diff --git "a/checkpoint-810/trainer_state.json" "b/checkpoint-810/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-810/trainer_state.json" @@ -0,0 +1,4889 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0163111668757843, + "eval_steps": 500, + "global_step": 810, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 2e-05, + "loss": 0.0, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 4e-05, + "loss": 0.0, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 6e-05, + "loss": 0.0, + "step": 3 + }, + { + "epoch": 0.01, + "learning_rate": 8e-05, + "loss": 0.0, + "step": 4 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 5 + }, + { + "epoch": 0.01, + "learning_rate": 0.00012, + "loss": 0.0, + "step": 6 + }, + { + "epoch": 0.01, + "learning_rate": 0.00014, + "loss": 0.0, + "step": 7 + }, + { + "epoch": 0.01, + "learning_rate": 0.00016, + "loss": 0.0, + "step": 8 + }, + { + "epoch": 0.01, + "learning_rate": 0.00018, + "loss": 0.0, + "step": 9 + }, + { + "epoch": 0.01, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 0.00019999980332042875, + "loss": 0.0, + "step": 11 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019999921328248872, + "loss": 0.0, + "step": 12 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019999822988850082, + "loss": 0.0, + "step": 13 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001999968531423333, + "loss": 0.0, + "step": 14 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001999950830494018, + "loss": 0.0, + "step": 15 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019999291961666908, + "loss": 0.0, + "step": 16 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019999036285264524, + "loss": 0.0, + "step": 17 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019998741276738754, + "loss": 0.0, + "step": 18 + }, + { + "epoch": 0.02, + "learning_rate": 0.00019998406937250034, + "loss": 0.0, + "step": 19 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019998033268113526, + "loss": 0.0, + "step": 20 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019997620270799092, + "loss": 0.0, + "step": 21 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001999716794693129, + "loss": 0.0, + "step": 22 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019996676298289384, + "loss": 0.0, + "step": 23 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019996145326807313, + "loss": 0.0, + "step": 24 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019995575034573705, + "loss": 0.0, + "step": 25 + }, + { + "epoch": 0.03, + "learning_rate": 0.00019994965423831854, + "loss": 0.0, + "step": 26 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001999431649697972, + "loss": 0.0, + "step": 27 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001999362825656992, + "loss": 0.0, + "step": 28 + }, + { + "epoch": 0.04, + "learning_rate": 0.00019992900705309704, + "loss": 0.0, + "step": 29 + }, + { + "epoch": 0.04, + "learning_rate": 0.00019992133846060968, + "loss": 0.0, + "step": 30 + }, + { + "epoch": 0.04, + "learning_rate": 0.00019991327681840218, + "loss": 0.0, + "step": 31 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001999048221581858, + "loss": 0.0, + "step": 32 + }, + { + "epoch": 0.04, + "learning_rate": 0.00019989597451321766, + "loss": 0.0, + "step": 33 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001998867339183008, + "loss": 0.0, + "step": 34 + }, + { + "epoch": 0.04, + "learning_rate": 0.00019987710040978395, + "loss": 0.0, + "step": 35 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019986707402556145, + "loss": 0.0, + "step": 36 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001998566548050729, + "loss": 0.0, + "step": 37 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001998458427893033, + "loss": 0.0, + "step": 38 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019983463802078272, + "loss": 0.0, + "step": 39 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019982304054358614, + "loss": 0.0, + "step": 40 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019981105040333327, + "loss": 0.0, + "step": 41 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019979866764718843, + "loss": 0.0, + "step": 42 + }, + { + "epoch": 0.05, + "learning_rate": 0.00019978589232386035, + "loss": 0.0, + "step": 43 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001997727244836019, + "loss": 0.0, + "step": 44 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019975916417821002, + "loss": 0.0, + "step": 45 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019974521146102537, + "loss": 0.0, + "step": 46 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019973086638693226, + "loss": 0.0, + "step": 47 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019971612901235832, + "loss": 0.0, + "step": 48 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001997009993952744, + "loss": 0.0, + "step": 49 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019968547759519425, + "loss": 0.0, + "step": 50 + }, + { + "epoch": 0.06, + "learning_rate": 0.00019966956367317425, + "loss": 0.0, + "step": 51 + }, + { + "epoch": 0.07, + "learning_rate": 0.00019965325769181325, + "loss": 0.0, + "step": 52 + }, + { + "epoch": 0.07, + "learning_rate": 0.00019963655971525233, + "loss": 0.0, + "step": 53 + }, + { + "epoch": 0.07, + "learning_rate": 0.00019961946980917456, + "loss": 0.0, + "step": 54 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001996019880408046, + "loss": 0.0, + "step": 55 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001995841144789086, + "loss": 0.0, + "step": 56 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001995658491937938, + "loss": 0.0, + "step": 57 + }, + { + "epoch": 0.07, + "learning_rate": 0.00019954719225730847, + "loss": 0.0, + "step": 58 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001995281437428413, + "loss": 0.0, + "step": 59 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019950870372532138, + "loss": 0.0, + "step": 60 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019948887228121777, + "loss": 0.0, + "step": 61 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019946864948853935, + "loss": 0.0, + "step": 62 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001994480354268342, + "loss": 0.0, + "step": 63 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019942703017718975, + "loss": 0.0, + "step": 64 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019940563382223197, + "loss": 0.0, + "step": 65 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019938384644612543, + "loss": 0.0, + "step": 66 + }, + { + "epoch": 0.08, + "learning_rate": 0.00019936166813457274, + "loss": 0.0, + "step": 67 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019933909897481433, + "loss": 0.0, + "step": 68 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019931613905562804, + "loss": 0.0, + "step": 69 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019929278846732884, + "loss": 0.0, + "step": 70 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001992690473017684, + "loss": 0.0, + "step": 71 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001992449156523347, + "loss": 0.0, + "step": 72 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019922039361395185, + "loss": 0.0, + "step": 73 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019919548128307954, + "loss": 0.0, + "step": 74 + }, + { + "epoch": 0.09, + "learning_rate": 0.00019917017875771266, + "loss": 0.0, + "step": 75 + }, + { + "epoch": 0.1, + "learning_rate": 0.00019914448613738106, + "loss": 0.0, + "step": 76 + }, + { + "epoch": 0.1, + "learning_rate": 0.00019911840352314895, + "loss": 0.0, + "step": 77 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001990919310176147, + "loss": 0.0, + "step": 78 + }, + { + "epoch": 0.1, + "learning_rate": 0.00019906506872491034, + "loss": 0.0, + "step": 79 + }, + { + "epoch": 0.1, + "learning_rate": 0.00019903781675070117, + "loss": 0.0, + "step": 80 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001990101752021853, + "loss": 0.0, + "step": 81 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001989821441880933, + "loss": 0.0, + "step": 82 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001989537238186877, + "loss": 0.0, + "step": 83 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019892491420576261, + "loss": 0.0, + "step": 84 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019889571546264335, + "loss": 0.0, + "step": 85 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019886612770418578, + "loss": 0.0, + "step": 86 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019883615104677608, + "loss": 0.0, + "step": 87 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019880578560833016, + "loss": 0.0, + "step": 88 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019877503150829326, + "loss": 0.0, + "step": 89 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019874388886763944, + "loss": 0.0, + "step": 90 + }, + { + "epoch": 0.11, + "learning_rate": 0.00019871235780887113, + "loss": 0.0, + "step": 91 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019868043845601863, + "loss": 0.0, + "step": 92 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019864813093463962, + "loss": 0.0, + "step": 93 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019861543537181867, + "loss": 0.0, + "step": 94 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019858235189616682, + "loss": 0.0, + "step": 95 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019854888063782088, + "loss": 0.0, + "step": 96 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019851502172844317, + "loss": 0.0, + "step": 97 + }, + { + "epoch": 0.12, + "learning_rate": 0.00019848077530122083, + "loss": 0.0, + "step": 98 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001984461414908652, + "loss": 0.0, + "step": 99 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001984111204336116, + "loss": 0.0, + "step": 100 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001983757122672186, + "loss": 0.0, + "step": 101 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001983399171309674, + "loss": 0.0, + "step": 102 + }, + { + "epoch": 0.13, + "learning_rate": 0.00019830373516566146, + "loss": 0.0, + "step": 103 + }, + { + "epoch": 0.13, + "learning_rate": 0.00019826716651362585, + "loss": 0.0, + "step": 104 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001982302113187067, + "loss": 0.0, + "step": 105 + }, + { + "epoch": 0.13, + "learning_rate": 0.00019819286972627066, + "loss": 0.0, + "step": 106 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001981551418832043, + "loss": 0.0, + "step": 107 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019811702793791355, + "loss": 0.0, + "step": 108 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019807852804032305, + "loss": 0.0, + "step": 109 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001980396423418757, + "loss": 0.0, + "step": 110 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019800037099553197, + "loss": 0.0, + "step": 111 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019796071415576925, + "loss": 0.0, + "step": 112 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019792067197858137, + "loss": 0.0, + "step": 113 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019788024462147788, + "loss": 0.0, + "step": 114 + }, + { + "epoch": 0.14, + "learning_rate": 0.00019783943224348352, + "loss": 0.0, + "step": 115 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019779823500513745, + "loss": 0.0, + "step": 116 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001977566530684928, + "loss": 0.0, + "step": 117 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019771468659711595, + "loss": 0.0, + "step": 118 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019767233575608583, + "loss": 0.0, + "step": 119 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019762960071199333, + "loss": 0.0, + "step": 120 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001975864816329407, + "loss": 0.0, + "step": 121 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019754297868854073, + "loss": 0.0, + "step": 122 + }, + { + "epoch": 0.15, + "learning_rate": 0.00019749909204991623, + "loss": 0.0, + "step": 123 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001974548218896993, + "loss": 0.0, + "step": 124 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019741016838203073, + "loss": 0.0, + "step": 125 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019736513170255911, + "loss": 0.0, + "step": 126 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019731971202844036, + "loss": 0.0, + "step": 127 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001972739095383369, + "loss": 0.0, + "step": 128 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019722772441241703, + "loss": 0.0, + "step": 129 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019718115683235417, + "loss": 0.0, + "step": 130 + }, + { + "epoch": 0.16, + "learning_rate": 0.00019713420698132614, + "loss": 0.0, + "step": 131 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001970868750440145, + "loss": 0.0, + "step": 132 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001970391612066037, + "loss": 0.0, + "step": 133 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001969910656567805, + "loss": 0.0, + "step": 134 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001969425885837332, + "loss": 0.0, + "step": 135 + }, + { + "epoch": 0.17, + "learning_rate": 0.00019689373017815073, + "loss": 0.0, + "step": 136 + }, + { + "epoch": 0.17, + "learning_rate": 0.00019684449063222213, + "loss": 0.0, + "step": 137 + }, + { + "epoch": 0.17, + "learning_rate": 0.00019679487013963564, + "loss": 0.0, + "step": 138 + }, + { + "epoch": 0.17, + "learning_rate": 0.000196744868895578, + "loss": 0.0, + "step": 139 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001966944870967337, + "loss": 0.0, + "step": 140 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019664372494128412, + "loss": 0.0, + "step": 141 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019659258262890683, + "loss": 0.0, + "step": 142 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019654106036077486, + "loss": 0.0, + "step": 143 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001964891583395557, + "loss": 0.0, + "step": 144 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019643687676941068, + "loss": 0.0, + "step": 145 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019638421585599423, + "loss": 0.0, + "step": 146 + }, + { + "epoch": 0.18, + "learning_rate": 0.00019633117580645278, + "loss": 0.0, + "step": 147 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001962777568294242, + "loss": 0.0, + "step": 148 + }, + { + "epoch": 0.19, + "learning_rate": 0.00019622395913503703, + "loss": 0.0, + "step": 149 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001961697829349093, + "loss": 0.0, + "step": 150 + }, + { + "epoch": 0.19, + "learning_rate": 0.00019611522844214813, + "loss": 0.0, + "step": 151 + }, + { + "epoch": 0.19, + "learning_rate": 0.00019606029587134854, + "loss": 0.0, + "step": 152 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001960049854385929, + "loss": 0.0, + "step": 153 + }, + { + "epoch": 0.19, + "learning_rate": 0.00019594929736144976, + "loss": 0.0, + "step": 154 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001958932318589733, + "loss": 0.0, + "step": 155 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019583678915170233, + "loss": 0.0, + "step": 156 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019577996946165937, + "loss": 0.0, + "step": 157 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019572277301234986, + "loss": 0.0, + "step": 158 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019566520002876133, + "loss": 0.0, + "step": 159 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019560725073736226, + "loss": 0.0, + "step": 160 + }, + { + "epoch": 0.2, + "learning_rate": 0.00019554892536610157, + "loss": 0.0, + "step": 161 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001954902241444074, + "loss": 0.0, + "step": 162 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001954311473031864, + "loss": 0.0, + "step": 163 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001953716950748227, + "loss": 0.0, + "step": 164 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019531186769317712, + "loss": 0.0, + "step": 165 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019525166539358606, + "loss": 0.0, + "step": 166 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019519108841286085, + "loss": 0.0, + "step": 167 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019513013698928652, + "loss": 0.0, + "step": 168 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019506881136262112, + "loss": 0.0, + "step": 169 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019500711177409454, + "loss": 0.0, + "step": 170 + }, + { + "epoch": 0.21, + "learning_rate": 0.00019494503846640782, + "loss": 0.0, + "step": 171 + }, + { + "epoch": 0.22, + "learning_rate": 0.00019488259168373197, + "loss": 0.0, + "step": 172 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001948197716717071, + "loss": 0.0, + "step": 173 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001947565786774415, + "loss": 0.0, + "step": 174 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001946930129495106, + "loss": 0.0, + "step": 175 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001946290747379559, + "loss": 0.0, + "step": 176 + }, + { + "epoch": 0.22, + "learning_rate": 0.00019456476429428437, + "loss": 0.0, + "step": 177 + }, + { + "epoch": 0.22, + "learning_rate": 0.00019450008187146684, + "loss": 0.0, + "step": 178 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001944350277239377, + "loss": 0.0, + "step": 179 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019436960210759326, + "loss": 0.0, + "step": 180 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019430380527979123, + "loss": 0.0, + "step": 181 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001942376374993494, + "loss": 0.0, + "step": 182 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019417109902654484, + "loss": 0.0, + "step": 183 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019410419012311268, + "loss": 0.0, + "step": 184 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019403691105224522, + "loss": 0.0, + "step": 185 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019396926207859084, + "loss": 0.0, + "step": 186 + }, + { + "epoch": 0.23, + "learning_rate": 0.00019390124346825295, + "loss": 0.0, + "step": 187 + }, + { + "epoch": 0.24, + "learning_rate": 0.00019383285548878898, + "loss": 0.0, + "step": 188 + }, + { + "epoch": 0.24, + "learning_rate": 0.00019376409840920933, + "loss": 0.0, + "step": 189 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001936949724999762, + "loss": 0.0, + "step": 190 + }, + { + "epoch": 0.24, + "learning_rate": 0.00019362547803300267, + "loss": 0.0, + "step": 191 + }, + { + "epoch": 0.24, + "learning_rate": 0.00019355561528165165, + "loss": 0.0, + "step": 192 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001934853845207346, + "loss": 0.0, + "step": 193 + }, + { + "epoch": 0.24, + "learning_rate": 0.00019341478602651069, + "loss": 0.0, + "step": 194 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001933438200766855, + "loss": 0.0, + "step": 195 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001932724869504101, + "loss": 0.0, + "step": 196 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019320078692827987, + "loss": 0.0, + "step": 197 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019312872029233339, + "loss": 0.0, + "step": 198 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019305628732605137, + "loss": 0.0, + "step": 199 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001929834883143555, + "loss": 0.0, + "step": 200 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019291032354360737, + "loss": 0.0, + "step": 201 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019283679330160726, + "loss": 0.0, + "step": 202 + }, + { + "epoch": 0.25, + "learning_rate": 0.00019276289787759315, + "loss": 0.0, + "step": 203 + }, + { + "epoch": 0.26, + "learning_rate": 0.00019268863756223938, + "loss": 0.0, + "step": 204 + }, + { + "epoch": 0.26, + "learning_rate": 0.00019261401264765573, + "loss": 0.0, + "step": 205 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001925390234273861, + "loss": 0.0, + "step": 206 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001924636701964075, + "loss": 0.0, + "step": 207 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001923879532511287, + "loss": 0.0, + "step": 208 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001923118728893892, + "loss": 0.0, + "step": 209 + }, + { + "epoch": 0.26, + "learning_rate": 0.00019223542941045817, + "loss": 0.0, + "step": 210 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001921586231150329, + "loss": 0.0, + "step": 211 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019208145430523805, + "loss": 0.0, + "step": 212 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019200392328462415, + "loss": 0.0, + "step": 213 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019192603035816656, + "loss": 0.0, + "step": 214 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019184777583226427, + "loss": 0.0, + "step": 215 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019176916001473857, + "loss": 0.0, + "step": 216 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019169018321483198, + "loss": 0.0, + "step": 217 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019161084574320696, + "loss": 0.0, + "step": 218 + }, + { + "epoch": 0.27, + "learning_rate": 0.00019153114791194473, + "loss": 0.0, + "step": 219 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019145109003454396, + "loss": 0.0, + "step": 220 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019137067242591964, + "loss": 0.0, + "step": 221 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019128989540240178, + "loss": 0.0, + "step": 222 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019120875928173422, + "loss": 0.0, + "step": 223 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019112726438307327, + "loss": 0.0, + "step": 224 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001910454110269866, + "loss": 0.0, + "step": 225 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019096319953545185, + "loss": 0.0, + "step": 226 + }, + { + "epoch": 0.28, + "learning_rate": 0.00019088063023185543, + "loss": 0.0, + "step": 227 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019079770344099126, + "loss": 0.0, + "step": 228 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019071441948905943, + "loss": 0.0, + "step": 229 + }, + { + "epoch": 0.29, + "learning_rate": 0.000190630778703665, + "loss": 0.0, + "step": 230 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019054678141381664, + "loss": 0.0, + "step": 231 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019046242794992538, + "loss": 0.0, + "step": 232 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001903777186438033, + "loss": 0.0, + "step": 233 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019029265382866214, + "loss": 0.0, + "step": 234 + }, + { + "epoch": 0.29, + "learning_rate": 0.00019020723383911215, + "loss": 0.0, + "step": 235 + }, + { + "epoch": 0.3, + "learning_rate": 0.00019012145901116072, + "loss": 0.0, + "step": 236 + }, + { + "epoch": 0.3, + "learning_rate": 0.00019003532968221089, + "loss": 0.0, + "step": 237 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018994884619106031, + "loss": 0.0, + "step": 238 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018986200887789972, + "loss": 0.0, + "step": 239 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018977481808431156, + "loss": 0.0, + "step": 240 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018968727415326884, + "loss": 0.0, + "step": 241 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018959937742913359, + "loss": 0.0, + "step": 242 + }, + { + "epoch": 0.3, + "learning_rate": 0.00018951112825765564, + "loss": 0.0, + "step": 243 + }, + { + "epoch": 0.31, + "learning_rate": 0.00018942252698597113, + "loss": 0.0, + "step": 244 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001893335739626013, + "loss": 0.0, + "step": 245 + }, + { + "epoch": 0.31, + "learning_rate": 0.000189244269537451, + "loss": 0.0, + "step": 246 + }, + { + "epoch": 0.31, + "learning_rate": 0.00018915461406180731, + "loss": 0.0, + "step": 247 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001890646078883383, + "loss": 0.0, + "step": 248 + }, + { + "epoch": 0.31, + "learning_rate": 0.00018897425137109143, + "loss": 0.0, + "step": 249 + }, + { + "epoch": 0.31, + "learning_rate": 0.00018888354486549237, + "loss": 0.0, + "step": 250 + }, + { + "epoch": 0.31, + "learning_rate": 0.00018879248872834338, + "loss": 0.0, + "step": 251 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018870108331782217, + "loss": 0.0, + "step": 252 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018860932899348028, + "loss": 0.0, + "step": 253 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018851722611624164, + "loss": 0.0, + "step": 254 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018842477504840149, + "loss": 0.0, + "step": 255 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001883319761536244, + "loss": 0.0, + "step": 256 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018823882979694342, + "loss": 0.0, + "step": 257 + }, + { + "epoch": 0.32, + "learning_rate": 0.00018814533634475822, + "loss": 0.0, + "step": 258 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001880514961648338, + "loss": 0.0, + "step": 259 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018795730962629917, + "loss": 0.0, + "step": 260 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018786277709964566, + "loss": 0.0, + "step": 261 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018776789895672558, + "loss": 0.0, + "step": 262 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001876726755707508, + "loss": 0.0, + "step": 263 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018757710731629116, + "loss": 0.0, + "step": 264 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018748119456927324, + "loss": 0.0, + "step": 265 + }, + { + "epoch": 0.33, + "learning_rate": 0.00018738493770697852, + "loss": 0.0, + "step": 266 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018728833710804214, + "loss": 0.0, + "step": 267 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018719139315245148, + "loss": 0.0, + "step": 268 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018709410622154437, + "loss": 0.0, + "step": 269 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001869964766980079, + "loss": 0.0, + "step": 270 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018689850496587674, + "loss": 0.0, + "step": 271 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018680019141053156, + "loss": 0.0, + "step": 272 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001867015364186978, + "loss": 0.0, + "step": 273 + }, + { + "epoch": 0.34, + "learning_rate": 0.00018660254037844388, + "loss": 0.0, + "step": 274 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018650320367917975, + "loss": 0.0, + "step": 275 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001864035267116554, + "loss": 0.0, + "step": 276 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018630350986795934, + "loss": 0.0, + "step": 277 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018620315354151695, + "loss": 0.0, + "step": 278 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018610245812708897, + "loss": 0.0, + "step": 279 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018600142402077006, + "loss": 0.0, + "step": 280 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018590005161998709, + "loss": 0.0, + "step": 281 + }, + { + "epoch": 0.35, + "learning_rate": 0.00018579834132349772, + "loss": 0.0, + "step": 282 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018569629353138863, + "loss": 0.0, + "step": 283 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018559390864507418, + "loss": 0.0, + "step": 284 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018549118706729468, + "loss": 0.0, + "step": 285 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001853881292021148, + "loss": 0.0, + "step": 286 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018528473545492217, + "loss": 0.0, + "step": 287 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018518100623242547, + "loss": 0.0, + "step": 288 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001850769419426531, + "loss": 0.0, + "step": 289 + }, + { + "epoch": 0.36, + "learning_rate": 0.00018497254299495146, + "loss": 0.0, + "step": 290 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018486780979998333, + "loss": 0.0, + "step": 291 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018476274276972636, + "loss": 0.0, + "step": 292 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001846573423174713, + "loss": 0.0, + "step": 293 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018455160885782045, + "loss": 0.0, + "step": 294 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018444554280668603, + "loss": 0.0, + "step": 295 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001843391445812886, + "loss": 0.0, + "step": 296 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018423241460015523, + "loss": 0.0, + "step": 297 + }, + { + "epoch": 0.37, + "learning_rate": 0.00018412535328311814, + "loss": 0.0, + "step": 298 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018401796105131275, + "loss": 0.0, + "step": 299 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018391023832717624, + "loss": 0.0, + "step": 300 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001838021855344458, + "loss": 0.0, + "step": 301 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018369380309815698, + "loss": 0.0, + "step": 302 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018358509144464198, + "loss": 0.0, + "step": 303 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018347605100152802, + "loss": 0.0, + "step": 304 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018336668219773569, + "loss": 0.0, + "step": 305 + }, + { + "epoch": 0.38, + "learning_rate": 0.00018325698546347715, + "loss": 0.0, + "step": 306 + }, + { + "epoch": 0.39, + "learning_rate": 0.00018314696123025454, + "loss": 0.0, + "step": 307 + }, + { + "epoch": 0.39, + "learning_rate": 0.00018303660993085826, + "loss": 0.0, + "step": 308 + }, + { + "epoch": 0.39, + "learning_rate": 0.00018292593199936518, + "loss": 0.0, + "step": 309 + }, + { + "epoch": 0.39, + "learning_rate": 0.00018281492787113708, + "loss": 0.0, + "step": 310 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001827035979828189, + "loss": 0.0, + "step": 311 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001825919427723369, + "loss": 0.0, + "step": 312 + }, + { + "epoch": 0.39, + "learning_rate": 0.00018247996267889706, + "loss": 0.0, + "step": 313 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001823676581429833, + "loss": 0.0, + "step": 314 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001822550296063558, + "loss": 0.0, + "step": 315 + }, + { + "epoch": 0.4, + "learning_rate": 0.00018214207751204918, + "loss": 0.0, + "step": 316 + }, + { + "epoch": 0.4, + "learning_rate": 0.00018202880230437085, + "loss": 0.0, + "step": 317 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001819152044288992, + "loss": 0.0, + "step": 318 + }, + { + "epoch": 0.4, + "learning_rate": 0.00018180128433248184, + "loss": 0.0, + "step": 319 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001816870424632339, + "loss": 0.0, + "step": 320 + }, + { + "epoch": 0.4, + "learning_rate": 0.00018157247927053622, + "loss": 0.0, + "step": 321 + }, + { + "epoch": 0.4, + "learning_rate": 0.00018145759520503358, + "loss": 0.0, + "step": 322 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018134239071863296, + "loss": 0.0, + "step": 323 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018122686626450174, + "loss": 0.0, + "step": 324 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001811110222970659, + "loss": 0.0, + "step": 325 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018099485927200836, + "loss": 0.0, + "step": 326 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001808783776462669, + "loss": 0.0, + "step": 327 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018076157787803268, + "loss": 0.0, + "step": 328 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018064446042674828, + "loss": 0.0, + "step": 329 + }, + { + "epoch": 0.41, + "learning_rate": 0.00018052702575310588, + "loss": 0.0, + "step": 330 + }, + { + "epoch": 0.42, + "learning_rate": 0.00018040927431904553, + "loss": 0.0, + "step": 331 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001802912065877532, + "loss": 0.0, + "step": 332 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001801728230236592, + "loss": 0.0, + "step": 333 + }, + { + "epoch": 0.42, + "learning_rate": 0.00018005412409243606, + "loss": 0.0, + "step": 334 + }, + { + "epoch": 0.42, + "learning_rate": 0.00017993511026099686, + "loss": 0.0, + "step": 335 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001798157819974934, + "loss": 0.0, + "step": 336 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001796961397713143, + "loss": 0.0, + "step": 337 + }, + { + "epoch": 0.42, + "learning_rate": 0.00017957618405308324, + "loss": 0.0, + "step": 338 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017945591531465694, + "loss": 0.0, + "step": 339 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017933533402912354, + "loss": 0.0, + "step": 340 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017921444067080052, + "loss": 0.0, + "step": 341 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017909323571523294, + "loss": 0.0, + "step": 342 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001789717196391916, + "loss": 0.0, + "step": 343 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001788498929206711, + "loss": 0.0, + "step": 344 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017872775603888793, + "loss": 0.0, + "step": 345 + }, + { + "epoch": 0.43, + "learning_rate": 0.00017860530947427875, + "loss": 0.0, + "step": 346 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017848255370849829, + "loss": 0.0, + "step": 347 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017835948922441755, + "loss": 0.0, + "step": 348 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017823611650612194, + "loss": 0.0, + "step": 349 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017811243603890934, + "loss": 0.0, + "step": 350 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017798844830928817, + "loss": 0.0, + "step": 351 + }, + { + "epoch": 0.44, + "learning_rate": 0.00017786415380497553, + "loss": 0.0, + "step": 352 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001777395530148952, + "loss": 0.0, + "step": 353 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001776146464291757, + "loss": 0.0, + "step": 354 + }, + { + "epoch": 0.45, + "learning_rate": 0.00017748943453914858, + "loss": 0.0, + "step": 355 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001773639178373463, + "loss": 0.0, + "step": 356 + }, + { + "epoch": 0.45, + "learning_rate": 0.00017723809681750017, + "loss": 0.0, + "step": 357 + }, + { + "epoch": 0.45, + "learning_rate": 0.00017711197197453878, + "loss": 0.0, + "step": 358 + }, + { + "epoch": 0.45, + "learning_rate": 0.00017698554380458567, + "loss": 0.0, + "step": 359 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001768588128049576, + "loss": 0.0, + "step": 360 + }, + { + "epoch": 0.45, + "learning_rate": 0.00017673177947416258, + "loss": 0.0, + "step": 361 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001766044443118978, + "loss": 0.0, + "step": 362 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017647680781904777, + "loss": 0.0, + "step": 363 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017634887049768237, + "loss": 0.0, + "step": 364 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017622063285105465, + "loss": 0.0, + "step": 365 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017609209538359917, + "loss": 0.0, + "step": 366 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017596325860092974, + "loss": 0.0, + "step": 367 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017583412300983776, + "loss": 0.0, + "step": 368 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017570468911828977, + "loss": 0.0, + "step": 369 + }, + { + "epoch": 0.46, + "learning_rate": 0.00017557495743542585, + "loss": 0.0, + "step": 370 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017544492847155742, + "loss": 0.0, + "step": 371 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017531460273816532, + "loss": 0.0, + "step": 372 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017518398074789775, + "loss": 0.0, + "step": 373 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017505306301456822, + "loss": 0.0, + "step": 374 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017492185005315363, + "loss": 0.0, + "step": 375 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001747903423797921, + "loss": 0.0, + "step": 376 + }, + { + "epoch": 0.47, + "learning_rate": 0.00017465854051178117, + "loss": 0.0, + "step": 377 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001745264449675755, + "loss": 0.0, + "step": 378 + }, + { + "epoch": 0.48, + "learning_rate": 0.00017439405626678496, + "loss": 0.0, + "step": 379 + }, + { + "epoch": 0.48, + "learning_rate": 0.00017426137493017265, + "loss": 0.0, + "step": 380 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001741284014796527, + "loss": 0.0, + "step": 381 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001739951364382884, + "loss": 0.0, + "step": 382 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001738615803302899, + "loss": 0.0, + "step": 383 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001737277336810124, + "loss": 0.0, + "step": 384 + }, + { + "epoch": 0.48, + "learning_rate": 0.00017359359701695396, + "loss": 0.0, + "step": 385 + }, + { + "epoch": 0.48, + "learning_rate": 0.00017345917086575332, + "loss": 0.0, + "step": 386 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017332445575618814, + "loss": 0.0, + "step": 387 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017318945221817255, + "loss": 0.0, + "step": 388 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017305416078275534, + "loss": 0.0, + "step": 389 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017291858198211773, + "loss": 0.0, + "step": 390 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001727827163495713, + "loss": 0.0, + "step": 391 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017264656441955603, + "loss": 0.0, + "step": 392 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001725101267276379, + "loss": 0.0, + "step": 393 + }, + { + "epoch": 0.49, + "learning_rate": 0.00017237340381050703, + "loss": 0.0, + "step": 394 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017223639620597556, + "loss": 0.0, + "step": 395 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017209910445297542, + "loss": 0.0, + "step": 396 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017196152909155628, + "loss": 0.0, + "step": 397 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017182367066288342, + "loss": 0.0, + "step": 398 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001716855297092355, + "loss": 0.0, + "step": 399 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017154710677400265, + "loss": 0.0, + "step": 400 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017140840240168417, + "loss": 0.0, + "step": 401 + }, + { + "epoch": 0.5, + "learning_rate": 0.00017126941713788632, + "loss": 0.0, + "step": 402 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017113015152932034, + "loss": 0.0, + "step": 403 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017099060612380028, + "loss": 0.0, + "step": 404 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017085078147024074, + "loss": 0.0, + "step": 405 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017071067811865476, + "loss": 0.0, + "step": 406 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017057029662015171, + "loss": 0.0, + "step": 407 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017042963752693502, + "loss": 0.0, + "step": 408 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001702887013923001, + "loss": 0.0, + "step": 409 + }, + { + "epoch": 0.51, + "learning_rate": 0.00017014748877063214, + "loss": 0.0, + "step": 410 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001700060002174039, + "loss": 0.0, + "step": 411 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016986423628917346, + "loss": 0.0, + "step": 412 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016972219754358232, + "loss": 0.0, + "step": 413 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016957988453935276, + "loss": 0.0, + "step": 414 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016943729783628608, + "loss": 0.0, + "step": 415 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016929443799526002, + "loss": 0.0, + "step": 416 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016915130557822695, + "loss": 0.0, + "step": 417 + }, + { + "epoch": 0.52, + "learning_rate": 0.00016900790114821122, + "loss": 0.0, + "step": 418 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016886422526930726, + "loss": 0.0, + "step": 419 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016872027850667735, + "loss": 0.0, + "step": 420 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016857606142654918, + "loss": 0.0, + "step": 421 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016843157459621384, + "loss": 0.0, + "step": 422 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016828681858402348, + "loss": 0.0, + "step": 423 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016814179395938913, + "loss": 0.0, + "step": 424 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016799650129277837, + "loss": 0.0, + "step": 425 + }, + { + "epoch": 0.53, + "learning_rate": 0.00016785094115571322, + "loss": 0.0, + "step": 426 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001677051141207678, + "loss": 0.0, + "step": 427 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016755902076156604, + "loss": 0.0, + "step": 428 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016741266165277952, + "loss": 0.0, + "step": 429 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016726603737012529, + "loss": 0.0, + "step": 430 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016711914849036318, + "loss": 0.0, + "step": 431 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001669719955912942, + "loss": 0.0, + "step": 432 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016682457925175763, + "loss": 0.0, + "step": 433 + }, + { + "epoch": 0.54, + "learning_rate": 0.00016667690005162916, + "loss": 0.0, + "step": 434 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016652895857181843, + "loss": 0.0, + "step": 435 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016638075539426677, + "loss": 0.0, + "step": 436 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001662322911019449, + "loss": 0.0, + "step": 437 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016608356627885072, + "loss": 0.0, + "step": 438 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016593458151000688, + "loss": 0.0, + "step": 439 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016578533738145868, + "loss": 0.0, + "step": 440 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016563583448027142, + "loss": 0.0, + "step": 441 + }, + { + "epoch": 0.55, + "learning_rate": 0.00016548607339452853, + "loss": 0.0, + "step": 442 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016533605471332887, + "loss": 0.0, + "step": 443 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016518577902678466, + "loss": 0.0, + "step": 444 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016503524692601908, + "loss": 0.0, + "step": 445 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016488445900316386, + "loss": 0.0, + "step": 446 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016473341585135712, + "loss": 0.0, + "step": 447 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016458211806474088, + "loss": 0.0, + "step": 448 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016443056623845884, + "loss": 0.0, + "step": 449 + }, + { + "epoch": 0.56, + "learning_rate": 0.00016427876096865394, + "loss": 0.0, + "step": 450 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001641267028524661, + "loss": 0.0, + "step": 451 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016397439248802984, + "loss": 0.0, + "step": 452 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016382183047447185, + "loss": 0.0, + "step": 453 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016366901741190882, + "loss": 0.0, + "step": 454 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001635159539014449, + "loss": 0.0, + "step": 455 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016336264054516933, + "loss": 0.0, + "step": 456 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016320907794615427, + "loss": 0.0, + "step": 457 + }, + { + "epoch": 0.57, + "learning_rate": 0.00016305526670845226, + "loss": 0.0, + "step": 458 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016290120743709383, + "loss": 0.0, + "step": 459 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001627469007380852, + "loss": 0.0, + "step": 460 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016259234721840591, + "loss": 0.0, + "step": 461 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016243754748600635, + "loss": 0.0, + "step": 462 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001622825021498054, + "loss": 0.0, + "step": 463 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016212721181968812, + "loss": 0.0, + "step": 464 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016197167710650315, + "loss": 0.0, + "step": 465 + }, + { + "epoch": 0.58, + "learning_rate": 0.00016181589862206052, + "loss": 0.0, + "step": 466 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001616598769791292, + "loss": 0.0, + "step": 467 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001615036127914345, + "loss": 0.0, + "step": 468 + }, + { + "epoch": 0.59, + "learning_rate": 0.00016134710667365596, + "loss": 0.0, + "step": 469 + }, + { + "epoch": 0.59, + "learning_rate": 0.00016119035924142466, + "loss": 0.0, + "step": 470 + }, + { + "epoch": 0.59, + "learning_rate": 0.000161033371111321, + "loss": 0.0, + "step": 471 + }, + { + "epoch": 0.59, + "learning_rate": 0.00016087614290087208, + "loss": 0.0, + "step": 472 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001607186752285495, + "loss": 0.0, + "step": 473 + }, + { + "epoch": 0.59, + "learning_rate": 0.00016056096871376667, + "loss": 0.0, + "step": 474 + }, + { + "epoch": 0.6, + "learning_rate": 0.00016040302397687665, + "loss": 0.0, + "step": 475 + }, + { + "epoch": 0.6, + "learning_rate": 0.00016024484163916952, + "loss": 0.0, + "step": 476 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001600864223228699, + "loss": 0.0, + "step": 477 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001599277666511347, + "loss": 0.0, + "step": 478 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001597688752480505, + "loss": 0.0, + "step": 479 + }, + { + "epoch": 0.6, + "learning_rate": 0.00015960974873863114, + "loss": 0.0, + "step": 480 + }, + { + "epoch": 0.6, + "learning_rate": 0.00015945038774881534, + "loss": 0.0, + "step": 481 + }, + { + "epoch": 0.6, + "learning_rate": 0.00015929079290546408, + "loss": 0.0, + "step": 482 + }, + { + "epoch": 0.61, + "learning_rate": 0.00015913096483635824, + "loss": 0.0, + "step": 483 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001589709041701962, + "loss": 0.0, + "step": 484 + }, + { + "epoch": 0.61, + "learning_rate": 0.00015881061153659126, + "loss": 0.0, + "step": 485 + }, + { + "epoch": 0.61, + "learning_rate": 0.00015865008756606904, + "loss": 0.0, + "step": 486 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001584893328900653, + "loss": 0.0, + "step": 487 + }, + { + "epoch": 0.61, + "learning_rate": 0.00015832834814092332, + "loss": 0.0, + "step": 488 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001581671339518912, + "loss": 0.0, + "step": 489 + }, + { + "epoch": 0.61, + "learning_rate": 0.00015800569095711982, + "loss": 0.0, + "step": 490 + }, + { + "epoch": 0.62, + "learning_rate": 0.00015784401979165988, + "loss": 0.0, + "step": 491 + }, + { + "epoch": 0.62, + "learning_rate": 0.00015768212109145975, + "loss": 0.0, + "step": 492 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001575199954933627, + "loss": 0.0, + "step": 493 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001573576436351046, + "loss": 0.0, + "step": 494 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001571950661553114, + "loss": 0.0, + "step": 495 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001570322636934964, + "loss": 0.0, + "step": 496 + }, + { + "epoch": 0.62, + "learning_rate": 0.00015686923689005804, + "loss": 0.0, + "step": 497 + }, + { + "epoch": 0.62, + "learning_rate": 0.00015670598638627706, + "loss": 0.0, + "step": 498 + }, + { + "epoch": 0.63, + "learning_rate": 0.00015654251282431437, + "loss": 0.0, + "step": 499 + }, + { + "epoch": 0.63, + "learning_rate": 0.00015637881684720805, + "loss": 0.0, + "step": 500 + }, + { + "epoch": 0.63, + "learning_rate": 0.00015621489909887126, + "loss": 0.0, + "step": 501 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001560507602240894, + "loss": 0.0, + "step": 502 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001558864008685178, + "loss": 0.0, + "step": 503 + }, + { + "epoch": 0.63, + "learning_rate": 0.000155721821678679, + "loss": 0.0, + "step": 504 + }, + { + "epoch": 0.63, + "learning_rate": 0.00015555702330196023, + "loss": 0.0, + "step": 505 + }, + { + "epoch": 0.63, + "learning_rate": 0.00015539200638661104, + "loss": 0.0, + "step": 506 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015522677158174054, + "loss": 0.0, + "step": 507 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001550613195373149, + "loss": 0.0, + "step": 508 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001548956509041549, + "loss": 0.0, + "step": 509 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015472976633393326, + "loss": 0.0, + "step": 510 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015456366647917207, + "loss": 0.0, + "step": 511 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015439735199324032, + "loss": 0.0, + "step": 512 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015423082353035125, + "loss": 0.0, + "step": 513 + }, + { + "epoch": 0.64, + "learning_rate": 0.00015406408174555976, + "loss": 0.0, + "step": 514 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015389712729475995, + "loss": 0.0, + "step": 515 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001537299608346824, + "loss": 0.0, + "step": 516 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015356258302289165, + "loss": 0.0, + "step": 517 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015339499451778364, + "loss": 0.0, + "step": 518 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015322719597858303, + "loss": 0.0, + "step": 519 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015305918806534082, + "loss": 0.0, + "step": 520 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015289097143893137, + "loss": 0.0, + "step": 521 + }, + { + "epoch": 0.65, + "learning_rate": 0.00015272254676105025, + "loss": 0.0, + "step": 522 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015255391469421128, + "loss": 0.0, + "step": 523 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015238507590174417, + "loss": 0.0, + "step": 524 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015221603104779166, + "loss": 0.0, + "step": 525 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015204678079730724, + "loss": 0.0, + "step": 526 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015187732581605217, + "loss": 0.0, + "step": 527 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001517076667705931, + "loss": 0.0, + "step": 528 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015153780432829945, + "loss": 0.0, + "step": 529 + }, + { + "epoch": 0.66, + "learning_rate": 0.00015136773915734066, + "loss": 0.0, + "step": 530 + }, + { + "epoch": 0.67, + "learning_rate": 0.00015119747192668357, + "loss": 0.0, + "step": 531 + }, + { + "epoch": 0.67, + "learning_rate": 0.00015102700330609, + "loss": 0.0, + "step": 532 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001508563339661137, + "loss": 0.0, + "step": 533 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001506854645780983, + "loss": 0.0, + "step": 534 + }, + { + "epoch": 0.67, + "learning_rate": 0.00015051439581417406, + "loss": 0.0, + "step": 535 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001503431283472556, + "loss": 0.0, + "step": 536 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001501716628510392, + "loss": 0.0, + "step": 537 + }, + { + "epoch": 0.68, + "learning_rate": 0.00015000000000000001, + "loss": 0.0, + "step": 538 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001498281404693896, + "loss": 0.0, + "step": 539 + }, + { + "epoch": 0.68, + "learning_rate": 0.00014965608493523313, + "loss": 0.0, + "step": 540 + }, + { + "epoch": 0.68, + "learning_rate": 0.00014948383407432678, + "loss": 0.0, + "step": 541 + }, + { + "epoch": 0.68, + "learning_rate": 0.00014931138856423502, + "loss": 0.0, + "step": 542 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001491387490832881, + "loss": 0.0, + "step": 543 + }, + { + "epoch": 0.68, + "learning_rate": 0.00014896591631057912, + "loss": 0.0, + "step": 544 + }, + { + "epoch": 0.68, + "learning_rate": 0.00014879289092596166, + "loss": 0.0, + "step": 545 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014861967361004687, + "loss": 0.0, + "step": 546 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014844626504420088, + "loss": 0.0, + "step": 547 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001482726659105422, + "loss": 0.0, + "step": 548 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014809887689193877, + "loss": 0.0, + "step": 549 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001479248986720057, + "loss": 0.0, + "step": 550 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014775073193510217, + "loss": 0.0, + "step": 551 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014757637736632898, + "loss": 0.0, + "step": 552 + }, + { + "epoch": 0.69, + "learning_rate": 0.00014740183565152575, + "loss": 0.0, + "step": 553 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001472271074772683, + "loss": 0.0, + "step": 554 + }, + { + "epoch": 0.7, + "learning_rate": 0.00014705219353086584, + "loss": 0.0, + "step": 555 + }, + { + "epoch": 0.7, + "learning_rate": 0.00014687709450035837, + "loss": 0.0, + "step": 556 + }, + { + "epoch": 0.7, + "learning_rate": 0.000146701811074514, + "loss": 0.0, + "step": 557 + }, + { + "epoch": 0.7, + "learning_rate": 0.00014652634394282608, + "loss": 0.0, + "step": 558 + }, + { + "epoch": 0.7, + "learning_rate": 0.00014635069379551055, + "loss": 0.0, + "step": 559 + }, + { + "epoch": 0.7, + "learning_rate": 0.00014617486132350343, + "loss": 0.0, + "step": 560 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001459988472184577, + "loss": 0.0, + "step": 561 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014582265217274104, + "loss": 0.0, + "step": 562 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014564627687943275, + "loss": 0.0, + "step": 563 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014546972203232112, + "loss": 0.0, + "step": 564 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014529298832590085, + "loss": 0.0, + "step": 565 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001451160764553701, + "loss": 0.0, + "step": 566 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014493898711662783, + "loss": 0.0, + "step": 567 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014476172100627127, + "loss": 0.0, + "step": 568 + }, + { + "epoch": 0.71, + "learning_rate": 0.00014458427882159278, + "loss": 0.0, + "step": 569 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014440666126057744, + "loss": 0.0, + "step": 570 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014422886902190014, + "loss": 0.0, + "step": 571 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014405090280492295, + "loss": 0.0, + "step": 572 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014387276330969222, + "loss": 0.0, + "step": 573 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014369445123693596, + "loss": 0.0, + "step": 574 + }, + { + "epoch": 0.72, + "learning_rate": 0.000143515967288061, + "loss": 0.0, + "step": 575 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014333731216515028, + "loss": 0.0, + "step": 576 + }, + { + "epoch": 0.72, + "learning_rate": 0.00014315848657096004, + "loss": 0.0, + "step": 577 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014297949120891718, + "loss": 0.0, + "step": 578 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014280032678311622, + "loss": 0.0, + "step": 579 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014262099399831683, + "loss": 0.0, + "step": 580 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014244149355994095, + "loss": 0.0, + "step": 581 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014226182617406996, + "loss": 0.0, + "step": 582 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014208199254744192, + "loss": 0.0, + "step": 583 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014190199338744887, + "loss": 0.0, + "step": 584 + }, + { + "epoch": 0.73, + "learning_rate": 0.00014172182940213395, + "loss": 0.0, + "step": 585 + }, + { + "epoch": 0.74, + "learning_rate": 0.00014154150130018866, + "loss": 0.0, + "step": 586 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001413610097909501, + "loss": 0.0, + "step": 587 + }, + { + "epoch": 0.74, + "learning_rate": 0.00014118035558439808, + "loss": 0.0, + "step": 588 + }, + { + "epoch": 0.74, + "learning_rate": 0.00014099953939115248, + "loss": 0.0, + "step": 589 + }, + { + "epoch": 0.74, + "learning_rate": 0.00014081856192247033, + "loss": 0.0, + "step": 590 + }, + { + "epoch": 0.74, + "learning_rate": 0.000140637423890243, + "loss": 0.0, + "step": 591 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001404561260069935, + "loss": 0.0, + "step": 592 + }, + { + "epoch": 0.74, + "learning_rate": 0.00014027466898587374, + "loss": 0.0, + "step": 593 + }, + { + "epoch": 0.75, + "learning_rate": 0.00014009305354066137, + "loss": 0.0, + "step": 594 + }, + { + "epoch": 0.75, + "learning_rate": 0.00013991128038575741, + "loss": 0.0, + "step": 595 + }, + { + "epoch": 0.75, + "learning_rate": 0.00013972935023618318, + "loss": 0.0, + "step": 596 + }, + { + "epoch": 0.75, + "learning_rate": 0.00013954726380757757, + "loss": 0.0, + "step": 597 + }, + { + "epoch": 0.75, + "learning_rate": 0.00013936502181619416, + "loss": 0.0, + "step": 598 + }, + { + "epoch": 0.75, + "learning_rate": 0.00013918262497889848, + "loss": 0.0, + "step": 599 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001390000740131652, + "loss": 0.0, + "step": 600 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001388173696370752, + "loss": 0.0, + "step": 601 + }, + { + "epoch": 0.76, + "learning_rate": 0.00013863451256931287, + "loss": 0.0, + "step": 602 + }, + { + "epoch": 0.76, + "learning_rate": 0.00013845150352916324, + "loss": 0.0, + "step": 603 + }, + { + "epoch": 0.76, + "learning_rate": 0.000138268343236509, + "loss": 0.0, + "step": 604 + }, + { + "epoch": 0.76, + "learning_rate": 0.00013808503241182795, + "loss": 0.0, + "step": 605 + }, + { + "epoch": 0.76, + "learning_rate": 0.00013790157177619004, + "loss": 0.0, + "step": 606 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001377179620512544, + "loss": 0.0, + "step": 607 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001375342039592667, + "loss": 0.0, + "step": 608 + }, + { + "epoch": 0.76, + "learning_rate": 0.00013735029822305614, + "loss": 0.0, + "step": 609 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013716624556603274, + "loss": 0.0, + "step": 610 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013698204671218456, + "loss": 0.0, + "step": 611 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001367977023860745, + "loss": 0.0, + "step": 612 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013661321331283796, + "loss": 0.0, + "step": 613 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013642858021817943, + "loss": 0.0, + "step": 614 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013624380382837016, + "loss": 0.0, + "step": 615 + }, + { + "epoch": 0.77, + "learning_rate": 0.000136058884870245, + "loss": 0.0, + "step": 616 + }, + { + "epoch": 0.77, + "learning_rate": 0.00013587382407119944, + "loss": 0.0, + "step": 617 + }, + { + "epoch": 0.78, + "learning_rate": 0.00013568862215918717, + "loss": 0.0, + "step": 618 + }, + { + "epoch": 0.78, + "learning_rate": 0.00013550327986271686, + "loss": 0.0, + "step": 619 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001353177979108493, + "loss": 0.0, + "step": 620 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001351321770331947, + "loss": 0.0, + "step": 621 + }, + { + "epoch": 0.78, + "learning_rate": 0.00013494641795990986, + "loss": 0.0, + "step": 622 + }, + { + "epoch": 0.78, + "learning_rate": 0.00013476052142169498, + "loss": 0.0, + "step": 623 + }, + { + "epoch": 0.78, + "learning_rate": 0.00013457448814979109, + "loss": 0.0, + "step": 624 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001343883188759771, + "loss": 0.0, + "step": 625 + }, + { + "epoch": 0.79, + "learning_rate": 0.00013420201433256689, + "loss": 0.0, + "step": 626 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001340155752524064, + "loss": 0.0, + "step": 627 + }, + { + "epoch": 0.79, + "learning_rate": 0.00013382900236887075, + "loss": 0.0, + "step": 628 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001336422964158615, + "loss": 0.0, + "step": 629 + }, + { + "epoch": 0.79, + "learning_rate": 0.00013345545812780353, + "loss": 0.0, + "step": 630 + }, + { + "epoch": 0.79, + "learning_rate": 0.00013326848823964243, + "loss": 0.0, + "step": 631 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001330813874868412, + "loss": 0.0, + "step": 632 + }, + { + "epoch": 0.79, + "learning_rate": 0.00013289415660537788, + "loss": 0.0, + "step": 633 + }, + { + "epoch": 0.8, + "learning_rate": 0.00013270679633174218, + "loss": 0.0, + "step": 634 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001325193074029329, + "loss": 0.0, + "step": 635 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001323316905564549, + "loss": 0.0, + "step": 636 + }, + { + "epoch": 0.8, + "learning_rate": 0.00013214394653031616, + "loss": 0.0, + "step": 637 + }, + { + "epoch": 0.8, + "learning_rate": 0.00013195607606302501, + "loss": 0.0, + "step": 638 + }, + { + "epoch": 0.8, + "learning_rate": 0.00013176807989358707, + "loss": 0.0, + "step": 639 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001315799587615025, + "loss": 0.0, + "step": 640 + }, + { + "epoch": 0.8, + "learning_rate": 0.00013139171340676296, + "loss": 0.0, + "step": 641 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001312033445698487, + "loss": 0.0, + "step": 642 + }, + { + "epoch": 0.81, + "learning_rate": 0.00013101485299172588, + "loss": 0.0, + "step": 643 + }, + { + "epoch": 0.81, + "learning_rate": 0.00013082623941384325, + "loss": 0.0, + "step": 644 + }, + { + "epoch": 0.81, + "learning_rate": 0.00013063750457812962, + "loss": 0.0, + "step": 645 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001304486492269907, + "loss": 0.0, + "step": 646 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001302596741033063, + "loss": 0.0, + "step": 647 + }, + { + "epoch": 0.81, + "learning_rate": 0.00013007057995042732, + "loss": 0.0, + "step": 648 + }, + { + "epoch": 0.81, + "learning_rate": 0.00012988136751217291, + "loss": 0.0, + "step": 649 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001296920375328275, + "loss": 0.0, + "step": 650 + }, + { + "epoch": 0.82, + "learning_rate": 0.00012950259075713788, + "loss": 0.0, + "step": 651 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001293130279303102, + "loss": 0.0, + "step": 652 + }, + { + "epoch": 0.82, + "learning_rate": 0.00012912334979800727, + "loss": 0.0, + "step": 653 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001289335571063453, + "loss": 0.0, + "step": 654 + }, + { + "epoch": 0.82, + "learning_rate": 0.00012874365060189117, + "loss": 0.0, + "step": 655 + }, + { + "epoch": 0.82, + "learning_rate": 0.00012855363103165957, + "loss": 0.0, + "step": 656 + }, + { + "epoch": 0.82, + "learning_rate": 0.00012836349914310978, + "loss": 0.0, + "step": 657 + }, + { + "epoch": 0.83, + "learning_rate": 0.00012817325568414297, + "loss": 0.0, + "step": 658 + }, + { + "epoch": 0.83, + "learning_rate": 0.00012798290140309923, + "loss": 0.0, + "step": 659 + }, + { + "epoch": 0.83, + "learning_rate": 0.00012779243704875444, + "loss": 0.0, + "step": 660 + }, + { + "epoch": 0.83, + "learning_rate": 0.00012760186337031762, + "loss": 0.0, + "step": 661 + }, + { + "epoch": 0.83, + "learning_rate": 0.00012741118111742777, + "loss": 0.0, + "step": 662 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001272203910401509, + "loss": 0.0, + "step": 663 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001270294938889773, + "loss": 0.0, + "step": 664 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001268384904148183, + "loss": 0.0, + "step": 665 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012664738136900348, + "loss": 0.0, + "step": 666 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001264561675032779, + "loss": 0.0, + "step": 667 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012626484956979866, + "loss": 0.0, + "step": 668 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012607342832113242, + "loss": 0.0, + "step": 669 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012588190451025207, + "loss": 0.0, + "step": 670 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012569027889053412, + "loss": 0.0, + "step": 671 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012549855221575542, + "loss": 0.0, + "step": 672 + }, + { + "epoch": 0.84, + "learning_rate": 0.00012530672524009036, + "loss": 0.0, + "step": 673 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001251147987181079, + "loss": 0.0, + "step": 674 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001249227734047686, + "loss": 0.0, + "step": 675 + }, + { + "epoch": 0.85, + "learning_rate": 0.00012473065005542155, + "loss": 0.0, + "step": 676 + }, + { + "epoch": 0.85, + "learning_rate": 0.00012453842942580155, + "loss": 0.0, + "step": 677 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001243461122720259, + "loss": 0.0, + "step": 678 + }, + { + "epoch": 0.85, + "learning_rate": 0.00012415369935059184, + "loss": 0.0, + "step": 679 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001239611914183731, + "loss": 0.0, + "step": 680 + }, + { + "epoch": 0.85, + "learning_rate": 0.00012376858923261733, + "loss": 0.0, + "step": 681 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012357589355094275, + "loss": 0.0, + "step": 682 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012338310513133543, + "loss": 0.0, + "step": 683 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012319022473214632, + "loss": 0.0, + "step": 684 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012299725311208808, + "loss": 0.0, + "step": 685 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012280419103023217, + "loss": 0.0, + "step": 686 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012261103924600598, + "loss": 0.0, + "step": 687 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001224177985191897, + "loss": 0.0, + "step": 688 + }, + { + "epoch": 0.86, + "learning_rate": 0.00012222446960991342, + "loss": 0.0, + "step": 689 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012203105327865407, + "loss": 0.0, + "step": 690 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012183755028623245, + "loss": 0.0, + "step": 691 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012164396139381029, + "loss": 0.0, + "step": 692 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012145028736288721, + "loss": 0.0, + "step": 693 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012125652895529766, + "loss": 0.0, + "step": 694 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012106268693320812, + "loss": 0.0, + "step": 695 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012086876205911389, + "loss": 0.0, + "step": 696 + }, + { + "epoch": 0.87, + "learning_rate": 0.00012067475509583616, + "loss": 0.0, + "step": 697 + }, + { + "epoch": 0.88, + "learning_rate": 0.00012048066680651908, + "loss": 0.0, + "step": 698 + }, + { + "epoch": 0.88, + "learning_rate": 0.00012028649795462668, + "loss": 0.0, + "step": 699 + }, + { + "epoch": 0.88, + "learning_rate": 0.00012009224930393988, + "loss": 0.0, + "step": 700 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001198979216185535, + "loss": 0.0, + "step": 701 + }, + { + "epoch": 0.88, + "learning_rate": 0.00011970351566287333, + "loss": 0.0, + "step": 702 + }, + { + "epoch": 0.88, + "learning_rate": 0.00011950903220161285, + "loss": 0.0, + "step": 703 + }, + { + "epoch": 0.88, + "learning_rate": 0.00011931447199979057, + "loss": 0.0, + "step": 704 + }, + { + "epoch": 0.88, + "learning_rate": 0.00011911983582272694, + "loss": 0.0, + "step": 705 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011892512443604102, + "loss": 0.0, + "step": 706 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011873033860564793, + "loss": 0.0, + "step": 707 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011853547909775553, + "loss": 0.0, + "step": 708 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011834054667886145, + "loss": 0.0, + "step": 709 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011814554211575027, + "loss": 0.0, + "step": 710 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011795046617549021, + "loss": 0.0, + "step": 711 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011775531962543036, + "loss": 0.0, + "step": 712 + }, + { + "epoch": 0.89, + "learning_rate": 0.00011756010323319744, + "loss": 0.0, + "step": 713 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011736481776669306, + "loss": 0.0, + "step": 714 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011716946399409039, + "loss": 0.0, + "step": 715 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011697404268383141, + "loss": 0.0, + "step": 716 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011677855460462366, + "loss": 0.0, + "step": 717 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011658300052543742, + "loss": 0.0, + "step": 718 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001163873812155025, + "loss": 0.0, + "step": 719 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011619169744430538, + "loss": 0.0, + "step": 720 + }, + { + "epoch": 0.9, + "learning_rate": 0.00011599594998158602, + "loss": 0.0, + "step": 721 + }, + { + "epoch": 0.91, + "learning_rate": 0.000115800139597335, + "loss": 0.0, + "step": 722 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011560426706179033, + "loss": 0.0, + "step": 723 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011540833314543458, + "loss": 0.0, + "step": 724 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011521233861899167, + "loss": 0.0, + "step": 725 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011501628425342403, + "loss": 0.0, + "step": 726 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011482017081992944, + "loss": 0.0, + "step": 727 + }, + { + "epoch": 0.91, + "learning_rate": 0.00011462399908993797, + "loss": 0.0, + "step": 728 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001144277698351091, + "loss": 0.0, + "step": 729 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011423148382732853, + "loss": 0.0, + "step": 730 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011403514183870521, + "loss": 0.0, + "step": 731 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001138387446415683, + "loss": 0.0, + "step": 732 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011364229300846417, + "loss": 0.0, + "step": 733 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011344578771215319, + "loss": 0.0, + "step": 734 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011324922952560702, + "loss": 0.0, + "step": 735 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011305261922200519, + "loss": 0.0, + "step": 736 + }, + { + "epoch": 0.92, + "learning_rate": 0.00011285595757473228, + "loss": 0.0, + "step": 737 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011265924535737493, + "loss": 0.0, + "step": 738 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001124624833437186, + "loss": 0.0, + "step": 739 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011226567230774467, + "loss": 0.0, + "step": 740 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001120688130236273, + "loss": 0.0, + "step": 741 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011187190626573052, + "loss": 0.0, + "step": 742 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011167495280860509, + "loss": 0.0, + "step": 743 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011147795342698537, + "loss": 0.0, + "step": 744 + }, + { + "epoch": 0.93, + "learning_rate": 0.00011128090889578648, + "loss": 0.0, + "step": 745 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011108381999010111, + "loss": 0.0, + "step": 746 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011088668748519647, + "loss": 0.0, + "step": 747 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011068951215651132, + "loss": 0.0, + "step": 748 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011049229477965275, + "loss": 0.0, + "step": 749 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011029503613039346, + "loss": 0.0, + "step": 750 + }, + { + "epoch": 0.94, + "learning_rate": 0.00011009773698466832, + "loss": 0.0, + "step": 751 + }, + { + "epoch": 0.94, + "learning_rate": 0.00010990039811857156, + "loss": 0.0, + "step": 752 + }, + { + "epoch": 0.94, + "learning_rate": 0.00010970302030835366, + "loss": 0.0, + "step": 753 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010950560433041826, + "loss": 0.0, + "step": 754 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010930815096131921, + "loss": 0.0, + "step": 755 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010911066097775735, + "loss": 0.0, + "step": 756 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001089131351565776, + "loss": 0.0, + "step": 757 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010871557427476583, + "loss": 0.0, + "step": 758 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010851797910944581, + "loss": 0.0, + "step": 759 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010832035043787625, + "loss": 0.0, + "step": 760 + }, + { + "epoch": 0.95, + "learning_rate": 0.00010812268903744753, + "loss": 0.0, + "step": 761 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010792499568567884, + "loss": 0.0, + "step": 762 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010772727116021513, + "loss": 0.0, + "step": 763 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010752951623882382, + "loss": 0.0, + "step": 764 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010733173169939199, + "loss": 0.0, + "step": 765 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010713391831992323, + "loss": 0.0, + "step": 766 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010693607687853455, + "loss": 0.0, + "step": 767 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010673820815345334, + "loss": 0.0, + "step": 768 + }, + { + "epoch": 0.96, + "learning_rate": 0.00010654031292301432, + "loss": 0.0, + "step": 769 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010634239196565646, + "loss": 0.0, + "step": 770 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010614444605991995, + "loss": 0.0, + "step": 771 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010594647598444312, + "loss": 0.0, + "step": 772 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010574848251795935, + "loss": 0.0, + "step": 773 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010555046643929403, + "loss": 0.0, + "step": 774 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010535242852736151, + "loss": 0.0, + "step": 775 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001051543695611621, + "loss": 0.0, + "step": 776 + }, + { + "epoch": 0.97, + "learning_rate": 0.00010495629031977872, + "loss": 0.0, + "step": 777 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010475819158237425, + "loss": 0.0, + "step": 778 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010456007412818818, + "loss": 0.0, + "step": 779 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010436193873653361, + "loss": 0.0, + "step": 780 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010416378618679425, + "loss": 0.0, + "step": 781 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010396561725842124, + "loss": 0.0, + "step": 782 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010376743273093017, + "loss": 0.0, + "step": 783 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010356923338389806, + "loss": 0.0, + "step": 784 + }, + { + "epoch": 0.98, + "learning_rate": 0.00010337101999696011, + "loss": 0.0, + "step": 785 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010317279334980678, + "loss": 0.0, + "step": 786 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010297455422218074, + "loss": 0.0, + "step": 787 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010277630339387374, + "loss": 0.0, + "step": 788 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010257804164472351, + "loss": 0.0, + "step": 789 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010237976975461075, + "loss": 0.0, + "step": 790 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010218148850345613, + "loss": 0.0, + "step": 791 + }, + { + "epoch": 0.99, + "learning_rate": 0.00010198319867121701, + "loss": 0.0, + "step": 792 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001017849010378846, + "loss": 0.0, + "step": 793 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010158659638348081, + "loss": 0.0, + "step": 794 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010138828548805506, + "loss": 0.0, + "step": 795 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010118996913168144, + "loss": 0.0, + "step": 796 + }, + { + "epoch": 1.0, + "learning_rate": 0.0001009916480944554, + "loss": 0.0, + "step": 797 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010079332315649097, + "loss": 0.0, + "step": 798 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010059499509791736, + "loss": 0.0, + "step": 799 + }, + { + "epoch": 1.0, + "learning_rate": 0.00010039666469887612, + "loss": 0.0, + "step": 800 + }, + { + "epoch": 1.01, + "learning_rate": 0.00010019833273951806, + "loss": 0.0, + "step": 801 + }, + { + "epoch": 1.01, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 802 + }, + { + "epoch": 1.01, + "learning_rate": 9.980166726048195e-05, + "loss": 0.0, + "step": 803 + }, + { + "epoch": 1.01, + "learning_rate": 9.960333530112386e-05, + "loss": 0.0, + "step": 804 + }, + { + "epoch": 1.01, + "learning_rate": 9.940500490208263e-05, + "loss": 0.0, + "step": 805 + }, + { + "epoch": 1.01, + "learning_rate": 9.920667684350905e-05, + "loss": 0.0, + "step": 806 + }, + { + "epoch": 1.01, + "learning_rate": 9.90083519055446e-05, + "loss": 0.0, + "step": 807 + }, + { + "epoch": 1.01, + "learning_rate": 9.881003086831859e-05, + "loss": 0.0, + "step": 808 + }, + { + "epoch": 1.02, + "learning_rate": 9.861171451194497e-05, + "loss": 0.0, + "step": 809 + }, + { + "epoch": 1.02, + "learning_rate": 9.84134036165192e-05, + "loss": 0.0, + "step": 810 + }, + { + "epoch": 1.02, + "eval_loss": NaN, + "eval_runtime": 1.6017, + "eval_samples_per_second": 26.847, + "eval_steps_per_second": 26.847, + "step": 810 + } + ], + "logging_steps": 1, + "max_steps": 1594, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 500, + "total_flos": 5791945174548480.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}