
| "grad_norm": 0.5047253966331482, | |
| "learning_rate": 0.00018108595808323736, | |
| "loss": 0.703, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.01682191896040541, | |
| "grad_norm": 0.44823339581489563, | |
| "learning_rate": 0.00018090169943749476, | |
| "loss": 0.71, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.016906028555207436, | |
| "grad_norm": 0.5810935497283936, | |
| "learning_rate": 0.00018071664232464002, | |
| "loss": 0.7506, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.01699013815000946, | |
| "grad_norm": 0.44463852047920227, | |
| "learning_rate": 0.0001805307885711122, | |
| "loss": 0.7075, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.01707424774481149, | |
| "grad_norm": 0.5899863839149475, | |
| "learning_rate": 0.00018034414001121278, | |
| "loss": 0.7882, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.017158357339613517, | |
| "grad_norm": 0.5106207132339478, | |
| "learning_rate": 0.00018015669848708767, | |
| "loss": 0.8387, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.017242466934415543, | |
| "grad_norm": 0.5090427994728088, | |
| "learning_rate": 0.00017996846584870908, | |
| "loss": 0.8835, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.01732657652921757, | |
| "grad_norm": 0.5673806071281433, | |
| "learning_rate": 0.0001797794439538571, | |
| "loss": 0.6051, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.0174106861240196, | |
| "grad_norm": 0.4103122651576996, | |
| "learning_rate": 0.0001795896346681016, | |
| "loss": 0.6471, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.017494795718821624, | |
| "grad_norm": 0.5361846089363098, | |
| "learning_rate": 0.00017939903986478355, | |
| "loss": 0.6601, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.01757890531362365, | |
| "grad_norm": 0.6034552454948425, | |
| "learning_rate": 0.00017920766142499672, | |
| "loss": 0.7289, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.01766301490842568, | |
| "grad_norm": 0.5352544784545898, | |
| "learning_rate": 0.00017901550123756906, | |
| "loss": 0.6441, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.017747124503227706, | |
| "grad_norm": 0.48942509293556213, | |
| "learning_rate": 0.00017882256119904403, | |
| "loss": 0.7517, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.01783123409802973, | |
| "grad_norm": 0.6330734491348267, | |
| "learning_rate": 0.00017862884321366188, | |
| "loss": 0.8312, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.01791534369283176, | |
| "grad_norm": 0.5588996410369873, | |
| "learning_rate": 0.000178434349193341, | |
| "loss": 0.8497, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.017999453287633787, | |
| "grad_norm": 0.4862910807132721, | |
| "learning_rate": 0.0001782390810576588, | |
| "loss": 0.7367, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.018083562882435813, | |
| "grad_norm": 0.45474931597709656, | |
| "learning_rate": 0.000178043040733833, | |
| "loss": 0.6998, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.018167672477237842, | |
| "grad_norm": 0.4923236072063446, | |
| "learning_rate": 0.00017784623015670238, | |
| "loss": 0.8464, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.018251782072039868, | |
| "grad_norm": 0.6574183702468872, | |
| "learning_rate": 0.00017764865126870786, | |
| "loss": 0.7606, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.018335891666841894, | |
| "grad_norm": 0.4556383192539215, | |
| "learning_rate": 0.00017745030601987337, | |
| "loss": 0.5689, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.018420001261643924, | |
| "grad_norm": 0.49504584074020386, | |
| "learning_rate": 0.00017725119636778644, | |
| "loss": 0.6945, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.01850411085644595, | |
| "grad_norm": 0.48922199010849, | |
| "learning_rate": 0.00017705132427757895, | |
| "loss": 0.6024, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.018588220451247976, | |
| "grad_norm": 0.4516255259513855, | |
| "learning_rate": 0.00017685069172190766, | |
| "loss": 0.6825, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.01867233004605, | |
| "grad_norm": 0.5373181104660034, | |
| "learning_rate": 0.00017664930068093498, | |
| "loss": 0.7165, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.01875643964085203, | |
| "grad_norm": 0.45531630516052246, | |
| "learning_rate": 0.00017644715314230918, | |
| "loss": 0.614, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.018840549235654057, | |
| "grad_norm": 0.38286691904067993, | |
| "learning_rate": 0.0001762442511011448, | |
| "loss": 0.6237, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.018924658830456083, | |
| "grad_norm": 0.8253074288368225, | |
| "learning_rate": 0.0001760405965600031, | |
| "loss": 0.7572, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.019008768425258112, | |
| "grad_norm": 0.49788135290145874, | |
| "learning_rate": 0.0001758361915288722, | |
| "loss": 0.836, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.019092878020060138, | |
| "grad_norm": 0.7003273367881775, | |
| "learning_rate": 0.0001756310380251472, | |
| "loss": 0.6137, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.019176987614862164, | |
| "grad_norm": 0.5549869537353516, | |
| "learning_rate": 0.00017542513807361037, | |
| "loss": 0.7504, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.019261097209664194, | |
| "grad_norm": 0.5484214425086975, | |
| "learning_rate": 0.00017521849370641114, | |
| "loss": 0.8757, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.01934520680446622, | |
| "grad_norm": 0.48561275005340576, | |
| "learning_rate": 0.00017501110696304596, | |
| "loss": 0.6577, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.019429316399268245, | |
| "grad_norm": 0.5252694487571716, | |
| "learning_rate": 0.00017480297989033825, | |
| "loss": 0.7415, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.019513425994070275, | |
| "grad_norm": 0.5625510811805725, | |
| "learning_rate": 0.00017459411454241822, | |
| "loss": 0.7563, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.0195975355888723, | |
| "grad_norm": 0.5950868725776672, | |
| "learning_rate": 0.00017438451298070252, | |
| "loss": 0.5738, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.019681645183674327, | |
| "grad_norm": 0.3883379399776459, | |
| "learning_rate": 0.00017417417727387394, | |
| "loss": 0.6791, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.019765754778476356, | |
| "grad_norm": 0.3784724175930023, | |
| "learning_rate": 0.000173963109497861, | |
| "loss": 0.6673, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.019849864373278382, | |
| "grad_norm": 0.5025336146354675, | |
| "learning_rate": 0.0001737513117358174, | |
| "loss": 0.678, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.019933973968080408, | |
| "grad_norm": 0.5078927874565125, | |
| "learning_rate": 0.0001735387860781016, | |
| "loss": 0.7093, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.020018083562882434, | |
| "grad_norm": 0.67873215675354, | |
| "learning_rate": 0.00017332553462225602, | |
| "loss": 0.7362, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.020102193157684464, | |
| "grad_norm": 0.6332508325576782, | |
| "learning_rate": 0.00017311155947298643, | |
| "loss": 0.848, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.02018630275248649, | |
| "grad_norm": 0.6372016668319702, | |
| "learning_rate": 0.00017289686274214118, | |
| "loss": 0.7129, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.020270412347288515, | |
| "grad_norm": 0.6516194939613342, | |
| "learning_rate": 0.0001726814465486903, | |
| "loss": 0.7255, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.020354521942090545, | |
| "grad_norm": 0.5692350268363953, | |
| "learning_rate": 0.0001724653130187047, | |
| "loss": 0.6044, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.02043863153689257, | |
| "grad_norm": 0.6217871308326721, | |
| "learning_rate": 0.00017224846428533499, | |
| "loss": 0.8463, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.020522741131694597, | |
| "grad_norm": 0.4982471466064453, | |
| "learning_rate": 0.0001720309024887907, | |
| "loss": 0.7304, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.020606850726496626, | |
| "grad_norm": 0.46791961789131165, | |
| "learning_rate": 0.00017181262977631888, | |
| "loss": 0.6765, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.020690960321298652, | |
| "grad_norm": 0.54993736743927, | |
| "learning_rate": 0.00017159364830218312, | |
| "loss": 0.5422, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.020775069916100678, | |
| "grad_norm": 0.557114839553833, | |
| "learning_rate": 0.00017137396022764214, | |
| "loss": 0.8051, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.020859179510902708, | |
| "grad_norm": 0.4503937065601349, | |
| "learning_rate": 0.00017115356772092857, | |
| "loss": 0.6993, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.020943289105704734, | |
| "grad_norm": 0.5274853706359863, | |
| "learning_rate": 0.0001709324729572274, | |
| "loss": 0.5102, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.02102739870050676, | |
| "grad_norm": 0.5487212538719177, | |
| "learning_rate": 0.00017071067811865476, | |
| "loss": 0.7262, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02111150829530879, | |
| "grad_norm": 0.4630710482597351, | |
| "learning_rate": 0.00017048818539423615, | |
| "loss": 0.6907, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.021195617890110815, | |
| "grad_norm": 0.4759877622127533, | |
| "learning_rate": 0.00017026499697988493, | |
| "loss": 0.6945, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.02127972748491284, | |
| "grad_norm": 0.44283753633499146, | |
| "learning_rate": 0.00017004111507838064, | |
| "loss": 0.491, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.021363837079714867, | |
| "grad_norm": 0.6089275479316711, | |
| "learning_rate": 0.00016981654189934727, | |
| "loss": 0.7725, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.021447946674516896, | |
| "grad_norm": 0.48466405272483826, | |
| "learning_rate": 0.00016959127965923142, | |
| "loss": 0.6735, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.021532056269318922, | |
| "grad_norm": 0.5322965383529663, | |
| "learning_rate": 0.0001693653305812805, | |
| "loss": 0.7648, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.021616165864120948, | |
| "grad_norm": 0.5427155494689941, | |
| "learning_rate": 0.00016913869689552064, | |
| "loss": 0.6212, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.021700275458922978, | |
| "grad_norm": 0.6860820651054382, | |
| "learning_rate": 0.00016891138083873487, | |
| "loss": 0.9867, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.021784385053725003, | |
| "grad_norm": 0.5238876342773438, | |
| "learning_rate": 0.00016868338465444085, | |
| "loss": 0.7087, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.02186849464852703, | |
| "grad_norm": 0.613838255405426, | |
| "learning_rate": 0.00016845471059286887, | |
| "loss": 0.6237, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.02195260424332906, | |
| "grad_norm": 0.6442722082138062, | |
| "learning_rate": 0.00016822536091093965, | |
| "loss": 0.6983, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.022036713838131085, | |
| "grad_norm": 0.5362198352813721, | |
| "learning_rate": 0.00016799533787224192, | |
| "loss": 0.6856, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.02212082343293311, | |
| "grad_norm": 0.5040967464447021, | |
| "learning_rate": 0.00016776464374701025, | |
| "loss": 0.5694, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.02220493302773514, | |
| "grad_norm": 0.6717795133590698, | |
| "learning_rate": 0.00016753328081210245, | |
| "loss": 0.7468, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.022289042622537166, | |
| "grad_norm": 0.5034369230270386, | |
| "learning_rate": 0.00016730125135097735, | |
| "loss": 0.7802, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.022373152217339192, | |
| "grad_norm": 0.49784526228904724, | |
| "learning_rate": 0.000167068557653672, | |
| "loss": 0.6088, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.02245726181214122, | |
| "grad_norm": 0.6437605619430542, | |
| "learning_rate": 0.0001668352020167793, | |
| "loss": 0.6883, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.022541371406943247, | |
| "grad_norm": 0.7116613984107971, | |
| "learning_rate": 0.00016660118674342517, | |
| "loss": 0.6185, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.022625481001745273, | |
| "grad_norm": 0.5817685723304749, | |
| "learning_rate": 0.00016636651414324587, | |
| "loss": 0.6651, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.022709590596547303, | |
| "grad_norm": 0.37707141041755676, | |
| "learning_rate": 0.00016613118653236518, | |
| "loss": 0.6047, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.02279370019134933, | |
| "grad_norm": 0.4310030937194824, | |
| "learning_rate": 0.0001658952062333717, | |
| "loss": 0.7464, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.022877809786151355, | |
| "grad_norm": 0.5687549710273743, | |
| "learning_rate": 0.00016565857557529566, | |
| "loss": 0.7018, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.02296191938095338, | |
| "grad_norm": 0.4826352894306183, | |
| "learning_rate": 0.00016542129689358612, | |
| "loss": 0.6731, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.02304602897575541, | |
| "grad_norm": 0.58961021900177, | |
| "learning_rate": 0.0001651833725300879, | |
| "loss": 0.7081, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.023130138570557436, | |
| "grad_norm": 0.3250476121902466, | |
| "learning_rate": 0.00016494480483301836, | |
| "loss": 0.3532, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.023214248165359462, | |
| "grad_norm": 0.6899309754371643, | |
| "learning_rate": 0.00016470559615694446, | |
| "loss": 0.7164, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.02329835776016149, | |
| "grad_norm": 0.5115370154380798, | |
| "learning_rate": 0.00016446574886275913, | |
| "loss": 0.8014, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.023382467354963517, | |
| "grad_norm": 0.4905194640159607, | |
| "learning_rate": 0.00016422526531765846, | |
| "loss": 0.5492, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.023466576949765543, | |
| "grad_norm": 0.5156686305999756, | |
| "learning_rate": 0.00016398414789511786, | |
| "loss": 0.5411, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.023550686544567573, | |
| "grad_norm": 0.6288424730300903, | |
| "learning_rate": 0.000163742398974869, | |
| "loss": 0.7018, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.0236347961393696, | |
| "grad_norm": 0.5037186145782471, | |
| "learning_rate": 0.00016350002094287609, | |
| "loss": 0.7827, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.023718905734171625, | |
| "grad_norm": 0.4114580750465393, | |
| "learning_rate": 0.00016325701619131246, | |
| "loss": 0.707, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.023803015328973654, | |
| "grad_norm": 0.4033624529838562, | |
| "learning_rate": 0.00016301338711853693, | |
| "loss": 0.7176, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.02388712492377568, | |
| "grad_norm": 0.7627992630004883, | |
| "learning_rate": 0.00016276913612907007, | |
| "loss": 0.7968, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.023971234518577706, | |
| "grad_norm": 0.5841469168663025, | |
| "learning_rate": 0.00016252426563357055, | |
| "loss": 0.7753, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.024055344113379735, | |
| "grad_norm": 0.7086002826690674, | |
| "learning_rate": 0.00016227877804881127, | |
| "loss": 0.701, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.02413945370818176, | |
| "grad_norm": 0.6601248979568481, | |
| "learning_rate": 0.00016203267579765563, | |
| "loss": 0.709, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.024223563302983787, | |
| "grad_norm": 0.6370972990989685, | |
| "learning_rate": 0.00016178596130903344, | |
| "loss": 0.7909, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.024307672897785813, | |
| "grad_norm": 0.5954873561859131, | |
| "learning_rate": 0.00016153863701791717, | |
| "loss": 0.5179, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.024391782492587843, | |
| "grad_norm": 0.6457350850105286, | |
| "learning_rate": 0.00016129070536529766, | |
| "loss": 0.8393, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.02447589208738987, | |
| "grad_norm": 0.6270333528518677, | |
| "learning_rate": 0.00016104216879816026, | |
| "loss": 0.6952, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.024560001682191895, | |
| "grad_norm": 0.4968658983707428, | |
| "learning_rate": 0.00016079302976946055, | |
| "loss": 0.8266, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.024644111276993924, | |
| "grad_norm": 0.6642244458198547, | |
| "learning_rate": 0.00016054329073810015, | |
| "loss": 0.7603, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.02472822087179595, | |
| "grad_norm": 0.4561289846897125, | |
| "learning_rate": 0.00016029295416890248, | |
| "loss": 0.5671, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.024812330466597976, | |
| "grad_norm": 0.5380001664161682, | |
| "learning_rate": 0.00016004202253258842, | |
| "loss": 0.7924, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.024896440061400005, | |
| "grad_norm": 0.48755741119384766, | |
| "learning_rate": 0.0001597904983057519, | |
| "loss": 0.5793, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.02498054965620203, | |
| "grad_norm": 0.4303657114505768, | |
| "learning_rate": 0.00015953838397083552, | |
| "loss": 0.5611, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.025064659251004057, | |
| "grad_norm": 0.5174483060836792, | |
| "learning_rate": 0.00015928568201610595, | |
| "loss": 0.8403, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.025148768845806087, | |
| "grad_norm": 0.4314514398574829, | |
| "learning_rate": 0.00015903239493562948, | |
| "loss": 0.6122, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.025232878440608113, | |
| "grad_norm": 0.5762454867362976, | |
| "learning_rate": 0.00015877852522924732, | |
| "loss": 0.6049, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02531698803541014, | |
| "grad_norm": 0.40056511759757996, | |
| "learning_rate": 0.00015852407540255104, | |
| "loss": 0.6657, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.025401097630212168, | |
| "grad_norm": 0.3158378303050995, | |
| "learning_rate": 0.00015826904796685762, | |
| "loss": 0.4039, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.025485207225014194, | |
| "grad_norm": 0.4131661653518677, | |
| "learning_rate": 0.00015801344543918495, | |
| "loss": 0.6172, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.02556931681981622, | |
| "grad_norm": 0.46788647770881653, | |
| "learning_rate": 0.00015775727034222675, | |
| "loss": 0.6595, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.025653426414618246, | |
| "grad_norm": 0.8410905003547668, | |
| "learning_rate": 0.00015750052520432787, | |
| "loss": 0.6009, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.025737536009420275, | |
| "grad_norm": 0.4922102987766266, | |
| "learning_rate": 0.0001572432125594591, | |
| "loss": 0.737, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.0258216456042223, | |
| "grad_norm": 0.48740994930267334, | |
| "learning_rate": 0.00015698533494719238, | |
| "loss": 0.6165, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.025905755199024327, | |
| "grad_norm": 0.6041740775108337, | |
| "learning_rate": 0.00015672689491267567, | |
| "loss": 0.5473, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.025989864793826357, | |
| "grad_norm": 0.5123183727264404, | |
| "learning_rate": 0.00015646789500660773, | |
| "loss": 0.7032, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.026073974388628383, | |
| "grad_norm": 0.4975537359714508, | |
| "learning_rate": 0.00015620833778521307, | |
| "loss": 0.7373, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.02615808398343041, | |
| "grad_norm": 0.41992077231407166, | |
| "learning_rate": 0.0001559482258102167, | |
| "loss": 0.7797, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.026242193578232438, | |
| "grad_norm": 0.5822920203208923, | |
| "learning_rate": 0.00015568756164881882, | |
| "loss": 0.6739, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.026326303173034464, | |
| "grad_norm": 0.5167087316513062, | |
| "learning_rate": 0.00015542634787366942, | |
| "loss": 0.6933, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.02641041276783649, | |
| "grad_norm": 0.5934262871742249, | |
| "learning_rate": 0.00015516458706284303, | |
| "loss": 0.5772, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.02649452236263852, | |
| "grad_norm": 0.43231528997421265, | |
| "learning_rate": 0.0001549022817998132, | |
| "loss": 0.7363, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.026578631957440545, | |
| "grad_norm": 0.6113074421882629, | |
| "learning_rate": 0.00015463943467342693, | |
| "loss": 0.8104, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.02666274155224257, | |
| "grad_norm": 0.6448655724525452, | |
| "learning_rate": 0.00015437604827787927, | |
| "loss": 0.7304, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.0267468511470446, | |
| "grad_norm": 0.630864143371582, | |
| "learning_rate": 0.00015411212521268758, | |
| "loss": 0.7209, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.026830960741846627, | |
| "grad_norm": 0.5093964338302612, | |
| "learning_rate": 0.00015384766808266602, | |
| "loss": 0.7396, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.026915070336648653, | |
| "grad_norm": 0.5301403999328613, | |
| "learning_rate": 0.00015358267949789966, | |
| "loss": 0.8367, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02699917993145068, | |
| "grad_norm": 0.5352559089660645, | |
| "learning_rate": 0.00015331716207371888, | |
| "loss": 0.6919, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.027083289526252708, | |
| "grad_norm": 0.4436148703098297, | |
| "learning_rate": 0.0001530511184306734, | |
| "loss": 0.6293, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.027167399121054734, | |
| "grad_norm": 0.5910261273384094, | |
| "learning_rate": 0.00015278455119450664, | |
| "loss": 0.7454, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.02725150871585676, | |
| "grad_norm": 0.5505343079566956, | |
| "learning_rate": 0.0001525174629961296, | |
| "loss": 0.7661, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.02733561831065879, | |
| "grad_norm": 0.5973290205001831, | |
| "learning_rate": 0.0001522498564715949, | |
| "loss": 0.754, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.027419727905460815, | |
| "grad_norm": 0.5219733715057373, | |
| "learning_rate": 0.00015198173426207094, | |
| "loss": 0.9089, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.02750383750026284, | |
| "grad_norm": 0.48295462131500244, | |
| "learning_rate": 0.00015171309901381572, | |
| "loss": 0.7115, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.02758794709506487, | |
| "grad_norm": 0.3980119228363037, | |
| "learning_rate": 0.00015144395337815064, | |
| "loss": 0.6346, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.027672056689866897, | |
| "grad_norm": 0.44824841618537903, | |
| "learning_rate": 0.00015117430001143452, | |
| "loss": 0.7356, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.027756166284668923, | |
| "grad_norm": 0.4944741129875183, | |
| "learning_rate": 0.00015090414157503714, | |
| "loss": 0.7034, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.027840275879470952, | |
| "grad_norm": 0.4512578845024109, | |
| "learning_rate": 0.00015063348073531324, | |
| "loss": 0.7027, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.027924385474272978, | |
| "grad_norm": 0.5666627287864685, | |
| "learning_rate": 0.0001503623201635761, | |
| "loss": 0.6587, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.028008495069075004, | |
| "grad_norm": 0.4888990521430969, | |
| "learning_rate": 0.000150090662536071, | |
| "loss": 0.5327, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.028092604663877033, | |
| "grad_norm": 0.5111712217330933, | |
| "learning_rate": 0.0001498185105339491, | |
| "loss": 0.6042, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.02817671425867906, | |
| "grad_norm": 0.4700281620025635, | |
| "learning_rate": 0.00014954586684324078, | |
| "loss": 0.583, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.028260823853481085, | |
| "grad_norm": 0.5448441505432129, | |
| "learning_rate": 0.00014927273415482915, | |
| "loss": 0.6319, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.02834493344828311, | |
| "grad_norm": 0.6051144599914551, | |
| "learning_rate": 0.00014899911516442365, | |
| "loss": 0.5532, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.02842904304308514, | |
| "grad_norm": 0.6583679914474487, | |
| "learning_rate": 0.00014872501257253323, | |
| "loss": 0.7324, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.028513152637887167, | |
| "grad_norm": 0.468326210975647, | |
| "learning_rate": 0.0001484504290844398, | |
| "loss": 0.6747, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.028597262232689193, | |
| "grad_norm": 0.5111216306686401, | |
| "learning_rate": 0.00014817536741017152, | |
| "loss": 0.8554, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.028681371827491222, | |
| "grad_norm": 0.42130234837532043, | |
| "learning_rate": 0.00014789983026447612, | |
| "loss": 0.5064, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.028765481422293248, | |
| "grad_norm": 0.6524533033370972, | |
| "learning_rate": 0.0001476238203667939, | |
| "loss": 0.7676, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.028849591017095274, | |
| "grad_norm": 0.5613992810249329, | |
| "learning_rate": 0.0001473473404412312, | |
| "loss": 0.563, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.028933700611897303, | |
| "grad_norm": 0.5069441795349121, | |
| "learning_rate": 0.0001470703932165333, | |
| "loss": 0.6787, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.02901781020669933, | |
| "grad_norm": 0.47367653250694275, | |
| "learning_rate": 0.00014679298142605734, | |
| "loss": 0.8746, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.029101919801501355, | |
| "grad_norm": 0.4816168546676636, | |
| "learning_rate": 0.00014651510780774583, | |
| "loss": 0.6131, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.029186029396303385, | |
| "grad_norm": 0.5362951159477234, | |
| "learning_rate": 0.00014623677510409918, | |
| "loss": 0.6675, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.02927013899110541, | |
| "grad_norm": 0.4960172772407532, | |
| "learning_rate": 0.00014595798606214882, | |
| "loss": 0.624, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.029354248585907437, | |
| "grad_norm": 0.3880994915962219, | |
| "learning_rate": 0.00014567874343342997, | |
| "loss": 0.4903, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.029438358180709466, | |
| "grad_norm": 0.5134441256523132, | |
| "learning_rate": 0.00014539904997395468, | |
| "loss": 0.7855, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.029522467775511492, | |
| "grad_norm": 0.5130655765533447, | |
| "learning_rate": 0.00014511890844418453, | |
| "loss": 0.723, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.029606577370313518, | |
| "grad_norm": 0.63141930103302, | |
| "learning_rate": 0.00014483832160900326, | |
| "loss": 0.8834, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.029690686965115544, | |
| "grad_norm": 0.5271434783935547, | |
| "learning_rate": 0.00014455729223768966, | |
| "loss": 0.5313, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.029774796559917573, | |
| "grad_norm": 0.44754379987716675, | |
| "learning_rate": 0.0001442758231038902, | |
| "loss": 0.6777, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.0298589061547196, | |
| "grad_norm": 0.5021183490753174, | |
| "learning_rate": 0.00014399391698559152, | |
| "loss": 0.476, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.029943015749521625, | |
| "grad_norm": 0.5433945059776306, | |
| "learning_rate": 0.0001437115766650933, | |
| "loss": 0.6773, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.030027125344323655, | |
| "grad_norm": 0.42916494607925415, | |
| "learning_rate": 0.00014342880492898048, | |
| "loss": 0.5919, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.03011123493912568, | |
| "grad_norm": 0.5809445381164551, | |
| "learning_rate": 0.0001431456045680959, | |
| "loss": 0.7717, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.030195344533927707, | |
| "grad_norm": 0.5388799905776978, | |
| "learning_rate": 0.00014286197837751286, | |
| "loss": 0.6828, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.030279454128729736, | |
| "grad_norm": 0.5621784329414368, | |
| "learning_rate": 0.00014257792915650728, | |
| "loss": 0.5758, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.030363563723531762, | |
| "grad_norm": 0.5903728604316711, | |
| "learning_rate": 0.00014229345970853032, | |
| "loss": 0.6692, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.030447673318333788, | |
| "grad_norm": 0.5247724652290344, | |
| "learning_rate": 0.00014200857284118066, | |
| "loss": 0.513, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.030531782913135817, | |
| "grad_norm": 0.6713283658027649, | |
| "learning_rate": 0.00014172327136617656, | |
| "loss": 0.6672, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.030615892507937843, | |
| "grad_norm": 0.4961901903152466, | |
| "learning_rate": 0.00014143755809932845, | |
| "loss": 0.5941, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.03070000210273987, | |
| "grad_norm": 0.5194029211997986, | |
| "learning_rate": 0.00014115143586051088, | |
| "loss": 0.7219, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.0307841116975419, | |
| "grad_norm": 0.5475531816482544, | |
| "learning_rate": 0.00014086490747363493, | |
| "loss": 0.7603, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.030868221292343925, | |
| "grad_norm": 0.6493870615959167, | |
| "learning_rate": 0.00014057797576662, | |
| "loss": 0.6935, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.03095233088714595, | |
| "grad_norm": 0.7089727520942688, | |
| "learning_rate": 0.00014029064357136628, | |
| "loss": 0.7187, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.031036440481947977, | |
| "grad_norm": 0.4623405337333679, | |
| "learning_rate": 0.00014000291372372647, | |
| "loss": 0.5853, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.031120550076750006, | |
| "grad_norm": 0.5741824507713318, | |
| "learning_rate": 0.00013971478906347806, | |
| "loss": 0.6912, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.031204659671552032, | |
| "grad_norm": 0.5672146081924438, | |
| "learning_rate": 0.00013942627243429512, | |
| "loss": 0.9029, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.03128876926635406, | |
| "grad_norm": 0.6760275959968567, | |
| "learning_rate": 0.00013913736668372026, | |
| "loss": 0.6691, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.031372878861156084, | |
| "grad_norm": 0.7026247978210449, | |
| "learning_rate": 0.00013884807466313663, | |
| "loss": 0.6567, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.03145698845595812, | |
| "grad_norm": 0.5599506497383118, | |
| "learning_rate": 0.00013855839922773968, | |
| "loss": 0.7411, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.03154109805076014, | |
| "grad_norm": 0.370534747838974, | |
| "learning_rate": 0.000138268343236509, | |
| "loss": 0.5276, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.03162520764556217, | |
| "grad_norm": 0.5061807036399841, | |
| "learning_rate": 0.00013797790955218014, | |
| "loss": 0.6831, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.031709317240364195, | |
| "grad_norm": 0.5783600807189941, | |
| "learning_rate": 0.00013768710104121627, | |
| "loss": 0.5448, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.03179342683516622, | |
| "grad_norm": 0.5063284039497375, | |
| "learning_rate": 0.00013739592057378003, | |
| "loss": 0.6389, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.031877536429968246, | |
| "grad_norm": 0.5093449950218201, | |
| "learning_rate": 0.0001371043710237051, | |
| "loss": 0.7035, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.03196164602477027, | |
| "grad_norm": 0.5934507250785828, | |
| "learning_rate": 0.00013681245526846783, | |
| "loss": 0.7793, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.032045755619572305, | |
| "grad_norm": 0.5645430088043213, | |
| "learning_rate": 0.0001365201761891588, | |
| "loss": 0.9189, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.03212986521437433, | |
| "grad_norm": 0.5648247599601746, | |
| "learning_rate": 0.00013622753667045457, | |
| "loss": 0.7319, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.03221397480917636, | |
| "grad_norm": 0.48350241780281067, | |
| "learning_rate": 0.00013593453960058908, | |
| "loss": 0.65, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.03229808440397838, | |
| "grad_norm": 0.4776511788368225, | |
| "learning_rate": 0.00013564118787132506, | |
| "loss": 0.5555, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.03238219399878041, | |
| "grad_norm": 0.339843213558197, | |
| "learning_rate": 0.00013534748437792573, | |
| "loss": 0.5663, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.032466303593582435, | |
| "grad_norm": 0.522289514541626, | |
| "learning_rate": 0.0001350534320191259, | |
| "loss": 0.7173, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.03255041318838447, | |
| "grad_norm": 0.6175903081893921, | |
| "learning_rate": 0.0001347590336971037, | |
| "loss": 0.6715, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.032634522783186494, | |
| "grad_norm": 0.5097800493240356, | |
| "learning_rate": 0.0001344642923174517, | |
| "loss": 0.7981, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.03271863237798852, | |
| "grad_norm": 0.45350587368011475, | |
| "learning_rate": 0.00013416921078914835, | |
| "loss": 0.7346, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.032802741972790546, | |
| "grad_norm": 0.7196130156517029, | |
| "learning_rate": 0.00013387379202452917, | |
| "loss": 0.6926, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.03288685156759257, | |
| "grad_norm": 0.44739460945129395, | |
| "learning_rate": 0.00013357803893925807, | |
| "loss": 0.6764, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.0329709611623946, | |
| "grad_norm": 0.42210137844085693, | |
| "learning_rate": 0.00013328195445229868, | |
| "loss": 0.5693, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.033055070757196624, | |
| "grad_norm": 0.25353190302848816, | |
| "learning_rate": 0.00013298554148588528, | |
| "loss": 0.2683, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.03313918035199866, | |
| "grad_norm": 0.5350023508071899, | |
| "learning_rate": 0.00013268880296549425, | |
| "loss": 0.6929, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.03322328994680068, | |
| "grad_norm": 0.5112295150756836, | |
| "learning_rate": 0.00013239174181981495, | |
| "loss": 0.6803, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.03330739954160271, | |
| "grad_norm": 0.5625605583190918, | |
| "learning_rate": 0.00013209436098072095, | |
| "loss": 0.7018, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.033391509136404735, | |
| "grad_norm": 0.6036564111709595, | |
| "learning_rate": 0.00013179666338324108, | |
| "loss": 0.6632, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.03347561873120676, | |
| "grad_norm": 0.6720706224441528, | |
| "learning_rate": 0.0001314986519655305, | |
| "loss": 0.6026, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.033559728326008786, | |
| "grad_norm": 0.5537442564964294, | |
| "learning_rate": 0.0001312003296688415, | |
| "loss": 0.7609, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.03364383792081082, | |
| "grad_norm": 0.567884087562561, | |
| "learning_rate": 0.00013090169943749476, | |
| "loss": 0.4756, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.033727947515612845, | |
| "grad_norm": 0.47909706830978394, | |
| "learning_rate": 0.0001306027642188501, | |
| "loss": 0.6392, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.03381205711041487, | |
| "grad_norm": 0.48807093501091003, | |
| "learning_rate": 0.00013030352696327742, | |
| "loss": 0.4313, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.0338961667052169, | |
| "grad_norm": 0.5797786712646484, | |
| "learning_rate": 0.00013000399062412763, | |
| "loss": 0.6879, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.03398027630001892, | |
| "grad_norm": 0.5691391229629517, | |
| "learning_rate": 0.0001297041581577035, | |
| "loss": 0.611, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.03406438589482095, | |
| "grad_norm": 0.491915762424469, | |
| "learning_rate": 0.0001294040325232304, | |
| "loss": 0.7652, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.03414849548962298, | |
| "grad_norm": 0.6337294578552246, | |
| "learning_rate": 0.00012910361668282719, | |
| "loss": 0.6599, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.03423260508442501, | |
| "grad_norm": 0.5192781686782837, | |
| "learning_rate": 0.00012880291360147693, | |
| "loss": 0.6136, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.034316714679227034, | |
| "grad_norm": 0.6465744376182556, | |
| "learning_rate": 0.0001285019262469976, | |
| "loss": 0.7522, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.03440082427402906, | |
| "grad_norm": 0.5028295516967773, | |
| "learning_rate": 0.00012820065759001293, | |
| "loss": 0.9119, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.034484933868831086, | |
| "grad_norm": 0.519190788269043, | |
| "learning_rate": 0.00012789911060392294, | |
| "loss": 0.4264, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.03456904346363311, | |
| "grad_norm": 0.6284279227256775, | |
| "learning_rate": 0.0001275972882648746, | |
| "loss": 0.7378, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.03465315305843514, | |
| "grad_norm": 0.679527997970581, | |
| "learning_rate": 0.00012729519355173254, | |
| "loss": 0.7436, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.03473726265323717, | |
| "grad_norm": 0.49145928025245667, | |
| "learning_rate": 0.00012699282944604967, | |
| "loss": 0.8936, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.0348213722480392, | |
| "grad_norm": 0.6400472521781921, | |
| "learning_rate": 0.00012669019893203759, | |
| "loss": 0.9139, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.03490548184284122, | |
| "grad_norm": 0.46638303995132446, | |
| "learning_rate": 0.0001263873049965373, | |
| "loss": 0.5398, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.03498959143764325, | |
| "grad_norm": 0.6132822632789612, | |
| "learning_rate": 0.00012608415062898972, | |
| "loss": 0.7835, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.035073701032445274, | |
| "grad_norm": 0.5918437838554382, | |
| "learning_rate": 0.000125780738821406, | |
| "loss": 0.6589, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.0351578106272473, | |
| "grad_norm": 0.45070680975914, | |
| "learning_rate": 0.00012547707256833823, | |
| "loss": 0.5931, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.03524192022204933, | |
| "grad_norm": 0.5677469372749329, | |
| "learning_rate": 0.00012517315486684972, | |
| "loss": 0.7679, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.03532602981685136, | |
| "grad_norm": 0.4486066997051239, | |
| "learning_rate": 0.0001248689887164855, | |
| "loss": 0.5029, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.035410139411653385, | |
| "grad_norm": 0.5143264532089233, | |
| "learning_rate": 0.00012456457711924266, | |
| "loss": 0.512, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.03549424900645541, | |
| "grad_norm": 0.5197947025299072, | |
| "learning_rate": 0.00012425992307954075, | |
| "loss": 0.6199, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.03557835860125744, | |
| "grad_norm": 0.5069385170936584, | |
| "learning_rate": 0.0001239550296041922, | |
| "loss": 0.5604, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.03566246819605946, | |
| "grad_norm": 0.5928438305854797, | |
| "learning_rate": 0.00012364989970237248, | |
| "loss": 0.8877, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.035746577790861496, | |
| "grad_norm": 0.5464084148406982, | |
| "learning_rate": 0.00012334453638559057, | |
| "loss": 0.7264, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.03583068738566352, | |
| "grad_norm": 0.5650874972343445, | |
| "learning_rate": 0.00012303894266765908, | |
| "loss": 0.6815, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.03591479698046555, | |
| "grad_norm": 0.5731722116470337, | |
| "learning_rate": 0.00012273312156466464, | |
| "loss": 0.9198, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.035998906575267574, | |
| "grad_norm": 0.3634185194969177, | |
| "learning_rate": 0.00012242707609493814, | |
| "loss": 0.6057, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.0360830161700696, | |
| "grad_norm": 0.5701162815093994, | |
| "learning_rate": 0.00012212080927902474, | |
| "loss": 0.6718, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.036167125764871626, | |
| "grad_norm": 0.5000714063644409, | |
| "learning_rate": 0.00012181432413965428, | |
| "loss": 0.7592, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.03625123535967365, | |
| "grad_norm": 0.5954747200012207, | |
| "learning_rate": 0.00012150762370171136, | |
| "loss": 0.7098, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.036335344954475685, | |
| "grad_norm": 0.5186465978622437, | |
| "learning_rate": 0.00012120071099220549, | |
| "loss": 0.5954, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.03641945454927771, | |
| "grad_norm": 0.5555379390716553, | |
| "learning_rate": 0.00012089358904024117, | |
| "loss": 0.7133, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.036503564144079736, | |
| "grad_norm": 0.46339845657348633, | |
| "learning_rate": 0.00012058626087698814, | |
| "loss": 0.7121, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.03658767373888176, | |
| "grad_norm": 0.45588961243629456, | |
| "learning_rate": 0.00012027872953565125, | |
| "loss": 0.705, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.03667178333368379, | |
| "grad_norm": 0.6387124061584473, | |
| "learning_rate": 0.00011997099805144069, | |
| "loss": 0.6375, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.036755892928485814, | |
| "grad_norm": 0.5656935572624207, | |
| "learning_rate": 0.000119663069461542, | |
| "loss": 0.7211, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.03684000252328785, | |
| "grad_norm": 0.4781017303466797, | |
| "learning_rate": 0.00011935494680508606, | |
| "loss": 0.6053, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.03692411211808987, | |
| "grad_norm": 0.5316019654273987, | |
| "learning_rate": 0.00011904663312311901, | |
| "loss": 0.6341, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.0370082217128919, | |
| "grad_norm": 0.535388171672821, | |
| "learning_rate": 0.00011873813145857249, | |
| "loss": 0.7128, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.037092331307693925, | |
| "grad_norm": 0.7049708366394043, | |
| "learning_rate": 0.00011842944485623335, | |
| "loss": 0.673, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.03717644090249595, | |
| "grad_norm": 0.5531265139579773, | |
| "learning_rate": 0.00011812057636271374, | |
| "loss": 0.7491, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.03726055049729798, | |
| "grad_norm": 0.5389276146888733, | |
| "learning_rate": 0.000117811529026421, | |
| "loss": 0.6979, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.0373446600921, | |
| "grad_norm": 0.3878541588783264, | |
| "learning_rate": 0.00011750230589752762, | |
| "loss": 0.4515, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.037428769686902036, | |
| "grad_norm": 0.7032809853553772, | |
| "learning_rate": 0.00011719291002794096, | |
| "loss": 0.6921, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.03751287928170406, | |
| "grad_norm": 0.5832718014717102, | |
| "learning_rate": 0.00011688334447127338, | |
| "loss": 0.508, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.03759698887650609, | |
| "grad_norm": 0.4945680499076843, | |
| "learning_rate": 0.00011657361228281199, | |
| "loss": 0.5809, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.037681098471308114, | |
| "grad_norm": 0.6034420728683472, | |
| "learning_rate": 0.00011626371651948838, | |
| "loss": 0.6125, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.03776520806611014, | |
| "grad_norm": 0.6533922553062439, | |
| "learning_rate": 0.00011595366023984864, | |
| "loss": 0.5984, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.037849317660912166, | |
| "grad_norm": 0.46035146713256836, | |
| "learning_rate": 0.0001156434465040231, | |
| "loss": 0.6094, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.0379334272557142, | |
| "grad_norm": 0.6323526501655579, | |
| "learning_rate": 0.00011533307837369607, | |
| "loss": 0.9116, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.038017536850516225, | |
| "grad_norm": 0.4355090260505676, | |
| "learning_rate": 0.00011502255891207572, | |
| "loss": 0.4778, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.03810164644531825, | |
| "grad_norm": 0.5190046429634094, | |
| "learning_rate": 0.00011471189118386375, | |
| "loss": 0.5997, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.038185756040120276, | |
| "grad_norm": 0.534466564655304, | |
| "learning_rate": 0.00011440107825522521, | |
| "loss": 0.7846, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.0382698656349223, | |
| "grad_norm": 0.6099439859390259, | |
| "learning_rate": 0.00011409012319375827, | |
| "loss": 0.5393, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.03835397522972433, | |
| "grad_norm": 0.4053705036640167, | |
| "learning_rate": 0.0001137790290684638, | |
| "loss": 0.547, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.03843808482452636, | |
| "grad_norm": 0.3879355788230896, | |
| "learning_rate": 0.00011346779894971527, | |
| "loss": 0.5772, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.03852219441932839, | |
| "grad_norm": 0.7428577542304993, | |
| "learning_rate": 0.00011315643590922827, | |
| "loss": 0.9059, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.03860630401413041, | |
| "grad_norm": 0.41162705421447754, | |
| "learning_rate": 0.0001128449430200303, | |
| "loss": 0.6041, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.03869041360893244, | |
| "grad_norm": 0.5828412175178528, | |
| "learning_rate": 0.00011253332335643043, | |
| "loss": 0.4467, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.038774523203734465, | |
| "grad_norm": 0.5897490978240967, | |
| "learning_rate": 0.00011222157999398895, | |
| "loss": 0.8586, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.03885863279853649, | |
| "grad_norm": 0.4602713882923126, | |
| "learning_rate": 0.00011190971600948699, | |
| "loss": 0.6802, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.03894274239333852, | |
| "grad_norm": 0.5070551633834839, | |
| "learning_rate": 0.00011159773448089614, | |
| "loss": 0.5585, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.03902685198814055, | |
| "grad_norm": 0.4254646897315979, | |
| "learning_rate": 0.00011128563848734816, | |
| "loss": 0.7244, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.039110961582942576, | |
| "grad_norm": 0.5656306147575378, | |
| "learning_rate": 0.00011097343110910452, | |
| "loss": 0.8105, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.0391950711777446, | |
| "grad_norm": 0.5550054311752319, | |
| "learning_rate": 0.000110661115427526, | |
| "loss": 0.4696, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.03927918077254663, | |
| "grad_norm": 0.47665438055992126, | |
| "learning_rate": 0.00011034869452504226, | |
| "loss": 0.4867, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.039363290367348654, | |
| "grad_norm": 0.6439712047576904, | |
| "learning_rate": 0.00011003617148512149, | |
| "loss": 0.6037, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.03944739996215068, | |
| "grad_norm": 0.49243077635765076, | |
| "learning_rate": 0.00010972354939223996, | |
| "loss": 0.646, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.03953150955695271, | |
| "grad_norm": 0.4704420268535614, | |
| "learning_rate": 0.00010941083133185146, | |
| "loss": 0.7801, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.03961561915175474, | |
| "grad_norm": 0.4875670075416565, | |
| "learning_rate": 0.00010909802039035701, | |
| "loss": 0.5314, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.039699728746556764, | |
| "grad_norm": 0.5254621505737305, | |
| "learning_rate": 0.00010878511965507434, | |
| "loss": 0.659, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.03978383834135879, | |
| "grad_norm": 0.5760181546211243, | |
| "learning_rate": 0.00010847213221420736, | |
| "loss": 0.9349, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.039867947936160816, | |
| "grad_norm": 0.4383930563926697, | |
| "learning_rate": 0.00010815906115681578, | |
| "loss": 0.8152, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.03995205753096284, | |
| "grad_norm": 0.5969526171684265, | |
| "learning_rate": 0.0001078459095727845, | |
| "loss": 0.7431, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.04003616712576487, | |
| "grad_norm": 0.5625801086425781, | |
| "learning_rate": 0.00010753268055279329, | |
| "loss": 0.7582, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.0401202767205669, | |
| "grad_norm": 0.5394108295440674, | |
| "learning_rate": 0.0001072193771882861, | |
| "loss": 0.6964, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.04020438631536893, | |
| "grad_norm": 0.8828211426734924, | |
| "learning_rate": 0.00010690600257144061, | |
| "loss": 0.6602, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.04028849591017095, | |
| "grad_norm": 0.6621891856193542, | |
| "learning_rate": 0.0001065925597951378, | |
| "loss": 0.6224, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.04037260550497298, | |
| "grad_norm": 0.5769515633583069, | |
| "learning_rate": 0.00010627905195293135, | |
| "loss": 0.7293, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.040456715099775005, | |
| "grad_norm": 0.47239354252815247, | |
| "learning_rate": 0.00010596548213901708, | |
| "loss": 0.6623, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.04054082469457703, | |
| "grad_norm": 0.5481030344963074, | |
| "learning_rate": 0.00010565185344820247, | |
| "loss": 0.8693, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.040624934289379064, | |
| "grad_norm": 0.5839731693267822, | |
| "learning_rate": 0.00010533816897587606, | |
| "loss": 0.7281, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.04070904388418109, | |
| "grad_norm": 0.5088500380516052, | |
| "learning_rate": 0.00010502443181797697, | |
| "loss": 0.6444, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.040793153478983116, | |
| "grad_norm": 0.514184832572937, | |
| "learning_rate": 0.00010471064507096426, | |
| "loss": 0.678, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.04087726307378514, | |
| "grad_norm": 0.48522424697875977, | |
| "learning_rate": 0.0001043968118317865, | |
| "loss": 0.6344, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.04096137266858717, | |
| "grad_norm": 0.5985432863235474, | |
| "learning_rate": 0.00010408293519785101, | |
| "loss": 0.6916, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.041045482263389194, | |
| "grad_norm": 0.5493951439857483, | |
| "learning_rate": 0.00010376901826699348, | |
| "loss": 0.6433, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.041129591858191226, | |
| "grad_norm": 0.5034871101379395, | |
| "learning_rate": 0.00010345506413744726, | |
| "loss": 0.6704, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.04121370145299325, | |
| "grad_norm": 0.5221031904220581, | |
| "learning_rate": 0.00010314107590781284, | |
| "loss": 0.4167, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.04129781104779528, | |
| "grad_norm": 0.5250301957130432, | |
| "learning_rate": 0.00010282705667702734, | |
| "loss": 0.5978, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.041381920642597304, | |
| "grad_norm": 0.5023308396339417, | |
| "learning_rate": 0.00010251300954433376, | |
| "loss": 0.448, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.04146603023739933, | |
| "grad_norm": 0.6943901181221008, | |
| "learning_rate": 0.00010219893760925052, | |
| "loss": 0.7664, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.041550139832201356, | |
| "grad_norm": 0.4787749946117401, | |
| "learning_rate": 0.00010188484397154084, | |
| "loss": 0.8114, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.04163424942700338, | |
| "grad_norm": 0.5617536902427673, | |
| "learning_rate": 0.00010157073173118208, | |
| "loss": 0.6003, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.041718359021805415, | |
| "grad_norm": 0.3669779300689697, | |
| "learning_rate": 0.00010125660398833528, | |
| "loss": 0.5983, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.04180246861660744, | |
| "grad_norm": 0.46090903878211975, | |
| "learning_rate": 0.00010094246384331442, | |
| "loss": 0.7085, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.04188657821140947, | |
| "grad_norm": 0.4907785952091217, | |
| "learning_rate": 0.00010062831439655591, | |
| "loss": 0.6197, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.04197068780621149, | |
| "grad_norm": 0.4900142252445221, | |
| "learning_rate": 0.00010031415874858797, | |
| "loss": 0.53, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.04205479740101352, | |
| "grad_norm": 0.5708168745040894, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7168, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.042138906995815545, | |
| "grad_norm": 0.5526877045631409, | |
| "learning_rate": 9.968584125141204e-05, | |
| "loss": 0.7179, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.04222301659061758, | |
| "grad_norm": 0.4733210802078247, | |
| "learning_rate": 9.937168560344412e-05, | |
| "loss": 0.4766, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.042307126185419604, | |
| "grad_norm": 0.5687675476074219, | |
| "learning_rate": 9.90575361566856e-05, | |
| "loss": 0.7382, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.04239123578022163, | |
| "grad_norm": 0.5211085677146912, | |
| "learning_rate": 9.874339601166473e-05, | |
| "loss": 0.7914, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.042475345375023656, | |
| "grad_norm": 0.6340366005897522, | |
| "learning_rate": 9.842926826881796e-05, | |
| "loss": 0.7195, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.04255945496982568, | |
| "grad_norm": 0.5366642475128174, | |
| "learning_rate": 9.81151560284592e-05, | |
| "loss": 0.6782, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.04264356456462771, | |
| "grad_norm": 0.6311143040657043, | |
| "learning_rate": 9.78010623907495e-05, | |
| "loss": 0.7104, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.042727674159429734, | |
| "grad_norm": 0.5550001859664917, | |
| "learning_rate": 9.748699045566626e-05, | |
| "loss": 0.834, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.042811783754231766, | |
| "grad_norm": 0.601676881313324, | |
| "learning_rate": 9.717294332297268e-05, | |
| "loss": 0.7551, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.04289589334903379, | |
| "grad_norm": 0.4123936891555786, | |
| "learning_rate": 9.685892409218717e-05, | |
| "loss": 0.7682, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.04298000294383582, | |
| "grad_norm": 0.7674761414527893, | |
| "learning_rate": 9.654493586255278e-05, | |
| "loss": 0.6967, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.043064112538637844, | |
| "grad_norm": 0.6111290454864502, | |
| "learning_rate": 9.623098173300654e-05, | |
| "loss": 0.5819, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.04314822213343987, | |
| "grad_norm": 0.43844571709632874, | |
| "learning_rate": 9.591706480214901e-05, | |
| "loss": 0.6972, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.043232331728241896, | |
| "grad_norm": 0.538962185382843, | |
| "learning_rate": 9.560318816821353e-05, | |
| "loss": 0.7395, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.04331644132304393, | |
| "grad_norm": 0.5768121480941772, | |
| "learning_rate": 9.528935492903575e-05, | |
| "loss": 0.8873, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.043400550917845955, | |
| "grad_norm": 0.492720365524292, | |
| "learning_rate": 9.497556818202306e-05, | |
| "loss": 0.5948, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.04348466051264798, | |
| "grad_norm": 0.516925036907196, | |
| "learning_rate": 9.466183102412395e-05, | |
| "loss": 0.6319, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.04356877010745001, | |
| "grad_norm": 0.6040381193161011, | |
| "learning_rate": 9.434814655179755e-05, | |
| "loss": 0.6935, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.04365287970225203, | |
| "grad_norm": 0.48684102296829224, | |
| "learning_rate": 9.403451786098294e-05, | |
| "loss": 0.5688, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.04373698929705406, | |
| "grad_norm": 0.5389447212219238, | |
| "learning_rate": 9.372094804706867e-05, | |
| "loss": 0.8248, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.04382109889185609, | |
| "grad_norm": 0.4723369777202606, | |
| "learning_rate": 9.340744020486222e-05, | |
| "loss": 0.5912, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.04390520848665812, | |
| "grad_norm": 0.5267740488052368, | |
| "learning_rate": 9.309399742855942e-05, | |
| "loss": 0.5262, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.043989318081460144, | |
| "grad_norm": 0.52064049243927, | |
| "learning_rate": 9.278062281171393e-05, | |
| "loss": 0.6672, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.04407342767626217, | |
| "grad_norm": 0.46649792790412903, | |
| "learning_rate": 9.246731944720675e-05, | |
| "loss": 0.7222, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.044157537271064196, | |
| "grad_norm": 0.5226630568504333, | |
| "learning_rate": 9.215409042721552e-05, | |
| "loss": 0.6677, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.04424164686586622, | |
| "grad_norm": 0.5565252304077148, | |
| "learning_rate": 9.184093884318425e-05, | |
| "loss": 0.5528, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.04432575646066825, | |
| "grad_norm": 0.5086897611618042, | |
| "learning_rate": 9.152786778579267e-05, | |
| "loss": 0.6953, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.04440986605547028, | |
| "grad_norm": 0.5770063400268555, | |
| "learning_rate": 9.121488034492569e-05, | |
| "loss": 0.7933, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.044493975650272306, | |
| "grad_norm": 0.6236185431480408, | |
| "learning_rate": 9.090197960964301e-05, | |
| "loss": 0.7398, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.04457808524507433, | |
| "grad_norm": 0.605636477470398, | |
| "learning_rate": 9.058916866814858e-05, | |
| "loss": 0.7508, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.04466219483987636, | |
| "grad_norm": 0.5256229043006897, | |
| "learning_rate": 9.027645060776006e-05, | |
| "loss": 0.7415, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.044746304434678384, | |
| "grad_norm": 0.4349639117717743, | |
| "learning_rate": 8.99638285148785e-05, | |
| "loss": 0.7829, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.04483041402948041, | |
| "grad_norm": 0.5248279571533203, | |
| "learning_rate": 8.965130547495776e-05, | |
| "loss": 0.6153, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.04491452362428244, | |
| "grad_norm": 0.47936955094337463, | |
| "learning_rate": 8.933888457247402e-05, | |
| "loss": 0.5984, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.04499863321908447, | |
| "grad_norm": 0.4791763722896576, | |
| "learning_rate": 8.902656889089548e-05, | |
| "loss": 0.6953, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.045082742813886495, | |
| "grad_norm": 0.47786253690719604, | |
| "learning_rate": 8.871436151265184e-05, | |
| "loss": 0.6712, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.04516685240868852, | |
| "grad_norm": 0.580772340297699, | |
| "learning_rate": 8.840226551910387e-05, | |
| "loss": 0.6813, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.04525096200349055, | |
| "grad_norm": 0.6560893058776855, | |
| "learning_rate": 8.809028399051302e-05, | |
| "loss": 0.5794, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.04533507159829257, | |
| "grad_norm": 0.479247510433197, | |
| "learning_rate": 8.777842000601105e-05, | |
| "loss": 0.6182, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.045419181193094606, | |
| "grad_norm": 0.5315003395080566, | |
| "learning_rate": 8.746667664356956e-05, | |
| "loss": 0.6151, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.04550329078789663, | |
| "grad_norm": 0.4729808568954468, | |
| "learning_rate": 8.715505697996971e-05, | |
| "loss": 0.6079, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.04558740038269866, | |
| "grad_norm": 0.5379468202590942, | |
| "learning_rate": 8.684356409077176e-05, | |
| "loss": 0.6553, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.045671509977500684, | |
| "grad_norm": 0.6452252864837646, | |
| "learning_rate": 8.653220105028474e-05, | |
| "loss": 0.7077, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.04575561957230271, | |
| "grad_norm": 0.5322025418281555, | |
| "learning_rate": 8.62209709315362e-05, | |
| "loss": 0.5957, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.045839729167104735, | |
| "grad_norm": 0.6063228249549866, | |
| "learning_rate": 8.590987680624174e-05, | |
| "loss": 0.8442, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.04592383876190676, | |
| "grad_norm": 0.6377241015434265, | |
| "learning_rate": 8.559892174477479e-05, | |
| "loss": 0.6145, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.046007948356708794, | |
| "grad_norm": 0.49134746193885803, | |
| "learning_rate": 8.528810881613626e-05, | |
| "loss": 0.7311, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.04609205795151082, | |
| "grad_norm": 0.7310553193092346, | |
| "learning_rate": 8.497744108792429e-05, | |
| "loss": 0.7093, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.046176167546312846, | |
| "grad_norm": 0.44871386885643005, | |
| "learning_rate": 8.466692162630392e-05, | |
| "loss": 0.488, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.04626027714111487, | |
| "grad_norm": 0.604282021522522, | |
| "learning_rate": 8.435655349597689e-05, | |
| "loss": 0.5964, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.0463443867359169, | |
| "grad_norm": 0.46552160382270813, | |
| "learning_rate": 8.404633976015134e-05, | |
| "loss": 0.7442, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.046428496330718924, | |
| "grad_norm": 0.5520333051681519, | |
| "learning_rate": 8.373628348051165e-05, | |
| "loss": 0.5222, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.04651260592552096, | |
| "grad_norm": 0.5013942718505859, | |
| "learning_rate": 8.342638771718802e-05, | |
| "loss": 0.6123, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.04659671552032298, | |
| "grad_norm": 0.5522478222846985, | |
| "learning_rate": 8.311665552872662e-05, | |
| "loss": 0.6076, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.04668082511512501, | |
| "grad_norm": 0.43584609031677246, | |
| "learning_rate": 8.280708997205904e-05, | |
| "loss": 0.5835, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.046764934709927035, | |
| "grad_norm": 0.6730960011482239, | |
| "learning_rate": 8.249769410247239e-05, | |
| "loss": 0.5367, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.04684904430472906, | |
| "grad_norm": 0.4698852002620697, | |
| "learning_rate": 8.218847097357898e-05, | |
| "loss": 0.7015, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.04693315389953109, | |
| "grad_norm": 0.6545222401618958, | |
| "learning_rate": 8.187942363728625e-05, | |
| "loss": 0.7149, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.04701726349433311, | |
| "grad_norm": 0.5913302898406982, | |
| "learning_rate": 8.157055514376666e-05, | |
| "loss": 0.634, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.047101373089135146, | |
| "grad_norm": 0.6566689610481262, | |
| "learning_rate": 8.126186854142752e-05, | |
| "loss": 0.6646, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.04718548268393717, | |
| "grad_norm": 0.5921803116798401, | |
| "learning_rate": 8.095336687688102e-05, | |
| "loss": 0.5967, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.0472695922787392, | |
| "grad_norm": 0.5838126540184021, | |
| "learning_rate": 8.064505319491398e-05, | |
| "loss": 0.6807, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.047353701873541224, | |
| "grad_norm": 0.5889870524406433, | |
| "learning_rate": 8.033693053845801e-05, | |
| "loss": 0.6156, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.04743781146834325, | |
| "grad_norm": 0.5544804334640503, | |
| "learning_rate": 8.002900194855932e-05, | |
| "loss": 0.7279, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.047521921063145275, | |
| "grad_norm": 0.5051852464675903, | |
| "learning_rate": 7.972127046434878e-05, | |
| "loss": 0.6375, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.04760603065794731, | |
| "grad_norm": 0.6946410536766052, | |
| "learning_rate": 7.941373912301189e-05, | |
| "loss": 0.8067, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.047690140252749334, | |
| "grad_norm": 0.5952439904212952, | |
| "learning_rate": 7.910641095975886e-05, | |
| "loss": 0.8346, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.04777424984755136, | |
| "grad_norm": 0.5868372321128845, | |
| "learning_rate": 7.879928900779456e-05, | |
| "loss": 0.8255, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.047858359442353386, | |
| "grad_norm": 0.6287057399749756, | |
| "learning_rate": 7.849237629828869e-05, | |
| "loss": 0.5474, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.04794246903715541, | |
| "grad_norm": 0.5471776127815247, | |
| "learning_rate": 7.818567586034577e-05, | |
| "loss": 0.5467, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.04802657863195744, | |
| "grad_norm": 0.8107778429985046, | |
| "learning_rate": 7.787919072097531e-05, | |
| "loss": 0.6715, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.04811068822675947, | |
| "grad_norm": 0.6035006046295166, | |
| "learning_rate": 7.75729239050619e-05, | |
| "loss": 0.6016, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.0481947978215615, | |
| "grad_norm": 0.6550392508506775, | |
| "learning_rate": 7.726687843533538e-05, | |
| "loss": 0.5897, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.04827890741636352, | |
| "grad_norm": 0.5324556827545166, | |
| "learning_rate": 7.696105733234098e-05, | |
| "loss": 0.5424, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.04836301701116555, | |
| "grad_norm": 0.5881845355033875, | |
| "learning_rate": 7.66554636144095e-05, | |
| "loss": 0.6065, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.048447126605967575, | |
| "grad_norm": 0.537804901599884, | |
| "learning_rate": 7.635010029762756e-05, | |
| "loss": 0.5698, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.0485312362007696, | |
| "grad_norm": 0.5901293754577637, | |
| "learning_rate": 7.604497039580785e-05, | |
| "loss": 0.6954, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.04861534579557163, | |
| "grad_norm": 0.5681110620498657, | |
| "learning_rate": 7.574007692045928e-05, | |
| "loss": 0.6799, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.04869945539037366, | |
| "grad_norm": 0.5399741530418396, | |
| "learning_rate": 7.543542288075739e-05, | |
| "loss": 0.6881, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.048783564985175686, | |
| "grad_norm": 0.605559229850769, | |
| "learning_rate": 7.513101128351454e-05, | |
| "loss": 0.608, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.04886767457997771, | |
| "grad_norm": 0.5992718935012817, | |
| "learning_rate": 7.48268451331503e-05, | |
| "loss": 0.6869, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.04895178417477974, | |
| "grad_norm": 0.8363056182861328, | |
| "learning_rate": 7.45229274316618e-05, | |
| "loss": 0.7678, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.04903589376958176, | |
| "grad_norm": 0.421335905790329, | |
| "learning_rate": 7.421926117859403e-05, | |
| "loss": 0.508, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.04912000336438379, | |
| "grad_norm": 0.4553971588611603, | |
| "learning_rate": 7.391584937101033e-05, | |
| "loss": 0.6228, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.04920411295918582, | |
| "grad_norm": 0.5710154175758362, | |
| "learning_rate": 7.361269500346274e-05, | |
| "loss": 0.6005, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.04928822255398785, | |
| "grad_norm": 0.5714388489723206, | |
| "learning_rate": 7.330980106796246e-05, | |
| "loss": 0.7007, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.049372332148789874, | |
| "grad_norm": 0.5537533760070801, | |
| "learning_rate": 7.300717055395039e-05, | |
| "loss": 0.6398, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.0494564417435919, | |
| "grad_norm": 0.43929553031921387, | |
| "learning_rate": 7.270480644826749e-05, | |
| "loss": 0.6056, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.049540551338393926, | |
| "grad_norm": 0.485586017370224, | |
| "learning_rate": 7.240271173512546e-05, | |
| "loss": 0.5453, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.04962466093319595, | |
| "grad_norm": 0.43666794896125793, | |
| "learning_rate": 7.210088939607708e-05, | |
| "loss": 0.5794, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.04970877052799798, | |
| "grad_norm": 0.5430508852005005, | |
| "learning_rate": 7.179934240998706e-05, | |
| "loss": 0.8037, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.04979288012280001, | |
| "grad_norm": 0.6481785774230957, | |
| "learning_rate": 7.149807375300239e-05, | |
| "loss": 0.7141, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.04987698971760204, | |
| "grad_norm": 0.5810648202896118, | |
| "learning_rate": 7.119708639852312e-05, | |
| "loss": 0.5569, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.04996109931240406, | |
| "grad_norm": 0.6031057238578796, | |
| "learning_rate": 7.089638331717284e-05, | |
| "loss": 0.6682, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.05004520890720609, | |
| "grad_norm": 0.48759695887565613, | |
| "learning_rate": 7.059596747676962e-05, | |
| "loss": 0.634, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.050129318502008115, | |
| "grad_norm": 0.4522048532962799, | |
| "learning_rate": 7.029584184229653e-05, | |
| "loss": 0.6403, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.05021342809681014, | |
| "grad_norm": 0.4637600779533386, | |
| "learning_rate": 6.999600937587239e-05, | |
| "loss": 0.798, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.050297537691612174, | |
| "grad_norm": 0.5411447882652283, | |
| "learning_rate": 6.969647303672262e-05, | |
| "loss": 0.6316, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.0503816472864142, | |
| "grad_norm": 0.48814845085144043, | |
| "learning_rate": 6.939723578114993e-05, | |
| "loss": 0.775, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.050465756881216225, | |
| "grad_norm": 0.6729063391685486, | |
| "learning_rate": 6.909830056250527e-05, | |
| "loss": 0.6383, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.05054986647601825, | |
| "grad_norm": 0.5428520441055298, | |
| "learning_rate": 6.879967033115853e-05, | |
| "loss": 0.5884, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.05063397607082028, | |
| "grad_norm": 0.5439213514328003, | |
| "learning_rate": 6.850134803446954e-05, | |
| "loss": 0.6811, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.0507180856656223, | |
| "grad_norm": 0.5737605690956116, | |
| "learning_rate": 6.820333661675893e-05, | |
| "loss": 0.559, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.050802195260424336, | |
| "grad_norm": 0.610083281993866, | |
| "learning_rate": 6.790563901927907e-05, | |
| "loss": 0.6291, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.05088630485522636, | |
| "grad_norm": 0.6092396378517151, | |
| "learning_rate": 6.760825818018508e-05, | |
| "loss": 0.8043, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.05097041445002839, | |
| "grad_norm": 0.43380871415138245, | |
| "learning_rate": 6.731119703450577e-05, | |
| "loss": 0.8899, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.051054524044830414, | |
| "grad_norm": 0.6090589761734009, | |
| "learning_rate": 6.701445851411472e-05, | |
| "loss": 0.7548, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.05113863363963244, | |
| "grad_norm": 0.5451867580413818, | |
| "learning_rate": 6.671804554770135e-05, | |
| "loss": 0.5946, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.051222743234434466, | |
| "grad_norm": 0.6997014880180359, | |
| "learning_rate": 6.642196106074194e-05, | |
| "loss": 0.6974, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.05130685282923649, | |
| "grad_norm": 0.5238434672355652, | |
| "learning_rate": 6.612620797547087e-05, | |
| "loss": 0.6132, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.051390962424038525, | |
| "grad_norm": 0.5040790438652039, | |
| "learning_rate": 6.583078921085167e-05, | |
| "loss": 0.7068, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.05147507201884055, | |
| "grad_norm": 0.5203388333320618, | |
| "learning_rate": 6.55357076825483e-05, | |
| "loss": 0.7418, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.05155918161364258, | |
| "grad_norm": 0.5750290751457214, | |
| "learning_rate": 6.52409663028963e-05, | |
| "loss": 0.5788, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.0516432912084446, | |
| "grad_norm": 0.4557915925979614, | |
| "learning_rate": 6.494656798087412e-05, | |
| "loss": 0.6357, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.05172740080324663, | |
| "grad_norm": 0.48326605558395386, | |
| "learning_rate": 6.465251562207431e-05, | |
| "loss": 0.6878, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.051811510398048655, | |
| "grad_norm": 0.5541282296180725, | |
| "learning_rate": 6.435881212867493e-05, | |
| "loss": 0.5612, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.05189561999285069, | |
| "grad_norm": 0.5726374983787537, | |
| "learning_rate": 6.406546039941094e-05, | |
| "loss": 0.5127, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.051979729587652714, | |
| "grad_norm": 0.4574257731437683, | |
| "learning_rate": 6.377246332954544e-05, | |
| "loss": 0.6469, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.05206383918245474, | |
| "grad_norm": 0.5137540698051453, | |
| "learning_rate": 6.347982381084123e-05, | |
| "loss": 0.6729, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.052147948777256765, | |
| "grad_norm": 0.5800307393074036, | |
| "learning_rate": 6.318754473153221e-05, | |
| "loss": 0.7838, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.05223205837205879, | |
| "grad_norm": 0.45078638195991516, | |
| "learning_rate": 6.289562897629492e-05, | |
| "loss": 0.4851, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.05231616796686082, | |
| "grad_norm": 0.45273396372795105, | |
| "learning_rate": 6.260407942621998e-05, | |
| "loss": 0.4557, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.05240027756166284, | |
| "grad_norm": 0.46480339765548706, | |
| "learning_rate": 6.231289895878375e-05, | |
| "loss": 0.5967, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.052484387156464876, | |
| "grad_norm": 0.5039330124855042, | |
| "learning_rate": 6.20220904478199e-05, | |
| "loss": 0.6384, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.0525684967512669, | |
| "grad_norm": 0.637080192565918, | |
| "learning_rate": 6.173165676349103e-05, | |
| "loss": 0.8325, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.05265260634606893, | |
| "grad_norm": 0.5003658533096313, | |
| "learning_rate": 6.144160077226036e-05, | |
| "loss": 0.5859, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.052736715940870954, | |
| "grad_norm": 0.6767755150794983, | |
| "learning_rate": 6.11519253368634e-05, | |
| "loss": 0.6984, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.05282082553567298, | |
| "grad_norm": 0.5123332142829895, | |
| "learning_rate": 6.086263331627976e-05, | |
| "loss": 0.5831, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.052904935130475006, | |
| "grad_norm": 0.6226670742034912, | |
| "learning_rate": 6.05737275657049e-05, | |
| "loss": 0.7954, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.05298904472527704, | |
| "grad_norm": 0.584923267364502, | |
| "learning_rate": 6.0285210936521955e-05, | |
| "loss": 0.7863, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.053073154320079065, | |
| "grad_norm": 0.5787730813026428, | |
| "learning_rate": 5.999708627627354e-05, | |
| "loss": 0.7289, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.05315726391488109, | |
| "grad_norm": 0.5147082805633545, | |
| "learning_rate": 5.9709356428633746e-05, | |
| "loss": 0.5297, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.05324137350968312, | |
| "grad_norm": 0.6487715840339661, | |
| "learning_rate": 5.9422024233380013e-05, | |
| "loss": 0.6826, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.05332548310448514, | |
| "grad_norm": 0.5044452548027039, | |
| "learning_rate": 5.913509252636511e-05, | |
| "loss": 0.5771, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.05340959269928717, | |
| "grad_norm": 0.6998575329780579, | |
| "learning_rate": 5.884856413948913e-05, | |
| "loss": 0.7294, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.0534937022940892, | |
| "grad_norm": 0.4688400626182556, | |
| "learning_rate": 5.856244190067159e-05, | |
| "loss": 0.6579, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.05357781188889123, | |
| "grad_norm": 0.4600168466567993, | |
| "learning_rate": 5.82767286338235e-05, | |
| "loss": 0.5962, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.05366192148369325, | |
| "grad_norm": 0.6258283257484436, | |
| "learning_rate": 5.799142715881938e-05, | |
| "loss": 0.6653, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.05374603107849528, | |
| "grad_norm": 0.5476671457290649, | |
| "learning_rate": 5.770654029146969e-05, | |
| "loss": 0.6163, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.053830140673297305, | |
| "grad_norm": 0.4993649125099182, | |
| "learning_rate": 5.7422070843492734e-05, | |
| "loss": 0.6052, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.05391425026809933, | |
| "grad_norm": 0.5785011053085327, | |
| "learning_rate": 5.713802162248718e-05, | |
| "loss": 0.6716, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.05399835986290136, | |
| "grad_norm": 0.6870988607406616, | |
| "learning_rate": 5.6854395431904094e-05, | |
| "loss": 0.6964, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.05408246945770339, | |
| "grad_norm": 0.40551093220710754, | |
| "learning_rate": 5.657119507101954e-05, | |
| "loss": 0.6014, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.054166579052505416, | |
| "grad_norm": 0.4204919934272766, | |
| "learning_rate": 5.6288423334906735e-05, | |
| "loss": 0.6235, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.05425068864730744, | |
| "grad_norm": 0.6323750019073486, | |
| "learning_rate": 5.6006083014408484e-05, | |
| "loss": 0.7131, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.05433479824210947, | |
| "grad_norm": 0.4721733033657074, | |
| "learning_rate": 5.572417689610987e-05, | |
| "loss": 0.5972, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.054418907836911494, | |
| "grad_norm": 0.5242834687232971, | |
| "learning_rate": 5.544270776231038e-05, | |
| "loss": 0.7438, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.05450301743171352, | |
| "grad_norm": 0.5273514986038208, | |
| "learning_rate": 5.5161678390996796e-05, | |
| "loss": 0.753, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.05458712702651555, | |
| "grad_norm": 0.5042054653167725, | |
| "learning_rate": 5.488109155581549e-05, | |
| "loss": 0.5714, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.05467123662131758, | |
| "grad_norm": 0.6366623640060425, | |
| "learning_rate": 5.4600950026045326e-05, | |
| "loss": 0.6925, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.054755346216119605, | |
| "grad_norm": 0.4337083101272583, | |
| "learning_rate": 5.4321256566570036e-05, | |
| "loss": 0.49, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.05483945581092163, | |
| "grad_norm": 0.5946835875511169, | |
| "learning_rate": 5.404201393785122e-05, | |
| "loss": 0.6322, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.05492356540572366, | |
| "grad_norm": 0.37681615352630615, | |
| "learning_rate": 5.3763224895900846e-05, | |
| "loss": 0.6986, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.05500767500052568, | |
| "grad_norm": 0.4949415326118469, | |
| "learning_rate": 5.348489219225416e-05, | |
| "loss": 0.5161, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.05509178459532771, | |
| "grad_norm": 0.5567456483840942, | |
| "learning_rate": 5.320701857394268e-05, | |
| "loss": 0.6031, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.05517589419012974, | |
| "grad_norm": 0.6035876870155334, | |
| "learning_rate": 5.292960678346675e-05, | |
| "loss": 0.6714, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.05526000378493177, | |
| "grad_norm": 0.830230712890625, | |
| "learning_rate": 5.265265955876879e-05, | |
| "loss": 0.6222, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.05534411337973379, | |
| "grad_norm": 0.5250204801559448, | |
| "learning_rate": 5.237617963320608e-05, | |
| "loss": 0.7947, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.05542822297453582, | |
| "grad_norm": 0.5689926147460938, | |
| "learning_rate": 5.210016973552391e-05, | |
| "loss": 0.752, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.055512332569337845, | |
| "grad_norm": 0.5850744843482971, | |
| "learning_rate": 5.182463258982846e-05, | |
| "loss": 0.758, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.05559644216413987, | |
| "grad_norm": 0.8335689306259155, | |
| "learning_rate": 5.1549570915560206e-05, | |
| "loss": 0.8418, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.055680551758941904, | |
| "grad_norm": 0.5002387762069702, | |
| "learning_rate": 5.127498742746675e-05, | |
| "loss": 0.5669, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.05576466135374393, | |
| "grad_norm": 0.5952647924423218, | |
| "learning_rate": 5.100088483557634e-05, | |
| "loss": 0.6905, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.055848770948545956, | |
| "grad_norm": 0.6531146764755249, | |
| "learning_rate": 5.072726584517086e-05, | |
| "loss": 0.527, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.05593288054334798, | |
| "grad_norm": 0.48260536789894104, | |
| "learning_rate": 5.045413315675924e-05, | |
| "loss": 0.6368, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.05601699013815001, | |
| "grad_norm": 0.5707049369812012, | |
| "learning_rate": 5.018148946605092e-05, | |
| "loss": 0.676, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.056101099732952034, | |
| "grad_norm": 0.5223572850227356, | |
| "learning_rate": 4.990933746392899e-05, | |
| "loss": 0.6228, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.05618520932775407, | |
| "grad_norm": 0.6292992234230042, | |
| "learning_rate": 4.9637679836423924e-05, | |
| "loss": 0.5522, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.05626931892255609, | |
| "grad_norm": 0.6064843535423279, | |
| "learning_rate": 4.9366519264686725e-05, | |
| "loss": 0.6125, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.05635342851735812, | |
| "grad_norm": 0.6772509813308716, | |
| "learning_rate": 4.909585842496287e-05, | |
| "loss": 0.7981, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.056437538112160145, | |
| "grad_norm": 0.5350778698921204, | |
| "learning_rate": 4.8825699988565485e-05, | |
| "loss": 0.8081, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.05652164770696217, | |
| "grad_norm": 0.4920620620250702, | |
| "learning_rate": 4.8556046621849346e-05, | |
| "loss": 0.4888, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.0566057573017642, | |
| "grad_norm": 0.5023487210273743, | |
| "learning_rate": 4.828690098618429e-05, | |
| "loss": 0.7385, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.05668986689656622, | |
| "grad_norm": 0.7149183750152588, | |
| "learning_rate": 4.8018265737929044e-05, | |
| "loss": 0.6377, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.056773976491368255, | |
| "grad_norm": 0.5488125681877136, | |
| "learning_rate": 4.7750143528405126e-05, | |
| "loss": 0.6302, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.05685808608617028, | |
| "grad_norm": 0.6204077005386353, | |
| "learning_rate": 4.748253700387042e-05, | |
| "loss": 0.7237, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.05694219568097231, | |
| "grad_norm": 0.4989658296108246, | |
| "learning_rate": 4.721544880549337e-05, | |
| "loss": 0.6335, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.05702630527577433, | |
| "grad_norm": 0.582874059677124, | |
| "learning_rate": 4.694888156932658e-05, | |
| "loss": 0.6944, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.05711041487057636, | |
| "grad_norm": 0.5402207374572754, | |
| "learning_rate": 4.668283792628114e-05, | |
| "loss": 0.752, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.057194524465378385, | |
| "grad_norm": 0.46678709983825684, | |
| "learning_rate": 4.6417320502100316e-05, | |
| "loss": 0.625, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.05727863406018042, | |
| "grad_norm": 0.5934203267097473, | |
| "learning_rate": 4.615233191733398e-05, | |
| "loss": 0.7576, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.057362743654982444, | |
| "grad_norm": 0.6489574909210205, | |
| "learning_rate": 4.588787478731242e-05, | |
| "loss": 0.6303, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.05744685324978447, | |
| "grad_norm": 0.503392219543457, | |
| "learning_rate": 4.5623951722120736e-05, | |
| "loss": 0.6418, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.057530962844586496, | |
| "grad_norm": 0.7234579920768738, | |
| "learning_rate": 4.5360565326573104e-05, | |
| "loss": 0.8319, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.05761507243938852, | |
| "grad_norm": 0.6499959826469421, | |
| "learning_rate": 4.5097718200186814e-05, | |
| "loss": 0.6247, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.05769918203419055, | |
| "grad_norm": 0.5063481330871582, | |
| "learning_rate": 4.483541293715698e-05, | |
| "loss": 0.6584, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.05778329162899258, | |
| "grad_norm": 0.5390834808349609, | |
| "learning_rate": 4.457365212633058e-05, | |
| "loss": 0.7036, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.05786740122379461, | |
| "grad_norm": 0.4784635305404663, | |
| "learning_rate": 4.431243835118124e-05, | |
| "loss": 0.4651, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.05795151081859663, | |
| "grad_norm": 0.5334429740905762, | |
| "learning_rate": 4.4051774189783315e-05, | |
| "loss": 0.5427, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.05803562041339866, | |
| "grad_norm": 0.5312195420265198, | |
| "learning_rate": 4.379166221478697e-05, | |
| "loss": 0.6366, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.058119730008200685, | |
| "grad_norm": 0.6458830237388611, | |
| "learning_rate": 4.3532104993392306e-05, | |
| "loss": 0.7941, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.05820383960300271, | |
| "grad_norm": 0.5415295362472534, | |
| "learning_rate": 4.327310508732437e-05, | |
| "loss": 0.5574, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.058287949197804736, | |
| "grad_norm": 0.4901331663131714, | |
| "learning_rate": 4.301466505280762e-05, | |
| "loss": 0.6664, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.05837205879260677, | |
| "grad_norm": 0.5454357266426086, | |
| "learning_rate": 4.2756787440540936e-05, | |
| "loss": 0.7233, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.058456168387408795, | |
| "grad_norm": 0.6961559057235718, | |
| "learning_rate": 4.249947479567218e-05, | |
| "loss": 0.707, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.05854027798221082, | |
| "grad_norm": 0.6365110874176025, | |
| "learning_rate": 4.224272965777326e-05, | |
| "loss": 0.7353, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.05862438757701285, | |
| "grad_norm": 0.527192234992981, | |
| "learning_rate": 4.1986554560815096e-05, | |
| "loss": 0.7661, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.05870849717181487, | |
| "grad_norm": 0.48179832100868225, | |
| "learning_rate": 4.173095203314241e-05, | |
| "loss": 0.665, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.0587926067666169, | |
| "grad_norm": 0.6591465473175049, | |
| "learning_rate": 4.1475924597449024e-05, | |
| "loss": 0.7179, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.05887671636141893, | |
| "grad_norm": 0.4673934876918793, | |
| "learning_rate": 4.12214747707527e-05, | |
| "loss": 0.6477, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.05896082595622096, | |
| "grad_norm": 0.8560356497764587, | |
| "learning_rate": 4.096760506437057e-05, | |
| "loss": 0.4083, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.059044935551022984, | |
| "grad_norm": 0.5286902189254761, | |
| "learning_rate": 4.071431798389408e-05, | |
| "loss": 0.7202, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.05912904514582501, | |
| "grad_norm": 0.5945466160774231, | |
| "learning_rate": 4.0461616029164526e-05, | |
| "loss": 0.8511, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.059213154740627036, | |
| "grad_norm": 0.7009603381156921, | |
| "learning_rate": 4.020950169424815e-05, | |
| "loss": 0.8503, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.05929726433542906, | |
| "grad_norm": 0.5323594808578491, | |
| "learning_rate": 3.9957977467411615e-05, | |
| "loss": 0.6064, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.05938137393023109, | |
| "grad_norm": 0.4955034852027893, | |
| "learning_rate": 3.9707045831097555e-05, | |
| "loss": 0.6115, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.05946548352503312, | |
| "grad_norm": 0.5311620831489563, | |
| "learning_rate": 3.945670926189987e-05, | |
| "loss": 0.6827, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.05954959311983515, | |
| "grad_norm": 0.432479590177536, | |
| "learning_rate": 3.920697023053949e-05, | |
| "loss": 0.5194, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.05963370271463717, | |
| "grad_norm": 0.544047474861145, | |
| "learning_rate": 3.895783120183976e-05, | |
| "loss": 0.6654, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.0597178123094392, | |
| "grad_norm": 0.4839647114276886, | |
| "learning_rate": 3.8709294634702376e-05, | |
| "loss": 0.5999, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.059801921904241225, | |
| "grad_norm": 0.6779318451881409, | |
| "learning_rate": 3.846136298208285e-05, | |
| "loss": 0.655, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.05988603149904325, | |
| "grad_norm": 0.5102113485336304, | |
| "learning_rate": 3.821403869096658e-05, | |
| "loss": 0.6759, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.05997014109384528, | |
| "grad_norm": 0.5493488311767578, | |
| "learning_rate": 3.796732420234443e-05, | |
| "loss": 0.5339, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.06005425068864731, | |
| "grad_norm": 0.5925408005714417, | |
| "learning_rate": 3.7721221951188765e-05, | |
| "loss": 0.751, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.060138360283449335, | |
| "grad_norm": 0.6988538503646851, | |
| "learning_rate": 3.747573436642951e-05, | |
| "loss": 0.759, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.06022246987825136, | |
| "grad_norm": 0.5451468825340271, | |
| "learning_rate": 3.7230863870929964e-05, | |
| "loss": 0.5549, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.06030657947305339, | |
| "grad_norm": 0.5933196544647217, | |
| "learning_rate": 3.698661288146311e-05, | |
| "loss": 0.5338, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.06039068906785541, | |
| "grad_norm": 0.5368034839630127, | |
| "learning_rate": 3.674298380868756e-05, | |
| "loss": 0.6672, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.060474798662657446, | |
| "grad_norm": 0.6387892961502075, | |
| "learning_rate": 3.649997905712396e-05, | |
| "loss": 0.6483, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.06055890825745947, | |
| "grad_norm": 0.5074757933616638, | |
| "learning_rate": 3.6257601025131026e-05, | |
| "loss": 0.5061, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.0606430178522615, | |
| "grad_norm": 0.7502937912940979, | |
| "learning_rate": 3.601585210488218e-05, | |
| "loss": 0.7917, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.060727127447063524, | |
| "grad_norm": 0.5781522989273071, | |
| "learning_rate": 3.577473468234156e-05, | |
| "loss": 0.825, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.06081123704186555, | |
| "grad_norm": 0.559719443321228, | |
| "learning_rate": 3.553425113724088e-05, | |
| "loss": 0.5209, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.060895346636667576, | |
| "grad_norm": 0.714248538017273, | |
| "learning_rate": 3.52944038430556e-05, | |
| "loss": 0.5418, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.0609794562314696, | |
| "grad_norm": 0.5317071080207825, | |
| "learning_rate": 3.5055195166981645e-05, | |
| "loss": 0.6363, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.061063565826271635, | |
| "grad_norm": 0.5349583625793457, | |
| "learning_rate": 3.481662746991214e-05, | |
| "loss": 0.5828, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.06114767542107366, | |
| "grad_norm": 0.5363706350326538, | |
| "learning_rate": 3.4578703106413904e-05, | |
| "loss": 0.5334, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.06123178501587569, | |
| "grad_norm": 0.4438410699367523, | |
| "learning_rate": 3.4341424424704375e-05, | |
| "loss": 0.7108, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.06131589461067771, | |
| "grad_norm": 0.6724744439125061, | |
| "learning_rate": 3.4104793766628304e-05, | |
| "loss": 0.6905, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.06140000420547974, | |
| "grad_norm": 0.5650711059570312, | |
| "learning_rate": 3.386881346763483e-05, | |
| "loss": 0.5404, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.061484113800281764, | |
| "grad_norm": 0.6993920803070068, | |
| "learning_rate": 3.363348585675414e-05, | |
| "loss": 0.6725, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.0615682233950838, | |
| "grad_norm": 0.5407055616378784, | |
| "learning_rate": 3.339881325657484e-05, | |
| "loss": 0.6004, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.06165233298988582, | |
| "grad_norm": 0.5904746055603027, | |
| "learning_rate": 3.316479798322072e-05, | |
| "loss": 0.6763, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.06173644258468785, | |
| "grad_norm": 0.5810348987579346, | |
| "learning_rate": 3.2931442346328004e-05, | |
| "loss": 0.8605, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.061820552179489875, | |
| "grad_norm": 0.48206856846809387, | |
| "learning_rate": 3.269874864902269e-05, | |
| "loss": 0.6533, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.0619046617742919, | |
| "grad_norm": 0.5678583383560181, | |
| "learning_rate": 3.246671918789755e-05, | |
| "loss": 0.5872, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.06198877136909393, | |
| "grad_norm": 0.5930237174034119, | |
| "learning_rate": 3.223535625298979e-05, | |
| "loss": 0.7073, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.06207288096389595, | |
| "grad_norm": 0.4858033061027527, | |
| "learning_rate": 3.200466212775808e-05, | |
| "loss": 0.5829, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.062156990558697986, | |
| "grad_norm": 0.6299439668655396, | |
| "learning_rate": 3.1774639089060363e-05, | |
| "loss": 0.4702, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.06224110015350001, | |
| "grad_norm": 0.5381211042404175, | |
| "learning_rate": 3.154528940713113e-05, | |
| "loss": 0.7213, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.06232520974830204, | |
| "grad_norm": 0.4895922839641571, | |
| "learning_rate": 3.1316615345559185e-05, | |
| "loss": 0.571, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.062409319343104064, | |
| "grad_norm": 0.507371723651886, | |
| "learning_rate": 3.108861916126518e-05, | |
| "loss": 0.6783, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.06249342893790609, | |
| "grad_norm": 0.5576531291007996, | |
| "learning_rate": 3.086130310447937e-05, | |
| "loss": 0.6561, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.06257753853270812, | |
| "grad_norm": 0.42016541957855225, | |
| "learning_rate": 3.063466941871952e-05, | |
| "loss": 0.5153, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.06266164812751014, | |
| "grad_norm": 0.7200675010681152, | |
| "learning_rate": 3.0408720340768572e-05, | |
| "loss": 0.6269, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.06274575772231217, | |
| "grad_norm": 0.5375575423240662, | |
| "learning_rate": 3.018345810065275e-05, | |
| "loss": 0.5603, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.0628298673171142, | |
| "grad_norm": 0.45930182933807373, | |
| "learning_rate": 2.9958884921619367e-05, | |
| "loss": 0.7863, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.06291397691191623, | |
| "grad_norm": 0.5124622583389282, | |
| "learning_rate": 2.9735003020115092e-05, | |
| "loss": 0.7125, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.06299808650671826, | |
| "grad_norm": 0.5122153162956238, | |
| "learning_rate": 2.9511814605763855e-05, | |
| "loss": 0.7104, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.06308219610152029, | |
| "grad_norm": 0.56404048204422, | |
| "learning_rate": 2.9289321881345254e-05, | |
| "loss": 0.6569, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.06316630569632231, | |
| "grad_norm": 0.4476301968097687, | |
| "learning_rate": 2.9067527042772636e-05, | |
| "loss": 0.6703, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.06325041529112434, | |
| "grad_norm": 0.432229220867157, | |
| "learning_rate": 2.8846432279071467e-05, | |
| "loss": 0.7372, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.06333452488592636, | |
| "grad_norm": 0.5445795059204102, | |
| "learning_rate": 2.8626039772357882e-05, | |
| "loss": 0.5326, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.06341863448072839, | |
| "grad_norm": 0.645452618598938, | |
| "learning_rate": 2.840635169781688e-05, | |
| "loss": 0.3862, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.06350274407553042, | |
| "grad_norm": 0.4100976288318634, | |
| "learning_rate": 2.8187370223681132e-05, | |
| "loss": 0.8218, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.06358685367033244, | |
| "grad_norm": 0.7901608347892761, | |
| "learning_rate": 2.7969097511209308e-05, | |
| "loss": 0.6172, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.06367096326513447, | |
| "grad_norm": 0.4609740376472473, | |
| "learning_rate": 2.775153571466502e-05, | |
| "loss": 0.7415, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.06375507285993649, | |
| "grad_norm": 0.684585452079773, | |
| "learning_rate": 2.753468698129533e-05, | |
| "loss": 0.696, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.06383918245473852, | |
| "grad_norm": 0.5767757296562195, | |
| "learning_rate": 2.7318553451309726e-05, | |
| "loss": 0.6223, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.06392329204954054, | |
| "grad_norm": 0.5188784003257751, | |
| "learning_rate": 2.7103137257858868e-05, | |
| "loss": 0.7564, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.06400740164434258, | |
| "grad_norm": 0.6304595470428467, | |
| "learning_rate": 2.688844052701359e-05, | |
| "loss": 0.7426, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.06409151123914461, | |
| "grad_norm": 0.5373560190200806, | |
| "learning_rate": 2.6674465377744017e-05, | |
| "loss": 0.7388, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.06417562083394664, | |
| "grad_norm": 0.5632065534591675, | |
| "learning_rate": 2.646121392189841e-05, | |
| "loss": 0.5892, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.06425973042874866, | |
| "grad_norm": 0.5747659802436829, | |
| "learning_rate": 2.624868826418262e-05, | |
| "loss": 0.4149, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.06434384002355069, | |
| "grad_norm": 0.7094728946685791, | |
| "learning_rate": 2.603689050213902e-05, | |
| "loss": 0.6391, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.06442794961835271, | |
| "grad_norm": 0.6141682863235474, | |
| "learning_rate": 2.582582272612609e-05, | |
| "loss": 0.5478, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.06451205921315474, | |
| "grad_norm": 0.5336157083511353, | |
| "learning_rate": 2.561548701929749e-05, | |
| "loss": 0.6272, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.06459616880795677, | |
| "grad_norm": 0.5744621753692627, | |
| "learning_rate": 2.540588545758179e-05, | |
| "loss": 0.6357, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.06468027840275879, | |
| "grad_norm": 0.5280835628509521, | |
| "learning_rate": 2.5197020109661772e-05, | |
| "loss": 0.6744, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.06476438799756082, | |
| "grad_norm": 0.5798782110214233, | |
| "learning_rate": 2.4988893036954043e-05, | |
| "loss": 0.9299, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.06484849759236284, | |
| "grad_norm": 0.7400964498519897, | |
| "learning_rate": 2.4781506293588873e-05, | |
| "loss": 0.7047, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.06493260718716487, | |
| "grad_norm": 0.4643491208553314, | |
| "learning_rate": 2.4574861926389615e-05, | |
| "loss": 0.5174, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.0650167167819669, | |
| "grad_norm": 0.4251709580421448, | |
| "learning_rate": 2.436896197485282e-05, | |
| "loss": 0.6295, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.06510082637676894, | |
| "grad_norm": 0.48561739921569824, | |
| "learning_rate": 2.4163808471127812e-05, | |
| "loss": 0.6543, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.06518493597157096, | |
| "grad_norm": 0.6305271983146667, | |
| "learning_rate": 2.3959403439996907e-05, | |
| "loss": 0.5729, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.06526904556637299, | |
| "grad_norm": 0.6219698190689087, | |
| "learning_rate": 2.37557488988552e-05, | |
| "loss": 0.7075, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.06535315516117501, | |
| "grad_norm": 0.5188342332839966, | |
| "learning_rate": 2.3552846857690846e-05, | |
| "loss": 0.7433, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.06543726475597704, | |
| "grad_norm": 1.439207673072815, | |
| "learning_rate": 2.3350699319065026e-05, | |
| "loss": 0.8646, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.06552137435077907, | |
| "grad_norm": 0.5281020402908325, | |
| "learning_rate": 2.3149308278092342e-05, | |
| "loss": 0.7782, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.06560548394558109, | |
| "grad_norm": 0.5641273260116577, | |
| "learning_rate": 2.2948675722421086e-05, | |
| "loss": 0.8682, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.06568959354038312, | |
| "grad_norm": 0.6752006411552429, | |
| "learning_rate": 2.2748803632213557e-05, | |
| "loss": 0.652, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.06577370313518514, | |
| "grad_norm": 0.48423561453819275, | |
| "learning_rate": 2.254969398012663e-05, | |
| "loss": 0.3943, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.06585781272998717, | |
| "grad_norm": 0.5390269756317139, | |
| "learning_rate": 2.235134873129213e-05, | |
| "loss": 0.657, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.0659419223247892, | |
| "grad_norm": 0.469672828912735, | |
| "learning_rate": 2.2153769843297667e-05, | |
| "loss": 0.5305, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.06602603191959122, | |
| "grad_norm": 0.6228986978530884, | |
| "learning_rate": 2.195695926616702e-05, | |
| "loss": 0.546, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.06611014151439325, | |
| "grad_norm": 0.537236750125885, | |
| "learning_rate": 2.1760918942341192e-05, | |
| "loss": 0.7306, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.06619425110919529, | |
| "grad_norm": 0.48184770345687866, | |
| "learning_rate": 2.1565650806658975e-05, | |
| "loss": 0.6454, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.06627836070399731, | |
| "grad_norm": 0.6069773435592651, | |
| "learning_rate": 2.137115678633811e-05, | |
| "loss": 0.6599, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.06636247029879934, | |
| "grad_norm": 0.5169489979743958, | |
| "learning_rate": 2.1177438800956007e-05, | |
| "loss": 0.5624, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.06644657989360137, | |
| "grad_norm": 0.4128231704235077, | |
| "learning_rate": 2.098449876243096e-05, | |
| "loss": 0.5663, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.06653068948840339, | |
| "grad_norm": 0.715587854385376, | |
| "learning_rate": 2.07923385750033e-05, | |
| "loss": 0.7287, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.06661479908320542, | |
| "grad_norm": 0.44507458806037903, | |
| "learning_rate": 2.0600960135216462e-05, | |
| "loss": 0.6717, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.06669890867800744, | |
| "grad_norm": 0.5170130729675293, | |
| "learning_rate": 2.0410365331898416e-05, | |
| "loss": 0.6406, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.06678301827280947, | |
| "grad_norm": 0.46590495109558105, | |
| "learning_rate": 2.0220556046142893e-05, | |
| "loss": 0.6061, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.0668671278676115, | |
| "grad_norm": 0.5688216686248779, | |
| "learning_rate": 2.0031534151290943e-05, | |
| "loss": 0.7645, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.06695123746241352, | |
| "grad_norm": 0.6071348786354065, | |
| "learning_rate": 1.9843301512912327e-05, | |
| "loss": 0.8112, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.06703534705721555, | |
| "grad_norm": 0.639324963092804, | |
| "learning_rate": 1.965585998878724e-05, | |
| "loss": 0.683, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.06711945665201757, | |
| "grad_norm": 0.5158023238182068, | |
| "learning_rate": 1.946921142888781e-05, | |
| "loss": 0.6532, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.06720356624681961, | |
| "grad_norm": 0.5088067650794983, | |
| "learning_rate": 1.928335767535997e-05, | |
| "loss": 0.6117, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.06728767584162164, | |
| "grad_norm": 0.4564967155456543, | |
| "learning_rate": 1.9098300562505266e-05, | |
| "loss": 0.6276, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.06737178543642366, | |
| "grad_norm": 0.620216429233551, | |
| "learning_rate": 1.891404191676265e-05, | |
| "loss": 0.655, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.06745589503122569, | |
| "grad_norm": 0.698749303817749, | |
| "learning_rate": 1.8730583556690605e-05, | |
| "loss": 0.5483, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.06754000462602772, | |
| "grad_norm": 0.48406967520713806, | |
| "learning_rate": 1.854792729294905e-05, | |
| "loss": 0.7465, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.06762411422082974, | |
| "grad_norm": 0.5097486972808838, | |
| "learning_rate": 1.8366074928281607e-05, | |
| "loss": 0.4625, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.06770822381563177, | |
| "grad_norm": 0.5742330551147461, | |
| "learning_rate": 1.818502825749764e-05, | |
| "loss": 0.6649, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.0677923334104338, | |
| "grad_norm": 0.49662643671035767, | |
| "learning_rate": 1.8004789067454764e-05, | |
| "loss": 0.7763, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.06787644300523582, | |
| "grad_norm": 0.6889787912368774, | |
| "learning_rate": 1.7825359137040988e-05, | |
| "loss": 0.6646, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.06796055260003785, | |
| "grad_norm": 0.47545164823532104, | |
| "learning_rate": 1.7646740237157256e-05, | |
| "loss": 0.6404, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.06804466219483987, | |
| "grad_norm": 0.5143394470214844, | |
| "learning_rate": 1.7468934130700044e-05, | |
| "loss": 0.6727, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.0681287717896419, | |
| "grad_norm": 0.5231197476387024, | |
| "learning_rate": 1.7291942572543807e-05, | |
| "loss": 0.6471, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.06821288138444392, | |
| "grad_norm": 0.5731273293495178, | |
| "learning_rate": 1.7115767309523812e-05, | |
| "loss": 0.7455, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.06829699097924596, | |
| "grad_norm": 0.5894562005996704, | |
| "learning_rate": 1.6940410080418723e-05, | |
| "loss": 0.6111, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.06838110057404799, | |
| "grad_norm": 0.5060394406318665, | |
| "learning_rate": 1.6765872615933677e-05, | |
| "loss": 0.5798, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.06846521016885002, | |
| "grad_norm": 0.5075900554656982, | |
| "learning_rate": 1.6592156638682886e-05, | |
| "loss": 0.6851, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.06854931976365204, | |
| "grad_norm": 0.40963247418403625, | |
| "learning_rate": 1.6419263863172997e-05, | |
| "loss": 0.53, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.06863342935845407, | |
| "grad_norm": 0.5541537404060364, | |
| "learning_rate": 1.6247195995785837e-05, | |
| "loss": 0.8987, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.0687175389532561, | |
| "grad_norm": 0.6535544395446777, | |
| "learning_rate": 1.6075954734761845e-05, | |
| "loss": 0.5659, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.06880164854805812, | |
| "grad_norm": 0.4968286454677582, | |
| "learning_rate": 1.5905541770183096e-05, | |
| "loss": 0.6928, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.06888575814286015, | |
| "grad_norm": 0.852871835231781, | |
| "learning_rate": 1.5735958783956794e-05, | |
| "loss": 0.5483, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.06896986773766217, | |
| "grad_norm": 0.6427825689315796, | |
| "learning_rate": 1.5567207449798515e-05, | |
| "loss": 0.5945, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.0690539773324642, | |
| "grad_norm": 0.6467059850692749, | |
| "learning_rate": 1.539928943321579e-05, | |
| "loss": 0.5874, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.06913808692726622, | |
| "grad_norm": 0.5707730054855347, | |
| "learning_rate": 1.5232206391491699e-05, | |
| "loss": 0.616, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.06922219652206825, | |
| "grad_norm": 0.6659408211708069, | |
| "learning_rate": 1.5065959973668353e-05, | |
| "loss": 0.782, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.06930630611687028, | |
| "grad_norm": 0.5345143675804138, | |
| "learning_rate": 1.4900551820530828e-05, | |
| "loss": 0.7579, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.06939041571167232, | |
| "grad_norm": 0.5110843777656555, | |
| "learning_rate": 1.4735983564590783e-05, | |
| "loss": 0.778, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.06947452530647434, | |
| "grad_norm": 0.4799653887748718, | |
| "learning_rate": 1.4572256830070497e-05, | |
| "loss": 0.6927, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.06955863490127637, | |
| "grad_norm": 0.5991430282592773, | |
| "learning_rate": 1.4409373232886702e-05, | |
| "loss": 0.6772, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.0696427444960784, | |
| "grad_norm": 0.6813725233078003, | |
| "learning_rate": 1.4247334380634792e-05, | |
| "loss": 0.6461, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.06972685409088042, | |
| "grad_norm": 0.6396465301513672, | |
| "learning_rate": 1.4086141872572789e-05, | |
| "loss": 0.7156, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.06981096368568245, | |
| "grad_norm": 0.6053589582443237, | |
| "learning_rate": 1.3925797299605647e-05, | |
| "loss": 0.8174, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.06989507328048447, | |
| "grad_norm": 0.5992446541786194, | |
| "learning_rate": 1.3766302244269624e-05, | |
| "loss": 0.7147, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.0699791828752865, | |
| "grad_norm": 0.5334646701812744, | |
| "learning_rate": 1.3607658280716473e-05, | |
| "loss": 0.5372, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.07006329247008852, | |
| "grad_norm": 0.4751766622066498, | |
| "learning_rate": 1.3449866974698122e-05, | |
| "loss": 0.4352, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.07014740206489055, | |
| "grad_norm": 0.4926972985267639, | |
| "learning_rate": 1.3292929883550998e-05, | |
| "loss": 0.6821, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.07023151165969257, | |
| "grad_norm": 0.4819171726703644, | |
| "learning_rate": 1.3136848556180892e-05, | |
| "loss": 0.6051, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.0703156212544946, | |
| "grad_norm": 0.48462870717048645, | |
| "learning_rate": 1.2981624533047432e-05, | |
| "loss": 0.5676, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.07039973084929663, | |
| "grad_norm": 0.5962949395179749, | |
| "learning_rate": 1.2827259346149122e-05, | |
| "loss": 0.5559, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.07048384044409867, | |
| "grad_norm": 0.5228509306907654, | |
| "learning_rate": 1.2673754519008008e-05, | |
| "loss": 0.6072, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.07056795003890069, | |
| "grad_norm": 0.6193320751190186, | |
| "learning_rate": 1.2521111566654731e-05, | |
| "loss": 0.6156, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.07065205963370272, | |
| "grad_norm": 0.48941028118133545, | |
| "learning_rate": 1.2369331995613665e-05, | |
| "loss": 0.5834, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.07073616922850474, | |
| "grad_norm": 0.5763342380523682, | |
| "learning_rate": 1.2218417303887842e-05, | |
| "loss": 0.6142, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.07082027882330677, | |
| "grad_norm": 0.5619479417800903, | |
| "learning_rate": 1.206836898094439e-05, | |
| "loss": 0.5918, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.0709043884181088, | |
| "grad_norm": 0.4731215536594391, | |
| "learning_rate": 1.191918850769964e-05, | |
| "loss": 0.4884, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.07098849801291082, | |
| "grad_norm": 0.7118493318557739, | |
| "learning_rate": 1.1770877356504683e-05, | |
| "loss": 0.6499, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.07107260760771285, | |
| "grad_norm": 0.7185307741165161, | |
| "learning_rate": 1.1623436991130654e-05, | |
| "loss": 0.7735, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.07115671720251487, | |
| "grad_norm": 0.5481952428817749, | |
| "learning_rate": 1.1476868866754486e-05, | |
| "loss": 0.919, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.0712408267973169, | |
| "grad_norm": 0.596553385257721, | |
| "learning_rate": 1.1331174429944347e-05, | |
| "loss": 0.6963, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.07132493639211893, | |
| "grad_norm": 0.5250246524810791, | |
| "learning_rate": 1.1186355118645554e-05, | |
| "loss": 0.6193, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.07140904598692095, | |
| "grad_norm": 0.5260514616966248, | |
| "learning_rate": 1.1042412362166222e-05, | |
| "loss": 0.6361, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.07149315558172299, | |
| "grad_norm": 0.5367104411125183, | |
| "learning_rate": 1.0899347581163221e-05, | |
| "loss": 0.5347, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.07157726517652502, | |
| "grad_norm": 0.5701801776885986, | |
| "learning_rate": 1.0757162187628222e-05, | |
| "loss": 0.6477, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.07166137477132704, | |
| "grad_norm": 0.494419127702713, | |
| "learning_rate": 1.0615857584873623e-05, | |
| "loss": 0.6831, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.07174548436612907, | |
| "grad_norm": 0.6777026057243347, | |
| "learning_rate": 1.0475435167518843e-05, | |
| "loss": 0.6998, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.0718295939609311, | |
| "grad_norm": 0.6337087154388428, | |
| "learning_rate": 1.0335896321476413e-05, | |
| "loss": 0.6884, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.07191370355573312, | |
| "grad_norm": 0.508190393447876, | |
| "learning_rate": 1.0197242423938446e-05, | |
| "loss": 0.6033, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.07199781315053515, | |
| "grad_norm": 0.588848888874054, | |
| "learning_rate": 1.0059474843362892e-05, | |
| "loss": 0.6572, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.07208192274533717, | |
| "grad_norm": 0.6991647481918335, | |
| "learning_rate": 9.922594939460194e-06, | |
| "loss": 0.5044, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.0721660323401392, | |
| "grad_norm": 0.6183068752288818, | |
| "learning_rate": 9.786604063179728e-06, | |
| "loss": 0.5294, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.07225014193494123, | |
| "grad_norm": 0.6094670295715332, | |
| "learning_rate": 9.651503556696516e-06, | |
| "loss": 0.4282, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.07233425152974325, | |
| "grad_norm": 0.5989966988563538, | |
| "learning_rate": 9.517294753398064e-06, | |
| "loss": 0.6284, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.07241836112454528, | |
| "grad_norm": 0.5924322009086609, | |
| "learning_rate": 9.383978977871021e-06, | |
| "loss": 0.4946, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.0725024707193473, | |
| "grad_norm": 0.6521950364112854, | |
| "learning_rate": 9.251557545888312e-06, | |
| "loss": 0.6351, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.07258658031414934, | |
| "grad_norm": 0.5028837323188782, | |
| "learning_rate": 9.120031764395987e-06, | |
| "loss": 0.6651, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.07267068990895137, | |
| "grad_norm": 0.5665683746337891, | |
| "learning_rate": 8.989402931500434e-06, | |
| "loss": 0.6975, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.0727547995037534, | |
| "grad_norm": 0.6868077516555786, | |
| "learning_rate": 8.85967233645547e-06, | |
| "loss": 0.6654, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.07283890909855542, | |
| "grad_norm": 0.4886159300804138, | |
| "learning_rate": 8.730841259649725e-06, | |
| "loss": 0.6435, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.07292301869335745, | |
| "grad_norm": 0.5298961997032166, | |
| "learning_rate": 8.602910972593892e-06, | |
| "loss": 0.6013, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.07300712828815947, | |
| "grad_norm": 0.5650634765625, | |
| "learning_rate": 8.475882737908248e-06, | |
| "loss": 0.6252, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.0730912378829615, | |
| "grad_norm": 0.6334890723228455, | |
| "learning_rate": 8.34975780931021e-06, | |
| "loss": 0.5932, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.07317534747776352, | |
| "grad_norm": 0.42107874155044556, | |
| "learning_rate": 8.224537431601886e-06, | |
| "loss": 0.7897, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.07325945707256555, | |
| "grad_norm": 0.4216519594192505, | |
| "learning_rate": 8.100222840657878e-06, | |
| "loss": 0.671, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.07334356666736758, | |
| "grad_norm": 0.4145600497722626, | |
| "learning_rate": 7.976815263412963e-06, | |
| "loss": 0.5159, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.0734276762621696, | |
| "grad_norm": 0.6570865511894226, | |
| "learning_rate": 7.854315917850163e-06, | |
| "loss": 0.685, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.07351178585697163, | |
| "grad_norm": 0.42912426590919495, | |
| "learning_rate": 7.73272601298851e-06, | |
| "loss": 0.5689, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.07359589545177365, | |
| "grad_norm": 0.5340107679367065, | |
| "learning_rate": 7.612046748871327e-06, | |
| "loss": 0.6738, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.0736800050465757, | |
| "grad_norm": 0.6763095855712891, | |
| "learning_rate": 7.492279316554207e-06, | |
| "loss": 0.6679, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.07376411464137772, | |
| "grad_norm": 0.6157491207122803, | |
| "learning_rate": 7.3734248980933395e-06, | |
| "loss": 0.7077, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.07384822423617975, | |
| "grad_norm": 0.6420180797576904, | |
| "learning_rate": 7.255484666533874e-06, | |
| "loss": 0.6328, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.07393233383098177, | |
| "grad_norm": 0.7011188268661499, | |
| "learning_rate": 7.138459785898266e-06, | |
| "loss": 0.6339, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.0740164434257838, | |
| "grad_norm": 0.5735622048377991, | |
| "learning_rate": 7.022351411174866e-06, | |
| "loss": 0.6621, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.07410055302058582, | |
| "grad_norm": 0.49764999747276306, | |
| "learning_rate": 6.907160688306425e-06, | |
| "loss": 0.5328, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.07418466261538785, | |
| "grad_norm": 0.6401441097259521, | |
| "learning_rate": 6.7928887541789055e-06, | |
| "loss": 0.6566, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.07426877221018988, | |
| "grad_norm": 0.44652143120765686, | |
| "learning_rate": 6.679536736610137e-06, | |
| "loss": 0.5332, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.0743528818049919, | |
| "grad_norm": 0.6002373695373535, | |
| "learning_rate": 6.5671057543387985e-06, | |
| "loss": 0.7656, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.07443699139979393, | |
| "grad_norm": 0.6372143030166626, | |
| "learning_rate": 6.455596917013273e-06, | |
| "loss": 0.7022, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.07452110099459595, | |
| "grad_norm": 0.4614960253238678, | |
| "learning_rate": 6.345011325180772e-06, | |
| "loss": 0.62, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.07460521058939798, | |
| "grad_norm": 0.6100625395774841, | |
| "learning_rate": 6.235350070276447e-06, | |
| "loss": 0.7091, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.0746893201842, | |
| "grad_norm": 0.6311874985694885, | |
| "learning_rate": 6.126614234612593e-06, | |
| "loss": 0.6988, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.07477342977900205, | |
| "grad_norm": 0.6123420000076294, | |
| "learning_rate": 6.018804891368035e-06, | |
| "loss": 0.6147, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.07485753937380407, | |
| "grad_norm": 0.5420098304748535, | |
| "learning_rate": 5.911923104577455e-06, | |
| "loss": 0.6469, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.0749416489686061, | |
| "grad_norm": 0.5997559428215027, | |
| "learning_rate": 5.805969929120947e-06, | |
| "loss": 0.6006, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.07502575856340812, | |
| "grad_norm": 0.801465630531311, | |
| "learning_rate": 5.700946410713548e-06, | |
| "loss": 0.7753, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.07510986815821015, | |
| "grad_norm": 0.5048848986625671, | |
| "learning_rate": 5.5968535858950345e-06, | |
| "loss": 0.6048, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.07519397775301218, | |
| "grad_norm": 0.6630212664604187, | |
| "learning_rate": 5.49369248201953e-06, | |
| "loss": 0.8072, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.0752780873478142, | |
| "grad_norm": 0.7513829469680786, | |
| "learning_rate": 5.39146411724547e-06, | |
| "loss": 0.6736, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.07536219694261623, | |
| "grad_norm": 1.04941987991333, | |
| "learning_rate": 5.290169500525577e-06, | |
| "loss": 0.6241, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.07544630653741825, | |
| "grad_norm": 0.4719288647174835, | |
| "learning_rate": 5.189809631596798e-06, | |
| "loss": 0.5016, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.07553041613222028, | |
| "grad_norm": 0.5207799673080444, | |
| "learning_rate": 5.0903855009705514e-06, | |
| "loss": 0.6855, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.0756145257270223, | |
| "grad_norm": 0.5883520245552063, | |
| "learning_rate": 4.991898089922819e-06, | |
| "loss": 0.841, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.07569863532182433, | |
| "grad_norm": 0.48410382866859436, | |
| "learning_rate": 4.8943483704846475e-06, | |
| "loss": 0.7806, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.07578274491662636, | |
| "grad_norm": 0.5785539150238037, | |
| "learning_rate": 4.797737305432337e-06, | |
| "loss": 0.7894, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.0758668545114284, | |
| "grad_norm": 0.6412384510040283, | |
| "learning_rate": 4.702065848278126e-06, | |
| "loss": 0.5901, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.07595096410623042, | |
| "grad_norm": 0.6484128832817078, | |
| "learning_rate": 4.607334943260655e-06, | |
| "loss": 0.5787, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.07603507370103245, | |
| "grad_norm": 0.6749901175498962, | |
| "learning_rate": 4.513545525335705e-06, | |
| "loss": 0.6713, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.07611918329583447, | |
| "grad_norm": 0.6375152468681335, | |
| "learning_rate": 4.420698520166988e-06, | |
| "loss": 0.5882, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.0762032928906365, | |
| "grad_norm": 0.5309505462646484, | |
| "learning_rate": 4.328794844116946e-06, | |
| "loss": 0.6259, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.07628740248543853, | |
| "grad_norm": 0.5349912047386169, | |
| "learning_rate": 4.237835404237778e-06, | |
| "loss": 0.7206, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.07637151208024055, | |
| "grad_norm": 0.5662460327148438, | |
| "learning_rate": 4.147821098262405e-06, | |
| "loss": 0.8227, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.07645562167504258, | |
| "grad_norm": 0.5176487565040588, | |
| "learning_rate": 4.0587528145957235e-06, | |
| "loss": 0.6219, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.0765397312698446, | |
| "grad_norm": 0.680101752281189, | |
| "learning_rate": 3.970631432305694e-06, | |
| "loss": 0.822, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.07662384086464663, | |
| "grad_norm": 0.6776043176651001, | |
| "learning_rate": 3.883457821114811e-06, | |
| "loss": 0.5682, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.07670795045944866, | |
| "grad_norm": 0.5205222368240356, | |
| "learning_rate": 3.797232841391407e-06, | |
| "loss": 0.6134, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.07679206005425068, | |
| "grad_norm": 0.5915586948394775, | |
| "learning_rate": 3.711957344141237e-06, | |
| "loss": 0.748, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.07687616964905272, | |
| "grad_norm": 0.5967119336128235, | |
| "learning_rate": 3.627632170999029e-06, | |
| "loss": 0.6232, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.07696027924385475, | |
| "grad_norm": 0.48944488167762756, | |
| "learning_rate": 3.5442581542201923e-06, | |
| "loss": 0.5955, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.07704438883865677, | |
| "grad_norm": 0.6073352098464966, | |
| "learning_rate": 3.461836116672612e-06, | |
| "loss": 0.7206, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.0771284984334588, | |
| "grad_norm": 0.6056821346282959, | |
| "learning_rate": 3.380366871828522e-06, | |
| "loss": 0.6869, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.07721260802826083, | |
| "grad_norm": 0.50712651014328, | |
| "learning_rate": 3.2998512237565005e-06, | |
| "loss": 0.5963, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.07729671762306285, | |
| "grad_norm": 0.5057475566864014, | |
| "learning_rate": 3.2202899671134546e-06, | |
| "loss": 0.7265, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.07738082721786488, | |
| "grad_norm": 0.4675261378288269, | |
| "learning_rate": 3.1416838871368924e-06, | |
| "loss": 0.4619, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.0774649368126669, | |
| "grad_norm": 0.5350446105003357, | |
| "learning_rate": 3.064033759637064e-06, | |
| "loss": 0.6337, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.07754904640746893, | |
| "grad_norm": 0.653113067150116, | |
| "learning_rate": 2.9873403509894203e-06, | |
| "loss": 0.6033, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.07763315600227096, | |
| "grad_norm": 0.6161055564880371, | |
| "learning_rate": 2.9116044181269007e-06, | |
| "loss": 0.5878, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.07771726559707298, | |
| "grad_norm": 0.569428563117981, | |
| "learning_rate": 2.836826708532603e-06, | |
| "loss": 0.6588, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.07780137519187501, | |
| "grad_norm": 0.7410022020339966, | |
| "learning_rate": 2.7630079602323442e-06, | |
| "loss": 0.4855, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.07788548478667703, | |
| "grad_norm": 0.6758790016174316, | |
| "learning_rate": 2.690148901787337e-06, | |
| "loss": 0.6881, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.07796959438147907, | |
| "grad_norm": 0.49047234654426575, | |
| "learning_rate": 2.618250252287113e-06, | |
| "loss": 0.5011, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.0780537039762811, | |
| "grad_norm": 0.5375692248344421, | |
| "learning_rate": 2.5473127213422763e-06, | |
| "loss": 0.7307, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.07813781357108313, | |
| "grad_norm": 0.581413745880127, | |
| "learning_rate": 2.4773370090776626e-06, | |
| "loss": 0.6392, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.07822192316588515, | |
| "grad_norm": 0.6516633033752441, | |
| "learning_rate": 2.4083238061252567e-06, | |
| "loss": 0.6592, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.07830603276068718, | |
| "grad_norm": 0.4725818634033203, | |
| "learning_rate": 2.3402737936175425e-06, | |
| "loss": 0.6819, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.0783901423554892, | |
| "grad_norm": 0.6635422110557556, | |
| "learning_rate": 2.273187643180652e-06, | |
| "loss": 0.6987, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.07847425195029123, | |
| "grad_norm": 0.5256288647651672, | |
| "learning_rate": 2.2070660169278166e-06, | |
| "loss": 0.5697, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.07855836154509326, | |
| "grad_norm": 0.4704425036907196, | |
| "learning_rate": 2.141909567452793e-06, | |
| "loss": 0.4441, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.07864247113989528, | |
| "grad_norm": 0.4934440553188324, | |
| "learning_rate": 2.0777189378234143e-06, | |
| "loss": 0.6377, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.07872658073469731, | |
| "grad_norm": 0.5220550298690796, | |
| "learning_rate": 2.014494761575314e-06, | |
| "loss": 0.5492, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.07881069032949933, | |
| "grad_norm": 0.7263022065162659, | |
| "learning_rate": 1.9522376627055583e-06, | |
| "loss": 0.6481, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.07889479992430136, | |
| "grad_norm": 0.6368206739425659, | |
| "learning_rate": 1.8909482556666024e-06, | |
| "loss": 0.5843, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.07897890951910339, | |
| "grad_norm": 0.522922933101654, | |
| "learning_rate": 1.8306271453601199e-06, | |
| "loss": 0.7028, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.07906301911390543, | |
| "grad_norm": 0.5848362445831299, | |
| "learning_rate": 1.771274927131139e-06, | |
| "loss": 0.7176, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.07914712870870745, | |
| "grad_norm": 0.5615982413291931, | |
| "learning_rate": 1.712892186762083e-06, | |
| "loss": 0.5697, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.07923123830350948, | |
| "grad_norm": 0.4720976948738098, | |
| "learning_rate": 1.6554795004670388e-06, | |
| "loss": 0.4611, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.0793153478983115, | |
| "grad_norm": 0.5845123529434204, | |
| "learning_rate": 1.5990374348860305e-06, | |
| "loss": 0.6374, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.07939945749311353, | |
| "grad_norm": 0.5101219415664673, | |
| "learning_rate": 1.543566547079467e-06, | |
| "loss": 0.551, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.07948356708791555, | |
| "grad_norm": 0.6655003428459167, | |
| "learning_rate": 1.4890673845226133e-06, | |
| "loss": 0.7873, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.07956767668271758, | |
| "grad_norm": 0.552952766418457, | |
| "learning_rate": 1.4355404851001952e-06, | |
| "loss": 0.7553, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.0796517862775196, | |
| "grad_norm": 0.37656185030937195, | |
| "learning_rate": 1.3829863771011253e-06, | |
| "loss": 0.5656, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.07973589587232163, | |
| "grad_norm": 0.7558304667472839, | |
| "learning_rate": 1.3314055792131964e-06, | |
| "loss": 0.6068, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.07982000546712366, | |
| "grad_norm": 0.584898054599762, | |
| "learning_rate": 1.280798600518085e-06, | |
| "loss": 0.5412, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.07990411506192568, | |
| "grad_norm": 0.542145848274231, | |
| "learning_rate": 1.231165940486234e-06, | |
| "loss": 0.563, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.07998822465672771, | |
| "grad_norm": 0.6324697732925415, | |
| "learning_rate": 1.1825080889719563e-06, | |
| "loss": 0.5728, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.08007233425152974, | |
| "grad_norm": 0.5715826749801636, | |
| "learning_rate": 1.134825526208605e-06, | |
| "loss": 0.6769, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.08015644384633178, | |
| "grad_norm": 0.5874225497245789, | |
| "learning_rate": 1.0881187228038215e-06, | |
| "loss": 0.579, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.0802405534411338, | |
| "grad_norm": 0.5252524018287659, | |
| "learning_rate": 1.0423881397349068e-06, | |
| "loss": 0.6326, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.08032466303593583, | |
| "grad_norm": 0.6097508668899536, | |
| "learning_rate": 9.976342283442463e-07, | |
| "loss": 0.6524, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.08040877263073785, | |
| "grad_norm": 0.41464993357658386, | |
| "learning_rate": 9.538574303348813e-07, | |
| "loss": 0.4743, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.08049288222553988, | |
| "grad_norm": 0.46832600235939026, | |
| "learning_rate": 9.110581777661331e-07, | |
| "loss": 0.5314, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.0805769918203419, | |
| "grad_norm": 0.5717182159423828, | |
| "learning_rate": 8.692368930493521e-07, | |
| "loss": 0.7329, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.08066110141514393, | |
| "grad_norm": 0.5042733550071716, | |
| "learning_rate": 8.283939889437209e-07, | |
| "loss": 0.7913, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.08074521100994596, | |
| "grad_norm": 0.626025915145874, | |
| "learning_rate": 7.885298685522235e-07, | |
| "loss": 0.7545, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.08082932060474798, | |
| "grad_norm": 0.5646237730979919, | |
| "learning_rate": 7.496449253176274e-07, | |
| "loss": 0.5513, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.08091343019955001, | |
| "grad_norm": 0.6173761487007141, | |
| "learning_rate": 7.117395430186414e-07, | |
| "loss": 0.6928, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.08099753979435204, | |
| "grad_norm": 0.5799218416213989, | |
| "learning_rate": 6.748140957660631e-07, | |
| "loss": 0.8141, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.08108164938915406, | |
| "grad_norm": 0.5109354853630066, | |
| "learning_rate": 6.388689479991605e-07, | |
| "loss": 0.8059, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.0811657589839561, | |
| "grad_norm": 0.5159839391708374, | |
| "learning_rate": 6.039044544820404e-07, | |
| "loss": 0.5749, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.08124986857875813, | |
| "grad_norm": 0.6682405471801758, | |
| "learning_rate": 5.699209603001076e-07, | |
| "loss": 0.7188, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.08133397817356015, | |
| "grad_norm": 0.8375939130783081, | |
| "learning_rate": 5.369188008567672e-07, | |
| "loss": 0.9425, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.08141808776836218, | |
| "grad_norm": 0.6056286692619324, | |
| "learning_rate": 5.048983018699827e-07, | |
| "loss": 0.6488, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.0815021973631642, | |
| "grad_norm": 0.5580015778541565, | |
| "learning_rate": 4.738597793691679e-07, | |
| "loss": 0.5895, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.08158630695796623, | |
| "grad_norm": 0.5345575213432312, | |
| "learning_rate": 4.438035396920004e-07, | |
| "loss": 0.6993, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.08167041655276826, | |
| "grad_norm": 0.5608043074607849, | |
| "learning_rate": 4.1472987948143473e-07, | |
| "loss": 0.7225, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.08175452614757028, | |
| "grad_norm": 0.5627851486206055, | |
| "learning_rate": 3.866390856827495e-07, | |
| "loss": 0.6505, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.08183863574237231, | |
| "grad_norm": 0.7156445384025574, | |
| "learning_rate": 3.595314355407609e-07, | |
| "loss": 0.8306, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.08192274533717434, | |
| "grad_norm": 0.8838444352149963, | |
| "learning_rate": 3.3340719659701313e-07, | |
| "loss": 0.5709, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.08200685493197636, | |
| "grad_norm": 0.5641219615936279, | |
| "learning_rate": 3.0826662668720364e-07, | |
| "loss": 0.7521, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.08209096452677839, | |
| "grad_norm": 0.636716902256012, | |
| "learning_rate": 2.841099739386066e-07, | |
| "loss": 0.6791, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.08217507412158041, | |
| "grad_norm": 0.5187475681304932, | |
| "learning_rate": 2.609374767676309e-07, | |
| "loss": 0.5493, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.08225918371638245, | |
| "grad_norm": 0.555397629737854, | |
| "learning_rate": 2.387493638774774e-07, | |
| "loss": 0.8756, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.08234329331118448, | |
| "grad_norm": 0.4511655271053314, | |
| "learning_rate": 2.175458542558517e-07, | |
| "loss": 0.43, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.0824274029059865, | |
| "grad_norm": 0.7089412212371826, | |
| "learning_rate": 1.973271571728441e-07, | |
| "loss": 0.6912, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.08251151250078853, | |
| "grad_norm": 0.46385011076927185, | |
| "learning_rate": 1.7809347217881966e-07, | |
| "loss": 0.6673, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.08259562209559056, | |
| "grad_norm": 0.5081963539123535, | |
| "learning_rate": 1.598449891024978e-07, | |
| "loss": 0.6091, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.08267973169039258, | |
| "grad_norm": 0.4684099853038788, | |
| "learning_rate": 1.425818880490315e-07, | |
| "loss": 0.6555, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.08276384128519461, | |
| "grad_norm": 0.5321608185768127, | |
| "learning_rate": 1.2630433939825327e-07, | |
| "loss": 0.6384, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.08284795087999663, | |
| "grad_norm": 0.5801177620887756, | |
| "learning_rate": 1.1101250380300965e-07, | |
| "loss": 0.6731, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.08293206047479866, | |
| "grad_norm": 0.5255969166755676, | |
| "learning_rate": 9.670653218752934e-08, | |
| "loss": 0.6411, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.08301617006960069, | |
| "grad_norm": 0.5361884236335754, | |
| "learning_rate": 8.33865657459909e-08, | |
| "loss": 0.4983, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.08310027966440271, | |
| "grad_norm": 0.42637842893600464, | |
| "learning_rate": 7.105273594107953e-08, | |
| "loss": 0.501, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.08318438925920474, | |
| "grad_norm": 0.5724627375602722, | |
| "learning_rate": 5.970516450271025e-08, | |
| "loss": 0.7244, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.08326849885400676, | |
| "grad_norm": 0.5623000264167786, | |
| "learning_rate": 4.934396342684e-08, | |
| "loss": 0.6702, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.0833526084488088, | |
| "grad_norm": 0.505596399307251, | |
| "learning_rate": 3.996923497434635e-08, | |
| "loss": 0.7967, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.08343671804361083, | |
| "grad_norm": 0.7859603762626648, | |
| "learning_rate": 3.1581071670006015e-08, | |
| "loss": 0.7663, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.08352082763841286, | |
| "grad_norm": 0.5909059643745422, | |
| "learning_rate": 2.417955630159563e-08, | |
| "loss": 0.6006, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.08360493723321488, | |
| "grad_norm": 0.4647238552570343, | |
| "learning_rate": 1.7764761919103477e-08, | |
| "loss": 0.6949, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.08368904682801691, | |
| "grad_norm": 0.5446833372116089, | |
| "learning_rate": 1.2336751833941229e-08, | |
| "loss": 0.5849, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.08377315642281893, | |
| "grad_norm": 0.5875367522239685, | |
| "learning_rate": 7.895579618388827e-09, | |
| "loss": 0.769, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.08385726601762096, | |
| "grad_norm": 0.769709587097168, | |
| "learning_rate": 4.4412891050171765e-09, | |
| "loss": 0.8165, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.08394137561242299, | |
| "grad_norm": 0.5166672468185425, | |
| "learning_rate": 1.973914386288467e-09, | |
| "loss": 0.696, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.08402548520722501, | |
| "grad_norm": 0.6133189797401428, | |
| "learning_rate": 4.934798141786879e-10, | |
| "loss": 0.7572, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.08410959480202704, | |
| "grad_norm": 0.4853799045085907, | |
| "learning_rate": 0.0, | |
| "loss": 0.6055, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.08410959480202704, | |
| "eval_loss": 0.6270027160644531, | |
| "eval_runtime": 118.2303, | |
| "eval_samples_per_second": 21.179, | |
| "eval_steps_per_second": 21.179, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "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": 5.774858388504576e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |