| { |
| "best_global_step": 166236, |
| "best_metric": 0.7059909142532652, |
| "best_model_checkpoint": "modernbert-heritage-classification/checkpoint-166236", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 249354, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0060155441661252675, |
| "grad_norm": 3.192460536956787, |
| "learning_rate": 1.996005678673693e-05, |
| "loss": 0.4511, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.012031088332250535, |
| "grad_norm": 2.3000173568725586, |
| "learning_rate": 1.992003336621831e-05, |
| "loss": 0.4198, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.018046632498375802, |
| "grad_norm": 6.538833141326904, |
| "learning_rate": 1.9879929738444142e-05, |
| "loss": 0.4086, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.02406217666450107, |
| "grad_norm": 0.18448229134082794, |
| "learning_rate": 1.9839826110669975e-05, |
| "loss": 0.3574, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.030077720830626337, |
| "grad_norm": 5.782385349273682, |
| "learning_rate": 1.9799722482895804e-05, |
| "loss": 0.3798, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.036093264996751605, |
| "grad_norm": 2.290853500366211, |
| "learning_rate": 1.9759699062377184e-05, |
| "loss": 0.38, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.042108809162876876, |
| "grad_norm": 4.156062126159668, |
| "learning_rate": 1.9719595434603017e-05, |
| "loss": 0.3838, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.04812435332900214, |
| "grad_norm": 2.1094439029693604, |
| "learning_rate": 1.967949180682885e-05, |
| "loss": 0.3625, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.05413989749512741, |
| "grad_norm": 8.78498363494873, |
| "learning_rate": 1.963938817905468e-05, |
| "loss": 0.3834, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.060155441661252675, |
| "grad_norm": 0.003325940575450659, |
| "learning_rate": 1.959928455128051e-05, |
| "loss": 0.3395, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.06617098582737795, |
| "grad_norm": 0.030467336997389793, |
| "learning_rate": 1.9559180923506344e-05, |
| "loss": 0.3605, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.07218652999350321, |
| "grad_norm": 0.7140659689903259, |
| "learning_rate": 1.9519077295732173e-05, |
| "loss": 0.352, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.07820207415962847, |
| "grad_norm": 0.013019168749451637, |
| "learning_rate": 1.9479053875213553e-05, |
| "loss": 0.3533, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.08421761832575375, |
| "grad_norm": 3.8174266815185547, |
| "learning_rate": 1.9438950247439386e-05, |
| "loss": 0.316, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.09023316249187902, |
| "grad_norm": 7.174667835235596, |
| "learning_rate": 1.939884661966522e-05, |
| "loss": 0.3495, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.09624870665800428, |
| "grad_norm": 1.6871285438537598, |
| "learning_rate": 1.9358742991891048e-05, |
| "loss": 0.3449, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.10226425082412956, |
| "grad_norm": 0.04142903909087181, |
| "learning_rate": 1.931863936411688e-05, |
| "loss": 0.3212, |
| "step": 8500 |
| }, |
| { |
| "epoch": 0.10827979499025482, |
| "grad_norm": 0.0040322355926036835, |
| "learning_rate": 1.9278535736342713e-05, |
| "loss": 0.3416, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.11429533915638009, |
| "grad_norm": 7.649967193603516, |
| "learning_rate": 1.9238512315824093e-05, |
| "loss": 0.3615, |
| "step": 9500 |
| }, |
| { |
| "epoch": 0.12031088332250535, |
| "grad_norm": 2.4331884384155273, |
| "learning_rate": 1.9198408688049922e-05, |
| "loss": 0.354, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.12632642748863063, |
| "grad_norm": 8.657504081726074, |
| "learning_rate": 1.9158305060275755e-05, |
| "loss": 0.3413, |
| "step": 10500 |
| }, |
| { |
| "epoch": 0.1323419716547559, |
| "grad_norm": 8.941267013549805, |
| "learning_rate": 1.9118201432501588e-05, |
| "loss": 0.3211, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.13835751582088116, |
| "grad_norm": 16.12137794494629, |
| "learning_rate": 1.9078178011982968e-05, |
| "loss": 0.3323, |
| "step": 11500 |
| }, |
| { |
| "epoch": 0.14437305998700642, |
| "grad_norm": 0.30122852325439453, |
| "learning_rate": 1.9038074384208797e-05, |
| "loss": 0.3163, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.15038860415313168, |
| "grad_norm": 0.2980821132659912, |
| "learning_rate": 1.899797075643463e-05, |
| "loss": 0.3264, |
| "step": 12500 |
| }, |
| { |
| "epoch": 0.15640414831925695, |
| "grad_norm": 0.006154663860797882, |
| "learning_rate": 1.8957867128660462e-05, |
| "loss": 0.3453, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.16241969248538224, |
| "grad_norm": 0.8030881285667419, |
| "learning_rate": 1.8917763500886292e-05, |
| "loss": 0.3152, |
| "step": 13500 |
| }, |
| { |
| "epoch": 0.1684352366515075, |
| "grad_norm": 17.813459396362305, |
| "learning_rate": 1.8877659873112124e-05, |
| "loss": 0.3189, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.17445078081763277, |
| "grad_norm": 1.1400656700134277, |
| "learning_rate": 1.8837556245337957e-05, |
| "loss": 0.3434, |
| "step": 14500 |
| }, |
| { |
| "epoch": 0.18046632498375803, |
| "grad_norm": 0.09848301857709885, |
| "learning_rate": 1.8797452617563786e-05, |
| "loss": 0.3087, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.1864818691498833, |
| "grad_norm": 0.04860129952430725, |
| "learning_rate": 1.875734898978962e-05, |
| "loss": 0.3289, |
| "step": 15500 |
| }, |
| { |
| "epoch": 0.19249741331600856, |
| "grad_norm": 0.012527902610599995, |
| "learning_rate": 1.8717325569271e-05, |
| "loss": 0.3324, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.19851295748213382, |
| "grad_norm": 5.626002788543701, |
| "learning_rate": 1.867730214875238e-05, |
| "loss": 0.3351, |
| "step": 16500 |
| }, |
| { |
| "epoch": 0.20452850164825911, |
| "grad_norm": 0.2671683430671692, |
| "learning_rate": 1.8637198520978212e-05, |
| "loss": 0.3467, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.21054404581438438, |
| "grad_norm": 0.4044207036495209, |
| "learning_rate": 1.859709489320404e-05, |
| "loss": 0.3248, |
| "step": 17500 |
| }, |
| { |
| "epoch": 0.21655958998050964, |
| "grad_norm": 0.19454579055309296, |
| "learning_rate": 1.8556991265429874e-05, |
| "loss": 0.339, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.2225751341466349, |
| "grad_norm": 13.63963508605957, |
| "learning_rate": 1.8516887637655703e-05, |
| "loss": 0.3252, |
| "step": 18500 |
| }, |
| { |
| "epoch": 0.22859067831276017, |
| "grad_norm": 3.8693294525146484, |
| "learning_rate": 1.8476784009881536e-05, |
| "loss": 0.3311, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.23460622247888543, |
| "grad_norm": 2.644678831100464, |
| "learning_rate": 1.8436760589362916e-05, |
| "loss": 0.3378, |
| "step": 19500 |
| }, |
| { |
| "epoch": 0.2406217666450107, |
| "grad_norm": 1.1504566669464111, |
| "learning_rate": 1.8396656961588745e-05, |
| "loss": 0.3162, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.24663731081113596, |
| "grad_norm": 0.33002209663391113, |
| "learning_rate": 1.8356553333814578e-05, |
| "loss": 0.3112, |
| "step": 20500 |
| }, |
| { |
| "epoch": 0.25265285497726125, |
| "grad_norm": 5.884070873260498, |
| "learning_rate": 1.831644970604041e-05, |
| "loss": 0.3375, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.2586683991433865, |
| "grad_norm": 0.004158430732786655, |
| "learning_rate": 1.827634607826624e-05, |
| "loss": 0.2711, |
| "step": 21500 |
| }, |
| { |
| "epoch": 0.2646839433095118, |
| "grad_norm": 6.601622581481934, |
| "learning_rate": 1.8236242450492072e-05, |
| "loss": 0.3031, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.27069948747563705, |
| "grad_norm": 4.164157390594482, |
| "learning_rate": 1.8196138822717905e-05, |
| "loss": 0.3228, |
| "step": 22500 |
| }, |
| { |
| "epoch": 0.2767150316417623, |
| "grad_norm": 34.43812942504883, |
| "learning_rate": 1.8156035194943734e-05, |
| "loss": 0.3061, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.2827305758078876, |
| "grad_norm": 0.02143733762204647, |
| "learning_rate": 1.8115931567169567e-05, |
| "loss": 0.3163, |
| "step": 23500 |
| }, |
| { |
| "epoch": 0.28874611997401284, |
| "grad_norm": 9.730279922485352, |
| "learning_rate": 1.8075908146650947e-05, |
| "loss": 0.2817, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.2947616641401381, |
| "grad_norm": 9.340431213378906, |
| "learning_rate": 1.803580451887678e-05, |
| "loss": 0.3105, |
| "step": 24500 |
| }, |
| { |
| "epoch": 0.30077720830626337, |
| "grad_norm": 9.22760009765625, |
| "learning_rate": 1.799570089110261e-05, |
| "loss": 0.309, |
| "step": 25000 |
| }, |
| { |
| "epoch": 0.30679275247238863, |
| "grad_norm": 3.7310335636138916, |
| "learning_rate": 1.795567747058399e-05, |
| "loss": 0.3075, |
| "step": 25500 |
| }, |
| { |
| "epoch": 0.3128082966385139, |
| "grad_norm": 4.968273639678955, |
| "learning_rate": 1.7915573842809822e-05, |
| "loss": 0.3056, |
| "step": 26000 |
| }, |
| { |
| "epoch": 0.3188238408046392, |
| "grad_norm": 15.309103012084961, |
| "learning_rate": 1.7875470215035654e-05, |
| "loss": 0.3063, |
| "step": 26500 |
| }, |
| { |
| "epoch": 0.3248393849707645, |
| "grad_norm": 0.028000958263874054, |
| "learning_rate": 1.7835366587261484e-05, |
| "loss": 0.3251, |
| "step": 27000 |
| }, |
| { |
| "epoch": 0.33085492913688974, |
| "grad_norm": 11.808887481689453, |
| "learning_rate": 1.7795262959487316e-05, |
| "loss": 0.3335, |
| "step": 27500 |
| }, |
| { |
| "epoch": 0.336870473303015, |
| "grad_norm": 6.1232123374938965, |
| "learning_rate": 1.775515933171315e-05, |
| "loss": 0.2928, |
| "step": 28000 |
| }, |
| { |
| "epoch": 0.34288601746914027, |
| "grad_norm": 2.071223258972168, |
| "learning_rate": 1.771505570393898e-05, |
| "loss": 0.3228, |
| "step": 28500 |
| }, |
| { |
| "epoch": 0.34890156163526553, |
| "grad_norm": 6.20293664932251, |
| "learning_rate": 1.767495207616481e-05, |
| "loss": 0.3115, |
| "step": 29000 |
| }, |
| { |
| "epoch": 0.3549171058013908, |
| "grad_norm": 0.5319348573684692, |
| "learning_rate": 1.7634848448390644e-05, |
| "loss": 0.3424, |
| "step": 29500 |
| }, |
| { |
| "epoch": 0.36093264996751606, |
| "grad_norm": 0.15102390944957733, |
| "learning_rate": 1.7594825027872024e-05, |
| "loss": 0.3198, |
| "step": 30000 |
| }, |
| { |
| "epoch": 0.3669481941336413, |
| "grad_norm": 10.45068359375, |
| "learning_rate": 1.7554721400097853e-05, |
| "loss": 0.3083, |
| "step": 30500 |
| }, |
| { |
| "epoch": 0.3729637382997666, |
| "grad_norm": 0.3780848979949951, |
| "learning_rate": 1.7514697979579233e-05, |
| "loss": 0.3079, |
| "step": 31000 |
| }, |
| { |
| "epoch": 0.37897928246589185, |
| "grad_norm": 0.02213098295032978, |
| "learning_rate": 1.7474674559060613e-05, |
| "loss": 0.301, |
| "step": 31500 |
| }, |
| { |
| "epoch": 0.3849948266320171, |
| "grad_norm": 2.598281145095825, |
| "learning_rate": 1.7434570931286446e-05, |
| "loss": 0.2878, |
| "step": 32000 |
| }, |
| { |
| "epoch": 0.3910103707981424, |
| "grad_norm": 0.14519445598125458, |
| "learning_rate": 1.739446730351228e-05, |
| "loss": 0.3231, |
| "step": 32500 |
| }, |
| { |
| "epoch": 0.39702591496426765, |
| "grad_norm": 0.04602254554629326, |
| "learning_rate": 1.7354363675738108e-05, |
| "loss": 0.2972, |
| "step": 33000 |
| }, |
| { |
| "epoch": 0.4030414591303929, |
| "grad_norm": 18.00914192199707, |
| "learning_rate": 1.731426004796394e-05, |
| "loss": 0.323, |
| "step": 33500 |
| }, |
| { |
| "epoch": 0.40905700329651823, |
| "grad_norm": 6.137303829193115, |
| "learning_rate": 1.7274156420189773e-05, |
| "loss": 0.3179, |
| "step": 34000 |
| }, |
| { |
| "epoch": 0.4150725474626435, |
| "grad_norm": 3.145047187805176, |
| "learning_rate": 1.7234052792415602e-05, |
| "loss": 0.306, |
| "step": 34500 |
| }, |
| { |
| "epoch": 0.42108809162876876, |
| "grad_norm": 4.087808609008789, |
| "learning_rate": 1.7193949164641435e-05, |
| "loss": 0.325, |
| "step": 35000 |
| }, |
| { |
| "epoch": 0.427103635794894, |
| "grad_norm": 0.009332289919257164, |
| "learning_rate": 1.7153845536867268e-05, |
| "loss": 0.2979, |
| "step": 35500 |
| }, |
| { |
| "epoch": 0.4331191799610193, |
| "grad_norm": 9.885307312011719, |
| "learning_rate": 1.7113741909093097e-05, |
| "loss": 0.3247, |
| "step": 36000 |
| }, |
| { |
| "epoch": 0.43913472412714455, |
| "grad_norm": 0.0056028529070317745, |
| "learning_rate": 1.707363828131893e-05, |
| "loss": 0.2949, |
| "step": 36500 |
| }, |
| { |
| "epoch": 0.4451502682932698, |
| "grad_norm": 8.444794654846191, |
| "learning_rate": 1.7033534653544762e-05, |
| "loss": 0.306, |
| "step": 37000 |
| }, |
| { |
| "epoch": 0.4511658124593951, |
| "grad_norm": 2.0681827068328857, |
| "learning_rate": 1.699343102577059e-05, |
| "loss": 0.2792, |
| "step": 37500 |
| }, |
| { |
| "epoch": 0.45718135662552034, |
| "grad_norm": 0.017593583092093468, |
| "learning_rate": 1.695340760525197e-05, |
| "loss": 0.3225, |
| "step": 38000 |
| }, |
| { |
| "epoch": 0.4631969007916456, |
| "grad_norm": 3.3854148387908936, |
| "learning_rate": 1.6913303977477804e-05, |
| "loss": 0.2962, |
| "step": 38500 |
| }, |
| { |
| "epoch": 0.46921244495777087, |
| "grad_norm": 0.4375062584877014, |
| "learning_rate": 1.6873200349703637e-05, |
| "loss": 0.3146, |
| "step": 39000 |
| }, |
| { |
| "epoch": 0.47522798912389613, |
| "grad_norm": 0.24493420124053955, |
| "learning_rate": 1.6833096721929466e-05, |
| "loss": 0.2745, |
| "step": 39500 |
| }, |
| { |
| "epoch": 0.4812435332900214, |
| "grad_norm": 5.30033540725708, |
| "learning_rate": 1.6793153508666397e-05, |
| "loss": 0.3242, |
| "step": 40000 |
| }, |
| { |
| "epoch": 0.48725907745614666, |
| "grad_norm": 0.08991783857345581, |
| "learning_rate": 1.6753049880892226e-05, |
| "loss": 0.2889, |
| "step": 40500 |
| }, |
| { |
| "epoch": 0.4932746216222719, |
| "grad_norm": 0.16233447194099426, |
| "learning_rate": 1.671294625311806e-05, |
| "loss": 0.3147, |
| "step": 41000 |
| }, |
| { |
| "epoch": 0.49929016578839724, |
| "grad_norm": 21.49981117248535, |
| "learning_rate": 1.6672842625343892e-05, |
| "loss": 0.3097, |
| "step": 41500 |
| }, |
| { |
| "epoch": 0.5053057099545225, |
| "grad_norm": 0.07683174312114716, |
| "learning_rate": 1.663281920482527e-05, |
| "loss": 0.3283, |
| "step": 42000 |
| }, |
| { |
| "epoch": 0.5113212541206478, |
| "grad_norm": 5.969061374664307, |
| "learning_rate": 1.65927155770511e-05, |
| "loss": 0.2973, |
| "step": 42500 |
| }, |
| { |
| "epoch": 0.517336798286773, |
| "grad_norm": 0.1165817528963089, |
| "learning_rate": 1.6552611949276934e-05, |
| "loss": 0.2865, |
| "step": 43000 |
| }, |
| { |
| "epoch": 0.5233523424528983, |
| "grad_norm": 4.852444648742676, |
| "learning_rate": 1.6512508321502766e-05, |
| "loss": 0.2995, |
| "step": 43500 |
| }, |
| { |
| "epoch": 0.5293678866190236, |
| "grad_norm": 0.3422677218914032, |
| "learning_rate": 1.6472404693728596e-05, |
| "loss": 0.3078, |
| "step": 44000 |
| }, |
| { |
| "epoch": 0.5353834307851488, |
| "grad_norm": 0.02138070948421955, |
| "learning_rate": 1.643230106595443e-05, |
| "loss": 0.3041, |
| "step": 44500 |
| }, |
| { |
| "epoch": 0.5413989749512741, |
| "grad_norm": 19.726158142089844, |
| "learning_rate": 1.639219743818026e-05, |
| "loss": 0.2747, |
| "step": 45000 |
| }, |
| { |
| "epoch": 0.5474145191173994, |
| "grad_norm": 0.8494102358818054, |
| "learning_rate": 1.635209381040609e-05, |
| "loss": 0.3165, |
| "step": 45500 |
| }, |
| { |
| "epoch": 0.5534300632835246, |
| "grad_norm": 69.28723907470703, |
| "learning_rate": 1.6311990182631923e-05, |
| "loss": 0.286, |
| "step": 46000 |
| }, |
| { |
| "epoch": 0.5594456074496499, |
| "grad_norm": 5.112706661224365, |
| "learning_rate": 1.6271886554857756e-05, |
| "loss": 0.3103, |
| "step": 46500 |
| }, |
| { |
| "epoch": 0.5654611516157751, |
| "grad_norm": 0.0373302698135376, |
| "learning_rate": 1.6231782927083585e-05, |
| "loss": 0.2966, |
| "step": 47000 |
| }, |
| { |
| "epoch": 0.5714766957819004, |
| "grad_norm": 0.3029526472091675, |
| "learning_rate": 1.6191679299309418e-05, |
| "loss": 0.2785, |
| "step": 47500 |
| }, |
| { |
| "epoch": 0.5774922399480257, |
| "grad_norm": 0.18373580276966095, |
| "learning_rate": 1.6151655878790798e-05, |
| "loss": 0.3328, |
| "step": 48000 |
| }, |
| { |
| "epoch": 0.5835077841141509, |
| "grad_norm": 0.8737627267837524, |
| "learning_rate": 1.611155225101663e-05, |
| "loss": 0.2909, |
| "step": 48500 |
| }, |
| { |
| "epoch": 0.5895233282802762, |
| "grad_norm": 2.7465710639953613, |
| "learning_rate": 1.607144862324246e-05, |
| "loss": 0.2991, |
| "step": 49000 |
| }, |
| { |
| "epoch": 0.5955388724464015, |
| "grad_norm": 19.38477325439453, |
| "learning_rate": 1.6031344995468292e-05, |
| "loss": 0.2785, |
| "step": 49500 |
| }, |
| { |
| "epoch": 0.6015544166125267, |
| "grad_norm": 0.013609528541564941, |
| "learning_rate": 1.5991241367694125e-05, |
| "loss": 0.3198, |
| "step": 50000 |
| }, |
| { |
| "epoch": 0.607569960778652, |
| "grad_norm": 6.444379806518555, |
| "learning_rate": 1.5951137739919954e-05, |
| "loss": 0.2816, |
| "step": 50500 |
| }, |
| { |
| "epoch": 0.6135855049447773, |
| "grad_norm": 6.260073661804199, |
| "learning_rate": 1.5911114319401334e-05, |
| "loss": 0.3193, |
| "step": 51000 |
| }, |
| { |
| "epoch": 0.6196010491109025, |
| "grad_norm": 8.19072151184082, |
| "learning_rate": 1.5871010691627167e-05, |
| "loss": 0.3021, |
| "step": 51500 |
| }, |
| { |
| "epoch": 0.6256165932770278, |
| "grad_norm": 5.27684211730957, |
| "learning_rate": 1.5830907063853e-05, |
| "loss": 0.2925, |
| "step": 52000 |
| }, |
| { |
| "epoch": 0.631632137443153, |
| "grad_norm": 0.031686898320913315, |
| "learning_rate": 1.579080343607883e-05, |
| "loss": 0.3063, |
| "step": 52500 |
| }, |
| { |
| "epoch": 0.6376476816092784, |
| "grad_norm": 0.15889908373355865, |
| "learning_rate": 1.575078001556021e-05, |
| "loss": 0.2893, |
| "step": 53000 |
| }, |
| { |
| "epoch": 0.6436632257754037, |
| "grad_norm": 5.396053314208984, |
| "learning_rate": 1.571067638778604e-05, |
| "loss": 0.3009, |
| "step": 53500 |
| }, |
| { |
| "epoch": 0.649678769941529, |
| "grad_norm": 0.15985211730003357, |
| "learning_rate": 1.5670572760011874e-05, |
| "loss": 0.2985, |
| "step": 54000 |
| }, |
| { |
| "epoch": 0.6556943141076542, |
| "grad_norm": 2.195845127105713, |
| "learning_rate": 1.5630469132237704e-05, |
| "loss": 0.2886, |
| "step": 54500 |
| }, |
| { |
| "epoch": 0.6617098582737795, |
| "grad_norm": 0.000236680411035195, |
| "learning_rate": 1.5590365504463536e-05, |
| "loss": 0.3041, |
| "step": 55000 |
| }, |
| { |
| "epoch": 0.6677254024399047, |
| "grad_norm": 0.012887202203273773, |
| "learning_rate": 1.555026187668937e-05, |
| "loss": 0.2848, |
| "step": 55500 |
| }, |
| { |
| "epoch": 0.67374094660603, |
| "grad_norm": 0.025558780878782272, |
| "learning_rate": 1.5510158248915198e-05, |
| "loss": 0.3037, |
| "step": 56000 |
| }, |
| { |
| "epoch": 0.6797564907721553, |
| "grad_norm": 13.303611755371094, |
| "learning_rate": 1.547013482839658e-05, |
| "loss": 0.3052, |
| "step": 56500 |
| }, |
| { |
| "epoch": 0.6857720349382805, |
| "grad_norm": 7.728808403015137, |
| "learning_rate": 1.543003120062241e-05, |
| "loss": 0.3076, |
| "step": 57000 |
| }, |
| { |
| "epoch": 0.6917875791044058, |
| "grad_norm": 7.381441593170166, |
| "learning_rate": 1.538992757284824e-05, |
| "loss": 0.2838, |
| "step": 57500 |
| }, |
| { |
| "epoch": 0.6978031232705311, |
| "grad_norm": 47.043766021728516, |
| "learning_rate": 1.5349823945074073e-05, |
| "loss": 0.2894, |
| "step": 58000 |
| }, |
| { |
| "epoch": 0.7038186674366563, |
| "grad_norm": 1.3084138631820679, |
| "learning_rate": 1.5309720317299902e-05, |
| "loss": 0.3089, |
| "step": 58500 |
| }, |
| { |
| "epoch": 0.7098342116027816, |
| "grad_norm": 0.28533193469047546, |
| "learning_rate": 1.5269616689525735e-05, |
| "loss": 0.3291, |
| "step": 59000 |
| }, |
| { |
| "epoch": 0.7158497557689069, |
| "grad_norm": 0.006308581214398146, |
| "learning_rate": 1.5229513061751566e-05, |
| "loss": 0.2842, |
| "step": 59500 |
| }, |
| { |
| "epoch": 0.7218652999350321, |
| "grad_norm": 0.18038663268089294, |
| "learning_rate": 1.5189409433977399e-05, |
| "loss": 0.2964, |
| "step": 60000 |
| }, |
| { |
| "epoch": 0.7278808441011574, |
| "grad_norm": 9.797381401062012, |
| "learning_rate": 1.514930580620323e-05, |
| "loss": 0.3303, |
| "step": 60500 |
| }, |
| { |
| "epoch": 0.7338963882672827, |
| "grad_norm": 0.27853924036026, |
| "learning_rate": 1.510920217842906e-05, |
| "loss": 0.2762, |
| "step": 61000 |
| }, |
| { |
| "epoch": 0.7399119324334079, |
| "grad_norm": 0.13307029008865356, |
| "learning_rate": 1.506917875791044e-05, |
| "loss": 0.2798, |
| "step": 61500 |
| }, |
| { |
| "epoch": 0.7459274765995332, |
| "grad_norm": 4.89689826965332, |
| "learning_rate": 1.5029075130136273e-05, |
| "loss": 0.3202, |
| "step": 62000 |
| }, |
| { |
| "epoch": 0.7519430207656584, |
| "grad_norm": 14.662346839904785, |
| "learning_rate": 1.4988971502362104e-05, |
| "loss": 0.3051, |
| "step": 62500 |
| }, |
| { |
| "epoch": 0.7579585649317837, |
| "grad_norm": 0.15676206350326538, |
| "learning_rate": 1.4948867874587935e-05, |
| "loss": 0.3065, |
| "step": 63000 |
| }, |
| { |
| "epoch": 0.763974109097909, |
| "grad_norm": 10.646651268005371, |
| "learning_rate": 1.4908764246813768e-05, |
| "loss": 0.2741, |
| "step": 63500 |
| }, |
| { |
| "epoch": 0.7699896532640342, |
| "grad_norm": 7.507605075836182, |
| "learning_rate": 1.4868740826295148e-05, |
| "loss": 0.3195, |
| "step": 64000 |
| }, |
| { |
| "epoch": 0.7760051974301595, |
| "grad_norm": 0.24443697929382324, |
| "learning_rate": 1.4828717405776526e-05, |
| "loss": 0.2662, |
| "step": 64500 |
| }, |
| { |
| "epoch": 0.7820207415962848, |
| "grad_norm": 7.655556678771973, |
| "learning_rate": 1.4788613778002359e-05, |
| "loss": 0.3168, |
| "step": 65000 |
| }, |
| { |
| "epoch": 0.78803628576241, |
| "grad_norm": 7.646127223968506, |
| "learning_rate": 1.474851015022819e-05, |
| "loss": 0.2803, |
| "step": 65500 |
| }, |
| { |
| "epoch": 0.7940518299285353, |
| "grad_norm": 0.0017321036430075765, |
| "learning_rate": 1.4708406522454023e-05, |
| "loss": 0.2779, |
| "step": 66000 |
| }, |
| { |
| "epoch": 0.8000673740946606, |
| "grad_norm": 0.004854683298617601, |
| "learning_rate": 1.4668302894679854e-05, |
| "loss": 0.3048, |
| "step": 66500 |
| }, |
| { |
| "epoch": 0.8060829182607858, |
| "grad_norm": 5.331120491027832, |
| "learning_rate": 1.4628199266905685e-05, |
| "loss": 0.3121, |
| "step": 67000 |
| }, |
| { |
| "epoch": 0.8120984624269111, |
| "grad_norm": 4.585981369018555, |
| "learning_rate": 1.4588095639131517e-05, |
| "loss": 0.2968, |
| "step": 67500 |
| }, |
| { |
| "epoch": 0.8181140065930365, |
| "grad_norm": 0.0019534584134817123, |
| "learning_rate": 1.4548072218612896e-05, |
| "loss": 0.2869, |
| "step": 68000 |
| }, |
| { |
| "epoch": 0.8241295507591617, |
| "grad_norm": 0.03927936032414436, |
| "learning_rate": 1.4507968590838728e-05, |
| "loss": 0.2942, |
| "step": 68500 |
| }, |
| { |
| "epoch": 0.830145094925287, |
| "grad_norm": 0.018528884276747704, |
| "learning_rate": 1.446786496306456e-05, |
| "loss": 0.2929, |
| "step": 69000 |
| }, |
| { |
| "epoch": 0.8361606390914122, |
| "grad_norm": 2.7401936054229736, |
| "learning_rate": 1.4427761335290392e-05, |
| "loss": 0.272, |
| "step": 69500 |
| }, |
| { |
| "epoch": 0.8421761832575375, |
| "grad_norm": 0.07152987271547318, |
| "learning_rate": 1.4387657707516223e-05, |
| "loss": 0.3162, |
| "step": 70000 |
| }, |
| { |
| "epoch": 0.8481917274236628, |
| "grad_norm": 5.40221643447876, |
| "learning_rate": 1.4347634286997603e-05, |
| "loss": 0.2717, |
| "step": 70500 |
| }, |
| { |
| "epoch": 0.854207271589788, |
| "grad_norm": 2.5662572383880615, |
| "learning_rate": 1.4307530659223434e-05, |
| "loss": 0.2913, |
| "step": 71000 |
| }, |
| { |
| "epoch": 0.8602228157559133, |
| "grad_norm": 6.399110794067383, |
| "learning_rate": 1.4267427031449267e-05, |
| "loss": 0.279, |
| "step": 71500 |
| }, |
| { |
| "epoch": 0.8662383599220386, |
| "grad_norm": 4.513882160186768, |
| "learning_rate": 1.4227323403675098e-05, |
| "loss": 0.3091, |
| "step": 72000 |
| }, |
| { |
| "epoch": 0.8722539040881638, |
| "grad_norm": 2.024488925933838, |
| "learning_rate": 1.4187219775900929e-05, |
| "loss": 0.2831, |
| "step": 72500 |
| }, |
| { |
| "epoch": 0.8782694482542891, |
| "grad_norm": 9.901945114135742, |
| "learning_rate": 1.4147116148126761e-05, |
| "loss": 0.2863, |
| "step": 73000 |
| }, |
| { |
| "epoch": 0.8842849924204144, |
| "grad_norm": 1.1856168508529663, |
| "learning_rate": 1.410709272760814e-05, |
| "loss": 0.2754, |
| "step": 73500 |
| }, |
| { |
| "epoch": 0.8903005365865396, |
| "grad_norm": 4.077794075012207, |
| "learning_rate": 1.4066989099833972e-05, |
| "loss": 0.298, |
| "step": 74000 |
| }, |
| { |
| "epoch": 0.8963160807526649, |
| "grad_norm": 0.2611056864261627, |
| "learning_rate": 1.4026885472059803e-05, |
| "loss": 0.2985, |
| "step": 74500 |
| }, |
| { |
| "epoch": 0.9023316249187902, |
| "grad_norm": 0.1290031224489212, |
| "learning_rate": 1.3986781844285636e-05, |
| "loss": 0.2798, |
| "step": 75000 |
| }, |
| { |
| "epoch": 0.9083471690849154, |
| "grad_norm": 0.1582382321357727, |
| "learning_rate": 1.3946678216511467e-05, |
| "loss": 0.2908, |
| "step": 75500 |
| }, |
| { |
| "epoch": 0.9143627132510407, |
| "grad_norm": 6.692793369293213, |
| "learning_rate": 1.3906574588737298e-05, |
| "loss": 0.2905, |
| "step": 76000 |
| }, |
| { |
| "epoch": 0.920378257417166, |
| "grad_norm": 12.920578002929688, |
| "learning_rate": 1.386647096096313e-05, |
| "loss": 0.306, |
| "step": 76500 |
| }, |
| { |
| "epoch": 0.9263938015832912, |
| "grad_norm": 0.14715702831745148, |
| "learning_rate": 1.3826367333188961e-05, |
| "loss": 0.2911, |
| "step": 77000 |
| }, |
| { |
| "epoch": 0.9324093457494165, |
| "grad_norm": 7.250746726989746, |
| "learning_rate": 1.3786343912670342e-05, |
| "loss": 0.2911, |
| "step": 77500 |
| }, |
| { |
| "epoch": 0.9384248899155417, |
| "grad_norm": 9.904165267944336, |
| "learning_rate": 1.3746240284896173e-05, |
| "loss": 0.2893, |
| "step": 78000 |
| }, |
| { |
| "epoch": 0.944440434081667, |
| "grad_norm": 1.3197027444839478, |
| "learning_rate": 1.3706136657122005e-05, |
| "loss": 0.295, |
| "step": 78500 |
| }, |
| { |
| "epoch": 0.9504559782477923, |
| "grad_norm": 0.5488787889480591, |
| "learning_rate": 1.3666033029347836e-05, |
| "loss": 0.2617, |
| "step": 79000 |
| }, |
| { |
| "epoch": 0.9564715224139175, |
| "grad_norm": 0.007599683478474617, |
| "learning_rate": 1.3625929401573667e-05, |
| "loss": 0.2874, |
| "step": 79500 |
| }, |
| { |
| "epoch": 0.9624870665800428, |
| "grad_norm": 28.689912796020508, |
| "learning_rate": 1.35858257737995e-05, |
| "loss": 0.2954, |
| "step": 80000 |
| }, |
| { |
| "epoch": 0.9685026107461681, |
| "grad_norm": 1.1681454181671143, |
| "learning_rate": 1.354572214602533e-05, |
| "loss": 0.2893, |
| "step": 80500 |
| }, |
| { |
| "epoch": 0.9745181549122933, |
| "grad_norm": 26.251056671142578, |
| "learning_rate": 1.3505618518251162e-05, |
| "loss": 0.2771, |
| "step": 81000 |
| }, |
| { |
| "epoch": 0.9805336990784186, |
| "grad_norm": 19.04422950744629, |
| "learning_rate": 1.3465514890476994e-05, |
| "loss": 0.2556, |
| "step": 81500 |
| }, |
| { |
| "epoch": 0.9865492432445438, |
| "grad_norm": 6.7204909324646, |
| "learning_rate": 1.3425411262702825e-05, |
| "loss": 0.2999, |
| "step": 82000 |
| }, |
| { |
| "epoch": 0.9925647874106691, |
| "grad_norm": 0.3453606069087982, |
| "learning_rate": 1.3385307634928656e-05, |
| "loss": 0.2795, |
| "step": 82500 |
| }, |
| { |
| "epoch": 0.9985803315767945, |
| "grad_norm": 2.7078826427459717, |
| "learning_rate": 1.3345204007154489e-05, |
| "loss": 0.2727, |
| "step": 83000 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.6738478320152713, |
| "eval_f1": 0.69947127247092, |
| "eval_loss": 0.287530779838562, |
| "eval_roc_auc": 0.8124488682847013, |
| "eval_runtime": 111.0668, |
| "eval_samples_per_second": 66.032, |
| "eval_steps_per_second": 66.032, |
| "step": 83118 |
| }, |
| { |
| "epoch": 1.0045958757429196, |
| "grad_norm": 8.754840850830078, |
| "learning_rate": 1.3305180586635869e-05, |
| "loss": 0.2439, |
| "step": 83500 |
| }, |
| { |
| "epoch": 1.010611419909045, |
| "grad_norm": 23.18345069885254, |
| "learning_rate": 1.3265237373372797e-05, |
| "loss": 0.2343, |
| "step": 84000 |
| }, |
| { |
| "epoch": 1.0166269640751702, |
| "grad_norm": 0.2795548141002655, |
| "learning_rate": 1.322513374559863e-05, |
| "loss": 0.2591, |
| "step": 84500 |
| }, |
| { |
| "epoch": 1.0226425082412955, |
| "grad_norm": 23.055418014526367, |
| "learning_rate": 1.318503011782446e-05, |
| "loss": 0.2403, |
| "step": 85000 |
| }, |
| { |
| "epoch": 1.0286580524074207, |
| "grad_norm": 0.0004251549835316837, |
| "learning_rate": 1.3144926490050291e-05, |
| "loss": 0.2575, |
| "step": 85500 |
| }, |
| { |
| "epoch": 1.034673596573546, |
| "grad_norm": 0.0705941691994667, |
| "learning_rate": 1.3104822862276124e-05, |
| "loss": 0.2298, |
| "step": 86000 |
| }, |
| { |
| "epoch": 1.0406891407396712, |
| "grad_norm": 0.14872965216636658, |
| "learning_rate": 1.3064719234501955e-05, |
| "loss": 0.2556, |
| "step": 86500 |
| }, |
| { |
| "epoch": 1.0467046849057966, |
| "grad_norm": 29.70933723449707, |
| "learning_rate": 1.3024615606727786e-05, |
| "loss": 0.246, |
| "step": 87000 |
| }, |
| { |
| "epoch": 1.0527202290719218, |
| "grad_norm": 0.12189821898937225, |
| "learning_rate": 1.2984511978953618e-05, |
| "loss": 0.2486, |
| "step": 87500 |
| }, |
| { |
| "epoch": 1.0587357732380471, |
| "grad_norm": 16.819997787475586, |
| "learning_rate": 1.2944488558434999e-05, |
| "loss": 0.2738, |
| "step": 88000 |
| }, |
| { |
| "epoch": 1.0647513174041723, |
| "grad_norm": 0.504358172416687, |
| "learning_rate": 1.290438493066083e-05, |
| "loss": 0.2512, |
| "step": 88500 |
| }, |
| { |
| "epoch": 1.0707668615702977, |
| "grad_norm": 33.048221588134766, |
| "learning_rate": 1.286428130288666e-05, |
| "loss": 0.2464, |
| "step": 89000 |
| }, |
| { |
| "epoch": 1.0767824057364228, |
| "grad_norm": 10.764643669128418, |
| "learning_rate": 1.2824177675112493e-05, |
| "loss": 0.2633, |
| "step": 89500 |
| }, |
| { |
| "epoch": 1.0827979499025482, |
| "grad_norm": 0.08640070259571075, |
| "learning_rate": 1.2784074047338324e-05, |
| "loss": 0.2406, |
| "step": 90000 |
| }, |
| { |
| "epoch": 1.0888134940686736, |
| "grad_norm": 14.491482734680176, |
| "learning_rate": 1.2744050626819704e-05, |
| "loss": 0.2914, |
| "step": 90500 |
| }, |
| { |
| "epoch": 1.0948290382347987, |
| "grad_norm": 6.458908557891846, |
| "learning_rate": 1.2703946999045535e-05, |
| "loss": 0.2768, |
| "step": 91000 |
| }, |
| { |
| "epoch": 1.100844582400924, |
| "grad_norm": 3.36252498626709, |
| "learning_rate": 1.2663923578526915e-05, |
| "loss": 0.2275, |
| "step": 91500 |
| }, |
| { |
| "epoch": 1.1068601265670492, |
| "grad_norm": 7.332001209259033, |
| "learning_rate": 1.2623819950752746e-05, |
| "loss": 0.2561, |
| "step": 92000 |
| }, |
| { |
| "epoch": 1.1128756707331746, |
| "grad_norm": 0.10880118608474731, |
| "learning_rate": 1.2583796530234126e-05, |
| "loss": 0.26, |
| "step": 92500 |
| }, |
| { |
| "epoch": 1.1188912148992998, |
| "grad_norm": 0.041975826025009155, |
| "learning_rate": 1.2543692902459959e-05, |
| "loss": 0.2588, |
| "step": 93000 |
| }, |
| { |
| "epoch": 1.1249067590654251, |
| "grad_norm": 19.071786880493164, |
| "learning_rate": 1.250358927468579e-05, |
| "loss": 0.2537, |
| "step": 93500 |
| }, |
| { |
| "epoch": 1.1309223032315503, |
| "grad_norm": 0.5620952248573303, |
| "learning_rate": 1.2463485646911621e-05, |
| "loss": 0.2694, |
| "step": 94000 |
| }, |
| { |
| "epoch": 1.1369378473976757, |
| "grad_norm": 0.4685472548007965, |
| "learning_rate": 1.2423382019137454e-05, |
| "loss": 0.2255, |
| "step": 94500 |
| }, |
| { |
| "epoch": 1.1429533915638008, |
| "grad_norm": 0.004374380223453045, |
| "learning_rate": 1.2383278391363285e-05, |
| "loss": 0.2979, |
| "step": 95000 |
| }, |
| { |
| "epoch": 1.1489689357299262, |
| "grad_norm": 0.21592481434345245, |
| "learning_rate": 1.2343174763589117e-05, |
| "loss": 0.2425, |
| "step": 95500 |
| }, |
| { |
| "epoch": 1.1549844798960514, |
| "grad_norm": 38.64509963989258, |
| "learning_rate": 1.2303071135814946e-05, |
| "loss": 0.2534, |
| "step": 96000 |
| }, |
| { |
| "epoch": 1.1610000240621767, |
| "grad_norm": 0.1709357500076294, |
| "learning_rate": 1.2262967508040777e-05, |
| "loss": 0.2277, |
| "step": 96500 |
| }, |
| { |
| "epoch": 1.1670155682283019, |
| "grad_norm": 0.02226692996919155, |
| "learning_rate": 1.2222863880266608e-05, |
| "loss": 0.248, |
| "step": 97000 |
| }, |
| { |
| "epoch": 1.1730311123944273, |
| "grad_norm": 0.11720846593379974, |
| "learning_rate": 1.2182760252492441e-05, |
| "loss": 0.2818, |
| "step": 97500 |
| }, |
| { |
| "epoch": 1.1790466565605524, |
| "grad_norm": 0.17743727564811707, |
| "learning_rate": 1.2142656624718272e-05, |
| "loss": 0.2625, |
| "step": 98000 |
| }, |
| { |
| "epoch": 1.1850622007266778, |
| "grad_norm": 0.5575046539306641, |
| "learning_rate": 1.2102633204199652e-05, |
| "loss": 0.274, |
| "step": 98500 |
| }, |
| { |
| "epoch": 1.191077744892803, |
| "grad_norm": 0.0813954770565033, |
| "learning_rate": 1.2062529576425483e-05, |
| "loss": 0.2764, |
| "step": 99000 |
| }, |
| { |
| "epoch": 1.1970932890589283, |
| "grad_norm": 6.563291549682617, |
| "learning_rate": 1.2022425948651316e-05, |
| "loss": 0.2426, |
| "step": 99500 |
| }, |
| { |
| "epoch": 1.2031088332250535, |
| "grad_norm": 4.957773126079701e-05, |
| "learning_rate": 1.1982322320877147e-05, |
| "loss": 0.2558, |
| "step": 100000 |
| }, |
| { |
| "epoch": 1.2091243773911788, |
| "grad_norm": 0.08407856523990631, |
| "learning_rate": 1.1942218693102978e-05, |
| "loss": 0.244, |
| "step": 100500 |
| }, |
| { |
| "epoch": 1.215139921557304, |
| "grad_norm": 0.04238196834921837, |
| "learning_rate": 1.1902195272584358e-05, |
| "loss": 0.2635, |
| "step": 101000 |
| }, |
| { |
| "epoch": 1.2211554657234294, |
| "grad_norm": 0.10858649015426636, |
| "learning_rate": 1.186209164481019e-05, |
| "loss": 0.2536, |
| "step": 101500 |
| }, |
| { |
| "epoch": 1.2271710098895545, |
| "grad_norm": 0.14419694244861603, |
| "learning_rate": 1.1821988017036021e-05, |
| "loss": 0.2803, |
| "step": 102000 |
| }, |
| { |
| "epoch": 1.23318655405568, |
| "grad_norm": 0.1416100561618805, |
| "learning_rate": 1.1781964596517402e-05, |
| "loss": 0.2198, |
| "step": 102500 |
| }, |
| { |
| "epoch": 1.239202098221805, |
| "grad_norm": 0.16191110014915466, |
| "learning_rate": 1.1741860968743232e-05, |
| "loss": 0.2687, |
| "step": 103000 |
| }, |
| { |
| "epoch": 1.2452176423879304, |
| "grad_norm": 3.2424659729003906, |
| "learning_rate": 1.1701757340969065e-05, |
| "loss": 0.2531, |
| "step": 103500 |
| }, |
| { |
| "epoch": 1.2512331865540558, |
| "grad_norm": 0.009034248068928719, |
| "learning_rate": 1.1661653713194896e-05, |
| "loss": 0.2144, |
| "step": 104000 |
| }, |
| { |
| "epoch": 1.257248730720181, |
| "grad_norm": 4.68690299987793, |
| "learning_rate": 1.1621550085420727e-05, |
| "loss": 0.2536, |
| "step": 104500 |
| }, |
| { |
| "epoch": 1.263264274886306, |
| "grad_norm": 0.04760544002056122, |
| "learning_rate": 1.158144645764656e-05, |
| "loss": 0.2394, |
| "step": 105000 |
| }, |
| { |
| "epoch": 1.2692798190524315, |
| "grad_norm": 0.027170976623892784, |
| "learning_rate": 1.154134282987239e-05, |
| "loss": 0.2644, |
| "step": 105500 |
| }, |
| { |
| "epoch": 1.2752953632185569, |
| "grad_norm": 29.608776092529297, |
| "learning_rate": 1.1501239202098222e-05, |
| "loss": 0.2456, |
| "step": 106000 |
| }, |
| { |
| "epoch": 1.281310907384682, |
| "grad_norm": 3.261889934539795, |
| "learning_rate": 1.1461135574324054e-05, |
| "loss": 0.2182, |
| "step": 106500 |
| }, |
| { |
| "epoch": 1.2873264515508072, |
| "grad_norm": 0.0033520832657814026, |
| "learning_rate": 1.1421031946549885e-05, |
| "loss": 0.2612, |
| "step": 107000 |
| }, |
| { |
| "epoch": 1.2933419957169325, |
| "grad_norm": 0.0770372673869133, |
| "learning_rate": 1.1380928318775716e-05, |
| "loss": 0.2517, |
| "step": 107500 |
| }, |
| { |
| "epoch": 1.299357539883058, |
| "grad_norm": 0.004508438985794783, |
| "learning_rate": 1.1340904898257096e-05, |
| "loss": 0.2552, |
| "step": 108000 |
| }, |
| { |
| "epoch": 1.305373084049183, |
| "grad_norm": 0.05816527456045151, |
| "learning_rate": 1.1300801270482929e-05, |
| "loss": 0.2577, |
| "step": 108500 |
| }, |
| { |
| "epoch": 1.3113886282153084, |
| "grad_norm": 8.914427757263184, |
| "learning_rate": 1.126069764270876e-05, |
| "loss": 0.2177, |
| "step": 109000 |
| }, |
| { |
| "epoch": 1.3174041723814336, |
| "grad_norm": 0.0006245987024158239, |
| "learning_rate": 1.1220594014934591e-05, |
| "loss": 0.2321, |
| "step": 109500 |
| }, |
| { |
| "epoch": 1.323419716547559, |
| "grad_norm": 10.717325210571289, |
| "learning_rate": 1.1180490387160424e-05, |
| "loss": 0.2532, |
| "step": 110000 |
| }, |
| { |
| "epoch": 1.3294352607136841, |
| "grad_norm": 3.988328218460083, |
| "learning_rate": 1.1140386759386255e-05, |
| "loss": 0.2733, |
| "step": 110500 |
| }, |
| { |
| "epoch": 1.3354508048798095, |
| "grad_norm": 9.221200942993164, |
| "learning_rate": 1.1100363338867635e-05, |
| "loss": 0.2611, |
| "step": 111000 |
| }, |
| { |
| "epoch": 1.3414663490459346, |
| "grad_norm": 0.08797607570886612, |
| "learning_rate": 1.1060259711093466e-05, |
| "loss": 0.2608, |
| "step": 111500 |
| }, |
| { |
| "epoch": 1.34748189321206, |
| "grad_norm": 0.018298862501978874, |
| "learning_rate": 1.1020156083319298e-05, |
| "loss": 0.2579, |
| "step": 112000 |
| }, |
| { |
| "epoch": 1.3534974373781852, |
| "grad_norm": 0.033742956817150116, |
| "learning_rate": 1.098005245554513e-05, |
| "loss": 0.2431, |
| "step": 112500 |
| }, |
| { |
| "epoch": 1.3595129815443106, |
| "grad_norm": 0.06899914145469666, |
| "learning_rate": 1.093994882777096e-05, |
| "loss": 0.253, |
| "step": 113000 |
| }, |
| { |
| "epoch": 1.3655285257104357, |
| "grad_norm": 0.1469135880470276, |
| "learning_rate": 1.0899845199996793e-05, |
| "loss": 0.244, |
| "step": 113500 |
| }, |
| { |
| "epoch": 1.371544069876561, |
| "grad_norm": 5.18328857421875, |
| "learning_rate": 1.0859821779478173e-05, |
| "loss": 0.2654, |
| "step": 114000 |
| }, |
| { |
| "epoch": 1.3775596140426862, |
| "grad_norm": 24.231332778930664, |
| "learning_rate": 1.0819718151704004e-05, |
| "loss": 0.237, |
| "step": 114500 |
| }, |
| { |
| "epoch": 1.3835751582088116, |
| "grad_norm": 0.2826213240623474, |
| "learning_rate": 1.0779614523929835e-05, |
| "loss": 0.2602, |
| "step": 115000 |
| }, |
| { |
| "epoch": 1.3895907023749368, |
| "grad_norm": 0.09144386649131775, |
| "learning_rate": 1.0739510896155668e-05, |
| "loss": 0.249, |
| "step": 115500 |
| }, |
| { |
| "epoch": 1.3956062465410621, |
| "grad_norm": 40.423797607421875, |
| "learning_rate": 1.0699407268381499e-05, |
| "loss": 0.2621, |
| "step": 116000 |
| }, |
| { |
| "epoch": 1.4016217907071873, |
| "grad_norm": 0.01249714009463787, |
| "learning_rate": 1.0659383847862879e-05, |
| "loss": 0.2451, |
| "step": 116500 |
| }, |
| { |
| "epoch": 1.4076373348733127, |
| "grad_norm": 0.14570143818855286, |
| "learning_rate": 1.061928022008871e-05, |
| "loss": 0.2481, |
| "step": 117000 |
| }, |
| { |
| "epoch": 1.413652879039438, |
| "grad_norm": 8.342629432678223, |
| "learning_rate": 1.0579176592314542e-05, |
| "loss": 0.2813, |
| "step": 117500 |
| }, |
| { |
| "epoch": 1.4196684232055632, |
| "grad_norm": 0.0001358153240289539, |
| "learning_rate": 1.0539072964540373e-05, |
| "loss": 0.2425, |
| "step": 118000 |
| }, |
| { |
| "epoch": 1.4256839673716883, |
| "grad_norm": 0.2505891025066376, |
| "learning_rate": 1.0498969336766204e-05, |
| "loss": 0.243, |
| "step": 118500 |
| }, |
| { |
| "epoch": 1.4316995115378137, |
| "grad_norm": 0.25133442878723145, |
| "learning_rate": 1.0458865708992037e-05, |
| "loss": 0.2313, |
| "step": 119000 |
| }, |
| { |
| "epoch": 1.437715055703939, |
| "grad_norm": 0.00044232659274712205, |
| "learning_rate": 1.0418762081217868e-05, |
| "loss": 0.2426, |
| "step": 119500 |
| }, |
| { |
| "epoch": 1.4437305998700642, |
| "grad_norm": 0.0011125517776235938, |
| "learning_rate": 1.0378658453443699e-05, |
| "loss": 0.2418, |
| "step": 120000 |
| }, |
| { |
| "epoch": 1.4497461440361894, |
| "grad_norm": 7.526780605316162, |
| "learning_rate": 1.0338635032925079e-05, |
| "loss": 0.2797, |
| "step": 120500 |
| }, |
| { |
| "epoch": 1.4557616882023148, |
| "grad_norm": 6.034137725830078, |
| "learning_rate": 1.0298531405150912e-05, |
| "loss": 0.2655, |
| "step": 121000 |
| }, |
| { |
| "epoch": 1.4617772323684401, |
| "grad_norm": 0.021329816430807114, |
| "learning_rate": 1.0258427777376743e-05, |
| "loss": 0.2633, |
| "step": 121500 |
| }, |
| { |
| "epoch": 1.4677927765345653, |
| "grad_norm": 0.013099109753966331, |
| "learning_rate": 1.0218324149602574e-05, |
| "loss": 0.2739, |
| "step": 122000 |
| }, |
| { |
| "epoch": 1.4738083207006905, |
| "grad_norm": 71.02497863769531, |
| "learning_rate": 1.0178300729083954e-05, |
| "loss": 0.254, |
| "step": 122500 |
| }, |
| { |
| "epoch": 1.4798238648668158, |
| "grad_norm": 15.551078796386719, |
| "learning_rate": 1.0138197101309786e-05, |
| "loss": 0.2602, |
| "step": 123000 |
| }, |
| { |
| "epoch": 1.4858394090329412, |
| "grad_norm": 4.560909655992873e-05, |
| "learning_rate": 1.0098093473535617e-05, |
| "loss": 0.2387, |
| "step": 123500 |
| }, |
| { |
| "epoch": 1.4918549531990664, |
| "grad_norm": 0.14415931701660156, |
| "learning_rate": 1.0057989845761448e-05, |
| "loss": 0.2823, |
| "step": 124000 |
| }, |
| { |
| "epoch": 1.4978704973651917, |
| "grad_norm": 43.61802673339844, |
| "learning_rate": 1.0017886217987281e-05, |
| "loss": 0.2378, |
| "step": 124500 |
| }, |
| { |
| "epoch": 1.5038860415313169, |
| "grad_norm": 0.20755548775196075, |
| "learning_rate": 9.977782590213112e-06, |
| "loss": 0.2433, |
| "step": 125000 |
| }, |
| { |
| "epoch": 1.5099015856974423, |
| "grad_norm": 0.2520395815372467, |
| "learning_rate": 9.937678962438943e-06, |
| "loss": 0.2539, |
| "step": 125500 |
| }, |
| { |
| "epoch": 1.5159171298635674, |
| "grad_norm": 0.3743610382080078, |
| "learning_rate": 9.897575334664774e-06, |
| "loss": 0.2789, |
| "step": 126000 |
| }, |
| { |
| "epoch": 1.5219326740296926, |
| "grad_norm": 0.0024790179450064898, |
| "learning_rate": 9.857551914146154e-06, |
| "loss": 0.2258, |
| "step": 126500 |
| }, |
| { |
| "epoch": 1.527948218195818, |
| "grad_norm": 21.557737350463867, |
| "learning_rate": 9.817448286371985e-06, |
| "loss": 0.2692, |
| "step": 127000 |
| }, |
| { |
| "epoch": 1.5339637623619433, |
| "grad_norm": 0.00022073234140407294, |
| "learning_rate": 9.777344658597818e-06, |
| "loss": 0.2674, |
| "step": 127500 |
| }, |
| { |
| "epoch": 1.5399793065280685, |
| "grad_norm": 0.0013155222404748201, |
| "learning_rate": 9.737241030823649e-06, |
| "loss": 0.242, |
| "step": 128000 |
| }, |
| { |
| "epoch": 1.5459948506941938, |
| "grad_norm": 0.2000940591096878, |
| "learning_rate": 9.69713740304948e-06, |
| "loss": 0.2487, |
| "step": 128500 |
| }, |
| { |
| "epoch": 1.5520103948603192, |
| "grad_norm": 0.0021518643479794264, |
| "learning_rate": 9.657033775275312e-06, |
| "loss": 0.2123, |
| "step": 129000 |
| }, |
| { |
| "epoch": 1.5580259390264444, |
| "grad_norm": 0.24494774639606476, |
| "learning_rate": 9.616930147501143e-06, |
| "loss": 0.2975, |
| "step": 129500 |
| }, |
| { |
| "epoch": 1.5640414831925695, |
| "grad_norm": 9.558631896972656, |
| "learning_rate": 9.576906726982523e-06, |
| "loss": 0.2485, |
| "step": 130000 |
| }, |
| { |
| "epoch": 1.570057027358695, |
| "grad_norm": 6.8279128074646, |
| "learning_rate": 9.536803099208354e-06, |
| "loss": 0.2589, |
| "step": 130500 |
| }, |
| { |
| "epoch": 1.5760725715248203, |
| "grad_norm": 0.018290262669324875, |
| "learning_rate": 9.496699471434187e-06, |
| "loss": 0.2605, |
| "step": 131000 |
| }, |
| { |
| "epoch": 1.5820881156909454, |
| "grad_norm": 0.19554129242897034, |
| "learning_rate": 9.456595843660018e-06, |
| "loss": 0.2405, |
| "step": 131500 |
| }, |
| { |
| "epoch": 1.5881036598570706, |
| "grad_norm": 26.12983512878418, |
| "learning_rate": 9.416492215885849e-06, |
| "loss": 0.2261, |
| "step": 132000 |
| }, |
| { |
| "epoch": 1.594119204023196, |
| "grad_norm": 19.12807273864746, |
| "learning_rate": 9.376388588111682e-06, |
| "loss": 0.2454, |
| "step": 132500 |
| }, |
| { |
| "epoch": 1.6001347481893213, |
| "grad_norm": 48.97768020629883, |
| "learning_rate": 9.336284960337513e-06, |
| "loss": 0.2488, |
| "step": 133000 |
| }, |
| { |
| "epoch": 1.6061502923554465, |
| "grad_norm": 0.10775639116764069, |
| "learning_rate": 9.296261539818893e-06, |
| "loss": 0.2712, |
| "step": 133500 |
| }, |
| { |
| "epoch": 1.6121658365215716, |
| "grad_norm": 18.01923942565918, |
| "learning_rate": 9.256157912044724e-06, |
| "loss": 0.2898, |
| "step": 134000 |
| }, |
| { |
| "epoch": 1.618181380687697, |
| "grad_norm": 0.057373084127902985, |
| "learning_rate": 9.216054284270556e-06, |
| "loss": 0.2235, |
| "step": 134500 |
| }, |
| { |
| "epoch": 1.6241969248538224, |
| "grad_norm": 0.0020849681459367275, |
| "learning_rate": 9.175950656496387e-06, |
| "loss": 0.2355, |
| "step": 135000 |
| }, |
| { |
| "epoch": 1.6302124690199475, |
| "grad_norm": 0.15205715596675873, |
| "learning_rate": 9.135847028722218e-06, |
| "loss": 0.2791, |
| "step": 135500 |
| }, |
| { |
| "epoch": 1.6362280131860727, |
| "grad_norm": 0.06727798283100128, |
| "learning_rate": 9.095823608203598e-06, |
| "loss": 0.2412, |
| "step": 136000 |
| }, |
| { |
| "epoch": 1.642243557352198, |
| "grad_norm": 0.19422870874404907, |
| "learning_rate": 9.055719980429431e-06, |
| "loss": 0.2431, |
| "step": 136500 |
| }, |
| { |
| "epoch": 1.6482591015183234, |
| "grad_norm": 0.06913313269615173, |
| "learning_rate": 9.015616352655262e-06, |
| "loss": 0.2687, |
| "step": 137000 |
| }, |
| { |
| "epoch": 1.6542746456844486, |
| "grad_norm": 0.06762461364269257, |
| "learning_rate": 8.975512724881093e-06, |
| "loss": 0.2498, |
| "step": 137500 |
| }, |
| { |
| "epoch": 1.6602901898505737, |
| "grad_norm": 0.1282985806465149, |
| "learning_rate": 8.935409097106926e-06, |
| "loss": 0.2503, |
| "step": 138000 |
| }, |
| { |
| "epoch": 1.6663057340166991, |
| "grad_norm": 0.6339439153671265, |
| "learning_rate": 8.895305469332756e-06, |
| "loss": 0.2454, |
| "step": 138500 |
| }, |
| { |
| "epoch": 1.6723212781828245, |
| "grad_norm": 0.15219150483608246, |
| "learning_rate": 8.855201841558587e-06, |
| "loss": 0.2845, |
| "step": 139000 |
| }, |
| { |
| "epoch": 1.6783368223489497, |
| "grad_norm": 5.600707054138184, |
| "learning_rate": 8.815178421039968e-06, |
| "loss": 0.2459, |
| "step": 139500 |
| }, |
| { |
| "epoch": 1.6843523665150748, |
| "grad_norm": 0.36313244700431824, |
| "learning_rate": 8.775155000521348e-06, |
| "loss": 0.2374, |
| "step": 140000 |
| }, |
| { |
| "epoch": 1.6903679106812002, |
| "grad_norm": 4.401086807250977, |
| "learning_rate": 8.73505137274718e-06, |
| "loss": 0.2425, |
| "step": 140500 |
| }, |
| { |
| "epoch": 1.6963834548473256, |
| "grad_norm": 0.00033720143255777657, |
| "learning_rate": 8.694947744973011e-06, |
| "loss": 0.2341, |
| "step": 141000 |
| }, |
| { |
| "epoch": 1.7023989990134507, |
| "grad_norm": 0.0821937769651413, |
| "learning_rate": 8.654844117198842e-06, |
| "loss": 0.2326, |
| "step": 141500 |
| }, |
| { |
| "epoch": 1.7084145431795759, |
| "grad_norm": 0.000721343734767288, |
| "learning_rate": 8.614740489424675e-06, |
| "loss": 0.2553, |
| "step": 142000 |
| }, |
| { |
| "epoch": 1.7144300873457012, |
| "grad_norm": 0.3059915602207184, |
| "learning_rate": 8.574636861650506e-06, |
| "loss": 0.2743, |
| "step": 142500 |
| }, |
| { |
| "epoch": 1.7204456315118266, |
| "grad_norm": 0.13650359213352203, |
| "learning_rate": 8.534533233876337e-06, |
| "loss": 0.255, |
| "step": 143000 |
| }, |
| { |
| "epoch": 1.7264611756779518, |
| "grad_norm": 0.16923360526561737, |
| "learning_rate": 8.49442960610217e-06, |
| "loss": 0.2832, |
| "step": 143500 |
| }, |
| { |
| "epoch": 1.7324767198440771, |
| "grad_norm": 0.0067064897157251835, |
| "learning_rate": 8.454325978328e-06, |
| "loss": 0.2425, |
| "step": 144000 |
| }, |
| { |
| "epoch": 1.7384922640102025, |
| "grad_norm": 0.33390843868255615, |
| "learning_rate": 8.41430255780938e-06, |
| "loss": 0.2577, |
| "step": 144500 |
| }, |
| { |
| "epoch": 1.7445078081763277, |
| "grad_norm": 0.11052338778972626, |
| "learning_rate": 8.374198930035212e-06, |
| "loss": 0.253, |
| "step": 145000 |
| }, |
| { |
| "epoch": 1.7505233523424528, |
| "grad_norm": 0.06620516628026962, |
| "learning_rate": 8.334095302261042e-06, |
| "loss": 0.2555, |
| "step": 145500 |
| }, |
| { |
| "epoch": 1.7565388965085782, |
| "grad_norm": 0.00559147447347641, |
| "learning_rate": 8.293991674486873e-06, |
| "loss": 0.2387, |
| "step": 146000 |
| }, |
| { |
| "epoch": 1.7625544406747036, |
| "grad_norm": 0.1951608508825302, |
| "learning_rate": 8.253888046712706e-06, |
| "loss": 0.2249, |
| "step": 146500 |
| }, |
| { |
| "epoch": 1.7685699848408287, |
| "grad_norm": 0.16532181203365326, |
| "learning_rate": 8.213784418938537e-06, |
| "loss": 0.2779, |
| "step": 147000 |
| }, |
| { |
| "epoch": 1.7745855290069539, |
| "grad_norm": 0.024670429527759552, |
| "learning_rate": 8.17368079116437e-06, |
| "loss": 0.24, |
| "step": 147500 |
| }, |
| { |
| "epoch": 1.7806010731730793, |
| "grad_norm": 0.7060034275054932, |
| "learning_rate": 8.1335771633902e-06, |
| "loss": 0.2646, |
| "step": 148000 |
| }, |
| { |
| "epoch": 1.7866166173392046, |
| "grad_norm": 0.16653361916542053, |
| "learning_rate": 8.093473535616032e-06, |
| "loss": 0.2829, |
| "step": 148500 |
| }, |
| { |
| "epoch": 1.7926321615053298, |
| "grad_norm": 9.015371322631836, |
| "learning_rate": 8.053369907841864e-06, |
| "loss": 0.2465, |
| "step": 149000 |
| }, |
| { |
| "epoch": 1.798647705671455, |
| "grad_norm": 0.08503203839063644, |
| "learning_rate": 8.013346487323243e-06, |
| "loss": 0.2476, |
| "step": 149500 |
| }, |
| { |
| "epoch": 1.8046632498375803, |
| "grad_norm": 0.31652989983558655, |
| "learning_rate": 7.973242859549075e-06, |
| "loss": 0.2807, |
| "step": 150000 |
| }, |
| { |
| "epoch": 1.8106787940037057, |
| "grad_norm": 20.23206329345703, |
| "learning_rate": 7.933139231774906e-06, |
| "loss": 0.2393, |
| "step": 150500 |
| }, |
| { |
| "epoch": 1.8166943381698308, |
| "grad_norm": 0.0023623614106327295, |
| "learning_rate": 7.893035604000739e-06, |
| "loss": 0.2606, |
| "step": 151000 |
| }, |
| { |
| "epoch": 1.822709882335956, |
| "grad_norm": 0.0003463963221292943, |
| "learning_rate": 7.85293197622657e-06, |
| "loss": 0.2377, |
| "step": 151500 |
| }, |
| { |
| "epoch": 1.8287254265020814, |
| "grad_norm": 22.177343368530273, |
| "learning_rate": 7.81290855570795e-06, |
| "loss": 0.225, |
| "step": 152000 |
| }, |
| { |
| "epoch": 1.8347409706682067, |
| "grad_norm": 0.16396382451057434, |
| "learning_rate": 7.772804927933781e-06, |
| "loss": 0.2749, |
| "step": 152500 |
| }, |
| { |
| "epoch": 1.840756514834332, |
| "grad_norm": 0.06666311621665955, |
| "learning_rate": 7.732701300159614e-06, |
| "loss": 0.2449, |
| "step": 153000 |
| }, |
| { |
| "epoch": 1.846772059000457, |
| "grad_norm": 0.2808719575405121, |
| "learning_rate": 7.692597672385445e-06, |
| "loss": 0.2201, |
| "step": 153500 |
| }, |
| { |
| "epoch": 1.8527876031665824, |
| "grad_norm": 0.02584654837846756, |
| "learning_rate": 7.652574251866825e-06, |
| "loss": 0.2555, |
| "step": 154000 |
| }, |
| { |
| "epoch": 1.8588031473327078, |
| "grad_norm": 6.074542999267578, |
| "learning_rate": 7.612470624092656e-06, |
| "loss": 0.2652, |
| "step": 154500 |
| }, |
| { |
| "epoch": 1.864818691498833, |
| "grad_norm": 0.007166345603764057, |
| "learning_rate": 7.572447203574036e-06, |
| "loss": 0.2777, |
| "step": 155000 |
| }, |
| { |
| "epoch": 1.870834235664958, |
| "grad_norm": 2.7224321365356445, |
| "learning_rate": 7.532423783055416e-06, |
| "loss": 0.2614, |
| "step": 155500 |
| }, |
| { |
| "epoch": 1.8768497798310835, |
| "grad_norm": 8.61175537109375, |
| "learning_rate": 7.492320155281248e-06, |
| "loss": 0.2716, |
| "step": 156000 |
| }, |
| { |
| "epoch": 1.8828653239972089, |
| "grad_norm": 0.4761998951435089, |
| "learning_rate": 7.452216527507079e-06, |
| "loss": 0.2617, |
| "step": 156500 |
| }, |
| { |
| "epoch": 1.888880868163334, |
| "grad_norm": 0.041520122438669205, |
| "learning_rate": 7.4121128997329105e-06, |
| "loss": 0.2233, |
| "step": 157000 |
| }, |
| { |
| "epoch": 1.8948964123294592, |
| "grad_norm": 0.3298840820789337, |
| "learning_rate": 7.372009271958742e-06, |
| "loss": 0.2905, |
| "step": 157500 |
| }, |
| { |
| "epoch": 1.9009119564955848, |
| "grad_norm": 0.14137232303619385, |
| "learning_rate": 7.331905644184573e-06, |
| "loss": 0.2489, |
| "step": 158000 |
| }, |
| { |
| "epoch": 1.90692750066171, |
| "grad_norm": 0.017581721767783165, |
| "learning_rate": 7.291802016410405e-06, |
| "loss": 0.2531, |
| "step": 158500 |
| }, |
| { |
| "epoch": 1.912943044827835, |
| "grad_norm": 0.10974815487861633, |
| "learning_rate": 7.251698388636237e-06, |
| "loss": 0.2387, |
| "step": 159000 |
| }, |
| { |
| "epoch": 1.9189585889939604, |
| "grad_norm": 0.09865211695432663, |
| "learning_rate": 7.211674968117617e-06, |
| "loss": 0.2364, |
| "step": 159500 |
| }, |
| { |
| "epoch": 1.9249741331600858, |
| "grad_norm": 0.004849567078053951, |
| "learning_rate": 7.171571340343448e-06, |
| "loss": 0.2751, |
| "step": 160000 |
| }, |
| { |
| "epoch": 1.930989677326211, |
| "grad_norm": 0.017855796962976456, |
| "learning_rate": 7.13146771256928e-06, |
| "loss": 0.2422, |
| "step": 160500 |
| }, |
| { |
| "epoch": 1.9370052214923361, |
| "grad_norm": 0.22124846279621124, |
| "learning_rate": 7.091364084795112e-06, |
| "loss": 0.2409, |
| "step": 161000 |
| }, |
| { |
| "epoch": 1.9430207656584615, |
| "grad_norm": 0.0023982899729162455, |
| "learning_rate": 7.051340664276492e-06, |
| "loss": 0.2699, |
| "step": 161500 |
| }, |
| { |
| "epoch": 1.9490363098245869, |
| "grad_norm": 8.00629997253418, |
| "learning_rate": 7.011237036502323e-06, |
| "loss": 0.2399, |
| "step": 162000 |
| }, |
| { |
| "epoch": 1.955051853990712, |
| "grad_norm": 7.468501091003418, |
| "learning_rate": 6.9711334087281545e-06, |
| "loss": 0.2535, |
| "step": 162500 |
| }, |
| { |
| "epoch": 1.9610673981568372, |
| "grad_norm": 0.4242821931838989, |
| "learning_rate": 6.931029780953986e-06, |
| "loss": 0.2506, |
| "step": 163000 |
| }, |
| { |
| "epoch": 1.9670829423229625, |
| "grad_norm": 0.020416131243109703, |
| "learning_rate": 6.890926153179817e-06, |
| "loss": 0.2482, |
| "step": 163500 |
| }, |
| { |
| "epoch": 1.973098486489088, |
| "grad_norm": 0.06718198210000992, |
| "learning_rate": 6.850822525405649e-06, |
| "loss": 0.2505, |
| "step": 164000 |
| }, |
| { |
| "epoch": 1.979114030655213, |
| "grad_norm": 0.7040701508522034, |
| "learning_rate": 6.81071889763148e-06, |
| "loss": 0.2196, |
| "step": 164500 |
| }, |
| { |
| "epoch": 1.9851295748213382, |
| "grad_norm": 0.1061306744813919, |
| "learning_rate": 6.770615269857311e-06, |
| "loss": 0.2699, |
| "step": 165000 |
| }, |
| { |
| "epoch": 1.9911451189874636, |
| "grad_norm": 0.1626858115196228, |
| "learning_rate": 6.730511642083143e-06, |
| "loss": 0.2428, |
| "step": 165500 |
| }, |
| { |
| "epoch": 1.997160663153589, |
| "grad_norm": 0.9182321429252625, |
| "learning_rate": 6.690408014308975e-06, |
| "loss": 0.2723, |
| "step": 166000 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_accuracy": 0.6772566130351786, |
| "eval_f1": 0.7059909142532652, |
| "eval_loss": 0.30793967843055725, |
| "eval_roc_auc": 0.8147532042541586, |
| "eval_runtime": 109.0185, |
| "eval_samples_per_second": 67.273, |
| "eval_steps_per_second": 67.273, |
| "step": 166236 |
| }, |
| { |
| "epoch": 2.003176207319714, |
| "grad_norm": 0.0028744570445269346, |
| "learning_rate": 6.650304386534806e-06, |
| "loss": 0.2161, |
| "step": 166500 |
| }, |
| { |
| "epoch": 2.0091917514858393, |
| "grad_norm": 0.10508907586336136, |
| "learning_rate": 6.6102007587606375e-06, |
| "loss": 0.1802, |
| "step": 167000 |
| }, |
| { |
| "epoch": 2.015207295651965, |
| "grad_norm": 0.023608388379216194, |
| "learning_rate": 6.570257545497566e-06, |
| "loss": 0.1646, |
| "step": 167500 |
| }, |
| { |
| "epoch": 2.02122283981809, |
| "grad_norm": 0.0004214489890728146, |
| "learning_rate": 6.530153917723398e-06, |
| "loss": 0.1425, |
| "step": 168000 |
| }, |
| { |
| "epoch": 2.027238383984215, |
| "grad_norm": 0.15282946825027466, |
| "learning_rate": 6.490050289949229e-06, |
| "loss": 0.1767, |
| "step": 168500 |
| }, |
| { |
| "epoch": 2.0332539281503403, |
| "grad_norm": 0.5594785213470459, |
| "learning_rate": 6.4499466621750605e-06, |
| "loss": 0.1431, |
| "step": 169000 |
| }, |
| { |
| "epoch": 2.039269472316466, |
| "grad_norm": 0.006397734861820936, |
| "learning_rate": 6.409843034400892e-06, |
| "loss": 0.1433, |
| "step": 169500 |
| }, |
| { |
| "epoch": 2.045285016482591, |
| "grad_norm": 0.02048441767692566, |
| "learning_rate": 6.369739406626724e-06, |
| "loss": 0.1751, |
| "step": 170000 |
| }, |
| { |
| "epoch": 2.0513005606487162, |
| "grad_norm": 0.08668813109397888, |
| "learning_rate": 6.329635778852555e-06, |
| "loss": 0.1904, |
| "step": 170500 |
| }, |
| { |
| "epoch": 2.0573161048148414, |
| "grad_norm": 1.3654146194458008, |
| "learning_rate": 6.289612358333935e-06, |
| "loss": 0.1477, |
| "step": 171000 |
| }, |
| { |
| "epoch": 2.063331648980967, |
| "grad_norm": 0.10657373815774918, |
| "learning_rate": 6.249508730559767e-06, |
| "loss": 0.1651, |
| "step": 171500 |
| }, |
| { |
| "epoch": 2.069347193147092, |
| "grad_norm": 0.0666826069355011, |
| "learning_rate": 6.209405102785599e-06, |
| "loss": 0.1896, |
| "step": 172000 |
| }, |
| { |
| "epoch": 2.0753627373132173, |
| "grad_norm": 0.03260161355137825, |
| "learning_rate": 6.16930147501143e-06, |
| "loss": 0.1595, |
| "step": 172500 |
| }, |
| { |
| "epoch": 2.0813782814793425, |
| "grad_norm": 11.851826667785645, |
| "learning_rate": 6.129197847237262e-06, |
| "loss": 0.1664, |
| "step": 173000 |
| }, |
| { |
| "epoch": 2.087393825645468, |
| "grad_norm": 0.0001777495490387082, |
| "learning_rate": 6.089094219463093e-06, |
| "loss": 0.1719, |
| "step": 173500 |
| }, |
| { |
| "epoch": 2.093409369811593, |
| "grad_norm": 0.001742048654705286, |
| "learning_rate": 6.048990591688924e-06, |
| "loss": 0.1336, |
| "step": 174000 |
| }, |
| { |
| "epoch": 2.0994249139777184, |
| "grad_norm": 0.233998104929924, |
| "learning_rate": 6.008886963914756e-06, |
| "loss": 0.1659, |
| "step": 174500 |
| }, |
| { |
| "epoch": 2.1054404581438435, |
| "grad_norm": 0.011589854955673218, |
| "learning_rate": 5.968863543396136e-06, |
| "loss": 0.1662, |
| "step": 175000 |
| }, |
| { |
| "epoch": 2.111456002309969, |
| "grad_norm": 0.003179445629939437, |
| "learning_rate": 5.928759915621968e-06, |
| "loss": 0.201, |
| "step": 175500 |
| }, |
| { |
| "epoch": 2.1174715464760943, |
| "grad_norm": 0.05272052437067032, |
| "learning_rate": 5.888656287847799e-06, |
| "loss": 0.1648, |
| "step": 176000 |
| }, |
| { |
| "epoch": 2.1234870906422194, |
| "grad_norm": 17.65900421142578, |
| "learning_rate": 5.848552660073631e-06, |
| "loss": 0.1499, |
| "step": 176500 |
| }, |
| { |
| "epoch": 2.1295026348083446, |
| "grad_norm": 0.017062697559595108, |
| "learning_rate": 5.808449032299463e-06, |
| "loss": 0.1812, |
| "step": 177000 |
| }, |
| { |
| "epoch": 2.13551817897447, |
| "grad_norm": 0.062195923179388046, |
| "learning_rate": 5.768505819036391e-06, |
| "loss": 0.1745, |
| "step": 177500 |
| }, |
| { |
| "epoch": 2.1415337231405953, |
| "grad_norm": 0.00937980879098177, |
| "learning_rate": 5.728402191262222e-06, |
| "loss": 0.138, |
| "step": 178000 |
| }, |
| { |
| "epoch": 2.1475492673067205, |
| "grad_norm": 13.196198463439941, |
| "learning_rate": 5.688298563488054e-06, |
| "loss": 0.1717, |
| "step": 178500 |
| }, |
| { |
| "epoch": 2.1535648114728456, |
| "grad_norm": 0.1440310925245285, |
| "learning_rate": 5.648194935713886e-06, |
| "loss": 0.1515, |
| "step": 179000 |
| }, |
| { |
| "epoch": 2.159580355638971, |
| "grad_norm": 0.032506223767995834, |
| "learning_rate": 5.608091307939717e-06, |
| "loss": 0.1401, |
| "step": 179500 |
| }, |
| { |
| "epoch": 2.1655958998050964, |
| "grad_norm": 6.104779458837584e-05, |
| "learning_rate": 5.5679876801655484e-06, |
| "loss": 0.1881, |
| "step": 180000 |
| }, |
| { |
| "epoch": 2.1716114439712215, |
| "grad_norm": 0.04752574488520622, |
| "learning_rate": 5.52788405239138e-06, |
| "loss": 0.1603, |
| "step": 180500 |
| }, |
| { |
| "epoch": 2.177626988137347, |
| "grad_norm": 0.005751811899244785, |
| "learning_rate": 5.487780424617212e-06, |
| "loss": 0.1983, |
| "step": 181000 |
| }, |
| { |
| "epoch": 2.1836425323034723, |
| "grad_norm": 0.03546688333153725, |
| "learning_rate": 5.447757004098591e-06, |
| "loss": 0.1831, |
| "step": 181500 |
| }, |
| { |
| "epoch": 2.1896580764695974, |
| "grad_norm": 0.21516333520412445, |
| "learning_rate": 5.407653376324423e-06, |
| "loss": 0.1708, |
| "step": 182000 |
| }, |
| { |
| "epoch": 2.1956736206357226, |
| "grad_norm": 0.061349764466285706, |
| "learning_rate": 5.367549748550255e-06, |
| "loss": 0.1737, |
| "step": 182500 |
| }, |
| { |
| "epoch": 2.201689164801848, |
| "grad_norm": 67.89110565185547, |
| "learning_rate": 5.327446120776087e-06, |
| "loss": 0.1465, |
| "step": 183000 |
| }, |
| { |
| "epoch": 2.2077047089679733, |
| "grad_norm": 0.0042543611489236355, |
| "learning_rate": 5.287342493001918e-06, |
| "loss": 0.2034, |
| "step": 183500 |
| }, |
| { |
| "epoch": 2.2137202531340985, |
| "grad_norm": 8.772153854370117, |
| "learning_rate": 5.247238865227749e-06, |
| "loss": 0.1354, |
| "step": 184000 |
| }, |
| { |
| "epoch": 2.2197357973002236, |
| "grad_norm": 0.46139106154441833, |
| "learning_rate": 5.207215444709128e-06, |
| "loss": 0.1838, |
| "step": 184500 |
| }, |
| { |
| "epoch": 2.2257513414663492, |
| "grad_norm": 0.007118485402315855, |
| "learning_rate": 5.16711181693496e-06, |
| "loss": 0.1702, |
| "step": 185000 |
| }, |
| { |
| "epoch": 2.2317668856324744, |
| "grad_norm": 0.01587226428091526, |
| "learning_rate": 5.127008189160792e-06, |
| "loss": 0.1566, |
| "step": 185500 |
| }, |
| { |
| "epoch": 2.2377824297985995, |
| "grad_norm": 0.006735761184245348, |
| "learning_rate": 5.086904561386623e-06, |
| "loss": 0.1644, |
| "step": 186000 |
| }, |
| { |
| "epoch": 2.2437979739647247, |
| "grad_norm": 111.12641143798828, |
| "learning_rate": 5.046881140868003e-06, |
| "loss": 0.1578, |
| "step": 186500 |
| }, |
| { |
| "epoch": 2.2498135181308503, |
| "grad_norm": 0.004494801629334688, |
| "learning_rate": 5.0067775130938344e-06, |
| "loss": 0.1625, |
| "step": 187000 |
| }, |
| { |
| "epoch": 2.2558290622969754, |
| "grad_norm": 61.42302703857422, |
| "learning_rate": 4.966673885319666e-06, |
| "loss": 0.1725, |
| "step": 187500 |
| }, |
| { |
| "epoch": 2.2618446064631006, |
| "grad_norm": 102.92018127441406, |
| "learning_rate": 4.926570257545498e-06, |
| "loss": 0.1503, |
| "step": 188000 |
| }, |
| { |
| "epoch": 2.2678601506292257, |
| "grad_norm": 0.15618078410625458, |
| "learning_rate": 4.886466629771329e-06, |
| "loss": 0.1586, |
| "step": 188500 |
| }, |
| { |
| "epoch": 2.2738756947953513, |
| "grad_norm": 0.1413143128156662, |
| "learning_rate": 4.846363001997161e-06, |
| "loss": 0.1985, |
| "step": 189000 |
| }, |
| { |
| "epoch": 2.2798912389614765, |
| "grad_norm": 40.41890335083008, |
| "learning_rate": 4.806259374222993e-06, |
| "loss": 0.1302, |
| "step": 189500 |
| }, |
| { |
| "epoch": 2.2859067831276016, |
| "grad_norm": 0.0008507597958669066, |
| "learning_rate": 4.766235953704372e-06, |
| "loss": 0.1409, |
| "step": 190000 |
| }, |
| { |
| "epoch": 2.291922327293727, |
| "grad_norm": 2.4075923647615127e-05, |
| "learning_rate": 4.726132325930204e-06, |
| "loss": 0.1463, |
| "step": 190500 |
| }, |
| { |
| "epoch": 2.2979378714598524, |
| "grad_norm": 0.05570465698838234, |
| "learning_rate": 4.6860286981560356e-06, |
| "loss": 0.1641, |
| "step": 191000 |
| }, |
| { |
| "epoch": 2.3039534156259776, |
| "grad_norm": 177.34120178222656, |
| "learning_rate": 4.645925070381867e-06, |
| "loss": 0.1766, |
| "step": 191500 |
| }, |
| { |
| "epoch": 2.3099689597921027, |
| "grad_norm": 25.356895446777344, |
| "learning_rate": 4.605821442607698e-06, |
| "loss": 0.1788, |
| "step": 192000 |
| }, |
| { |
| "epoch": 2.3159845039582283, |
| "grad_norm": 26.419334411621094, |
| "learning_rate": 4.56571781483353e-06, |
| "loss": 0.1875, |
| "step": 192500 |
| }, |
| { |
| "epoch": 2.3220000481243535, |
| "grad_norm": 0.05582762509584427, |
| "learning_rate": 4.525614187059362e-06, |
| "loss": 0.1357, |
| "step": 193000 |
| }, |
| { |
| "epoch": 2.3280155922904786, |
| "grad_norm": 0.01142974104732275, |
| "learning_rate": 4.485510559285193e-06, |
| "loss": 0.148, |
| "step": 193500 |
| }, |
| { |
| "epoch": 2.3340311364566038, |
| "grad_norm": 0.013201882131397724, |
| "learning_rate": 4.445487138766573e-06, |
| "loss": 0.1279, |
| "step": 194000 |
| }, |
| { |
| "epoch": 2.340046680622729, |
| "grad_norm": 0.007936575450003147, |
| "learning_rate": 4.405383510992405e-06, |
| "loss": 0.1492, |
| "step": 194500 |
| }, |
| { |
| "epoch": 2.3460622247888545, |
| "grad_norm": 0.030919533222913742, |
| "learning_rate": 4.365360090473785e-06, |
| "loss": 0.1298, |
| "step": 195000 |
| }, |
| { |
| "epoch": 2.3520777689549797, |
| "grad_norm": 0.0013994334731251001, |
| "learning_rate": 4.325256462699616e-06, |
| "loss": 0.1555, |
| "step": 195500 |
| }, |
| { |
| "epoch": 2.358093313121105, |
| "grad_norm": 0.004653325304389, |
| "learning_rate": 4.285152834925448e-06, |
| "loss": 0.1612, |
| "step": 196000 |
| }, |
| { |
| "epoch": 2.3641088572872304, |
| "grad_norm": 0.034009434282779694, |
| "learning_rate": 4.2450492071512796e-06, |
| "loss": 0.1543, |
| "step": 196500 |
| }, |
| { |
| "epoch": 2.3701244014533556, |
| "grad_norm": 0.6855669021606445, |
| "learning_rate": 4.20502578663266e-06, |
| "loss": 0.2072, |
| "step": 197000 |
| }, |
| { |
| "epoch": 2.3761399456194807, |
| "grad_norm": 45.97907257080078, |
| "learning_rate": 4.164922158858491e-06, |
| "loss": 0.1637, |
| "step": 197500 |
| }, |
| { |
| "epoch": 2.382155489785606, |
| "grad_norm": 0.36125409603118896, |
| "learning_rate": 4.1248185310843216e-06, |
| "loss": 0.142, |
| "step": 198000 |
| }, |
| { |
| "epoch": 2.388171033951731, |
| "grad_norm": 0.013704453594982624, |
| "learning_rate": 4.084714903310153e-06, |
| "loss": 0.1702, |
| "step": 198500 |
| }, |
| { |
| "epoch": 2.3941865781178566, |
| "grad_norm": 28.00567626953125, |
| "learning_rate": 4.044611275535985e-06, |
| "loss": 0.1641, |
| "step": 199000 |
| }, |
| { |
| "epoch": 2.4002021222839818, |
| "grad_norm": 0.4930565655231476, |
| "learning_rate": 4.0046680622729135e-06, |
| "loss": 0.147, |
| "step": 199500 |
| }, |
| { |
| "epoch": 2.406217666450107, |
| "grad_norm": 1.68142831325531, |
| "learning_rate": 3.964564434498745e-06, |
| "loss": 0.1543, |
| "step": 200000 |
| }, |
| { |
| "epoch": 2.4122332106162325, |
| "grad_norm": 0.00046357992687262595, |
| "learning_rate": 3.924460806724576e-06, |
| "loss": 0.1383, |
| "step": 200500 |
| }, |
| { |
| "epoch": 2.4182487547823577, |
| "grad_norm": 0.08767001330852509, |
| "learning_rate": 3.884357178950408e-06, |
| "loss": 0.1697, |
| "step": 201000 |
| }, |
| { |
| "epoch": 2.424264298948483, |
| "grad_norm": 14.962833404541016, |
| "learning_rate": 3.84425355117624e-06, |
| "loss": 0.1919, |
| "step": 201500 |
| }, |
| { |
| "epoch": 2.430279843114608, |
| "grad_norm": 0.0750858262181282, |
| "learning_rate": 3.8041499234020714e-06, |
| "loss": 0.1238, |
| "step": 202000 |
| }, |
| { |
| "epoch": 2.4362953872807336, |
| "grad_norm": 0.011100132018327713, |
| "learning_rate": 3.7640462956279028e-06, |
| "loss": 0.1584, |
| "step": 202500 |
| }, |
| { |
| "epoch": 2.4423109314468587, |
| "grad_norm": 0.0002675870491657406, |
| "learning_rate": 3.723942667853734e-06, |
| "loss": 0.1393, |
| "step": 203000 |
| }, |
| { |
| "epoch": 2.448326475612984, |
| "grad_norm": 22.56096839904785, |
| "learning_rate": 3.683839040079566e-06, |
| "loss": 0.1284, |
| "step": 203500 |
| }, |
| { |
| "epoch": 2.454342019779109, |
| "grad_norm": 0.0010695152450352907, |
| "learning_rate": 3.643815619560946e-06, |
| "loss": 0.1636, |
| "step": 204000 |
| }, |
| { |
| "epoch": 2.4603575639452346, |
| "grad_norm": 0.05840376392006874, |
| "learning_rate": 3.6037119917867774e-06, |
| "loss": 0.1824, |
| "step": 204500 |
| }, |
| { |
| "epoch": 2.46637310811136, |
| "grad_norm": 0.0005692685954272747, |
| "learning_rate": 3.563608364012609e-06, |
| "loss": 0.1189, |
| "step": 205000 |
| }, |
| { |
| "epoch": 2.472388652277485, |
| "grad_norm": 12.676384925842285, |
| "learning_rate": 3.5235047362384407e-06, |
| "loss": 0.1589, |
| "step": 205500 |
| }, |
| { |
| "epoch": 2.47840419644361, |
| "grad_norm": 0.024767233058810234, |
| "learning_rate": 3.483401108464272e-06, |
| "loss": 0.179, |
| "step": 206000 |
| }, |
| { |
| "epoch": 2.4844197406097357, |
| "grad_norm": 0.014465268701314926, |
| "learning_rate": 3.443297480690104e-06, |
| "loss": 0.134, |
| "step": 206500 |
| }, |
| { |
| "epoch": 2.490435284775861, |
| "grad_norm": 46.07574462890625, |
| "learning_rate": 3.4031938529159353e-06, |
| "loss": 0.1839, |
| "step": 207000 |
| }, |
| { |
| "epoch": 2.496450828941986, |
| "grad_norm": 2.9257161617279053, |
| "learning_rate": 3.3630902251417667e-06, |
| "loss": 0.1522, |
| "step": 207500 |
| }, |
| { |
| "epoch": 2.5024663731081116, |
| "grad_norm": 0.010684625245630741, |
| "learning_rate": 3.3230668046231463e-06, |
| "loss": 0.137, |
| "step": 208000 |
| }, |
| { |
| "epoch": 2.5084819172742367, |
| "grad_norm": 0.023651426658034325, |
| "learning_rate": 3.2829631768489777e-06, |
| "loss": 0.1893, |
| "step": 208500 |
| }, |
| { |
| "epoch": 2.514497461440362, |
| "grad_norm": 5.37974214553833, |
| "learning_rate": 3.242939756330358e-06, |
| "loss": 0.1688, |
| "step": 209000 |
| }, |
| { |
| "epoch": 2.520513005606487, |
| "grad_norm": 0.21090950071811676, |
| "learning_rate": 3.202836128556189e-06, |
| "loss": 0.173, |
| "step": 209500 |
| }, |
| { |
| "epoch": 2.526528549772612, |
| "grad_norm": 24.091960906982422, |
| "learning_rate": 3.162732500782021e-06, |
| "loss": 0.1614, |
| "step": 210000 |
| }, |
| { |
| "epoch": 2.532544093938738, |
| "grad_norm": 0.032513245940208435, |
| "learning_rate": 3.1226288730078524e-06, |
| "loss": 0.1816, |
| "step": 210500 |
| }, |
| { |
| "epoch": 2.538559638104863, |
| "grad_norm": 0.045939259231090546, |
| "learning_rate": 3.082525245233684e-06, |
| "loss": 0.1646, |
| "step": 211000 |
| }, |
| { |
| "epoch": 2.544575182270988, |
| "grad_norm": 0.7573652863502502, |
| "learning_rate": 3.0424216174595156e-06, |
| "loss": 0.1224, |
| "step": 211500 |
| }, |
| { |
| "epoch": 2.5505907264371137, |
| "grad_norm": 3.404059648513794, |
| "learning_rate": 3.0023981969408957e-06, |
| "loss": 0.1475, |
| "step": 212000 |
| }, |
| { |
| "epoch": 2.556606270603239, |
| "grad_norm": 0.0026354603469371796, |
| "learning_rate": 2.962294569166727e-06, |
| "loss": 0.1591, |
| "step": 212500 |
| }, |
| { |
| "epoch": 2.562621814769364, |
| "grad_norm": 21.002397537231445, |
| "learning_rate": 2.9221909413925585e-06, |
| "loss": 0.133, |
| "step": 213000 |
| }, |
| { |
| "epoch": 2.568637358935489, |
| "grad_norm": 0.020220952108502388, |
| "learning_rate": 2.8820873136183903e-06, |
| "loss": 0.144, |
| "step": 213500 |
| }, |
| { |
| "epoch": 2.5746529031016143, |
| "grad_norm": 0.02407023496925831, |
| "learning_rate": 2.84206389309977e-06, |
| "loss": 0.1522, |
| "step": 214000 |
| }, |
| { |
| "epoch": 2.58066844726774, |
| "grad_norm": 0.05026541277766228, |
| "learning_rate": 2.8019602653256018e-06, |
| "loss": 0.1751, |
| "step": 214500 |
| }, |
| { |
| "epoch": 2.586683991433865, |
| "grad_norm": 0.0013814900303259492, |
| "learning_rate": 2.761856637551433e-06, |
| "loss": 0.1606, |
| "step": 215000 |
| }, |
| { |
| "epoch": 2.5926995355999902, |
| "grad_norm": 0.09587771445512772, |
| "learning_rate": 2.721753009777265e-06, |
| "loss": 0.1553, |
| "step": 215500 |
| }, |
| { |
| "epoch": 2.598715079766116, |
| "grad_norm": 0.014226296916604042, |
| "learning_rate": 2.6816493820030964e-06, |
| "loss": 0.1732, |
| "step": 216000 |
| }, |
| { |
| "epoch": 2.604730623932241, |
| "grad_norm": 0.020662061870098114, |
| "learning_rate": 2.641545754228928e-06, |
| "loss": 0.1851, |
| "step": 216500 |
| }, |
| { |
| "epoch": 2.610746168098366, |
| "grad_norm": 0.020077265799045563, |
| "learning_rate": 2.6015223337103074e-06, |
| "loss": 0.1672, |
| "step": 217000 |
| }, |
| { |
| "epoch": 2.6167617122644913, |
| "grad_norm": 17.917278289794922, |
| "learning_rate": 2.561418705936139e-06, |
| "loss": 0.1757, |
| "step": 217500 |
| }, |
| { |
| "epoch": 2.622777256430617, |
| "grad_norm": 0.010099813342094421, |
| "learning_rate": 2.5213150781619707e-06, |
| "loss": 0.154, |
| "step": 218000 |
| }, |
| { |
| "epoch": 2.628792800596742, |
| "grad_norm": 50.070430755615234, |
| "learning_rate": 2.4812114503878025e-06, |
| "loss": 0.1557, |
| "step": 218500 |
| }, |
| { |
| "epoch": 2.634808344762867, |
| "grad_norm": 0.0430847629904747, |
| "learning_rate": 2.441107822613634e-06, |
| "loss": 0.1578, |
| "step": 219000 |
| }, |
| { |
| "epoch": 2.640823888928993, |
| "grad_norm": 45.390384674072266, |
| "learning_rate": 2.4010844020950135e-06, |
| "loss": 0.1584, |
| "step": 219500 |
| }, |
| { |
| "epoch": 2.646839433095118, |
| "grad_norm": 0.01655266433954239, |
| "learning_rate": 2.361141188831942e-06, |
| "loss": 0.1926, |
| "step": 220000 |
| }, |
| { |
| "epoch": 2.652854977261243, |
| "grad_norm": 0.006697576493024826, |
| "learning_rate": 2.3210375610577732e-06, |
| "loss": 0.1469, |
| "step": 220500 |
| }, |
| { |
| "epoch": 2.6588705214273682, |
| "grad_norm": 0.055724818259477615, |
| "learning_rate": 2.280933933283605e-06, |
| "loss": 0.1548, |
| "step": 221000 |
| }, |
| { |
| "epoch": 2.6648860655934934, |
| "grad_norm": 0.09359436482191086, |
| "learning_rate": 2.2408303055094365e-06, |
| "loss": 0.1523, |
| "step": 221500 |
| }, |
| { |
| "epoch": 2.670901609759619, |
| "grad_norm": 0.005888832733035088, |
| "learning_rate": 2.2007266777352683e-06, |
| "loss": 0.1896, |
| "step": 222000 |
| }, |
| { |
| "epoch": 2.676917153925744, |
| "grad_norm": 0.018979087471961975, |
| "learning_rate": 2.1606230499610997e-06, |
| "loss": 0.123, |
| "step": 222500 |
| }, |
| { |
| "epoch": 2.6829326980918693, |
| "grad_norm": 0.11329194158315659, |
| "learning_rate": 2.1205194221869315e-06, |
| "loss": 0.1556, |
| "step": 223000 |
| }, |
| { |
| "epoch": 2.688948242257995, |
| "grad_norm": 0.018361905589699745, |
| "learning_rate": 2.080415794412763e-06, |
| "loss": 0.1621, |
| "step": 223500 |
| }, |
| { |
| "epoch": 2.69496378642412, |
| "grad_norm": 0.00052923389011994, |
| "learning_rate": 2.0403121666385943e-06, |
| "loss": 0.1309, |
| "step": 224000 |
| }, |
| { |
| "epoch": 2.700979330590245, |
| "grad_norm": 0.030753634870052338, |
| "learning_rate": 2.0002085388644257e-06, |
| "loss": 0.1401, |
| "step": 224500 |
| }, |
| { |
| "epoch": 2.7069948747563704, |
| "grad_norm": 0.01107688806951046, |
| "learning_rate": 1.9601049110902575e-06, |
| "loss": 0.1444, |
| "step": 225000 |
| }, |
| { |
| "epoch": 2.7130104189224955, |
| "grad_norm": 0.0049855210818350315, |
| "learning_rate": 1.920001283316089e-06, |
| "loss": 0.1398, |
| "step": 225500 |
| }, |
| { |
| "epoch": 2.719025963088621, |
| "grad_norm": 52.67742156982422, |
| "learning_rate": 1.8798976555419205e-06, |
| "loss": 0.1743, |
| "step": 226000 |
| }, |
| { |
| "epoch": 2.7250415072547463, |
| "grad_norm": 0.17802126705646515, |
| "learning_rate": 1.839794027767752e-06, |
| "loss": 0.1774, |
| "step": 226500 |
| }, |
| { |
| "epoch": 2.7310570514208714, |
| "grad_norm": 0.03584326431155205, |
| "learning_rate": 1.7996903999935835e-06, |
| "loss": 0.1282, |
| "step": 227000 |
| }, |
| { |
| "epoch": 2.737072595586997, |
| "grad_norm": 0.08680440485477448, |
| "learning_rate": 1.7595867722194151e-06, |
| "loss": 0.1281, |
| "step": 227500 |
| }, |
| { |
| "epoch": 2.743088139753122, |
| "grad_norm": 0.2684645354747772, |
| "learning_rate": 1.719563351700795e-06, |
| "loss": 0.1414, |
| "step": 228000 |
| }, |
| { |
| "epoch": 2.7491036839192473, |
| "grad_norm": 0.027800705283880234, |
| "learning_rate": 1.679539931182175e-06, |
| "loss": 0.1622, |
| "step": 228500 |
| }, |
| { |
| "epoch": 2.7551192280853725, |
| "grad_norm": 11.01578140258789, |
| "learning_rate": 1.6394363034080062e-06, |
| "loss": 0.1396, |
| "step": 229000 |
| }, |
| { |
| "epoch": 2.7611347722514976, |
| "grad_norm": 0.0035239350982010365, |
| "learning_rate": 1.5993326756338379e-06, |
| "loss": 0.1867, |
| "step": 229500 |
| }, |
| { |
| "epoch": 2.767150316417623, |
| "grad_norm": 0.04345840960741043, |
| "learning_rate": 1.5592290478596695e-06, |
| "loss": 0.1404, |
| "step": 230000 |
| }, |
| { |
| "epoch": 2.7731658605837484, |
| "grad_norm": 0.003262243466451764, |
| "learning_rate": 1.5192056273410493e-06, |
| "loss": 0.1605, |
| "step": 230500 |
| }, |
| { |
| "epoch": 2.7791814047498735, |
| "grad_norm": 0.4090683162212372, |
| "learning_rate": 1.479101999566881e-06, |
| "loss": 0.1547, |
| "step": 231000 |
| }, |
| { |
| "epoch": 2.785196948915999, |
| "grad_norm": 0.012451679445803165, |
| "learning_rate": 1.4389983717927125e-06, |
| "loss": 0.1147, |
| "step": 231500 |
| }, |
| { |
| "epoch": 2.7912124930821243, |
| "grad_norm": 0.008698958903551102, |
| "learning_rate": 1.3988947440185442e-06, |
| "loss": 0.1869, |
| "step": 232000 |
| }, |
| { |
| "epoch": 2.7972280372482494, |
| "grad_norm": 1.2418529987335205, |
| "learning_rate": 1.3587911162443755e-06, |
| "loss": 0.1428, |
| "step": 232500 |
| }, |
| { |
| "epoch": 2.8032435814143746, |
| "grad_norm": 0.24010685086250305, |
| "learning_rate": 1.3186874884702072e-06, |
| "loss": 0.172, |
| "step": 233000 |
| }, |
| { |
| "epoch": 2.8092591255805, |
| "grad_norm": 0.008175286464393139, |
| "learning_rate": 1.2785838606960388e-06, |
| "loss": 0.1932, |
| "step": 233500 |
| }, |
| { |
| "epoch": 2.8152746697466253, |
| "grad_norm": 0.009441105648875237, |
| "learning_rate": 1.2384802329218702e-06, |
| "loss": 0.1202, |
| "step": 234000 |
| }, |
| { |
| "epoch": 2.8212902139127505, |
| "grad_norm": 0.0015565038193017244, |
| "learning_rate": 1.1984568124032502e-06, |
| "loss": 0.1215, |
| "step": 234500 |
| }, |
| { |
| "epoch": 2.827305758078876, |
| "grad_norm": 71.70852661132812, |
| "learning_rate": 1.1583531846290816e-06, |
| "loss": 0.1498, |
| "step": 235000 |
| }, |
| { |
| "epoch": 2.8333213022450012, |
| "grad_norm": 4.693429470062256, |
| "learning_rate": 1.1182495568549132e-06, |
| "loss": 0.1463, |
| "step": 235500 |
| }, |
| { |
| "epoch": 2.8393368464111264, |
| "grad_norm": 0.04238261282444, |
| "learning_rate": 1.0781459290807446e-06, |
| "loss": 0.1874, |
| "step": 236000 |
| }, |
| { |
| "epoch": 2.8453523905772515, |
| "grad_norm": 0.091603122651577, |
| "learning_rate": 1.0380423013065762e-06, |
| "loss": 0.1495, |
| "step": 236500 |
| }, |
| { |
| "epoch": 2.8513679347433767, |
| "grad_norm": 26.191734313964844, |
| "learning_rate": 9.979386735324079e-07, |
| "loss": 0.1246, |
| "step": 237000 |
| }, |
| { |
| "epoch": 2.8573834789095023, |
| "grad_norm": 0.0009205593378283083, |
| "learning_rate": 9.579152530137877e-07, |
| "loss": 0.1417, |
| "step": 237500 |
| }, |
| { |
| "epoch": 2.8633990230756274, |
| "grad_norm": 0.0035299675073474646, |
| "learning_rate": 9.178116252396192e-07, |
| "loss": 0.1665, |
| "step": 238000 |
| }, |
| { |
| "epoch": 2.8694145672417526, |
| "grad_norm": 0.004722926300019026, |
| "learning_rate": 8.777079974654508e-07, |
| "loss": 0.144, |
| "step": 238500 |
| }, |
| { |
| "epoch": 2.875430111407878, |
| "grad_norm": 0.007760610897094011, |
| "learning_rate": 8.376043696912823e-07, |
| "loss": 0.1325, |
| "step": 239000 |
| }, |
| { |
| "epoch": 2.8814456555740033, |
| "grad_norm": 0.0008993456140160561, |
| "learning_rate": 7.975007419171139e-07, |
| "loss": 0.1572, |
| "step": 239500 |
| }, |
| { |
| "epoch": 2.8874611997401285, |
| "grad_norm": 0.18450585007667542, |
| "learning_rate": 7.574773213984937e-07, |
| "loss": 0.1473, |
| "step": 240000 |
| }, |
| { |
| "epoch": 2.8934767439062536, |
| "grad_norm": 0.0713871493935585, |
| "learning_rate": 7.173736936243253e-07, |
| "loss": 0.1216, |
| "step": 240500 |
| }, |
| { |
| "epoch": 2.899492288072379, |
| "grad_norm": 0.0023451121523976326, |
| "learning_rate": 6.773502731057053e-07, |
| "loss": 0.1409, |
| "step": 241000 |
| }, |
| { |
| "epoch": 2.9055078322385044, |
| "grad_norm": 0.014280580915510654, |
| "learning_rate": 6.372466453315368e-07, |
| "loss": 0.1946, |
| "step": 241500 |
| }, |
| { |
| "epoch": 2.9115233764046295, |
| "grad_norm": 0.008721483871340752, |
| "learning_rate": 5.971430175573683e-07, |
| "loss": 0.1555, |
| "step": 242000 |
| }, |
| { |
| "epoch": 2.9175389205707547, |
| "grad_norm": 0.039840217679739, |
| "learning_rate": 5.570393897831999e-07, |
| "loss": 0.1427, |
| "step": 242500 |
| }, |
| { |
| "epoch": 2.9235544647368803, |
| "grad_norm": 0.002139493590220809, |
| "learning_rate": 5.170159692645797e-07, |
| "loss": 0.1436, |
| "step": 243000 |
| }, |
| { |
| "epoch": 2.9295700089030055, |
| "grad_norm": 0.0007610457250848413, |
| "learning_rate": 4.769123414904113e-07, |
| "loss": 0.1393, |
| "step": 243500 |
| }, |
| { |
| "epoch": 2.9355855530691306, |
| "grad_norm": 0.05628383159637451, |
| "learning_rate": 4.368087137162428e-07, |
| "loss": 0.1699, |
| "step": 244000 |
| }, |
| { |
| "epoch": 2.9416010972352558, |
| "grad_norm": 0.02916884608566761, |
| "learning_rate": 3.9670508594207435e-07, |
| "loss": 0.1401, |
| "step": 244500 |
| }, |
| { |
| "epoch": 2.947616641401381, |
| "grad_norm": 0.11827978491783142, |
| "learning_rate": 3.5660145816790586e-07, |
| "loss": 0.168, |
| "step": 245000 |
| }, |
| { |
| "epoch": 2.9536321855675065, |
| "grad_norm": 0.0027151680551469326, |
| "learning_rate": 3.165780376492858e-07, |
| "loss": 0.1561, |
| "step": 245500 |
| }, |
| { |
| "epoch": 2.9596477297336317, |
| "grad_norm": 0.040678609162569046, |
| "learning_rate": 2.764744098751173e-07, |
| "loss": 0.1305, |
| "step": 246000 |
| }, |
| { |
| "epoch": 2.965663273899757, |
| "grad_norm": 7.518092632293701, |
| "learning_rate": 2.3637078210094888e-07, |
| "loss": 0.2053, |
| "step": 246500 |
| }, |
| { |
| "epoch": 2.9716788180658824, |
| "grad_norm": 0.18074116110801697, |
| "learning_rate": 1.962671543267804e-07, |
| "loss": 0.1548, |
| "step": 247000 |
| }, |
| { |
| "epoch": 2.9776943622320076, |
| "grad_norm": 7.7608208656311035, |
| "learning_rate": 1.5616352655261197e-07, |
| "loss": 0.1676, |
| "step": 247500 |
| }, |
| { |
| "epoch": 2.9837099063981327, |
| "grad_norm": 0.00906237680464983, |
| "learning_rate": 1.1605989877844351e-07, |
| "loss": 0.1534, |
| "step": 248000 |
| }, |
| { |
| "epoch": 2.989725450564258, |
| "grad_norm": 179.03021240234375, |
| "learning_rate": 7.595627100427505e-08, |
| "loss": 0.1176, |
| "step": 248500 |
| }, |
| { |
| "epoch": 2.9957409947303835, |
| "grad_norm": 0.00344484462402761, |
| "learning_rate": 3.58526432301066e-08, |
| "loss": 0.1711, |
| "step": 249000 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_accuracy": 0.6820289064630488, |
| "eval_f1": 0.7003498950314906, |
| "eval_loss": 0.4521113932132721, |
| "eval_roc_auc": 0.8145350422688847, |
| "eval_runtime": 110.6722, |
| "eval_samples_per_second": 66.268, |
| "eval_steps_per_second": 66.268, |
| "step": 249354 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 249354, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.6370896223238963e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
