| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.944596361907027, | |
| "eval_steps": 500, | |
| "global_step": 50000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.059445963619070265, | |
| "grad_norm": 2.0723960399627686, | |
| "learning_rate": 4.985138509095233e-05, | |
| "loss": 3.6788, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11889192723814053, | |
| "grad_norm": 1.9278995990753174, | |
| "learning_rate": 4.970277018190465e-05, | |
| "loss": 3.4742, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1783378908572108, | |
| "grad_norm": 1.4848977327346802, | |
| "learning_rate": 4.955415527285698e-05, | |
| "loss": 3.3942, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.23778385447628106, | |
| "grad_norm": 1.3492341041564941, | |
| "learning_rate": 4.94055403638093e-05, | |
| "loss": 3.3358, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2972298180953513, | |
| "grad_norm": 1.212128758430481, | |
| "learning_rate": 4.925692545476163e-05, | |
| "loss": 3.2851, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.3566757817144216, | |
| "grad_norm": 1.1597293615341187, | |
| "learning_rate": 4.9108310545713945e-05, | |
| "loss": 3.2331, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.41612174533349183, | |
| "grad_norm": 0.9653922319412231, | |
| "learning_rate": 4.8959695636666275e-05, | |
| "loss": 3.2339, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.4755677089525621, | |
| "grad_norm": 1.0085793733596802, | |
| "learning_rate": 4.88110807276186e-05, | |
| "loss": 3.1856, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.5350136725716323, | |
| "grad_norm": 1.0556505918502808, | |
| "learning_rate": 4.866246581857092e-05, | |
| "loss": 3.1748, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.5944596361907026, | |
| "grad_norm": 0.9526228904724121, | |
| "learning_rate": 4.851385090952324e-05, | |
| "loss": 3.1529, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6539055998097729, | |
| "grad_norm": 0.984980046749115, | |
| "learning_rate": 4.836523600047557e-05, | |
| "loss": 3.1378, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.7133515634288432, | |
| "grad_norm": 1.0135027170181274, | |
| "learning_rate": 4.8216621091427895e-05, | |
| "loss": 3.0848, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.7727975270479135, | |
| "grad_norm": 0.9454924464225769, | |
| "learning_rate": 4.806800618238022e-05, | |
| "loss": 3.0916, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.8322434906669837, | |
| "grad_norm": 0.9793129563331604, | |
| "learning_rate": 4.791939127333254e-05, | |
| "loss": 3.0642, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.891689454286054, | |
| "grad_norm": 0.9016062617301941, | |
| "learning_rate": 4.777077636428487e-05, | |
| "loss": 3.0657, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.9511354179051242, | |
| "grad_norm": 0.8690605163574219, | |
| "learning_rate": 4.762216145523719e-05, | |
| "loss": 3.0281, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.0105813815241944, | |
| "grad_norm": 0.891808271408081, | |
| "learning_rate": 4.7473546546189516e-05, | |
| "loss": 3.0155, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.0700273451432647, | |
| "grad_norm": 0.9521974325180054, | |
| "learning_rate": 4.732493163714184e-05, | |
| "loss": 2.9713, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.129473308762335, | |
| "grad_norm": 0.9132643938064575, | |
| "learning_rate": 4.717631672809417e-05, | |
| "loss": 2.9663, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.1889192723814053, | |
| "grad_norm": 0.909182608127594, | |
| "learning_rate": 4.702770181904649e-05, | |
| "loss": 2.9616, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.2483652360004756, | |
| "grad_norm": 0.912726104259491, | |
| "learning_rate": 4.687908690999881e-05, | |
| "loss": 2.9653, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.3078111996195458, | |
| "grad_norm": 0.8568936586380005, | |
| "learning_rate": 4.6730472000951136e-05, | |
| "loss": 2.9486, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.3672571632386161, | |
| "grad_norm": 0.9120291471481323, | |
| "learning_rate": 4.6581857091903465e-05, | |
| "loss": 2.932, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.4267031268576864, | |
| "grad_norm": 0.981961190700531, | |
| "learning_rate": 4.643324218285579e-05, | |
| "loss": 2.9345, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.4861490904767567, | |
| "grad_norm": 0.9763424396514893, | |
| "learning_rate": 4.628462727380811e-05, | |
| "loss": 2.9193, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.545595054095827, | |
| "grad_norm": 0.8868328332901001, | |
| "learning_rate": 4.6136012364760434e-05, | |
| "loss": 2.9164, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.605041017714897, | |
| "grad_norm": 0.9175488352775574, | |
| "learning_rate": 4.598739745571276e-05, | |
| "loss": 2.8932, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.6644869813339676, | |
| "grad_norm": 0.890186607837677, | |
| "learning_rate": 4.583878254666508e-05, | |
| "loss": 2.8933, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.7239329449530376, | |
| "grad_norm": 0.9198343753814697, | |
| "learning_rate": 4.569016763761741e-05, | |
| "loss": 2.881, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.783378908572108, | |
| "grad_norm": 0.9706104397773743, | |
| "learning_rate": 4.554155272856973e-05, | |
| "loss": 2.8705, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.8428248721911782, | |
| "grad_norm": 0.9355807304382324, | |
| "learning_rate": 4.539293781952206e-05, | |
| "loss": 2.8601, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.9022708358102485, | |
| "grad_norm": 0.8972137570381165, | |
| "learning_rate": 4.524432291047438e-05, | |
| "loss": 2.8632, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.9617167994293188, | |
| "grad_norm": 0.8553013801574707, | |
| "learning_rate": 4.5095708001426706e-05, | |
| "loss": 2.8696, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.021162763048389, | |
| "grad_norm": 0.8952363133430481, | |
| "learning_rate": 4.494709309237903e-05, | |
| "loss": 2.8541, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.0806087266674593, | |
| "grad_norm": 0.8947279453277588, | |
| "learning_rate": 4.479847818333135e-05, | |
| "loss": 2.8203, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.1400546902865294, | |
| "grad_norm": 0.8680304884910583, | |
| "learning_rate": 4.4649863274283674e-05, | |
| "loss": 2.8088, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.1995006539056, | |
| "grad_norm": 0.8425644040107727, | |
| "learning_rate": 4.4501248365236004e-05, | |
| "loss": 2.8064, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.25894661752467, | |
| "grad_norm": 0.9474213719367981, | |
| "learning_rate": 4.4352633456188327e-05, | |
| "loss": 2.7851, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.3183925811437405, | |
| "grad_norm": 0.9292487502098083, | |
| "learning_rate": 4.420401854714065e-05, | |
| "loss": 2.8062, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.3778385447628105, | |
| "grad_norm": 0.8527488708496094, | |
| "learning_rate": 4.405540363809297e-05, | |
| "loss": 2.7851, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.437284508381881, | |
| "grad_norm": 0.9439261555671692, | |
| "learning_rate": 4.39067887290453e-05, | |
| "loss": 2.7873, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.496730472000951, | |
| "grad_norm": 0.9343836903572083, | |
| "learning_rate": 4.3758173819997624e-05, | |
| "loss": 2.7611, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.5561764356200216, | |
| "grad_norm": 0.9050599932670593, | |
| "learning_rate": 4.360955891094995e-05, | |
| "loss": 2.767, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.6156223992390917, | |
| "grad_norm": 0.9053699374198914, | |
| "learning_rate": 4.346094400190227e-05, | |
| "loss": 2.7873, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.6750683628581617, | |
| "grad_norm": 0.9282116293907166, | |
| "learning_rate": 4.33123290928546e-05, | |
| "loss": 2.7607, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.7345143264772322, | |
| "grad_norm": 0.9617480635643005, | |
| "learning_rate": 4.316371418380692e-05, | |
| "loss": 2.7678, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.7939602900963023, | |
| "grad_norm": 0.9725137948989868, | |
| "learning_rate": 4.3015099274759244e-05, | |
| "loss": 2.7665, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.853406253715373, | |
| "grad_norm": 0.9514666199684143, | |
| "learning_rate": 4.286648436571157e-05, | |
| "loss": 2.7534, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.912852217334443, | |
| "grad_norm": 0.9485461115837097, | |
| "learning_rate": 4.27178694566639e-05, | |
| "loss": 2.7306, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.9722981809535134, | |
| "grad_norm": 1.014106035232544, | |
| "learning_rate": 4.256925454761622e-05, | |
| "loss": 2.736, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 3.0317441445725835, | |
| "grad_norm": 0.9117903113365173, | |
| "learning_rate": 4.242063963856854e-05, | |
| "loss": 2.7278, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 3.091190108191654, | |
| "grad_norm": 0.8904880881309509, | |
| "learning_rate": 4.2272024729520865e-05, | |
| "loss": 2.7156, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 3.150636071810724, | |
| "grad_norm": 0.8653568625450134, | |
| "learning_rate": 4.2123409820473194e-05, | |
| "loss": 2.7137, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 3.210082035429794, | |
| "grad_norm": 0.9386480450630188, | |
| "learning_rate": 4.197479491142551e-05, | |
| "loss": 2.7021, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 3.2695279990488646, | |
| "grad_norm": 1.0122427940368652, | |
| "learning_rate": 4.182618000237784e-05, | |
| "loss": 2.699, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 3.3289739626679347, | |
| "grad_norm": 0.9319558143615723, | |
| "learning_rate": 4.167756509333017e-05, | |
| "loss": 2.689, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 3.388419926287005, | |
| "grad_norm": 0.9281746745109558, | |
| "learning_rate": 4.152895018428249e-05, | |
| "loss": 2.7027, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 3.4478658899060752, | |
| "grad_norm": 0.9750462770462036, | |
| "learning_rate": 4.1380335275234815e-05, | |
| "loss": 2.6947, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 3.5073118535251457, | |
| "grad_norm": 0.8887720704078674, | |
| "learning_rate": 4.123172036618714e-05, | |
| "loss": 2.6864, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 3.566757817144216, | |
| "grad_norm": 0.9884176254272461, | |
| "learning_rate": 4.108310545713947e-05, | |
| "loss": 2.6893, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 3.6262037807632863, | |
| "grad_norm": 0.9995080828666687, | |
| "learning_rate": 4.093449054809178e-05, | |
| "loss": 2.6734, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 3.6856497443823564, | |
| "grad_norm": 1.0068608522415161, | |
| "learning_rate": 4.078587563904411e-05, | |
| "loss": 2.6766, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 3.7450957080014264, | |
| "grad_norm": 1.0225422382354736, | |
| "learning_rate": 4.0637260729996435e-05, | |
| "loss": 2.6757, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 3.804541671620497, | |
| "grad_norm": 0.9354658126831055, | |
| "learning_rate": 4.0488645820948765e-05, | |
| "loss": 2.6593, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 3.8639876352395675, | |
| "grad_norm": 0.9209592938423157, | |
| "learning_rate": 4.034003091190108e-05, | |
| "loss": 2.6547, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 3.9234335988586375, | |
| "grad_norm": 0.8945015668869019, | |
| "learning_rate": 4.019141600285341e-05, | |
| "loss": 2.6719, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 3.9828795624777076, | |
| "grad_norm": 0.9823748469352722, | |
| "learning_rate": 4.004280109380573e-05, | |
| "loss": 2.6781, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 4.042325526096778, | |
| "grad_norm": 1.0186822414398193, | |
| "learning_rate": 3.989418618475806e-05, | |
| "loss": 2.6469, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 4.101771489715849, | |
| "grad_norm": 0.9255732297897339, | |
| "learning_rate": 3.974557127571038e-05, | |
| "loss": 2.6296, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 4.161217453334919, | |
| "grad_norm": 1.0235294103622437, | |
| "learning_rate": 3.959695636666271e-05, | |
| "loss": 2.6358, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 4.220663416953989, | |
| "grad_norm": 0.911547064781189, | |
| "learning_rate": 3.944834145761503e-05, | |
| "loss": 2.6354, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 4.280109380573059, | |
| "grad_norm": 1.0124516487121582, | |
| "learning_rate": 3.929972654856735e-05, | |
| "loss": 2.6416, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 4.33955534419213, | |
| "grad_norm": 1.0222316980361938, | |
| "learning_rate": 3.9151111639519676e-05, | |
| "loss": 2.6188, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 4.3990013078112, | |
| "grad_norm": 0.9710135459899902, | |
| "learning_rate": 3.9002496730472005e-05, | |
| "loss": 2.6228, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 4.45844727143027, | |
| "grad_norm": 1.0287182331085205, | |
| "learning_rate": 3.885388182142433e-05, | |
| "loss": 2.6067, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 4.51789323504934, | |
| "grad_norm": 0.9699456095695496, | |
| "learning_rate": 3.870526691237665e-05, | |
| "loss": 2.6385, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 4.57733919866841, | |
| "grad_norm": 0.9066009521484375, | |
| "learning_rate": 3.855665200332897e-05, | |
| "loss": 2.6284, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 4.636785162287481, | |
| "grad_norm": 0.8537769317626953, | |
| "learning_rate": 3.84080370942813e-05, | |
| "loss": 2.6135, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 4.696231125906551, | |
| "grad_norm": 1.0666980743408203, | |
| "learning_rate": 3.8259422185233626e-05, | |
| "loss": 2.6312, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 4.755677089525621, | |
| "grad_norm": 1.0641474723815918, | |
| "learning_rate": 3.811080727618595e-05, | |
| "loss": 2.6127, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 4.815123053144691, | |
| "grad_norm": 1.076323390007019, | |
| "learning_rate": 3.796219236713827e-05, | |
| "loss": 2.6184, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 4.874569016763762, | |
| "grad_norm": 0.8963558077812195, | |
| "learning_rate": 3.78135774580906e-05, | |
| "loss": 2.6165, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 4.934014980382832, | |
| "grad_norm": 0.968908429145813, | |
| "learning_rate": 3.766496254904292e-05, | |
| "loss": 2.6009, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 4.993460944001902, | |
| "grad_norm": 0.9362033605575562, | |
| "learning_rate": 3.7516347639995246e-05, | |
| "loss": 2.5956, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 5.052906907620972, | |
| "grad_norm": 1.1101199388504028, | |
| "learning_rate": 3.736773273094757e-05, | |
| "loss": 2.5755, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 5.112352871240043, | |
| "grad_norm": 1.2178868055343628, | |
| "learning_rate": 3.72191178218999e-05, | |
| "loss": 2.5724, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 5.171798834859113, | |
| "grad_norm": 1.0143418312072754, | |
| "learning_rate": 3.707050291285222e-05, | |
| "loss": 2.5834, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 5.231244798478183, | |
| "grad_norm": 0.9720271825790405, | |
| "learning_rate": 3.6921888003804544e-05, | |
| "loss": 2.586, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 5.290690762097253, | |
| "grad_norm": 0.8847070932388306, | |
| "learning_rate": 3.6773273094756866e-05, | |
| "loss": 2.5953, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 5.3501367257163235, | |
| "grad_norm": 0.9654759764671326, | |
| "learning_rate": 3.6624658185709196e-05, | |
| "loss": 2.5777, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 5.409582689335394, | |
| "grad_norm": 0.9272730350494385, | |
| "learning_rate": 3.647604327666151e-05, | |
| "loss": 2.5774, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 5.4690286529544645, | |
| "grad_norm": 0.9674676656723022, | |
| "learning_rate": 3.632742836761384e-05, | |
| "loss": 2.5779, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 5.528474616573535, | |
| "grad_norm": 1.0238367319107056, | |
| "learning_rate": 3.6178813458566164e-05, | |
| "loss": 2.5683, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 5.587920580192605, | |
| "grad_norm": 1.1663753986358643, | |
| "learning_rate": 3.603019854951849e-05, | |
| "loss": 2.5802, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 5.647366543811675, | |
| "grad_norm": 0.8961432576179504, | |
| "learning_rate": 3.588158364047081e-05, | |
| "loss": 2.5726, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 5.706812507430746, | |
| "grad_norm": 1.1115467548370361, | |
| "learning_rate": 3.573296873142314e-05, | |
| "loss": 2.5719, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 5.766258471049816, | |
| "grad_norm": 1.00434148311615, | |
| "learning_rate": 3.558435382237546e-05, | |
| "loss": 2.556, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 5.825704434668886, | |
| "grad_norm": 1.1120518445968628, | |
| "learning_rate": 3.5435738913327784e-05, | |
| "loss": 2.5627, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 5.885150398287957, | |
| "grad_norm": 0.9611983299255371, | |
| "learning_rate": 3.528712400428011e-05, | |
| "loss": 2.5568, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 5.944596361907027, | |
| "grad_norm": 1.1176481246948242, | |
| "learning_rate": 3.5138509095232436e-05, | |
| "loss": 2.5634, | |
| "step": 50000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 168220, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 2.090205609984e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |