diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4334 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 1, + "global_step": 61188, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.004902922141596391, + "grad_norm": 0.5712823271751404, + "learning_rate": 4.902922141596392e-07, + "loss": 0.3911, + "step": 100 + }, + { + "epoch": 0.009805844283192783, + "grad_norm": 0.7707542181015015, + "learning_rate": 9.805844283192783e-07, + "loss": 0.3924, + "step": 200 + }, + { + "epoch": 0.014708766424789174, + "grad_norm": 0.7901243567466736, + "learning_rate": 1.4708766424789175e-06, + "loss": 0.338, + "step": 300 + }, + { + "epoch": 0.019611688566385566, + "grad_norm": 0.4191139042377472, + "learning_rate": 1.9611688566385566e-06, + "loss": 0.2563, + "step": 400 + }, + { + "epoch": 0.024514610707981957, + "grad_norm": 0.5541927218437195, + "learning_rate": 2.4514610707981958e-06, + "loss": 0.1763, + "step": 500 + }, + { + "epoch": 0.02941753284957835, + "grad_norm": 0.5028297305107117, + "learning_rate": 2.941753284957835e-06, + "loss": 0.1511, + "step": 600 + }, + { + "epoch": 0.03432045499117474, + "grad_norm": 0.6625041961669922, + "learning_rate": 3.432045499117474e-06, + "loss": 0.142, + "step": 700 + }, + { + "epoch": 0.03922337713277113, + "grad_norm": 0.42951473593711853, + "learning_rate": 3.922337713277113e-06, + "loss": 0.1381, + "step": 800 + }, + { + "epoch": 0.04412629927436752, + "grad_norm": 0.6672629714012146, + "learning_rate": 4.412629927436753e-06, + "loss": 0.132, + "step": 900 + }, + { + "epoch": 0.049029221415963914, + "grad_norm": 0.4937724173069, + "learning_rate": 4.9029221415963915e-06, + "loss": 0.1418, + "step": 1000 + }, + { + "epoch": 0.053932143557560305, + "grad_norm": 0.3915172815322876, + "learning_rate": 5.393214355756031e-06, + "loss": 0.1254, + "step": 1100 + }, + { + "epoch": 0.0588350656991567, + "grad_norm": 0.39539510011672974, + "learning_rate": 5.88350656991567e-06, + "loss": 0.1252, + "step": 1200 + }, + { + "epoch": 0.06373798784075309, + "grad_norm": 0.7038196325302124, + "learning_rate": 6.373798784075309e-06, + "loss": 0.121, + "step": 1300 + }, + { + "epoch": 0.06864090998234948, + "grad_norm": 0.6034905910491943, + "learning_rate": 6.864090998234948e-06, + "loss": 0.1231, + "step": 1400 + }, + { + "epoch": 0.07354383212394587, + "grad_norm": 0.5692676305770874, + "learning_rate": 7.354383212394588e-06, + "loss": 0.1138, + "step": 1500 + }, + { + "epoch": 0.07844675426554226, + "grad_norm": 0.8084454536437988, + "learning_rate": 7.844675426554226e-06, + "loss": 0.1149, + "step": 1600 + }, + { + "epoch": 0.08334967640713865, + "grad_norm": 0.4154703915119171, + "learning_rate": 8.334967640713865e-06, + "loss": 0.1069, + "step": 1700 + }, + { + "epoch": 0.08825259854873505, + "grad_norm": 0.8582517504692078, + "learning_rate": 8.825259854873506e-06, + "loss": 0.1121, + "step": 1800 + }, + { + "epoch": 0.09315552069033144, + "grad_norm": 0.5646160244941711, + "learning_rate": 9.315552069033144e-06, + "loss": 0.1018, + "step": 1900 + }, + { + "epoch": 0.09805844283192783, + "grad_norm": 0.5309118032455444, + "learning_rate": 9.805844283192783e-06, + "loss": 0.1027, + "step": 2000 + }, + { + "epoch": 0.10296136497352422, + "grad_norm": 0.6853476166725159, + "learning_rate": 1.0296136497352422e-05, + "loss": 0.1084, + "step": 2100 + }, + { + "epoch": 0.10786428711512061, + "grad_norm": 0.6206139922142029, + "learning_rate": 1.0786428711512062e-05, + "loss": 0.1008, + "step": 2200 + }, + { + "epoch": 0.112767209256717, + "grad_norm": 0.5461515784263611, + "learning_rate": 1.1276720925671701e-05, + "loss": 0.0988, + "step": 2300 + }, + { + "epoch": 0.1176701313983134, + "grad_norm": 0.8931940793991089, + "learning_rate": 1.176701313983134e-05, + "loss": 0.0915, + "step": 2400 + }, + { + "epoch": 0.12257305353990978, + "grad_norm": 0.779230535030365, + "learning_rate": 1.2257305353990978e-05, + "loss": 0.0978, + "step": 2500 + }, + { + "epoch": 0.12747597568150618, + "grad_norm": 0.6850101947784424, + "learning_rate": 1.2747597568150619e-05, + "loss": 0.095, + "step": 2600 + }, + { + "epoch": 0.13237889782310258, + "grad_norm": 0.9090009331703186, + "learning_rate": 1.323788978231026e-05, + "loss": 0.0915, + "step": 2700 + }, + { + "epoch": 0.13728181996469896, + "grad_norm": 0.7030491828918457, + "learning_rate": 1.3728181996469896e-05, + "loss": 0.0884, + "step": 2800 + }, + { + "epoch": 0.14218474210629536, + "grad_norm": 0.7005199790000916, + "learning_rate": 1.4218474210629537e-05, + "loss": 0.0846, + "step": 2900 + }, + { + "epoch": 0.14708766424789174, + "grad_norm": 0.8862398266792297, + "learning_rate": 1.4708766424789175e-05, + "loss": 0.0871, + "step": 3000 + }, + { + "epoch": 0.15199058638948815, + "grad_norm": 0.667682409286499, + "learning_rate": 1.5199058638948816e-05, + "loss": 0.0852, + "step": 3100 + }, + { + "epoch": 0.15689350853108452, + "grad_norm": 0.6546317338943481, + "learning_rate": 1.5689350853108453e-05, + "loss": 0.0843, + "step": 3200 + }, + { + "epoch": 0.16179643067268093, + "grad_norm": 1.0147327184677124, + "learning_rate": 1.6179643067268095e-05, + "loss": 0.0824, + "step": 3300 + }, + { + "epoch": 0.1666993528142773, + "grad_norm": 0.9607380628585815, + "learning_rate": 1.666993528142773e-05, + "loss": 0.0783, + "step": 3400 + }, + { + "epoch": 0.1716022749558737, + "grad_norm": 0.6415452361106873, + "learning_rate": 1.7160227495587372e-05, + "loss": 0.0804, + "step": 3500 + }, + { + "epoch": 0.1765051970974701, + "grad_norm": 0.5318363904953003, + "learning_rate": 1.765051970974701e-05, + "loss": 0.0808, + "step": 3600 + }, + { + "epoch": 0.1814081192390665, + "grad_norm": 0.7230063080787659, + "learning_rate": 1.814081192390665e-05, + "loss": 0.0775, + "step": 3700 + }, + { + "epoch": 0.18631104138066287, + "grad_norm": 0.9152194857597351, + "learning_rate": 1.863110413806629e-05, + "loss": 0.0746, + "step": 3800 + }, + { + "epoch": 0.19121396352225928, + "grad_norm": 0.769027054309845, + "learning_rate": 1.9121396352225927e-05, + "loss": 0.0774, + "step": 3900 + }, + { + "epoch": 0.19611688566385566, + "grad_norm": 0.8306779861450195, + "learning_rate": 1.9611688566385566e-05, + "loss": 0.0735, + "step": 4000 + }, + { + "epoch": 0.20101980780545206, + "grad_norm": 0.502835750579834, + "learning_rate": 2.0101980780545208e-05, + "loss": 0.076, + "step": 4100 + }, + { + "epoch": 0.20592272994704844, + "grad_norm": 0.7486907839775085, + "learning_rate": 2.0592272994704843e-05, + "loss": 0.0786, + "step": 4200 + }, + { + "epoch": 0.21082565208864484, + "grad_norm": 0.6800108551979065, + "learning_rate": 2.1082565208864486e-05, + "loss": 0.0724, + "step": 4300 + }, + { + "epoch": 0.21572857423024122, + "grad_norm": 0.8575353026390076, + "learning_rate": 2.1572857423024124e-05, + "loss": 0.0726, + "step": 4400 + }, + { + "epoch": 0.22063149637183763, + "grad_norm": 1.0145362615585327, + "learning_rate": 2.2063149637183763e-05, + "loss": 0.0659, + "step": 4500 + }, + { + "epoch": 0.225534418513434, + "grad_norm": 0.7558209896087646, + "learning_rate": 2.2553441851343402e-05, + "loss": 0.0688, + "step": 4600 + }, + { + "epoch": 0.2304373406550304, + "grad_norm": 0.8182774186134338, + "learning_rate": 2.304373406550304e-05, + "loss": 0.0718, + "step": 4700 + }, + { + "epoch": 0.2353402627966268, + "grad_norm": 0.5569434762001038, + "learning_rate": 2.353402627966268e-05, + "loss": 0.0704, + "step": 4800 + }, + { + "epoch": 0.2402431849382232, + "grad_norm": 0.7403717041015625, + "learning_rate": 2.402431849382232e-05, + "loss": 0.0729, + "step": 4900 + }, + { + "epoch": 0.24514610707981957, + "grad_norm": 1.4814640283584595, + "learning_rate": 2.4514610707981957e-05, + "loss": 0.0731, + "step": 5000 + }, + { + "epoch": 0.250049029221416, + "grad_norm": 0.5557782649993896, + "learning_rate": 2.50049029221416e-05, + "loss": 0.0676, + "step": 5100 + }, + { + "epoch": 0.25495195136301235, + "grad_norm": 0.5674753785133362, + "learning_rate": 2.5495195136301238e-05, + "loss": 0.0675, + "step": 5200 + }, + { + "epoch": 0.25985487350460873, + "grad_norm": 0.8610398769378662, + "learning_rate": 2.5985487350460873e-05, + "loss": 0.0676, + "step": 5300 + }, + { + "epoch": 0.26475779564620516, + "grad_norm": 0.6987388134002686, + "learning_rate": 2.647577956462052e-05, + "loss": 0.0683, + "step": 5400 + }, + { + "epoch": 0.26966071778780154, + "grad_norm": 1.1044350862503052, + "learning_rate": 2.6966071778780154e-05, + "loss": 0.0684, + "step": 5500 + }, + { + "epoch": 0.2745636399293979, + "grad_norm": 0.7976281046867371, + "learning_rate": 2.7456363992939792e-05, + "loss": 0.0666, + "step": 5600 + }, + { + "epoch": 0.2794665620709943, + "grad_norm": 0.7889923453330994, + "learning_rate": 2.794665620709943e-05, + "loss": 0.0688, + "step": 5700 + }, + { + "epoch": 0.28436948421259073, + "grad_norm": 1.111566424369812, + "learning_rate": 2.8436948421259073e-05, + "loss": 0.0637, + "step": 5800 + }, + { + "epoch": 0.2892724063541871, + "grad_norm": 0.8932578563690186, + "learning_rate": 2.8927240635418712e-05, + "loss": 0.0636, + "step": 5900 + }, + { + "epoch": 0.2941753284957835, + "grad_norm": 1.0494166612625122, + "learning_rate": 2.941753284957835e-05, + "loss": 0.0641, + "step": 6000 + }, + { + "epoch": 0.29907825063737986, + "grad_norm": 0.7820980548858643, + "learning_rate": 2.9907825063737986e-05, + "loss": 0.0715, + "step": 6100 + }, + { + "epoch": 0.3039811727789763, + "grad_norm": 0.9710949659347534, + "learning_rate": 3.039811727789763e-05, + "loss": 0.066, + "step": 6200 + }, + { + "epoch": 0.30888409492057267, + "grad_norm": 0.807854950428009, + "learning_rate": 3.088840949205727e-05, + "loss": 0.0629, + "step": 6300 + }, + { + "epoch": 0.31378701706216905, + "grad_norm": 0.8420621156692505, + "learning_rate": 3.1378701706216906e-05, + "loss": 0.0692, + "step": 6400 + }, + { + "epoch": 0.3186899392037654, + "grad_norm": 0.776466965675354, + "learning_rate": 3.1868993920376544e-05, + "loss": 0.0625, + "step": 6500 + }, + { + "epoch": 0.32359286134536186, + "grad_norm": 0.8149374127388, + "learning_rate": 3.235928613453619e-05, + "loss": 0.0612, + "step": 6600 + }, + { + "epoch": 0.32849578348695824, + "grad_norm": 1.0844323635101318, + "learning_rate": 3.284957834869583e-05, + "loss": 0.0641, + "step": 6700 + }, + { + "epoch": 0.3333987056285546, + "grad_norm": 0.6500579118728638, + "learning_rate": 3.333987056285546e-05, + "loss": 0.0597, + "step": 6800 + }, + { + "epoch": 0.338301627770151, + "grad_norm": 0.7218449115753174, + "learning_rate": 3.38301627770151e-05, + "loss": 0.0603, + "step": 6900 + }, + { + "epoch": 0.3432045499117474, + "grad_norm": 1.0777571201324463, + "learning_rate": 3.4320454991174745e-05, + "loss": 0.0572, + "step": 7000 + }, + { + "epoch": 0.3481074720533438, + "grad_norm": 0.786324143409729, + "learning_rate": 3.4810747205334384e-05, + "loss": 0.06, + "step": 7100 + }, + { + "epoch": 0.3530103941949402, + "grad_norm": 0.44759607315063477, + "learning_rate": 3.530103941949402e-05, + "loss": 0.064, + "step": 7200 + }, + { + "epoch": 0.35791331633653656, + "grad_norm": 0.4074469208717346, + "learning_rate": 3.5791331633653654e-05, + "loss": 0.06, + "step": 7300 + }, + { + "epoch": 0.362816238478133, + "grad_norm": 0.9225257039070129, + "learning_rate": 3.62816238478133e-05, + "loss": 0.061, + "step": 7400 + }, + { + "epoch": 0.36771916061972937, + "grad_norm": 0.6077998280525208, + "learning_rate": 3.677191606197294e-05, + "loss": 0.053, + "step": 7500 + }, + { + "epoch": 0.37262208276132575, + "grad_norm": 0.9564809799194336, + "learning_rate": 3.726220827613258e-05, + "loss": 0.0558, + "step": 7600 + }, + { + "epoch": 0.3775250049029221, + "grad_norm": 0.6612119078636169, + "learning_rate": 3.7752500490292216e-05, + "loss": 0.0621, + "step": 7700 + }, + { + "epoch": 0.38242792704451856, + "grad_norm": 0.7367020845413208, + "learning_rate": 3.8242792704451855e-05, + "loss": 0.0596, + "step": 7800 + }, + { + "epoch": 0.38733084918611493, + "grad_norm": 0.595927357673645, + "learning_rate": 3.873308491861149e-05, + "loss": 0.0576, + "step": 7900 + }, + { + "epoch": 0.3922337713277113, + "grad_norm": 0.7036033272743225, + "learning_rate": 3.922337713277113e-05, + "loss": 0.059, + "step": 8000 + }, + { + "epoch": 0.3971366934693077, + "grad_norm": 0.8807573914527893, + "learning_rate": 3.971366934693077e-05, + "loss": 0.0587, + "step": 8100 + }, + { + "epoch": 0.4020396156109041, + "grad_norm": 0.8724229335784912, + "learning_rate": 4.0203961561090416e-05, + "loss": 0.0588, + "step": 8200 + }, + { + "epoch": 0.4069425377525005, + "grad_norm": 0.7618996500968933, + "learning_rate": 4.0694253775250055e-05, + "loss": 0.0576, + "step": 8300 + }, + { + "epoch": 0.4118454598940969, + "grad_norm": 0.743467390537262, + "learning_rate": 4.118454598940969e-05, + "loss": 0.0581, + "step": 8400 + }, + { + "epoch": 0.41674838203569325, + "grad_norm": 0.6151107549667358, + "learning_rate": 4.1674838203569326e-05, + "loss": 0.0633, + "step": 8500 + }, + { + "epoch": 0.4216513041772897, + "grad_norm": 1.3272536993026733, + "learning_rate": 4.216513041772897e-05, + "loss": 0.0579, + "step": 8600 + }, + { + "epoch": 0.42655422631888606, + "grad_norm": 0.7883033752441406, + "learning_rate": 4.265542263188861e-05, + "loss": 0.0579, + "step": 8700 + }, + { + "epoch": 0.43145714846048244, + "grad_norm": 0.6852431297302246, + "learning_rate": 4.314571484604825e-05, + "loss": 0.0594, + "step": 8800 + }, + { + "epoch": 0.4363600706020788, + "grad_norm": 0.5753522515296936, + "learning_rate": 4.363600706020789e-05, + "loss": 0.0564, + "step": 8900 + }, + { + "epoch": 0.44126299274367525, + "grad_norm": 0.510944128036499, + "learning_rate": 4.4126299274367526e-05, + "loss": 0.0571, + "step": 9000 + }, + { + "epoch": 0.44616591488527163, + "grad_norm": 0.7688953876495361, + "learning_rate": 4.4616591488527165e-05, + "loss": 0.0544, + "step": 9100 + }, + { + "epoch": 0.451068837026868, + "grad_norm": 0.6209575533866882, + "learning_rate": 4.5106883702686804e-05, + "loss": 0.054, + "step": 9200 + }, + { + "epoch": 0.4559717591684644, + "grad_norm": 0.42720097303390503, + "learning_rate": 4.559717591684644e-05, + "loss": 0.0581, + "step": 9300 + }, + { + "epoch": 0.4608746813100608, + "grad_norm": 0.43942296504974365, + "learning_rate": 4.608746813100608e-05, + "loss": 0.0527, + "step": 9400 + }, + { + "epoch": 0.4657776034516572, + "grad_norm": 0.8481646776199341, + "learning_rate": 4.657776034516572e-05, + "loss": 0.0553, + "step": 9500 + }, + { + "epoch": 0.4706805255932536, + "grad_norm": 0.670299232006073, + "learning_rate": 4.706805255932536e-05, + "loss": 0.0474, + "step": 9600 + }, + { + "epoch": 0.47558344773484995, + "grad_norm": 0.7548401951789856, + "learning_rate": 4.7558344773485e-05, + "loss": 0.0527, + "step": 9700 + }, + { + "epoch": 0.4804863698764464, + "grad_norm": 0.49902454018592834, + "learning_rate": 4.804863698764464e-05, + "loss": 0.0563, + "step": 9800 + }, + { + "epoch": 0.48538929201804276, + "grad_norm": 0.6898229718208313, + "learning_rate": 4.853892920180428e-05, + "loss": 0.054, + "step": 9900 + }, + { + "epoch": 0.49029221415963914, + "grad_norm": 0.7754454612731934, + "learning_rate": 4.902922141596391e-05, + "loss": 0.0569, + "step": 10000 + }, + { + "epoch": 0.4951951363012355, + "grad_norm": 0.5083482265472412, + "learning_rate": 4.951951363012355e-05, + "loss": 0.0524, + "step": 10100 + }, + { + "epoch": 0.500098058442832, + "grad_norm": 0.5904260277748108, + "learning_rate": 5.00098058442832e-05, + "loss": 0.0544, + "step": 10200 + }, + { + "epoch": 0.5050009805844283, + "grad_norm": 0.3116986155509949, + "learning_rate": 5.0500098058442836e-05, + "loss": 0.0569, + "step": 10300 + }, + { + "epoch": 0.5099039027260247, + "grad_norm": 0.4503382444381714, + "learning_rate": 5.0990390272602475e-05, + "loss": 0.0493, + "step": 10400 + }, + { + "epoch": 0.5148068248676211, + "grad_norm": 0.4754750728607178, + "learning_rate": 5.1480682486762114e-05, + "loss": 0.0549, + "step": 10500 + }, + { + "epoch": 0.5197097470092175, + "grad_norm": 0.5342922806739807, + "learning_rate": 5.1970974700921746e-05, + "loss": 0.0544, + "step": 10600 + }, + { + "epoch": 0.5246126691508138, + "grad_norm": 0.8138884902000427, + "learning_rate": 5.2461266915081384e-05, + "loss": 0.0524, + "step": 10700 + }, + { + "epoch": 0.5295155912924103, + "grad_norm": 0.8454159498214722, + "learning_rate": 5.295155912924104e-05, + "loss": 0.0508, + "step": 10800 + }, + { + "epoch": 0.5344185134340067, + "grad_norm": 0.5879427790641785, + "learning_rate": 5.3441851343400675e-05, + "loss": 0.0512, + "step": 10900 + }, + { + "epoch": 0.5393214355756031, + "grad_norm": 0.6766416430473328, + "learning_rate": 5.393214355756031e-05, + "loss": 0.0517, + "step": 11000 + }, + { + "epoch": 0.5442243577171995, + "grad_norm": 0.5555431842803955, + "learning_rate": 5.4422435771719946e-05, + "loss": 0.0528, + "step": 11100 + }, + { + "epoch": 0.5491272798587958, + "grad_norm": 0.43258973956108093, + "learning_rate": 5.4912727985879585e-05, + "loss": 0.0536, + "step": 11200 + }, + { + "epoch": 0.5540302020003922, + "grad_norm": 0.3768501877784729, + "learning_rate": 5.5403020200039224e-05, + "loss": 0.0514, + "step": 11300 + }, + { + "epoch": 0.5589331241419886, + "grad_norm": 0.4958087205886841, + "learning_rate": 5.589331241419886e-05, + "loss": 0.0524, + "step": 11400 + }, + { + "epoch": 0.563836046283585, + "grad_norm": 0.5648083686828613, + "learning_rate": 5.63836046283585e-05, + "loss": 0.0556, + "step": 11500 + }, + { + "epoch": 0.5687389684251815, + "grad_norm": 0.39568981528282166, + "learning_rate": 5.6873896842518147e-05, + "loss": 0.0497, + "step": 11600 + }, + { + "epoch": 0.5736418905667778, + "grad_norm": 0.3252398371696472, + "learning_rate": 5.7364189056677785e-05, + "loss": 0.0547, + "step": 11700 + }, + { + "epoch": 0.5785448127083742, + "grad_norm": 0.24218697845935822, + "learning_rate": 5.7854481270837424e-05, + "loss": 0.0448, + "step": 11800 + }, + { + "epoch": 0.5834477348499706, + "grad_norm": 0.4660034775733948, + "learning_rate": 5.834477348499706e-05, + "loss": 0.0509, + "step": 11900 + }, + { + "epoch": 0.588350656991567, + "grad_norm": 0.14836396276950836, + "learning_rate": 5.88350656991567e-05, + "loss": 0.0494, + "step": 12000 + }, + { + "epoch": 0.5932535791331633, + "grad_norm": 0.3045461177825928, + "learning_rate": 5.932535791331634e-05, + "loss": 0.0524, + "step": 12100 + }, + { + "epoch": 0.5981565012747597, + "grad_norm": 0.3134712874889374, + "learning_rate": 5.981565012747597e-05, + "loss": 0.0461, + "step": 12200 + }, + { + "epoch": 0.6030594234163561, + "grad_norm": 0.3141169250011444, + "learning_rate": 6.030594234163561e-05, + "loss": 0.0511, + "step": 12300 + }, + { + "epoch": 0.6079623455579526, + "grad_norm": 0.22318390011787415, + "learning_rate": 6.079623455579526e-05, + "loss": 0.0495, + "step": 12400 + }, + { + "epoch": 0.612865267699549, + "grad_norm": 0.5111287832260132, + "learning_rate": 6.12865267699549e-05, + "loss": 0.0532, + "step": 12500 + }, + { + "epoch": 0.6177681898411453, + "grad_norm": 0.3595598042011261, + "learning_rate": 6.177681898411453e-05, + "loss": 0.0482, + "step": 12600 + }, + { + "epoch": 0.6226711119827417, + "grad_norm": 0.294145405292511, + "learning_rate": 6.226711119827417e-05, + "loss": 0.047, + "step": 12700 + }, + { + "epoch": 0.6275740341243381, + "grad_norm": 0.39084282517433167, + "learning_rate": 6.275740341243381e-05, + "loss": 0.0509, + "step": 12800 + }, + { + "epoch": 0.6324769562659345, + "grad_norm": 0.36329174041748047, + "learning_rate": 6.324769562659345e-05, + "loss": 0.0466, + "step": 12900 + }, + { + "epoch": 0.6373798784075309, + "grad_norm": 0.48088952898979187, + "learning_rate": 6.373798784075309e-05, + "loss": 0.049, + "step": 13000 + }, + { + "epoch": 0.6422828005491272, + "grad_norm": 0.3504548966884613, + "learning_rate": 6.422828005491273e-05, + "loss": 0.0565, + "step": 13100 + }, + { + "epoch": 0.6471857226907237, + "grad_norm": 0.4509230852127075, + "learning_rate": 6.471857226907238e-05, + "loss": 0.0476, + "step": 13200 + }, + { + "epoch": 0.6520886448323201, + "grad_norm": 0.49520692229270935, + "learning_rate": 6.520886448323202e-05, + "loss": 0.0495, + "step": 13300 + }, + { + "epoch": 0.6569915669739165, + "grad_norm": 0.2963676154613495, + "learning_rate": 6.569915669739166e-05, + "loss": 0.0486, + "step": 13400 + }, + { + "epoch": 0.6618944891155129, + "grad_norm": 0.279095321893692, + "learning_rate": 6.618944891155128e-05, + "loss": 0.0495, + "step": 13500 + }, + { + "epoch": 0.6667974112571092, + "grad_norm": 0.3021456003189087, + "learning_rate": 6.667974112571092e-05, + "loss": 0.0497, + "step": 13600 + }, + { + "epoch": 0.6717003333987056, + "grad_norm": 0.49535706639289856, + "learning_rate": 6.717003333987056e-05, + "loss": 0.0501, + "step": 13700 + }, + { + "epoch": 0.676603255540302, + "grad_norm": 0.38932666182518005, + "learning_rate": 6.76603255540302e-05, + "loss": 0.0492, + "step": 13800 + }, + { + "epoch": 0.6815061776818984, + "grad_norm": 0.33593711256980896, + "learning_rate": 6.815061776818984e-05, + "loss": 0.0466, + "step": 13900 + }, + { + "epoch": 0.6864090998234949, + "grad_norm": 0.5009463429450989, + "learning_rate": 6.864090998234949e-05, + "loss": 0.0496, + "step": 14000 + }, + { + "epoch": 0.6913120219650912, + "grad_norm": 0.28159981966018677, + "learning_rate": 6.913120219650913e-05, + "loss": 0.0495, + "step": 14100 + }, + { + "epoch": 0.6962149441066876, + "grad_norm": 0.18720287084579468, + "learning_rate": 6.962149441066877e-05, + "loss": 0.0495, + "step": 14200 + }, + { + "epoch": 0.701117866248284, + "grad_norm": 0.3543257713317871, + "learning_rate": 7.01117866248284e-05, + "loss": 0.0474, + "step": 14300 + }, + { + "epoch": 0.7060207883898804, + "grad_norm": 0.3448733389377594, + "learning_rate": 7.060207883898804e-05, + "loss": 0.0465, + "step": 14400 + }, + { + "epoch": 0.7109237105314767, + "grad_norm": 0.47061529755592346, + "learning_rate": 7.109237105314768e-05, + "loss": 0.0479, + "step": 14500 + }, + { + "epoch": 0.7158266326730731, + "grad_norm": 0.18680918216705322, + "learning_rate": 7.158266326730731e-05, + "loss": 0.0478, + "step": 14600 + }, + { + "epoch": 0.7207295548146695, + "grad_norm": 0.2860680818557739, + "learning_rate": 7.207295548146695e-05, + "loss": 0.0416, + "step": 14700 + }, + { + "epoch": 0.725632476956266, + "grad_norm": 0.5014829635620117, + "learning_rate": 7.25632476956266e-05, + "loss": 0.0515, + "step": 14800 + }, + { + "epoch": 0.7305353990978624, + "grad_norm": 0.624320924282074, + "learning_rate": 7.305353990978624e-05, + "loss": 0.0448, + "step": 14900 + }, + { + "epoch": 0.7354383212394587, + "grad_norm": 0.37180984020233154, + "learning_rate": 7.354383212394588e-05, + "loss": 0.0445, + "step": 15000 + }, + { + "epoch": 0.7403412433810551, + "grad_norm": 0.3159758448600769, + "learning_rate": 7.403412433810552e-05, + "loss": 0.0462, + "step": 15100 + }, + { + "epoch": 0.7452441655226515, + "grad_norm": 0.4232478439807892, + "learning_rate": 7.452441655226515e-05, + "loss": 0.0455, + "step": 15200 + }, + { + "epoch": 0.7501470876642479, + "grad_norm": 0.18599086999893188, + "learning_rate": 7.501470876642479e-05, + "loss": 0.0452, + "step": 15300 + }, + { + "epoch": 0.7550500098058442, + "grad_norm": 0.37140703201293945, + "learning_rate": 7.550500098058443e-05, + "loss": 0.048, + "step": 15400 + }, + { + "epoch": 0.7599529319474406, + "grad_norm": 0.45977580547332764, + "learning_rate": 7.599529319474407e-05, + "loss": 0.0449, + "step": 15500 + }, + { + "epoch": 0.7648558540890371, + "grad_norm": 0.2783525586128235, + "learning_rate": 7.648558540890371e-05, + "loss": 0.0408, + "step": 15600 + }, + { + "epoch": 0.7697587762306335, + "grad_norm": 0.3637586832046509, + "learning_rate": 7.697587762306335e-05, + "loss": 0.0461, + "step": 15700 + }, + { + "epoch": 0.7746616983722299, + "grad_norm": 0.3235738277435303, + "learning_rate": 7.746616983722299e-05, + "loss": 0.0462, + "step": 15800 + }, + { + "epoch": 0.7795646205138262, + "grad_norm": 0.28826725482940674, + "learning_rate": 7.795646205138263e-05, + "loss": 0.0446, + "step": 15900 + }, + { + "epoch": 0.7844675426554226, + "grad_norm": 0.4955115020275116, + "learning_rate": 7.844675426554226e-05, + "loss": 0.0469, + "step": 16000 + }, + { + "epoch": 0.789370464797019, + "grad_norm": 0.4371892809867859, + "learning_rate": 7.89370464797019e-05, + "loss": 0.0446, + "step": 16100 + }, + { + "epoch": 0.7942733869386154, + "grad_norm": 0.42605656385421753, + "learning_rate": 7.942733869386154e-05, + "loss": 0.0488, + "step": 16200 + }, + { + "epoch": 0.7991763090802118, + "grad_norm": 0.2385357916355133, + "learning_rate": 7.991763090802118e-05, + "loss": 0.0466, + "step": 16300 + }, + { + "epoch": 0.8040792312218082, + "grad_norm": 0.34842029213905334, + "learning_rate": 8.040792312218083e-05, + "loss": 0.0473, + "step": 16400 + }, + { + "epoch": 0.8089821533634046, + "grad_norm": 0.348670095205307, + "learning_rate": 8.089821533634047e-05, + "loss": 0.046, + "step": 16500 + }, + { + "epoch": 0.813885075505001, + "grad_norm": 0.0661313384771347, + "learning_rate": 8.138850755050011e-05, + "loss": 0.0424, + "step": 16600 + }, + { + "epoch": 0.8187879976465974, + "grad_norm": 0.27380064129829407, + "learning_rate": 8.187879976465974e-05, + "loss": 0.0396, + "step": 16700 + }, + { + "epoch": 0.8236909197881938, + "grad_norm": 0.2240746170282364, + "learning_rate": 8.236909197881937e-05, + "loss": 0.043, + "step": 16800 + }, + { + "epoch": 0.8285938419297901, + "grad_norm": 0.39140066504478455, + "learning_rate": 8.285938419297901e-05, + "loss": 0.0403, + "step": 16900 + }, + { + "epoch": 0.8334967640713865, + "grad_norm": 0.31604424118995667, + "learning_rate": 8.334967640713865e-05, + "loss": 0.0458, + "step": 17000 + }, + { + "epoch": 0.8383996862129829, + "grad_norm": 0.29236578941345215, + "learning_rate": 8.383996862129829e-05, + "loss": 0.0477, + "step": 17100 + }, + { + "epoch": 0.8433026083545794, + "grad_norm": 0.40906059741973877, + "learning_rate": 8.433026083545794e-05, + "loss": 0.0401, + "step": 17200 + }, + { + "epoch": 0.8482055304961758, + "grad_norm": 0.2863566279411316, + "learning_rate": 8.482055304961758e-05, + "loss": 0.0388, + "step": 17300 + }, + { + "epoch": 0.8531084526377721, + "grad_norm": 0.437642902135849, + "learning_rate": 8.531084526377722e-05, + "loss": 0.045, + "step": 17400 + }, + { + "epoch": 0.8580113747793685, + "grad_norm": 0.29014888405799866, + "learning_rate": 8.580113747793686e-05, + "loss": 0.0418, + "step": 17500 + }, + { + "epoch": 0.8629142969209649, + "grad_norm": 0.27143239974975586, + "learning_rate": 8.62914296920965e-05, + "loss": 0.0443, + "step": 17600 + }, + { + "epoch": 0.8678172190625613, + "grad_norm": 0.14953571557998657, + "learning_rate": 8.678172190625614e-05, + "loss": 0.0431, + "step": 17700 + }, + { + "epoch": 0.8727201412041576, + "grad_norm": 0.37678760290145874, + "learning_rate": 8.727201412041577e-05, + "loss": 0.0441, + "step": 17800 + }, + { + "epoch": 0.877623063345754, + "grad_norm": 0.22045595943927765, + "learning_rate": 8.77623063345754e-05, + "loss": 0.0469, + "step": 17900 + }, + { + "epoch": 0.8825259854873505, + "grad_norm": 0.2513475716114044, + "learning_rate": 8.825259854873505e-05, + "loss": 0.0397, + "step": 18000 + }, + { + "epoch": 0.8874289076289469, + "grad_norm": 0.21822652220726013, + "learning_rate": 8.874289076289469e-05, + "loss": 0.0461, + "step": 18100 + }, + { + "epoch": 0.8923318297705433, + "grad_norm": 0.31887462735176086, + "learning_rate": 8.923318297705433e-05, + "loss": 0.0404, + "step": 18200 + }, + { + "epoch": 0.8972347519121396, + "grad_norm": 0.3159906268119812, + "learning_rate": 8.972347519121397e-05, + "loss": 0.041, + "step": 18300 + }, + { + "epoch": 0.902137674053736, + "grad_norm": 0.20168788731098175, + "learning_rate": 9.021376740537361e-05, + "loss": 0.0437, + "step": 18400 + }, + { + "epoch": 0.9070405961953324, + "grad_norm": 0.45709195733070374, + "learning_rate": 9.070405961953325e-05, + "loss": 0.0425, + "step": 18500 + }, + { + "epoch": 0.9119435183369288, + "grad_norm": 0.2685320973396301, + "learning_rate": 9.119435183369288e-05, + "loss": 0.0418, + "step": 18600 + }, + { + "epoch": 0.9168464404785251, + "grad_norm": 0.14851762354373932, + "learning_rate": 9.168464404785252e-05, + "loss": 0.042, + "step": 18700 + }, + { + "epoch": 0.9217493626201216, + "grad_norm": 0.1795952171087265, + "learning_rate": 9.217493626201216e-05, + "loss": 0.0417, + "step": 18800 + }, + { + "epoch": 0.926652284761718, + "grad_norm": 0.11718897521495819, + "learning_rate": 9.26652284761718e-05, + "loss": 0.0456, + "step": 18900 + }, + { + "epoch": 0.9315552069033144, + "grad_norm": 0.11578471958637238, + "learning_rate": 9.315552069033144e-05, + "loss": 0.0421, + "step": 19000 + }, + { + "epoch": 0.9364581290449108, + "grad_norm": 0.2343079149723053, + "learning_rate": 9.364581290449108e-05, + "loss": 0.0406, + "step": 19100 + }, + { + "epoch": 0.9413610511865071, + "grad_norm": 0.29279735684394836, + "learning_rate": 9.413610511865072e-05, + "loss": 0.0395, + "step": 19200 + }, + { + "epoch": 0.9462639733281035, + "grad_norm": 0.2973973751068115, + "learning_rate": 9.462639733281036e-05, + "loss": 0.0343, + "step": 19300 + }, + { + "epoch": 0.9511668954696999, + "grad_norm": 0.4062161147594452, + "learning_rate": 9.511668954697e-05, + "loss": 0.0442, + "step": 19400 + }, + { + "epoch": 0.9560698176112963, + "grad_norm": 0.4501809775829315, + "learning_rate": 9.560698176112963e-05, + "loss": 0.0425, + "step": 19500 + }, + { + "epoch": 0.9609727397528928, + "grad_norm": 0.2221497744321823, + "learning_rate": 9.609727397528929e-05, + "loss": 0.0415, + "step": 19600 + }, + { + "epoch": 0.9658756618944891, + "grad_norm": 0.2941090762615204, + "learning_rate": 9.658756618944892e-05, + "loss": 0.0409, + "step": 19700 + }, + { + "epoch": 0.9707785840360855, + "grad_norm": 0.32935085892677307, + "learning_rate": 9.707785840360856e-05, + "loss": 0.0436, + "step": 19800 + }, + { + "epoch": 0.9756815061776819, + "grad_norm": 0.31291815638542175, + "learning_rate": 9.756815061776819e-05, + "loss": 0.0407, + "step": 19900 + }, + { + "epoch": 0.9805844283192783, + "grad_norm": 0.3670090138912201, + "learning_rate": 9.805844283192783e-05, + "loss": 0.0415, + "step": 20000 + }, + { + "epoch": 0.9854873504608747, + "grad_norm": 0.29986053705215454, + "learning_rate": 9.854873504608747e-05, + "loss": 0.0453, + "step": 20100 + }, + { + "epoch": 0.990390272602471, + "grad_norm": 0.4767267405986786, + "learning_rate": 9.90390272602471e-05, + "loss": 0.0422, + "step": 20200 + }, + { + "epoch": 0.9952931947440674, + "grad_norm": 0.45798006653785706, + "learning_rate": 9.952931947440674e-05, + "loss": 0.0422, + "step": 20300 + }, + { + "epoch": 1.0, + "eval_loss": 0.03229685500264168, + "eval_runtime": 221.6122, + "eval_samples_per_second": 22.562, + "eval_steps_per_second": 22.562, + "step": 20396 + }, + { + "epoch": 1.000196116885664, + "grad_norm": 0.3639116585254669, + "learning_rate": 9.999999988283857e-05, + "loss": 0.0414, + "step": 20400 + }, + { + "epoch": 1.0050990390272603, + "grad_norm": 0.2748640179634094, + "learning_rate": 9.999992079888559e-05, + "loss": 0.0335, + "step": 20500 + }, + { + "epoch": 1.0100019611688567, + "grad_norm": 0.1961342841386795, + "learning_rate": 9.999969526339681e-05, + "loss": 0.0269, + "step": 20600 + }, + { + "epoch": 1.014904883310453, + "grad_norm": 0.15799719095230103, + "learning_rate": 9.999932327703287e-05, + "loss": 0.0286, + "step": 20700 + }, + { + "epoch": 1.0198078054520494, + "grad_norm": 0.26234665513038635, + "learning_rate": 9.999880484088331e-05, + "loss": 0.0293, + "step": 20800 + }, + { + "epoch": 1.0247107275936458, + "grad_norm": 0.24818925559520721, + "learning_rate": 9.999813995646665e-05, + "loss": 0.0305, + "step": 20900 + }, + { + "epoch": 1.0296136497352422, + "grad_norm": 0.39123162627220154, + "learning_rate": 9.999732862573036e-05, + "loss": 0.0309, + "step": 21000 + }, + { + "epoch": 1.0345165718768385, + "grad_norm": 0.27162662148475647, + "learning_rate": 9.999637085105086e-05, + "loss": 0.0292, + "step": 21100 + }, + { + "epoch": 1.039419494018435, + "grad_norm": 0.22811737656593323, + "learning_rate": 9.99952666352335e-05, + "loss": 0.0257, + "step": 21200 + }, + { + "epoch": 1.0443224161600313, + "grad_norm": 0.19570066034793854, + "learning_rate": 9.999401598151257e-05, + "loss": 0.0301, + "step": 21300 + }, + { + "epoch": 1.0492253383016277, + "grad_norm": 0.2861987054347992, + "learning_rate": 9.999261889355128e-05, + "loss": 0.0311, + "step": 21400 + }, + { + "epoch": 1.0541282604432243, + "grad_norm": 0.1091461256146431, + "learning_rate": 9.999107537544174e-05, + "loss": 0.0295, + "step": 21500 + }, + { + "epoch": 1.0590311825848207, + "grad_norm": 0.1704551875591278, + "learning_rate": 9.998938543170499e-05, + "loss": 0.0303, + "step": 21600 + }, + { + "epoch": 1.063934104726417, + "grad_norm": 0.25216275453567505, + "learning_rate": 9.998754906729093e-05, + "loss": 0.0342, + "step": 21700 + }, + { + "epoch": 1.0688370268680134, + "grad_norm": 0.17220348119735718, + "learning_rate": 9.998556628757832e-05, + "loss": 0.0293, + "step": 21800 + }, + { + "epoch": 1.0737399490096098, + "grad_norm": 0.46171247959136963, + "learning_rate": 9.998343709837481e-05, + "loss": 0.0298, + "step": 21900 + }, + { + "epoch": 1.0786428711512062, + "grad_norm": 0.23441363871097565, + "learning_rate": 9.998116150591685e-05, + "loss": 0.0295, + "step": 22000 + }, + { + "epoch": 1.0835457932928025, + "grad_norm": 0.3156551718711853, + "learning_rate": 9.997873951686976e-05, + "loss": 0.0286, + "step": 22100 + }, + { + "epoch": 1.088448715434399, + "grad_norm": 0.43607401847839355, + "learning_rate": 9.99761711383276e-05, + "loss": 0.029, + "step": 22200 + }, + { + "epoch": 1.0933516375759953, + "grad_norm": 0.11727584898471832, + "learning_rate": 9.997345637781325e-05, + "loss": 0.0288, + "step": 22300 + }, + { + "epoch": 1.0982545597175917, + "grad_norm": 0.09930656105279922, + "learning_rate": 9.997059524327837e-05, + "loss": 0.0298, + "step": 22400 + }, + { + "epoch": 1.103157481859188, + "grad_norm": 0.28237900137901306, + "learning_rate": 9.996758774310326e-05, + "loss": 0.0305, + "step": 22500 + }, + { + "epoch": 1.1080604040007844, + "grad_norm": 0.318491667509079, + "learning_rate": 9.996443388609706e-05, + "loss": 0.0315, + "step": 22600 + }, + { + "epoch": 1.1129633261423808, + "grad_norm": 0.24862946569919586, + "learning_rate": 9.996113368149751e-05, + "loss": 0.0308, + "step": 22700 + }, + { + "epoch": 1.1178662482839772, + "grad_norm": 0.2209358662366867, + "learning_rate": 9.9957687138971e-05, + "loss": 0.0272, + "step": 22800 + }, + { + "epoch": 1.1227691704255736, + "grad_norm": 0.20178958773612976, + "learning_rate": 9.995409426861261e-05, + "loss": 0.0301, + "step": 22900 + }, + { + "epoch": 1.12767209256717, + "grad_norm": 0.05385458096861839, + "learning_rate": 9.995035508094596e-05, + "loss": 0.0293, + "step": 23000 + }, + { + "epoch": 1.1325750147087663, + "grad_norm": 0.42040759325027466, + "learning_rate": 9.994646958692327e-05, + "loss": 0.0278, + "step": 23100 + }, + { + "epoch": 1.1374779368503627, + "grad_norm": 0.2339663952589035, + "learning_rate": 9.99424377979253e-05, + "loss": 0.0276, + "step": 23200 + }, + { + "epoch": 1.1423808589919593, + "grad_norm": 0.3861044645309448, + "learning_rate": 9.993825972576127e-05, + "loss": 0.0288, + "step": 23300 + }, + { + "epoch": 1.1472837811335557, + "grad_norm": 0.27347856760025024, + "learning_rate": 9.993393538266892e-05, + "loss": 0.0301, + "step": 23400 + }, + { + "epoch": 1.152186703275152, + "grad_norm": 0.19038136303424835, + "learning_rate": 9.992946478131444e-05, + "loss": 0.0303, + "step": 23500 + }, + { + "epoch": 1.1570896254167484, + "grad_norm": 0.23598672449588776, + "learning_rate": 9.992484793479233e-05, + "loss": 0.0309, + "step": 23600 + }, + { + "epoch": 1.1619925475583448, + "grad_norm": 0.28980758786201477, + "learning_rate": 9.992008485662552e-05, + "loss": 0.0296, + "step": 23700 + }, + { + "epoch": 1.1668954696999412, + "grad_norm": 0.19490975141525269, + "learning_rate": 9.991517556076522e-05, + "loss": 0.0296, + "step": 23800 + }, + { + "epoch": 1.1717983918415376, + "grad_norm": 0.12678834795951843, + "learning_rate": 9.991012006159094e-05, + "loss": 0.0265, + "step": 23900 + }, + { + "epoch": 1.176701313983134, + "grad_norm": 0.29768842458724976, + "learning_rate": 9.990491837391042e-05, + "loss": 0.03, + "step": 24000 + }, + { + "epoch": 1.1816042361247303, + "grad_norm": 0.33209484815597534, + "learning_rate": 9.989957051295957e-05, + "loss": 0.0275, + "step": 24100 + }, + { + "epoch": 1.1865071582663267, + "grad_norm": 0.24054765701293945, + "learning_rate": 9.989407649440249e-05, + "loss": 0.0262, + "step": 24200 + }, + { + "epoch": 1.191410080407923, + "grad_norm": 0.1172015443444252, + "learning_rate": 9.988843633433134e-05, + "loss": 0.0306, + "step": 24300 + }, + { + "epoch": 1.1963130025495194, + "grad_norm": 0.22456437349319458, + "learning_rate": 9.988265004926633e-05, + "loss": 0.0308, + "step": 24400 + }, + { + "epoch": 1.2012159246911158, + "grad_norm": 0.17739421129226685, + "learning_rate": 9.987671765615574e-05, + "loss": 0.0311, + "step": 24500 + }, + { + "epoch": 1.2061188468327124, + "grad_norm": 0.39153727889060974, + "learning_rate": 9.987063917237572e-05, + "loss": 0.0318, + "step": 24600 + }, + { + "epoch": 1.2110217689743088, + "grad_norm": 0.1307838261127472, + "learning_rate": 9.986441461573036e-05, + "loss": 0.0267, + "step": 24700 + }, + { + "epoch": 1.2159246911159052, + "grad_norm": 0.35021230578422546, + "learning_rate": 9.985804400445163e-05, + "loss": 0.0314, + "step": 24800 + }, + { + "epoch": 1.2208276132575016, + "grad_norm": 0.20624665915966034, + "learning_rate": 9.985152735719927e-05, + "loss": 0.0332, + "step": 24900 + }, + { + "epoch": 1.225730535399098, + "grad_norm": 0.32110336422920227, + "learning_rate": 9.984486469306078e-05, + "loss": 0.0281, + "step": 25000 + }, + { + "epoch": 1.2306334575406943, + "grad_norm": 0.2653130888938904, + "learning_rate": 9.983805603155131e-05, + "loss": 0.0287, + "step": 25100 + }, + { + "epoch": 1.2355363796822907, + "grad_norm": 0.11658957600593567, + "learning_rate": 9.983110139261369e-05, + "loss": 0.0315, + "step": 25200 + }, + { + "epoch": 1.240439301823887, + "grad_norm": 0.5068228244781494, + "learning_rate": 9.98240007966183e-05, + "loss": 0.0251, + "step": 25300 + }, + { + "epoch": 1.2453422239654834, + "grad_norm": 0.27341774106025696, + "learning_rate": 9.981675426436304e-05, + "loss": 0.0311, + "step": 25400 + }, + { + "epoch": 1.2502451461070798, + "grad_norm": 0.22564496099948883, + "learning_rate": 9.980936181707326e-05, + "loss": 0.027, + "step": 25500 + }, + { + "epoch": 1.2551480682486762, + "grad_norm": 0.33587944507598877, + "learning_rate": 9.980182347640168e-05, + "loss": 0.0285, + "step": 25600 + }, + { + "epoch": 1.2600509903902726, + "grad_norm": 0.19802583754062653, + "learning_rate": 9.97941392644284e-05, + "loss": 0.0251, + "step": 25700 + }, + { + "epoch": 1.264953912531869, + "grad_norm": 0.36870622634887695, + "learning_rate": 9.978630920366072e-05, + "loss": 0.0294, + "step": 25800 + }, + { + "epoch": 1.2698568346734653, + "grad_norm": 0.2961844801902771, + "learning_rate": 9.977833331703317e-05, + "loss": 0.0294, + "step": 25900 + }, + { + "epoch": 1.2747597568150617, + "grad_norm": 0.31916147470474243, + "learning_rate": 9.977021162790743e-05, + "loss": 0.0318, + "step": 26000 + }, + { + "epoch": 1.279662678956658, + "grad_norm": 0.27534112334251404, + "learning_rate": 9.976194416007217e-05, + "loss": 0.0297, + "step": 26100 + }, + { + "epoch": 1.2845656010982545, + "grad_norm": 0.30862635374069214, + "learning_rate": 9.975353093774313e-05, + "loss": 0.028, + "step": 26200 + }, + { + "epoch": 1.2894685232398508, + "grad_norm": 0.364801824092865, + "learning_rate": 9.974497198556292e-05, + "loss": 0.0286, + "step": 26300 + }, + { + "epoch": 1.2943714453814472, + "grad_norm": 0.39874985814094543, + "learning_rate": 9.973626732860102e-05, + "loss": 0.0302, + "step": 26400 + }, + { + "epoch": 1.2992743675230438, + "grad_norm": 0.3267800509929657, + "learning_rate": 9.972741699235368e-05, + "loss": 0.0314, + "step": 26500 + }, + { + "epoch": 1.3041772896646402, + "grad_norm": 0.24671390652656555, + "learning_rate": 9.971842100274384e-05, + "loss": 0.0258, + "step": 26600 + }, + { + "epoch": 1.3090802118062366, + "grad_norm": 0.3735731840133667, + "learning_rate": 9.97092793861211e-05, + "loss": 0.0291, + "step": 26700 + }, + { + "epoch": 1.313983133947833, + "grad_norm": 0.13338430225849152, + "learning_rate": 9.969999216926151e-05, + "loss": 0.0301, + "step": 26800 + }, + { + "epoch": 1.3188860560894293, + "grad_norm": 0.23581305146217346, + "learning_rate": 9.969055937936773e-05, + "loss": 0.0309, + "step": 26900 + }, + { + "epoch": 1.3237889782310257, + "grad_norm": 0.09043537825345993, + "learning_rate": 9.96809810440687e-05, + "loss": 0.0274, + "step": 27000 + }, + { + "epoch": 1.328691900372622, + "grad_norm": 0.36778169870376587, + "learning_rate": 9.96712571914197e-05, + "loss": 0.0306, + "step": 27100 + }, + { + "epoch": 1.3335948225142185, + "grad_norm": 0.12855975329875946, + "learning_rate": 9.966138784990227e-05, + "loss": 0.023, + "step": 27200 + }, + { + "epoch": 1.3384977446558148, + "grad_norm": 0.5465484261512756, + "learning_rate": 9.965137304842402e-05, + "loss": 0.0287, + "step": 27300 + }, + { + "epoch": 1.3434006667974112, + "grad_norm": 0.22656790912151337, + "learning_rate": 9.964121281631866e-05, + "loss": 0.0303, + "step": 27400 + }, + { + "epoch": 1.3483035889390076, + "grad_norm": 0.334923654794693, + "learning_rate": 9.96309071833459e-05, + "loss": 0.0265, + "step": 27500 + }, + { + "epoch": 1.353206511080604, + "grad_norm": 0.26427406072616577, + "learning_rate": 9.96204561796913e-05, + "loss": 0.0299, + "step": 27600 + }, + { + "epoch": 1.3581094332222006, + "grad_norm": 0.1943475604057312, + "learning_rate": 9.960985983596618e-05, + "loss": 0.028, + "step": 27700 + }, + { + "epoch": 1.363012355363797, + "grad_norm": 0.3269706070423126, + "learning_rate": 9.959911818320764e-05, + "loss": 0.0271, + "step": 27800 + }, + { + "epoch": 1.3679152775053933, + "grad_norm": 0.2163461148738861, + "learning_rate": 9.958823125287836e-05, + "loss": 0.0295, + "step": 27900 + }, + { + "epoch": 1.3728181996469897, + "grad_norm": 0.24624493718147278, + "learning_rate": 9.957719907686654e-05, + "loss": 0.0343, + "step": 28000 + }, + { + "epoch": 1.377721121788586, + "grad_norm": 0.15970952808856964, + "learning_rate": 9.95660216874858e-05, + "loss": 0.0301, + "step": 28100 + }, + { + "epoch": 1.3826240439301825, + "grad_norm": 0.521918773651123, + "learning_rate": 9.955469911747511e-05, + "loss": 0.0314, + "step": 28200 + }, + { + "epoch": 1.3875269660717788, + "grad_norm": 0.17727439105510712, + "learning_rate": 9.954323139999871e-05, + "loss": 0.0276, + "step": 28300 + }, + { + "epoch": 1.3924298882133752, + "grad_norm": 0.3565279245376587, + "learning_rate": 9.95316185686459e-05, + "loss": 0.0259, + "step": 28400 + }, + { + "epoch": 1.3973328103549716, + "grad_norm": 0.17430347204208374, + "learning_rate": 9.951986065743111e-05, + "loss": 0.03, + "step": 28500 + }, + { + "epoch": 1.402235732496568, + "grad_norm": 0.13704228401184082, + "learning_rate": 9.950795770079368e-05, + "loss": 0.0274, + "step": 28600 + }, + { + "epoch": 1.4071386546381643, + "grad_norm": 0.23465342819690704, + "learning_rate": 9.949590973359778e-05, + "loss": 0.0266, + "step": 28700 + }, + { + "epoch": 1.4120415767797607, + "grad_norm": 0.18756377696990967, + "learning_rate": 9.948371679113229e-05, + "loss": 0.0274, + "step": 28800 + }, + { + "epoch": 1.416944498921357, + "grad_norm": 0.17243963479995728, + "learning_rate": 9.947137890911084e-05, + "loss": 0.0254, + "step": 28900 + }, + { + "epoch": 1.4218474210629535, + "grad_norm": 0.14038296043872833, + "learning_rate": 9.945889612367149e-05, + "loss": 0.0252, + "step": 29000 + }, + { + "epoch": 1.4267503432045499, + "grad_norm": 0.25634902715682983, + "learning_rate": 9.944626847137677e-05, + "loss": 0.0288, + "step": 29100 + }, + { + "epoch": 1.4316532653461462, + "grad_norm": 0.4887700080871582, + "learning_rate": 9.94334959892135e-05, + "loss": 0.0288, + "step": 29200 + }, + { + "epoch": 1.4365561874877426, + "grad_norm": 0.33462828397750854, + "learning_rate": 9.942057871459275e-05, + "loss": 0.0295, + "step": 29300 + }, + { + "epoch": 1.441459109629339, + "grad_norm": 0.3429461419582367, + "learning_rate": 9.940751668534966e-05, + "loss": 0.0272, + "step": 29400 + }, + { + "epoch": 1.4463620317709354, + "grad_norm": 0.3318476676940918, + "learning_rate": 9.939430993974339e-05, + "loss": 0.0288, + "step": 29500 + }, + { + "epoch": 1.4512649539125317, + "grad_norm": 0.3497527241706848, + "learning_rate": 9.938095851645696e-05, + "loss": 0.0314, + "step": 29600 + }, + { + "epoch": 1.4561678760541283, + "grad_norm": 0.38459303975105286, + "learning_rate": 9.936746245459716e-05, + "loss": 0.0284, + "step": 29700 + }, + { + "epoch": 1.4610707981957247, + "grad_norm": 0.34776678681373596, + "learning_rate": 9.935382179369442e-05, + "loss": 0.0254, + "step": 29800 + }, + { + "epoch": 1.465973720337321, + "grad_norm": 0.13150684535503387, + "learning_rate": 9.934003657370275e-05, + "loss": 0.0274, + "step": 29900 + }, + { + "epoch": 1.4708766424789175, + "grad_norm": 0.3473518192768097, + "learning_rate": 9.932610683499951e-05, + "loss": 0.0287, + "step": 30000 + }, + { + "epoch": 1.4757795646205139, + "grad_norm": 0.10194269567728043, + "learning_rate": 9.931203261838542e-05, + "loss": 0.0265, + "step": 30100 + }, + { + "epoch": 1.4806824867621102, + "grad_norm": 0.2663326859474182, + "learning_rate": 9.929781396508435e-05, + "loss": 0.0256, + "step": 30200 + }, + { + "epoch": 1.4855854089037066, + "grad_norm": 0.37899428606033325, + "learning_rate": 9.928345091674324e-05, + "loss": 0.0287, + "step": 30300 + }, + { + "epoch": 1.490488331045303, + "grad_norm": 0.16388994455337524, + "learning_rate": 9.926894351543196e-05, + "loss": 0.0281, + "step": 30400 + }, + { + "epoch": 1.4953912531868994, + "grad_norm": 0.1827380210161209, + "learning_rate": 9.92542918036432e-05, + "loss": 0.0279, + "step": 30500 + }, + { + "epoch": 1.5002941753284957, + "grad_norm": 0.3441345691680908, + "learning_rate": 9.923949582429237e-05, + "loss": 0.0299, + "step": 30600 + }, + { + "epoch": 1.5051970974700923, + "grad_norm": 0.22477801144123077, + "learning_rate": 9.92245556207174e-05, + "loss": 0.0263, + "step": 30700 + }, + { + "epoch": 1.5101000196116887, + "grad_norm": 0.2690620422363281, + "learning_rate": 9.920947123667866e-05, + "loss": 0.0281, + "step": 30800 + }, + { + "epoch": 1.515002941753285, + "grad_norm": 0.3723180592060089, + "learning_rate": 9.919424271635884e-05, + "loss": 0.0275, + "step": 30900 + }, + { + "epoch": 1.5199058638948815, + "grad_norm": 0.4014623165130615, + "learning_rate": 9.917887010436285e-05, + "loss": 0.0259, + "step": 31000 + }, + { + "epoch": 1.5248087860364778, + "grad_norm": 0.29620611667633057, + "learning_rate": 9.916335344571758e-05, + "loss": 0.0281, + "step": 31100 + }, + { + "epoch": 1.5297117081780742, + "grad_norm": 0.3609066605567932, + "learning_rate": 9.914769278587189e-05, + "loss": 0.0249, + "step": 31200 + }, + { + "epoch": 1.5346146303196706, + "grad_norm": 0.20832450687885284, + "learning_rate": 9.913188817069641e-05, + "loss": 0.0306, + "step": 31300 + }, + { + "epoch": 1.539517552461267, + "grad_norm": 0.14223641157150269, + "learning_rate": 9.91159396464834e-05, + "loss": 0.0261, + "step": 31400 + }, + { + "epoch": 1.5444204746028634, + "grad_norm": 0.2740347981452942, + "learning_rate": 9.909984725994666e-05, + "loss": 0.0306, + "step": 31500 + }, + { + "epoch": 1.5493233967444597, + "grad_norm": 0.2499512881040573, + "learning_rate": 9.908361105822134e-05, + "loss": 0.0271, + "step": 31600 + }, + { + "epoch": 1.5542263188860561, + "grad_norm": 0.32326480746269226, + "learning_rate": 9.90672310888639e-05, + "loss": 0.0273, + "step": 31700 + }, + { + "epoch": 1.5591292410276525, + "grad_norm": 0.3028701841831207, + "learning_rate": 9.90507073998518e-05, + "loss": 0.0277, + "step": 31800 + }, + { + "epoch": 1.5640321631692489, + "grad_norm": 0.3670022487640381, + "learning_rate": 9.903404003958352e-05, + "loss": 0.026, + "step": 31900 + }, + { + "epoch": 1.5689350853108452, + "grad_norm": 0.19097840785980225, + "learning_rate": 9.901722905687835e-05, + "loss": 0.0245, + "step": 32000 + }, + { + "epoch": 1.5738380074524416, + "grad_norm": 0.3168416917324066, + "learning_rate": 9.900027450097629e-05, + "loss": 0.0273, + "step": 32100 + }, + { + "epoch": 1.578740929594038, + "grad_norm": 0.3735883831977844, + "learning_rate": 9.898317642153776e-05, + "loss": 0.0285, + "step": 32200 + }, + { + "epoch": 1.5836438517356344, + "grad_norm": 0.17389927804470062, + "learning_rate": 9.896593486864369e-05, + "loss": 0.0279, + "step": 32300 + }, + { + "epoch": 1.5885467738772308, + "grad_norm": 0.402938574552536, + "learning_rate": 9.89485498927952e-05, + "loss": 0.0271, + "step": 32400 + }, + { + "epoch": 1.5934496960188271, + "grad_norm": 0.11623658984899521, + "learning_rate": 9.893102154491348e-05, + "loss": 0.0282, + "step": 32500 + }, + { + "epoch": 1.5983526181604235, + "grad_norm": 0.4419660270214081, + "learning_rate": 9.891334987633969e-05, + "loss": 0.0274, + "step": 32600 + }, + { + "epoch": 1.6032555403020199, + "grad_norm": 0.19593150913715363, + "learning_rate": 9.889553493883476e-05, + "loss": 0.0276, + "step": 32700 + }, + { + "epoch": 1.6081584624436163, + "grad_norm": 0.313765287399292, + "learning_rate": 9.887757678457927e-05, + "loss": 0.0302, + "step": 32800 + }, + { + "epoch": 1.6130613845852126, + "grad_norm": 0.15113359689712524, + "learning_rate": 9.885947546617332e-05, + "loss": 0.0289, + "step": 32900 + }, + { + "epoch": 1.617964306726809, + "grad_norm": 0.16863977909088135, + "learning_rate": 9.884123103663629e-05, + "loss": 0.0249, + "step": 33000 + }, + { + "epoch": 1.6228672288684056, + "grad_norm": 0.4099023640155792, + "learning_rate": 9.882284354940674e-05, + "loss": 0.0272, + "step": 33100 + }, + { + "epoch": 1.627770151010002, + "grad_norm": 0.18381527066230774, + "learning_rate": 9.88043130583423e-05, + "loss": 0.0256, + "step": 33200 + }, + { + "epoch": 1.6326730731515984, + "grad_norm": 0.21517693996429443, + "learning_rate": 9.878563961771942e-05, + "loss": 0.0276, + "step": 33300 + }, + { + "epoch": 1.6375759952931948, + "grad_norm": 0.23215888440608978, + "learning_rate": 9.876682328223329e-05, + "loss": 0.0231, + "step": 33400 + }, + { + "epoch": 1.6424789174347911, + "grad_norm": 0.31246307492256165, + "learning_rate": 9.87478641069976e-05, + "loss": 0.0248, + "step": 33500 + }, + { + "epoch": 1.6473818395763875, + "grad_norm": 0.2109220176935196, + "learning_rate": 9.872876214754442e-05, + "loss": 0.026, + "step": 33600 + }, + { + "epoch": 1.6522847617179839, + "grad_norm": 0.17266134917736053, + "learning_rate": 9.870951745982411e-05, + "loss": 0.0305, + "step": 33700 + }, + { + "epoch": 1.6571876838595803, + "grad_norm": 0.24855448305606842, + "learning_rate": 9.869013010020504e-05, + "loss": 0.0268, + "step": 33800 + }, + { + "epoch": 1.6620906060011769, + "grad_norm": 0.4860318899154663, + "learning_rate": 9.867060012547343e-05, + "loss": 0.0279, + "step": 33900 + }, + { + "epoch": 1.6669935281427732, + "grad_norm": 0.17674511671066284, + "learning_rate": 9.86509275928333e-05, + "loss": 0.0284, + "step": 34000 + }, + { + "epoch": 1.6718964502843696, + "grad_norm": 0.25708097219467163, + "learning_rate": 9.863111255990618e-05, + "loss": 0.028, + "step": 34100 + }, + { + "epoch": 1.676799372425966, + "grad_norm": 0.18099327385425568, + "learning_rate": 9.861115508473102e-05, + "loss": 0.0277, + "step": 34200 + }, + { + "epoch": 1.6817022945675624, + "grad_norm": 0.6304388046264648, + "learning_rate": 9.859105522576395e-05, + "loss": 0.0275, + "step": 34300 + }, + { + "epoch": 1.6866052167091587, + "grad_norm": 0.3071998655796051, + "learning_rate": 9.857081304187817e-05, + "loss": 0.03, + "step": 34400 + }, + { + "epoch": 1.6915081388507551, + "grad_norm": 0.2949463725090027, + "learning_rate": 9.855042859236376e-05, + "loss": 0.0256, + "step": 34500 + }, + { + "epoch": 1.6964110609923515, + "grad_norm": 0.22837311029434204, + "learning_rate": 9.852990193692748e-05, + "loss": 0.0249, + "step": 34600 + }, + { + "epoch": 1.7013139831339479, + "grad_norm": 0.21440072357654572, + "learning_rate": 9.850923313569266e-05, + "loss": 0.0261, + "step": 34700 + }, + { + "epoch": 1.7062169052755443, + "grad_norm": 0.2580870985984802, + "learning_rate": 9.848842224919891e-05, + "loss": 0.0257, + "step": 34800 + }, + { + "epoch": 1.7111198274171406, + "grad_norm": 0.23171992599964142, + "learning_rate": 9.846746933840209e-05, + "loss": 0.0284, + "step": 34900 + }, + { + "epoch": 1.716022749558737, + "grad_norm": 0.18120083212852478, + "learning_rate": 9.844637446467399e-05, + "loss": 0.0262, + "step": 35000 + }, + { + "epoch": 1.7209256717003334, + "grad_norm": 0.2012002021074295, + "learning_rate": 9.842513768980226e-05, + "loss": 0.0255, + "step": 35100 + }, + { + "epoch": 1.7258285938419298, + "grad_norm": 0.2966020405292511, + "learning_rate": 9.840375907599013e-05, + "loss": 0.025, + "step": 35200 + }, + { + "epoch": 1.7307315159835261, + "grad_norm": 0.1654832810163498, + "learning_rate": 9.838223868585637e-05, + "loss": 0.0265, + "step": 35300 + }, + { + "epoch": 1.7356344381251225, + "grad_norm": 0.21913449466228485, + "learning_rate": 9.836057658243491e-05, + "loss": 0.0282, + "step": 35400 + }, + { + "epoch": 1.740537360266719, + "grad_norm": 0.21052640676498413, + "learning_rate": 9.833877282917485e-05, + "loss": 0.0247, + "step": 35500 + }, + { + "epoch": 1.7454402824083153, + "grad_norm": 0.27234575152397156, + "learning_rate": 9.831682748994013e-05, + "loss": 0.0279, + "step": 35600 + }, + { + "epoch": 1.7503432045499117, + "grad_norm": 0.30390360951423645, + "learning_rate": 9.829474062900945e-05, + "loss": 0.0281, + "step": 35700 + }, + { + "epoch": 1.755246126691508, + "grad_norm": 0.3458337187767029, + "learning_rate": 9.827251231107599e-05, + "loss": 0.024, + "step": 35800 + }, + { + "epoch": 1.7601490488331044, + "grad_norm": 0.26846686005592346, + "learning_rate": 9.82501426012473e-05, + "loss": 0.0316, + "step": 35900 + }, + { + "epoch": 1.7650519709747008, + "grad_norm": 0.21699777245521545, + "learning_rate": 9.822763156504502e-05, + "loss": 0.0274, + "step": 36000 + }, + { + "epoch": 1.7699548931162972, + "grad_norm": 0.25235772132873535, + "learning_rate": 9.82049792684048e-05, + "loss": 0.0273, + "step": 36100 + }, + { + "epoch": 1.7748578152578935, + "grad_norm": 0.06357789784669876, + "learning_rate": 9.818218577767599e-05, + "loss": 0.0233, + "step": 36200 + }, + { + "epoch": 1.7797607373994901, + "grad_norm": 0.25367268919944763, + "learning_rate": 9.815925115962155e-05, + "loss": 0.0249, + "step": 36300 + }, + { + "epoch": 1.7846636595410865, + "grad_norm": 0.23693925142288208, + "learning_rate": 9.813617548141776e-05, + "loss": 0.0248, + "step": 36400 + }, + { + "epoch": 1.789566581682683, + "grad_norm": 0.08688795566558838, + "learning_rate": 9.811295881065415e-05, + "loss": 0.0251, + "step": 36500 + }, + { + "epoch": 1.7944695038242793, + "grad_norm": 0.3065630793571472, + "learning_rate": 9.808960121533312e-05, + "loss": 0.0284, + "step": 36600 + }, + { + "epoch": 1.7993724259658757, + "grad_norm": 0.20038124918937683, + "learning_rate": 9.806610276386992e-05, + "loss": 0.028, + "step": 36700 + }, + { + "epoch": 1.804275348107472, + "grad_norm": 0.25850433111190796, + "learning_rate": 9.804246352509233e-05, + "loss": 0.0274, + "step": 36800 + }, + { + "epoch": 1.8091782702490684, + "grad_norm": 0.21738532185554504, + "learning_rate": 9.801868356824051e-05, + "loss": 0.0309, + "step": 36900 + }, + { + "epoch": 1.8140811923906648, + "grad_norm": 0.19677628576755524, + "learning_rate": 9.799476296296682e-05, + "loss": 0.0273, + "step": 37000 + }, + { + "epoch": 1.8189841145322614, + "grad_norm": 0.5938333868980408, + "learning_rate": 9.797070177933554e-05, + "loss": 0.025, + "step": 37100 + }, + { + "epoch": 1.8238870366738578, + "grad_norm": 0.38806554675102234, + "learning_rate": 9.794650008782273e-05, + "loss": 0.0281, + "step": 37200 + }, + { + "epoch": 1.8287899588154541, + "grad_norm": 0.3352005183696747, + "learning_rate": 9.792215795931601e-05, + "loss": 0.0265, + "step": 37300 + }, + { + "epoch": 1.8336928809570505, + "grad_norm": 0.281361848115921, + "learning_rate": 9.789767546511433e-05, + "loss": 0.0301, + "step": 37400 + }, + { + "epoch": 1.838595803098647, + "grad_norm": 0.265520304441452, + "learning_rate": 9.787305267692778e-05, + "loss": 0.0257, + "step": 37500 + }, + { + "epoch": 1.8434987252402433, + "grad_norm": 0.1471584439277649, + "learning_rate": 9.784828966687736e-05, + "loss": 0.0245, + "step": 37600 + }, + { + "epoch": 1.8484016473818397, + "grad_norm": 0.12242023646831512, + "learning_rate": 9.782338650749483e-05, + "loss": 0.0254, + "step": 37700 + }, + { + "epoch": 1.853304569523436, + "grad_norm": 0.22480016946792603, + "learning_rate": 9.779834327172241e-05, + "loss": 0.0254, + "step": 37800 + }, + { + "epoch": 1.8582074916650324, + "grad_norm": 0.4627992808818817, + "learning_rate": 9.777316003291262e-05, + "loss": 0.0251, + "step": 37900 + }, + { + "epoch": 1.8631104138066288, + "grad_norm": 0.3330112397670746, + "learning_rate": 9.774783686482807e-05, + "loss": 0.025, + "step": 38000 + }, + { + "epoch": 1.8680133359482252, + "grad_norm": 0.46279048919677734, + "learning_rate": 9.772237384164121e-05, + "loss": 0.0283, + "step": 38100 + }, + { + "epoch": 1.8729162580898215, + "grad_norm": 0.39556947350502014, + "learning_rate": 9.769677103793411e-05, + "loss": 0.0245, + "step": 38200 + }, + { + "epoch": 1.877819180231418, + "grad_norm": 0.27834552526474, + "learning_rate": 9.767102852869832e-05, + "loss": 0.0271, + "step": 38300 + }, + { + "epoch": 1.8827221023730143, + "grad_norm": 0.16157115995883942, + "learning_rate": 9.764514638933455e-05, + "loss": 0.0292, + "step": 38400 + }, + { + "epoch": 1.8876250245146107, + "grad_norm": 0.1439429223537445, + "learning_rate": 9.761912469565251e-05, + "loss": 0.0282, + "step": 38500 + }, + { + "epoch": 1.892527946656207, + "grad_norm": 0.3553975224494934, + "learning_rate": 9.759296352387063e-05, + "loss": 0.0265, + "step": 38600 + }, + { + "epoch": 1.8974308687978034, + "grad_norm": 0.37897151708602905, + "learning_rate": 9.756666295061593e-05, + "loss": 0.0272, + "step": 38700 + }, + { + "epoch": 1.9023337909393998, + "grad_norm": 0.2812596261501312, + "learning_rate": 9.754022305292372e-05, + "loss": 0.0289, + "step": 38800 + }, + { + "epoch": 1.9072367130809962, + "grad_norm": 0.39995622634887695, + "learning_rate": 9.751364390823741e-05, + "loss": 0.025, + "step": 38900 + }, + { + "epoch": 1.9121396352225926, + "grad_norm": 0.4930385649204254, + "learning_rate": 9.748692559440822e-05, + "loss": 0.0287, + "step": 39000 + }, + { + "epoch": 1.917042557364189, + "grad_norm": 0.2677227854728699, + "learning_rate": 9.746006818969504e-05, + "loss": 0.025, + "step": 39100 + }, + { + "epoch": 1.9219454795057853, + "grad_norm": 0.23274803161621094, + "learning_rate": 9.743307177276419e-05, + "loss": 0.0277, + "step": 39200 + }, + { + "epoch": 1.9268484016473817, + "grad_norm": 0.6990579962730408, + "learning_rate": 9.74059364226891e-05, + "loss": 0.0259, + "step": 39300 + }, + { + "epoch": 1.931751323788978, + "grad_norm": 0.19638150930404663, + "learning_rate": 9.737866221895018e-05, + "loss": 0.0249, + "step": 39400 + }, + { + "epoch": 1.9366542459305747, + "grad_norm": 0.22524696588516235, + "learning_rate": 9.735124924143455e-05, + "loss": 0.0239, + "step": 39500 + }, + { + "epoch": 1.941557168072171, + "grad_norm": 0.2928408682346344, + "learning_rate": 9.732369757043576e-05, + "loss": 0.0275, + "step": 39600 + }, + { + "epoch": 1.9464600902137674, + "grad_norm": 0.26414749026298523, + "learning_rate": 9.729600728665365e-05, + "loss": 0.0264, + "step": 39700 + }, + { + "epoch": 1.9513630123553638, + "grad_norm": 0.36181434988975525, + "learning_rate": 9.726817847119403e-05, + "loss": 0.0284, + "step": 39800 + }, + { + "epoch": 1.9562659344969602, + "grad_norm": 0.18077948689460754, + "learning_rate": 9.72402112055685e-05, + "loss": 0.028, + "step": 39900 + }, + { + "epoch": 1.9611688566385566, + "grad_norm": 0.20345261693000793, + "learning_rate": 9.721210557169416e-05, + "loss": 0.0256, + "step": 40000 + }, + { + "epoch": 1.966071778780153, + "grad_norm": 0.1142786517739296, + "learning_rate": 9.71838616518934e-05, + "loss": 0.0258, + "step": 40100 + }, + { + "epoch": 1.9709747009217493, + "grad_norm": 0.18993543088436127, + "learning_rate": 9.715547952889366e-05, + "loss": 0.0276, + "step": 40200 + }, + { + "epoch": 1.975877623063346, + "grad_norm": 0.21492281556129456, + "learning_rate": 9.712695928582718e-05, + "loss": 0.026, + "step": 40300 + }, + { + "epoch": 1.9807805452049423, + "grad_norm": 0.3433995246887207, + "learning_rate": 9.709830100623078e-05, + "loss": 0.0261, + "step": 40400 + }, + { + "epoch": 1.9856834673465387, + "grad_norm": 0.2835942506790161, + "learning_rate": 9.706950477404555e-05, + "loss": 0.0278, + "step": 40500 + }, + { + "epoch": 1.990586389488135, + "grad_norm": 0.1534029096364975, + "learning_rate": 9.70405706736167e-05, + "loss": 0.0299, + "step": 40600 + }, + { + "epoch": 1.9954893116297314, + "grad_norm": 0.24719446897506714, + "learning_rate": 9.70114987896932e-05, + "loss": 0.0272, + "step": 40700 + }, + { + "epoch": 2.0, + "eval_loss": 0.029068127274513245, + "eval_runtime": 233.9259, + "eval_samples_per_second": 21.374, + "eval_steps_per_second": 21.374, + "step": 40792 + }, + { + "epoch": 2.000392233771328, + "grad_norm": 0.14692184329032898, + "learning_rate": 9.698228920742765e-05, + "loss": 0.0282, + "step": 40800 + }, + { + "epoch": 2.005295155912924, + "grad_norm": 0.14744846522808075, + "learning_rate": 9.695294201237593e-05, + "loss": 0.0088, + "step": 40900 + }, + { + "epoch": 2.0101980780545206, + "grad_norm": 0.03153422847390175, + "learning_rate": 9.692345729049703e-05, + "loss": 0.0099, + "step": 41000 + }, + { + "epoch": 2.015101000196117, + "grad_norm": 0.17189723253250122, + "learning_rate": 9.689383512815273e-05, + "loss": 0.0102, + "step": 41100 + }, + { + "epoch": 2.0200039223377133, + "grad_norm": 0.056081388145685196, + "learning_rate": 9.686407561210741e-05, + "loss": 0.01, + "step": 41200 + }, + { + "epoch": 2.0249068444793097, + "grad_norm": 0.18959540128707886, + "learning_rate": 9.683417882952773e-05, + "loss": 0.0118, + "step": 41300 + }, + { + "epoch": 2.029809766620906, + "grad_norm": 0.1800815314054489, + "learning_rate": 9.68041448679824e-05, + "loss": 0.0108, + "step": 41400 + }, + { + "epoch": 2.0347126887625024, + "grad_norm": 0.024594638496637344, + "learning_rate": 9.6773973815442e-05, + "loss": 0.0089, + "step": 41500 + }, + { + "epoch": 2.039615610904099, + "grad_norm": 0.12218187749385834, + "learning_rate": 9.674366576027859e-05, + "loss": 0.0101, + "step": 41600 + }, + { + "epoch": 2.044518533045695, + "grad_norm": 0.009762736968696117, + "learning_rate": 9.671322079126553e-05, + "loss": 0.0087, + "step": 41700 + }, + { + "epoch": 2.0494214551872916, + "grad_norm": 0.23141342401504517, + "learning_rate": 9.668263899757722e-05, + "loss": 0.0103, + "step": 41800 + }, + { + "epoch": 2.054324377328888, + "grad_norm": 0.22465147078037262, + "learning_rate": 9.665192046878879e-05, + "loss": 0.011, + "step": 41900 + }, + { + "epoch": 2.0592272994704843, + "grad_norm": 0.2304316610097885, + "learning_rate": 9.662106529487593e-05, + "loss": 0.01, + "step": 42000 + }, + { + "epoch": 2.0641302216120807, + "grad_norm": 0.253606915473938, + "learning_rate": 9.659007356621453e-05, + "loss": 0.01, + "step": 42100 + }, + { + "epoch": 2.069033143753677, + "grad_norm": 0.09148263186216354, + "learning_rate": 9.655894537358045e-05, + "loss": 0.0104, + "step": 42200 + }, + { + "epoch": 2.0739360658952735, + "grad_norm": 0.20658960938453674, + "learning_rate": 9.652768080814927e-05, + "loss": 0.0103, + "step": 42300 + }, + { + "epoch": 2.07883898803687, + "grad_norm": 0.0633637011051178, + "learning_rate": 9.6496279961496e-05, + "loss": 0.0112, + "step": 42400 + }, + { + "epoch": 2.083741910178466, + "grad_norm": 0.20948578417301178, + "learning_rate": 9.646474292559483e-05, + "loss": 0.0097, + "step": 42500 + }, + { + "epoch": 2.0886448323200626, + "grad_norm": 0.24738560616970062, + "learning_rate": 9.643306979281885e-05, + "loss": 0.0093, + "step": 42600 + }, + { + "epoch": 2.093547754461659, + "grad_norm": 0.11444352567195892, + "learning_rate": 9.640126065593981e-05, + "loss": 0.0093, + "step": 42700 + }, + { + "epoch": 2.0984506766032553, + "grad_norm": 0.20413364470005035, + "learning_rate": 9.636931560812777e-05, + "loss": 0.0116, + "step": 42800 + }, + { + "epoch": 2.1033535987448517, + "grad_norm": 0.051684655249118805, + "learning_rate": 9.633723474295092e-05, + "loss": 0.0082, + "step": 42900 + }, + { + "epoch": 2.1082565208864485, + "grad_norm": 0.1661156564950943, + "learning_rate": 9.630501815437525e-05, + "loss": 0.0112, + "step": 43000 + }, + { + "epoch": 2.113159443028045, + "grad_norm": 0.11993784457445145, + "learning_rate": 9.627266593676427e-05, + "loss": 0.0089, + "step": 43100 + }, + { + "epoch": 2.1180623651696413, + "grad_norm": 0.21582569181919098, + "learning_rate": 9.624017818487879e-05, + "loss": 0.0103, + "step": 43200 + }, + { + "epoch": 2.1229652873112377, + "grad_norm": 0.3271758556365967, + "learning_rate": 9.620755499387658e-05, + "loss": 0.0104, + "step": 43300 + }, + { + "epoch": 2.127868209452834, + "grad_norm": 0.08778655529022217, + "learning_rate": 9.61747964593121e-05, + "loss": 0.012, + "step": 43400 + }, + { + "epoch": 2.1327711315944304, + "grad_norm": 0.32174956798553467, + "learning_rate": 9.614190267713626e-05, + "loss": 0.0108, + "step": 43500 + }, + { + "epoch": 2.137674053736027, + "grad_norm": 0.03531063720583916, + "learning_rate": 9.610887374369612e-05, + "loss": 0.0113, + "step": 43600 + }, + { + "epoch": 2.142576975877623, + "grad_norm": 0.31483060121536255, + "learning_rate": 9.607570975573462e-05, + "loss": 0.0098, + "step": 43700 + }, + { + "epoch": 2.1474798980192196, + "grad_norm": 0.0525469034910202, + "learning_rate": 9.604241081039021e-05, + "loss": 0.0095, + "step": 43800 + }, + { + "epoch": 2.152382820160816, + "grad_norm": 0.11407173424959183, + "learning_rate": 9.600897700519668e-05, + "loss": 0.0104, + "step": 43900 + }, + { + "epoch": 2.1572857423024123, + "grad_norm": 0.14029061794281006, + "learning_rate": 9.597540843808286e-05, + "loss": 0.0108, + "step": 44000 + }, + { + "epoch": 2.1621886644440087, + "grad_norm": 0.21073657274246216, + "learning_rate": 9.594170520737227e-05, + "loss": 0.0112, + "step": 44100 + }, + { + "epoch": 2.167091586585605, + "grad_norm": 0.05304975062608719, + "learning_rate": 9.590786741178281e-05, + "loss": 0.0109, + "step": 44200 + }, + { + "epoch": 2.1719945087272015, + "grad_norm": 0.04336349293589592, + "learning_rate": 9.587389515042664e-05, + "loss": 0.0107, + "step": 44300 + }, + { + "epoch": 2.176897430868798, + "grad_norm": 0.1172260120511055, + "learning_rate": 9.58397885228097e-05, + "loss": 0.0111, + "step": 44400 + }, + { + "epoch": 2.181800353010394, + "grad_norm": 0.18478664755821228, + "learning_rate": 9.580554762883148e-05, + "loss": 0.0091, + "step": 44500 + }, + { + "epoch": 2.1867032751519906, + "grad_norm": 0.09167050570249557, + "learning_rate": 9.57711725687848e-05, + "loss": 0.0114, + "step": 44600 + }, + { + "epoch": 2.191606197293587, + "grad_norm": 0.01435750164091587, + "learning_rate": 9.573666344335541e-05, + "loss": 0.0101, + "step": 44700 + }, + { + "epoch": 2.1965091194351833, + "grad_norm": 0.01749005913734436, + "learning_rate": 9.570202035362177e-05, + "loss": 0.0107, + "step": 44800 + }, + { + "epoch": 2.2014120415767797, + "grad_norm": 0.27203837037086487, + "learning_rate": 9.56672434010547e-05, + "loss": 0.0118, + "step": 44900 + }, + { + "epoch": 2.206314963718376, + "grad_norm": 0.29238438606262207, + "learning_rate": 9.563233268751712e-05, + "loss": 0.0103, + "step": 45000 + }, + { + "epoch": 2.2112178858599725, + "grad_norm": 0.026590676978230476, + "learning_rate": 9.559728831526378e-05, + "loss": 0.0128, + "step": 45100 + }, + { + "epoch": 2.216120808001569, + "grad_norm": 0.2632576525211334, + "learning_rate": 9.556211038694084e-05, + "loss": 0.0103, + "step": 45200 + }, + { + "epoch": 2.2210237301431652, + "grad_norm": 0.096591517329216, + "learning_rate": 9.552679900558574e-05, + "loss": 0.0082, + "step": 45300 + }, + { + "epoch": 2.2259266522847616, + "grad_norm": 0.10889133810997009, + "learning_rate": 9.549135427462672e-05, + "loss": 0.0105, + "step": 45400 + }, + { + "epoch": 2.230829574426358, + "grad_norm": 0.44789671897888184, + "learning_rate": 9.545577629788269e-05, + "loss": 0.0114, + "step": 45500 + }, + { + "epoch": 2.2357324965679544, + "grad_norm": 0.46069878339767456, + "learning_rate": 9.542006517956277e-05, + "loss": 0.0119, + "step": 45600 + }, + { + "epoch": 2.2406354187095507, + "grad_norm": 0.18736761808395386, + "learning_rate": 9.53842210242661e-05, + "loss": 0.0106, + "step": 45700 + }, + { + "epoch": 2.245538340851147, + "grad_norm": 0.1695462465286255, + "learning_rate": 9.534824393698149e-05, + "loss": 0.0124, + "step": 45800 + }, + { + "epoch": 2.2504412629927435, + "grad_norm": 0.22079171240329742, + "learning_rate": 9.531213402308709e-05, + "loss": 0.01, + "step": 45900 + }, + { + "epoch": 2.25534418513434, + "grad_norm": 0.18455089628696442, + "learning_rate": 9.527589138835011e-05, + "loss": 0.0118, + "step": 46000 + }, + { + "epoch": 2.2602471072759363, + "grad_norm": 0.08626440167427063, + "learning_rate": 9.52395161389265e-05, + "loss": 0.0098, + "step": 46100 + }, + { + "epoch": 2.2651500294175326, + "grad_norm": 0.15146349370479584, + "learning_rate": 9.520300838136069e-05, + "loss": 0.0122, + "step": 46200 + }, + { + "epoch": 2.270052951559129, + "grad_norm": 0.06669634580612183, + "learning_rate": 9.516636822258514e-05, + "loss": 0.011, + "step": 46300 + }, + { + "epoch": 2.2749558737007254, + "grad_norm": 0.121702179312706, + "learning_rate": 9.512959576992022e-05, + "loss": 0.0125, + "step": 46400 + }, + { + "epoch": 2.279858795842322, + "grad_norm": 0.10746506601572037, + "learning_rate": 9.509269113107371e-05, + "loss": 0.0111, + "step": 46500 + }, + { + "epoch": 2.2847617179839186, + "grad_norm": 0.30326399207115173, + "learning_rate": 9.505565441414061e-05, + "loss": 0.0105, + "step": 46600 + }, + { + "epoch": 2.289664640125515, + "grad_norm": 0.12522533535957336, + "learning_rate": 9.50184857276028e-05, + "loss": 0.013, + "step": 46700 + }, + { + "epoch": 2.2945675622671113, + "grad_norm": 0.19467896223068237, + "learning_rate": 9.498118518032864e-05, + "loss": 0.0099, + "step": 46800 + }, + { + "epoch": 2.2994704844087077, + "grad_norm": 0.22523294389247894, + "learning_rate": 9.494375288157276e-05, + "loss": 0.0094, + "step": 46900 + }, + { + "epoch": 2.304373406550304, + "grad_norm": 0.07772246748209, + "learning_rate": 9.490618894097569e-05, + "loss": 0.0098, + "step": 47000 + }, + { + "epoch": 2.3092763286919005, + "grad_norm": 0.40480080246925354, + "learning_rate": 9.486849346856355e-05, + "loss": 0.0102, + "step": 47100 + }, + { + "epoch": 2.314179250833497, + "grad_norm": 0.15440988540649414, + "learning_rate": 9.483066657474771e-05, + "loss": 0.0113, + "step": 47200 + }, + { + "epoch": 2.3190821729750932, + "grad_norm": 0.24852220714092255, + "learning_rate": 9.479270837032446e-05, + "loss": 0.011, + "step": 47300 + }, + { + "epoch": 2.3239850951166896, + "grad_norm": 0.04837888851761818, + "learning_rate": 9.475461896647475e-05, + "loss": 0.0099, + "step": 47400 + }, + { + "epoch": 2.328888017258286, + "grad_norm": 0.10291013866662979, + "learning_rate": 9.471639847476377e-05, + "loss": 0.01, + "step": 47500 + }, + { + "epoch": 2.3337909393998824, + "grad_norm": 0.1631251573562622, + "learning_rate": 9.46780470071407e-05, + "loss": 0.0114, + "step": 47600 + }, + { + "epoch": 2.3386938615414787, + "grad_norm": 0.10871821641921997, + "learning_rate": 9.463956467593833e-05, + "loss": 0.0109, + "step": 47700 + }, + { + "epoch": 2.343596783683075, + "grad_norm": 0.2547306716442108, + "learning_rate": 9.460095159387279e-05, + "loss": 0.0121, + "step": 47800 + }, + { + "epoch": 2.3484997058246715, + "grad_norm": 0.33700889348983765, + "learning_rate": 9.456220787404315e-05, + "loss": 0.0118, + "step": 47900 + }, + { + "epoch": 2.353402627966268, + "grad_norm": 0.21864359080791473, + "learning_rate": 9.452333362993115e-05, + "loss": 0.0113, + "step": 48000 + }, + { + "epoch": 2.3583055501078642, + "grad_norm": 0.30565133690834045, + "learning_rate": 9.44843289754008e-05, + "loss": 0.0119, + "step": 48100 + }, + { + "epoch": 2.3632084722494606, + "grad_norm": 0.3426356911659241, + "learning_rate": 9.444519402469813e-05, + "loss": 0.013, + "step": 48200 + }, + { + "epoch": 2.368111394391057, + "grad_norm": 0.260628342628479, + "learning_rate": 9.44059288924508e-05, + "loss": 0.0119, + "step": 48300 + }, + { + "epoch": 2.3730143165326534, + "grad_norm": 0.12320883572101593, + "learning_rate": 9.436653369366777e-05, + "loss": 0.0117, + "step": 48400 + }, + { + "epoch": 2.3779172386742498, + "grad_norm": 0.3849811851978302, + "learning_rate": 9.432700854373894e-05, + "loss": 0.0118, + "step": 48500 + }, + { + "epoch": 2.382820160815846, + "grad_norm": 0.27962300181388855, + "learning_rate": 9.42873535584349e-05, + "loss": 0.0126, + "step": 48600 + }, + { + "epoch": 2.3877230829574425, + "grad_norm": 0.08513534814119339, + "learning_rate": 9.424756885390649e-05, + "loss": 0.0116, + "step": 48700 + }, + { + "epoch": 2.392626005099039, + "grad_norm": 0.0896671712398529, + "learning_rate": 9.420765454668454e-05, + "loss": 0.0108, + "step": 48800 + }, + { + "epoch": 2.3975289272406353, + "grad_norm": 0.2822389304637909, + "learning_rate": 9.416761075367944e-05, + "loss": 0.0104, + "step": 48900 + }, + { + "epoch": 2.4024318493822316, + "grad_norm": 0.15120819211006165, + "learning_rate": 9.412743759218089e-05, + "loss": 0.0112, + "step": 49000 + }, + { + "epoch": 2.407334771523828, + "grad_norm": 0.09544061869382858, + "learning_rate": 9.408713517985748e-05, + "loss": 0.0109, + "step": 49100 + }, + { + "epoch": 2.412237693665425, + "grad_norm": 0.2666862905025482, + "learning_rate": 9.404670363475645e-05, + "loss": 0.0105, + "step": 49200 + }, + { + "epoch": 2.417140615807021, + "grad_norm": 0.2836065888404846, + "learning_rate": 9.400614307530316e-05, + "loss": 0.0105, + "step": 49300 + }, + { + "epoch": 2.4220435379486176, + "grad_norm": 0.110966756939888, + "learning_rate": 9.396545362030096e-05, + "loss": 0.0119, + "step": 49400 + }, + { + "epoch": 2.426946460090214, + "grad_norm": 0.09849441796541214, + "learning_rate": 9.392463538893071e-05, + "loss": 0.0106, + "step": 49500 + }, + { + "epoch": 2.4318493822318104, + "grad_norm": 0.21830490231513977, + "learning_rate": 9.388368850075043e-05, + "loss": 0.0101, + "step": 49600 + }, + { + "epoch": 2.4367523043734067, + "grad_norm": 0.07172509282827377, + "learning_rate": 9.3842613075695e-05, + "loss": 0.01, + "step": 49700 + }, + { + "epoch": 2.441655226515003, + "grad_norm": 0.21159067749977112, + "learning_rate": 9.380140923407583e-05, + "loss": 0.0117, + "step": 49800 + }, + { + "epoch": 2.4465581486565995, + "grad_norm": 0.046259116381406784, + "learning_rate": 9.376007709658039e-05, + "loss": 0.0113, + "step": 49900 + }, + { + "epoch": 2.451461070798196, + "grad_norm": 0.2556672692298889, + "learning_rate": 9.371861678427198e-05, + "loss": 0.0111, + "step": 50000 + }, + { + "epoch": 2.4563639929397922, + "grad_norm": 0.1805562973022461, + "learning_rate": 9.367702841858935e-05, + "loss": 0.0117, + "step": 50100 + }, + { + "epoch": 2.4612669150813886, + "grad_norm": 0.05842866376042366, + "learning_rate": 9.363531212134625e-05, + "loss": 0.0109, + "step": 50200 + }, + { + "epoch": 2.466169837222985, + "grad_norm": 0.20916195213794708, + "learning_rate": 9.359346801473121e-05, + "loss": 0.0108, + "step": 50300 + }, + { + "epoch": 2.4710727593645814, + "grad_norm": 0.17823529243469238, + "learning_rate": 9.35514962213071e-05, + "loss": 0.0118, + "step": 50400 + }, + { + "epoch": 2.4759756815061778, + "grad_norm": 0.14857697486877441, + "learning_rate": 9.350939686401077e-05, + "loss": 0.0126, + "step": 50500 + }, + { + "epoch": 2.480878603647774, + "grad_norm": 0.13901618123054504, + "learning_rate": 9.346717006615276e-05, + "loss": 0.013, + "step": 50600 + }, + { + "epoch": 2.4857815257893705, + "grad_norm": 0.21642690896987915, + "learning_rate": 9.342481595141683e-05, + "loss": 0.0094, + "step": 50700 + }, + { + "epoch": 2.490684447930967, + "grad_norm": 0.0339026153087616, + "learning_rate": 9.338233464385968e-05, + "loss": 0.0106, + "step": 50800 + }, + { + "epoch": 2.4955873700725633, + "grad_norm": 0.06651703268289566, + "learning_rate": 9.333972626791057e-05, + "loss": 0.0118, + "step": 50900 + }, + { + "epoch": 2.5004902922141596, + "grad_norm": 0.22575177252292633, + "learning_rate": 9.329699094837092e-05, + "loss": 0.0097, + "step": 51000 + }, + { + "epoch": 2.505393214355756, + "grad_norm": 0.16404572129249573, + "learning_rate": 9.325412881041402e-05, + "loss": 0.0122, + "step": 51100 + }, + { + "epoch": 2.5102961364973524, + "grad_norm": 0.14393766224384308, + "learning_rate": 9.321113997958458e-05, + "loss": 0.013, + "step": 51200 + }, + { + "epoch": 2.5151990586389488, + "grad_norm": 0.26015493273735046, + "learning_rate": 9.316802458179839e-05, + "loss": 0.011, + "step": 51300 + }, + { + "epoch": 2.520101980780545, + "grad_norm": 0.21623893082141876, + "learning_rate": 9.312478274334197e-05, + "loss": 0.012, + "step": 51400 + }, + { + "epoch": 2.5250049029221415, + "grad_norm": 0.035357438027858734, + "learning_rate": 9.308141459087222e-05, + "loss": 0.0111, + "step": 51500 + }, + { + "epoch": 2.529907825063738, + "grad_norm": 0.19075345993041992, + "learning_rate": 9.303792025141597e-05, + "loss": 0.0119, + "step": 51600 + }, + { + "epoch": 2.5348107472053343, + "grad_norm": 0.13520535826683044, + "learning_rate": 9.29942998523697e-05, + "loss": 0.0105, + "step": 51700 + }, + { + "epoch": 2.5397136693469307, + "grad_norm": 0.07859195023775101, + "learning_rate": 9.295055352149909e-05, + "loss": 0.0125, + "step": 51800 + }, + { + "epoch": 2.544616591488527, + "grad_norm": 0.23639413714408875, + "learning_rate": 9.290668138693867e-05, + "loss": 0.0111, + "step": 51900 + }, + { + "epoch": 2.5495195136301234, + "grad_norm": 0.2909085750579834, + "learning_rate": 9.286268357719152e-05, + "loss": 0.0123, + "step": 52000 + }, + { + "epoch": 2.55442243577172, + "grad_norm": 0.0753180980682373, + "learning_rate": 9.281856022112874e-05, + "loss": 0.0132, + "step": 52100 + }, + { + "epoch": 2.559325357913316, + "grad_norm": 0.21286705136299133, + "learning_rate": 9.27743114479892e-05, + "loss": 0.0115, + "step": 52200 + }, + { + "epoch": 2.5642282800549125, + "grad_norm": 0.1657351553440094, + "learning_rate": 9.272993738737914e-05, + "loss": 0.0118, + "step": 52300 + }, + { + "epoch": 2.569131202196509, + "grad_norm": 0.18879827857017517, + "learning_rate": 9.268543816927174e-05, + "loss": 0.013, + "step": 52400 + }, + { + "epoch": 2.5740341243381053, + "grad_norm": 0.1545262485742569, + "learning_rate": 9.264081392400679e-05, + "loss": 0.0107, + "step": 52500 + }, + { + "epoch": 2.5789370464797017, + "grad_norm": 0.1411045491695404, + "learning_rate": 9.259606478229028e-05, + "loss": 0.0121, + "step": 52600 + }, + { + "epoch": 2.583839968621298, + "grad_norm": 0.15814360976219177, + "learning_rate": 9.255119087519404e-05, + "loss": 0.0114, + "step": 52700 + }, + { + "epoch": 2.5887428907628944, + "grad_norm": 0.10805512964725494, + "learning_rate": 9.25061923341553e-05, + "loss": 0.0103, + "step": 52800 + }, + { + "epoch": 2.593645812904491, + "grad_norm": 0.10512258112430573, + "learning_rate": 9.24610692909764e-05, + "loss": 0.0106, + "step": 52900 + }, + { + "epoch": 2.5985487350460876, + "grad_norm": 0.17147375643253326, + "learning_rate": 9.241582187782434e-05, + "loss": 0.0114, + "step": 53000 + }, + { + "epoch": 2.603451657187684, + "grad_norm": 0.23781518638134003, + "learning_rate": 9.237045022723036e-05, + "loss": 0.0101, + "step": 53100 + }, + { + "epoch": 2.6083545793292804, + "grad_norm": 0.2328936904668808, + "learning_rate": 9.232495447208967e-05, + "loss": 0.013, + "step": 53200 + }, + { + "epoch": 2.6132575014708768, + "grad_norm": 0.39143115282058716, + "learning_rate": 9.22793347456609e-05, + "loss": 0.0117, + "step": 53300 + }, + { + "epoch": 2.618160423612473, + "grad_norm": 0.19932937622070312, + "learning_rate": 9.223359118156587e-05, + "loss": 0.0125, + "step": 53400 + }, + { + "epoch": 2.6230633457540695, + "grad_norm": 0.31379079818725586, + "learning_rate": 9.218772391378908e-05, + "loss": 0.0099, + "step": 53500 + }, + { + "epoch": 2.627966267895666, + "grad_norm": 0.3086315095424652, + "learning_rate": 9.214173307667739e-05, + "loss": 0.0122, + "step": 53600 + }, + { + "epoch": 2.6328691900372623, + "grad_norm": 0.1653462052345276, + "learning_rate": 9.209561880493961e-05, + "loss": 0.0157, + "step": 53700 + }, + { + "epoch": 2.6377721121788587, + "grad_norm": 0.2647700309753418, + "learning_rate": 9.204938123364604e-05, + "loss": 0.0118, + "step": 53800 + }, + { + "epoch": 2.642675034320455, + "grad_norm": 0.042278312146663666, + "learning_rate": 9.200302049822815e-05, + "loss": 0.0143, + "step": 53900 + }, + { + "epoch": 2.6475779564620514, + "grad_norm": 0.162938192486763, + "learning_rate": 9.195653673447822e-05, + "loss": 0.013, + "step": 54000 + }, + { + "epoch": 2.652480878603648, + "grad_norm": 0.06934090703725815, + "learning_rate": 9.19099300785488e-05, + "loss": 0.0109, + "step": 54100 + }, + { + "epoch": 2.657383800745244, + "grad_norm": 0.32144516706466675, + "learning_rate": 9.186320066695244e-05, + "loss": 0.0122, + "step": 54200 + }, + { + "epoch": 2.6622867228868405, + "grad_norm": 0.1582040935754776, + "learning_rate": 9.181634863656125e-05, + "loss": 0.0108, + "step": 54300 + }, + { + "epoch": 2.667189645028437, + "grad_norm": 0.05845705047249794, + "learning_rate": 9.176937412460647e-05, + "loss": 0.0109, + "step": 54400 + }, + { + "epoch": 2.6720925671700333, + "grad_norm": 0.32942378520965576, + "learning_rate": 9.172227726867811e-05, + "loss": 0.0131, + "step": 54500 + }, + { + "epoch": 2.6769954893116297, + "grad_norm": 0.1683381050825119, + "learning_rate": 9.167505820672453e-05, + "loss": 0.0114, + "step": 54600 + }, + { + "epoch": 2.681898411453226, + "grad_norm": 0.061455611139535904, + "learning_rate": 9.162771707705204e-05, + "loss": 0.0114, + "step": 54700 + }, + { + "epoch": 2.6868013335948224, + "grad_norm": 0.25896042585372925, + "learning_rate": 9.158025401832447e-05, + "loss": 0.0117, + "step": 54800 + }, + { + "epoch": 2.691704255736419, + "grad_norm": 0.18199358880519867, + "learning_rate": 9.15326691695628e-05, + "loss": 0.0118, + "step": 54900 + }, + { + "epoch": 2.696607177878015, + "grad_norm": 0.44476819038391113, + "learning_rate": 9.148496267014472e-05, + "loss": 0.0106, + "step": 55000 + }, + { + "epoch": 2.7015101000196116, + "grad_norm": 0.16368268430233002, + "learning_rate": 9.143713465980428e-05, + "loss": 0.0112, + "step": 55100 + }, + { + "epoch": 2.706413022161208, + "grad_norm": 0.23616598546504974, + "learning_rate": 9.13891852786314e-05, + "loss": 0.0126, + "step": 55200 + }, + { + "epoch": 2.7113159443028043, + "grad_norm": 0.2906864285469055, + "learning_rate": 9.134111466707151e-05, + "loss": 0.0127, + "step": 55300 + }, + { + "epoch": 2.716218866444401, + "grad_norm": 0.14189335703849792, + "learning_rate": 9.129292296592514e-05, + "loss": 0.0111, + "step": 55400 + }, + { + "epoch": 2.7211217885859975, + "grad_norm": 0.08627419173717499, + "learning_rate": 9.124461031634748e-05, + "loss": 0.0108, + "step": 55500 + }, + { + "epoch": 2.726024710727594, + "grad_norm": 0.13704799115657806, + "learning_rate": 9.1196176859848e-05, + "loss": 0.0129, + "step": 55600 + }, + { + "epoch": 2.7309276328691903, + "grad_norm": 0.05801895260810852, + "learning_rate": 9.114762273828998e-05, + "loss": 0.011, + "step": 55700 + }, + { + "epoch": 2.7358305550107866, + "grad_norm": 0.31558868288993835, + "learning_rate": 9.10989480938902e-05, + "loss": 0.0125, + "step": 55800 + }, + { + "epoch": 2.740733477152383, + "grad_norm": 0.06790540367364883, + "learning_rate": 9.105015306921836e-05, + "loss": 0.0111, + "step": 55900 + }, + { + "epoch": 2.7456363992939794, + "grad_norm": 0.3000267744064331, + "learning_rate": 9.100123780719685e-05, + "loss": 0.0121, + "step": 56000 + }, + { + "epoch": 2.750539321435576, + "grad_norm": 0.22068212926387787, + "learning_rate": 9.095220245110019e-05, + "loss": 0.0106, + "step": 56100 + }, + { + "epoch": 2.755442243577172, + "grad_norm": 0.18099111318588257, + "learning_rate": 9.090304714455469e-05, + "loss": 0.0122, + "step": 56200 + }, + { + "epoch": 2.7603451657187685, + "grad_norm": 0.2899315059185028, + "learning_rate": 9.085377203153795e-05, + "loss": 0.0124, + "step": 56300 + }, + { + "epoch": 2.765248087860365, + "grad_norm": 0.3247213363647461, + "learning_rate": 9.080437725637854e-05, + "loss": 0.0122, + "step": 56400 + }, + { + "epoch": 2.7701510100019613, + "grad_norm": 0.07788939774036407, + "learning_rate": 9.07548629637555e-05, + "loss": 0.0106, + "step": 56500 + }, + { + "epoch": 2.7750539321435577, + "grad_norm": 0.13057078421115875, + "learning_rate": 9.070522929869796e-05, + "loss": 0.0111, + "step": 56600 + }, + { + "epoch": 2.779956854285154, + "grad_norm": 0.0548156201839447, + "learning_rate": 9.065547640658462e-05, + "loss": 0.0118, + "step": 56700 + }, + { + "epoch": 2.7848597764267504, + "grad_norm": 0.24302823841571808, + "learning_rate": 9.060560443314353e-05, + "loss": 0.0116, + "step": 56800 + }, + { + "epoch": 2.789762698568347, + "grad_norm": 0.27404043078422546, + "learning_rate": 9.055561352445143e-05, + "loss": 0.013, + "step": 56900 + }, + { + "epoch": 2.794665620709943, + "grad_norm": 0.5266895890235901, + "learning_rate": 9.050550382693346e-05, + "loss": 0.01, + "step": 57000 + }, + { + "epoch": 2.7995685428515396, + "grad_norm": 0.1247442364692688, + "learning_rate": 9.04552754873627e-05, + "loss": 0.011, + "step": 57100 + }, + { + "epoch": 2.804471464993136, + "grad_norm": 0.24913443624973297, + "learning_rate": 9.040492865285973e-05, + "loss": 0.0119, + "step": 57200 + }, + { + "epoch": 2.8093743871347323, + "grad_norm": 0.11120960861444473, + "learning_rate": 9.035446347089221e-05, + "loss": 0.012, + "step": 57300 + }, + { + "epoch": 2.8142773092763287, + "grad_norm": 0.2342406064271927, + "learning_rate": 9.030388008927444e-05, + "loss": 0.0128, + "step": 57400 + }, + { + "epoch": 2.819180231417925, + "grad_norm": 0.19678667187690735, + "learning_rate": 9.025317865616693e-05, + "loss": 0.0142, + "step": 57500 + }, + { + "epoch": 2.8240831535595214, + "grad_norm": 0.342568576335907, + "learning_rate": 9.020235932007597e-05, + "loss": 0.0104, + "step": 57600 + }, + { + "epoch": 2.828986075701118, + "grad_norm": 0.27162832021713257, + "learning_rate": 9.015142222985322e-05, + "loss": 0.0105, + "step": 57700 + }, + { + "epoch": 2.833888997842714, + "grad_norm": 0.3186455965042114, + "learning_rate": 9.010036753469518e-05, + "loss": 0.0121, + "step": 57800 + }, + { + "epoch": 2.8387919199843106, + "grad_norm": 0.3207502067089081, + "learning_rate": 9.004919538414287e-05, + "loss": 0.0124, + "step": 57900 + }, + { + "epoch": 2.843694842125907, + "grad_norm": 0.42707347869873047, + "learning_rate": 8.999790592808133e-05, + "loss": 0.0126, + "step": 58000 + }, + { + "epoch": 2.8485977642675033, + "grad_norm": 0.24902117252349854, + "learning_rate": 8.99464993167392e-05, + "loss": 0.0113, + "step": 58100 + }, + { + "epoch": 2.8535006864090997, + "grad_norm": 0.16693954169750214, + "learning_rate": 8.989497570068824e-05, + "loss": 0.0126, + "step": 58200 + }, + { + "epoch": 2.858403608550696, + "grad_norm": 0.41618144512176514, + "learning_rate": 8.984333523084298e-05, + "loss": 0.0132, + "step": 58300 + }, + { + "epoch": 2.8633065306922925, + "grad_norm": 0.36843499541282654, + "learning_rate": 8.979157805846013e-05, + "loss": 0.0119, + "step": 58400 + }, + { + "epoch": 2.868209452833889, + "grad_norm": 0.3611968755722046, + "learning_rate": 8.973970433513832e-05, + "loss": 0.0104, + "step": 58500 + }, + { + "epoch": 2.873112374975485, + "grad_norm": 0.18182240426540375, + "learning_rate": 8.968771421281749e-05, + "loss": 0.0106, + "step": 58600 + }, + { + "epoch": 2.8780152971170816, + "grad_norm": 0.21805822849273682, + "learning_rate": 8.963560784377858e-05, + "loss": 0.0117, + "step": 58700 + }, + { + "epoch": 2.882918219258678, + "grad_norm": 0.1938275694847107, + "learning_rate": 8.958338538064297e-05, + "loss": 0.0122, + "step": 58800 + }, + { + "epoch": 2.8878211414002743, + "grad_norm": 0.27438417077064514, + "learning_rate": 8.953104697637209e-05, + "loss": 0.0131, + "step": 58900 + }, + { + "epoch": 2.8927240635418707, + "grad_norm": 0.21847118437290192, + "learning_rate": 8.9478592784267e-05, + "loss": 0.011, + "step": 59000 + }, + { + "epoch": 2.897626985683467, + "grad_norm": 0.3260138928890228, + "learning_rate": 8.942602295796785e-05, + "loss": 0.0135, + "step": 59100 + }, + { + "epoch": 2.9025299078250635, + "grad_norm": 0.2678786814212799, + "learning_rate": 8.937333765145355e-05, + "loss": 0.013, + "step": 59200 + }, + { + "epoch": 2.90743282996666, + "grad_norm": 0.352421373128891, + "learning_rate": 8.932053701904122e-05, + "loss": 0.0131, + "step": 59300 + }, + { + "epoch": 2.9123357521082567, + "grad_norm": 0.25367945432662964, + "learning_rate": 8.926762121538577e-05, + "loss": 0.0125, + "step": 59400 + }, + { + "epoch": 2.917238674249853, + "grad_norm": 0.07467200607061386, + "learning_rate": 8.921459039547946e-05, + "loss": 0.0105, + "step": 59500 + }, + { + "epoch": 2.9221415963914494, + "grad_norm": 0.42252108454704285, + "learning_rate": 8.916144471465144e-05, + "loss": 0.012, + "step": 59600 + }, + { + "epoch": 2.927044518533046, + "grad_norm": 0.14323803782463074, + "learning_rate": 8.910818432856728e-05, + "loss": 0.0109, + "step": 59700 + }, + { + "epoch": 2.931947440674642, + "grad_norm": 0.18532708287239075, + "learning_rate": 8.905480939322855e-05, + "loss": 0.0128, + "step": 59800 + }, + { + "epoch": 2.9368503628162386, + "grad_norm": 0.30199986696243286, + "learning_rate": 8.900132006497231e-05, + "loss": 0.0123, + "step": 59900 + }, + { + "epoch": 2.941753284957835, + "grad_norm": 0.0945155918598175, + "learning_rate": 8.894771650047072e-05, + "loss": 0.0135, + "step": 60000 + }, + { + "epoch": 2.9466562070994313, + "grad_norm": 0.1907569169998169, + "learning_rate": 8.889399885673046e-05, + "loss": 0.0133, + "step": 60100 + }, + { + "epoch": 2.9515591292410277, + "grad_norm": 0.048369161784648895, + "learning_rate": 8.884016729109246e-05, + "loss": 0.0121, + "step": 60200 + }, + { + "epoch": 2.956462051382624, + "grad_norm": 0.28431326150894165, + "learning_rate": 8.878622196123127e-05, + "loss": 0.0141, + "step": 60300 + }, + { + "epoch": 2.9613649735242205, + "grad_norm": 0.3524870276451111, + "learning_rate": 8.873216302515463e-05, + "loss": 0.0121, + "step": 60400 + }, + { + "epoch": 2.966267895665817, + "grad_norm": 0.10910056531429291, + "learning_rate": 8.867799064120313e-05, + "loss": 0.011, + "step": 60500 + }, + { + "epoch": 2.971170817807413, + "grad_norm": 0.5855691432952881, + "learning_rate": 8.862370496804958e-05, + "loss": 0.0112, + "step": 60600 + }, + { + "epoch": 2.9760737399490096, + "grad_norm": 0.2968946099281311, + "learning_rate": 8.856930616469861e-05, + "loss": 0.0095, + "step": 60700 + }, + { + "epoch": 2.980976662090606, + "grad_norm": 0.07704410701990128, + "learning_rate": 8.851479439048624e-05, + "loss": 0.0109, + "step": 60800 + }, + { + "epoch": 2.9858795842322023, + "grad_norm": 0.11921142786741257, + "learning_rate": 8.846016980507942e-05, + "loss": 0.0131, + "step": 60900 + }, + { + "epoch": 2.9907825063737987, + "grad_norm": 0.1547405868768692, + "learning_rate": 8.840543256847548e-05, + "loss": 0.0131, + "step": 61000 + }, + { + "epoch": 2.995685428515395, + "grad_norm": 0.1697314977645874, + "learning_rate": 8.83505828410017e-05, + "loss": 0.0127, + "step": 61100 + }, + { + "epoch": 3.0, + "eval_loss": 0.030867407098412514, + "eval_runtime": 223.9786, + "eval_samples_per_second": 22.324, + "eval_steps_per_second": 22.324, + "step": 61188 + } + ], + "logging_steps": 100, + "max_steps": 203960, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 6.448056809124618e+18, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}