Training in progress, epoch 1, checkpoint

last-checkpoint/added_tokens.json

@@ -0,0 +1,1034 @@
+{
+  "</assistant>": 50268,
+  "</final>": 50272,
+  "</think>": 50270,
+  "</user>": 50266,
+  "<assistant>": 50267,
+  "<final>": 50271,
+  "<think>": 50269,
+  "<user>": 50265,
+  "[class_0_0]": 50273,
+  "[class_0_1000]": 51273,
+  "[class_0_1001]": 51274,
+  "[class_0_1002]": 51275,
+  "[class_0_1003]": 51276,
+  "[class_0_1004]": 51277,
+  "[class_0_1005]": 51278,
+  "[class_0_1006]": 51279,
+  "[class_0_1007]": 51280,
+  "[class_0_1008]": 51281,
+  "[class_0_1009]": 51282,
+  "[class_0_100]": 50373,
+  "[class_0_1010]": 51283,
+  "[class_0_1011]": 51284,
+  "[class_0_1012]": 51285,
+  "[class_0_1013]": 51286,
+  "[class_0_1014]": 51287,
+  "[class_0_1015]": 51288,
+  "[class_0_1016]": 51289,
+  "[class_0_1017]": 51290,
+  "[class_0_1018]": 51291,
+  "[class_0_1019]": 51292,
+  "[class_0_101]": 50374,
+  "[class_0_1020]": 51293,
+  "[class_0_1021]": 51294,
+  "[class_0_1022]": 51295,
+  "[class_0_1023]": 51296,
+  "[class_0_102]": 50375,
+  "[class_0_103]": 50376,
+  "[class_0_104]": 50377,
+  "[class_0_105]": 50378,
+  "[class_0_106]": 50379,
+  "[class_0_107]": 50380,
+  "[class_0_108]": 50381,
+  "[class_0_109]": 50382,
+  "[class_0_10]": 50283,
+  "[class_0_110]": 50383,
+  "[class_0_111]": 50384,
+  "[class_0_112]": 50385,
+  "[class_0_113]": 50386,
+  "[class_0_114]": 50387,
+  "[class_0_115]": 50388,
+  "[class_0_116]": 50389,
+  "[class_0_117]": 50390,
+  "[class_0_118]": 50391,
+  "[class_0_119]": 50392,
+  "[class_0_11]": 50284,
+  "[class_0_120]": 50393,
+  "[class_0_121]": 50394,
+  "[class_0_122]": 50395,
+  "[class_0_123]": 50396,
+  "[class_0_124]": 50397,
+  "[class_0_125]": 50398,
+  "[class_0_126]": 50399,
+  "[class_0_127]": 50400,
+  "[class_0_128]": 50401,
+  "[class_0_129]": 50402,
+  "[class_0_12]": 50285,
+  "[class_0_130]": 50403,
+  "[class_0_131]": 50404,
+  "[class_0_132]": 50405,
+  "[class_0_133]": 50406,
+  "[class_0_134]": 50407,
+  "[class_0_135]": 50408,
+  "[class_0_136]": 50409,
+  "[class_0_137]": 50410,
+  "[class_0_138]": 50411,
+  "[class_0_139]": 50412,
+  "[class_0_13]": 50286,
+  "[class_0_140]": 50413,
+  "[class_0_141]": 50414,
+  "[class_0_142]": 50415,
+  "[class_0_143]": 50416,
+  "[class_0_144]": 50417,
+  "[class_0_145]": 50418,
+  "[class_0_146]": 50419,
+  "[class_0_147]": 50420,
+  "[class_0_148]": 50421,
+  "[class_0_149]": 50422,
+  "[class_0_14]": 50287,
+  "[class_0_150]": 50423,
+  "[class_0_151]": 50424,
+  "[class_0_152]": 50425,
+  "[class_0_153]": 50426,
+  "[class_0_154]": 50427,
+  "[class_0_155]": 50428,
+  "[class_0_156]": 50429,
+  "[class_0_157]": 50430,
+  "[class_0_158]": 50431,
+  "[class_0_159]": 50432,
+  "[class_0_15]": 50288,
+  "[class_0_160]": 50433,
+  "[class_0_161]": 50434,
+  "[class_0_162]": 50435,
+  "[class_0_163]": 50436,
+  "[class_0_164]": 50437,
+  "[class_0_165]": 50438,
+  "[class_0_166]": 50439,
+  "[class_0_167]": 50440,
+  "[class_0_168]": 50441,
+  "[class_0_169]": 50442,
+  "[class_0_16]": 50289,
+  "[class_0_170]": 50443,
+  "[class_0_171]": 50444,
+  "[class_0_172]": 50445,
+  "[class_0_173]": 50446,
+  "[class_0_174]": 50447,
+  "[class_0_175]": 50448,
+  "[class_0_176]": 50449,
+  "[class_0_177]": 50450,
+  "[class_0_178]": 50451,
+  "[class_0_179]": 50452,
+  "[class_0_17]": 50290,
+  "[class_0_180]": 50453,
+  "[class_0_181]": 50454,
+  "[class_0_182]": 50455,
+  "[class_0_183]": 50456,
+  "[class_0_184]": 50457,
+  "[class_0_185]": 50458,
+  "[class_0_186]": 50459,
+  "[class_0_187]": 50460,
+  "[class_0_188]": 50461,
+  "[class_0_189]": 50462,
+  "[class_0_18]": 50291,
+  "[class_0_190]": 50463,
+  "[class_0_191]": 50464,
+  "[class_0_192]": 50465,
+  "[class_0_193]": 50466,
+  "[class_0_194]": 50467,
+  "[class_0_195]": 50468,
+  "[class_0_196]": 50469,
+  "[class_0_197]": 50470,
+  "[class_0_198]": 50471,
+  "[class_0_199]": 50472,
+  "[class_0_19]": 50292,
+  "[class_0_1]": 50274,
+  "[class_0_200]": 50473,
+  "[class_0_201]": 50474,
+  "[class_0_202]": 50475,
+  "[class_0_203]": 50476,
+  "[class_0_204]": 50477,
+  "[class_0_205]": 50478,
+  "[class_0_206]": 50479,
+  "[class_0_207]": 50480,
+  "[class_0_208]": 50481,
+  "[class_0_209]": 50482,
+  "[class_0_20]": 50293,
+  "[class_0_210]": 50483,
+  "[class_0_211]": 50484,
+  "[class_0_212]": 50485,
+  "[class_0_213]": 50486,
+  "[class_0_214]": 50487,
+  "[class_0_215]": 50488,
+  "[class_0_216]": 50489,
+  "[class_0_217]": 50490,
+  "[class_0_218]": 50491,
+  "[class_0_219]": 50492,
+  "[class_0_21]": 50294,
+  "[class_0_220]": 50493,
+  "[class_0_221]": 50494,
+  "[class_0_222]": 50495,
+  "[class_0_223]": 50496,
+  "[class_0_224]": 50497,
+  "[class_0_225]": 50498,
+  "[class_0_226]": 50499,
+  "[class_0_227]": 50500,
+  "[class_0_228]": 50501,
+  "[class_0_229]": 50502,
+  "[class_0_22]": 50295,
+  "[class_0_230]": 50503,
+  "[class_0_231]": 50504,
+  "[class_0_232]": 50505,
+  "[class_0_233]": 50506,
+  "[class_0_234]": 50507,
+  "[class_0_235]": 50508,
+  "[class_0_236]": 50509,
+  "[class_0_237]": 50510,
+  "[class_0_238]": 50511,
+  "[class_0_239]": 50512,
+  "[class_0_23]": 50296,
+  "[class_0_240]": 50513,
+  "[class_0_241]": 50514,
+  "[class_0_242]": 50515,
+  "[class_0_243]": 50516,
+  "[class_0_244]": 50517,
+  "[class_0_245]": 50518,
+  "[class_0_246]": 50519,
+  "[class_0_247]": 50520,
+  "[class_0_248]": 50521,
+  "[class_0_249]": 50522,
+  "[class_0_24]": 50297,
+  "[class_0_250]": 50523,
+  "[class_0_251]": 50524,
+  "[class_0_252]": 50525,
+  "[class_0_253]": 50526,
+  "[class_0_254]": 50527,
+  "[class_0_255]": 50528,
+  "[class_0_256]": 50529,
+  "[class_0_257]": 50530,
+  "[class_0_258]": 50531,
+  "[class_0_259]": 50532,
+  "[class_0_25]": 50298,
+  "[class_0_260]": 50533,
+  "[class_0_261]": 50534,
+  "[class_0_262]": 50535,
+  "[class_0_263]": 50536,
+  "[class_0_264]": 50537,
+  "[class_0_265]": 50538,
+  "[class_0_266]": 50539,
+  "[class_0_267]": 50540,
+  "[class_0_268]": 50541,
+  "[class_0_269]": 50542,
+  "[class_0_26]": 50299,
+  "[class_0_270]": 50543,
+  "[class_0_271]": 50544,
+  "[class_0_272]": 50545,
+  "[class_0_273]": 50546,
+  "[class_0_274]": 50547,
+  "[class_0_275]": 50548,
+  "[class_0_276]": 50549,
+  "[class_0_277]": 50550,
+  "[class_0_278]": 50551,
+  "[class_0_279]": 50552,
+  "[class_0_27]": 50300,
+  "[class_0_280]": 50553,
+  "[class_0_281]": 50554,
+  "[class_0_282]": 50555,
+  "[class_0_283]": 50556,
+  "[class_0_284]": 50557,
+  "[class_0_285]": 50558,
+  "[class_0_286]": 50559,
+  "[class_0_287]": 50560,
+  "[class_0_288]": 50561,
+  "[class_0_289]": 50562,
+  "[class_0_28]": 50301,
+  "[class_0_290]": 50563,
+  "[class_0_291]": 50564,
+  "[class_0_292]": 50565,
+  "[class_0_293]": 50566,
+  "[class_0_294]": 50567,
+  "[class_0_295]": 50568,
+  "[class_0_296]": 50569,
+  "[class_0_297]": 50570,
+  "[class_0_298]": 50571,
+  "[class_0_299]": 50572,
+  "[class_0_29]": 50302,
+  "[class_0_2]": 50275,
+  "[class_0_300]": 50573,
+  "[class_0_301]": 50574,
+  "[class_0_302]": 50575,
+  "[class_0_303]": 50576,
+  "[class_0_304]": 50577,
+  "[class_0_305]": 50578,
+  "[class_0_306]": 50579,
+  "[class_0_307]": 50580,
+  "[class_0_308]": 50581,
+  "[class_0_309]": 50582,
+  "[class_0_30]": 50303,
+  "[class_0_310]": 50583,
+  "[class_0_311]": 50584,
+  "[class_0_312]": 50585,
+  "[class_0_313]": 50586,
+  "[class_0_314]": 50587,
+  "[class_0_315]": 50588,
+  "[class_0_316]": 50589,
+  "[class_0_317]": 50590,
+  "[class_0_318]": 50591,
+  "[class_0_319]": 50592,
+  "[class_0_31]": 50304,
+  "[class_0_320]": 50593,
+  "[class_0_321]": 50594,
+  "[class_0_322]": 50595,
+  "[class_0_323]": 50596,
+  "[class_0_324]": 50597,
+  "[class_0_325]": 50598,
+  "[class_0_326]": 50599,
+  "[class_0_327]": 50600,
+  "[class_0_328]": 50601,
+  "[class_0_329]": 50602,
+  "[class_0_32]": 50305,
+  "[class_0_330]": 50603,
+  "[class_0_331]": 50604,
+  "[class_0_332]": 50605,
+  "[class_0_333]": 50606,
+  "[class_0_334]": 50607,
+  "[class_0_335]": 50608,
+  "[class_0_336]": 50609,
+  "[class_0_337]": 50610,
+  "[class_0_338]": 50611,
+  "[class_0_339]": 50612,
+  "[class_0_33]": 50306,
+  "[class_0_340]": 50613,
+  "[class_0_341]": 50614,
+  "[class_0_342]": 50615,
+  "[class_0_343]": 50616,
+  "[class_0_344]": 50617,
+  "[class_0_345]": 50618,
+  "[class_0_346]": 50619,
+  "[class_0_347]": 50620,
+  "[class_0_348]": 50621,
+  "[class_0_349]": 50622,
+  "[class_0_34]": 50307,
+  "[class_0_350]": 50623,
+  "[class_0_351]": 50624,
+  "[class_0_352]": 50625,
+  "[class_0_353]": 50626,
+  "[class_0_354]": 50627,
+  "[class_0_355]": 50628,
+  "[class_0_356]": 50629,
+  "[class_0_357]": 50630,
+  "[class_0_358]": 50631,
+  "[class_0_359]": 50632,
+  "[class_0_35]": 50308,
+  "[class_0_360]": 50633,
+  "[class_0_361]": 50634,
+  "[class_0_362]": 50635,
+  "[class_0_363]": 50636,
+  "[class_0_364]": 50637,
+  "[class_0_365]": 50638,
+  "[class_0_366]": 50639,
+  "[class_0_367]": 50640,
+  "[class_0_368]": 50641,
+  "[class_0_369]": 50642,
+  "[class_0_36]": 50309,
+  "[class_0_370]": 50643,
+  "[class_0_371]": 50644,
+  "[class_0_372]": 50645,
+  "[class_0_373]": 50646,
+  "[class_0_374]": 50647,
+  "[class_0_375]": 50648,
+  "[class_0_376]": 50649,
+  "[class_0_377]": 50650,
+  "[class_0_378]": 50651,
+  "[class_0_379]": 50652,
+  "[class_0_37]": 50310,
+  "[class_0_380]": 50653,
+  "[class_0_381]": 50654,
+  "[class_0_382]": 50655,
+  "[class_0_383]": 50656,
+  "[class_0_384]": 50657,
+  "[class_0_385]": 50658,
+  "[class_0_386]": 50659,
+  "[class_0_387]": 50660,
+  "[class_0_388]": 50661,
+  "[class_0_389]": 50662,
+  "[class_0_38]": 50311,
+  "[class_0_390]": 50663,
+  "[class_0_391]": 50664,
+  "[class_0_392]": 50665,
+  "[class_0_393]": 50666,
+  "[class_0_394]": 50667,
+  "[class_0_395]": 50668,
+  "[class_0_396]": 50669,
+  "[class_0_397]": 50670,
+  "[class_0_398]": 50671,
+  "[class_0_399]": 50672,
+  "[class_0_39]": 50312,
+  "[class_0_3]": 50276,
+  "[class_0_400]": 50673,
+  "[class_0_401]": 50674,
+  "[class_0_402]": 50675,
+  "[class_0_403]": 50676,
+  "[class_0_404]": 50677,
+  "[class_0_405]": 50678,
+  "[class_0_406]": 50679,
+  "[class_0_407]": 50680,
+  "[class_0_408]": 50681,
+  "[class_0_409]": 50682,
+  "[class_0_40]": 50313,
+  "[class_0_410]": 50683,
+  "[class_0_411]": 50684,
+  "[class_0_412]": 50685,
+  "[class_0_413]": 50686,
+  "[class_0_414]": 50687,
+  "[class_0_415]": 50688,
+  "[class_0_416]": 50689,
+  "[class_0_417]": 50690,
+  "[class_0_418]": 50691,
+  "[class_0_419]": 50692,
+  "[class_0_41]": 50314,
+  "[class_0_420]": 50693,
+  "[class_0_421]": 50694,
+  "[class_0_422]": 50695,
+  "[class_0_423]": 50696,
+  "[class_0_424]": 50697,
+  "[class_0_425]": 50698,
+  "[class_0_426]": 50699,
+  "[class_0_427]": 50700,
+  "[class_0_428]": 50701,
+  "[class_0_429]": 50702,
+  "[class_0_42]": 50315,
+  "[class_0_430]": 50703,
+  "[class_0_431]": 50704,
+  "[class_0_432]": 50705,
+  "[class_0_433]": 50706,
+  "[class_0_434]": 50707,
+  "[class_0_435]": 50708,
+  "[class_0_436]": 50709,
+  "[class_0_437]": 50710,
+  "[class_0_438]": 50711,
+  "[class_0_439]": 50712,
+  "[class_0_43]": 50316,
+  "[class_0_440]": 50713,
+  "[class_0_441]": 50714,
+  "[class_0_442]": 50715,
+  "[class_0_443]": 50716,
+  "[class_0_444]": 50717,
+  "[class_0_445]": 50718,
+  "[class_0_446]": 50719,
+  "[class_0_447]": 50720,
+  "[class_0_448]": 50721,
+  "[class_0_449]": 50722,
+  "[class_0_44]": 50317,
+  "[class_0_450]": 50723,
+  "[class_0_451]": 50724,
+  "[class_0_452]": 50725,
+  "[class_0_453]": 50726,
+  "[class_0_454]": 50727,
+  "[class_0_455]": 50728,
+  "[class_0_456]": 50729,
+  "[class_0_457]": 50730,
+  "[class_0_458]": 50731,
+  "[class_0_459]": 50732,
+  "[class_0_45]": 50318,
+  "[class_0_460]": 50733,
+  "[class_0_461]": 50734,
+  "[class_0_462]": 50735,
+  "[class_0_463]": 50736,
+  "[class_0_464]": 50737,
+  "[class_0_465]": 50738,
+  "[class_0_466]": 50739,
+  "[class_0_467]": 50740,
+  "[class_0_468]": 50741,
+  "[class_0_469]": 50742,
+  "[class_0_46]": 50319,
+  "[class_0_470]": 50743,
+  "[class_0_471]": 50744,
+  "[class_0_472]": 50745,
+  "[class_0_473]": 50746,
+  "[class_0_474]": 50747,
+  "[class_0_475]": 50748,
+  "[class_0_476]": 50749,
+  "[class_0_477]": 50750,
+  "[class_0_478]": 50751,
+  "[class_0_479]": 50752,
+  "[class_0_47]": 50320,
+  "[class_0_480]": 50753,
+  "[class_0_481]": 50754,
+  "[class_0_482]": 50755,
+  "[class_0_483]": 50756,
+  "[class_0_484]": 50757,
+  "[class_0_485]": 50758,
+  "[class_0_486]": 50759,
+  "[class_0_487]": 50760,
+  "[class_0_488]": 50761,
+  "[class_0_489]": 50762,
+  "[class_0_48]": 50321,
+  "[class_0_490]": 50763,
+  "[class_0_491]": 50764,
+  "[class_0_492]": 50765,
+  "[class_0_493]": 50766,
+  "[class_0_494]": 50767,
+  "[class_0_495]": 50768,
+  "[class_0_496]": 50769,
+  "[class_0_497]": 50770,
+  "[class_0_498]": 50771,
+  "[class_0_499]": 50772,
+  "[class_0_49]": 50322,
+  "[class_0_4]": 50277,
+  "[class_0_500]": 50773,
+  "[class_0_501]": 50774,
+  "[class_0_502]": 50775,
+  "[class_0_503]": 50776,
+  "[class_0_504]": 50777,
+  "[class_0_505]": 50778,
+  "[class_0_506]": 50779,
+  "[class_0_507]": 50780,
+  "[class_0_508]": 50781,
+  "[class_0_509]": 50782,
+  "[class_0_50]": 50323,
+  "[class_0_510]": 50783,
+  "[class_0_511]": 50784,
+  "[class_0_512]": 50785,
+  "[class_0_513]": 50786,
+  "[class_0_514]": 50787,
+  "[class_0_515]": 50788,
+  "[class_0_516]": 50789,
+  "[class_0_517]": 50790,
+  "[class_0_518]": 50791,
+  "[class_0_519]": 50792,
+  "[class_0_51]": 50324,
+  "[class_0_520]": 50793,
+  "[class_0_521]": 50794,
+  "[class_0_522]": 50795,
+  "[class_0_523]": 50796,
+  "[class_0_524]": 50797,
+  "[class_0_525]": 50798,
+  "[class_0_526]": 50799,
+  "[class_0_527]": 50800,
+  "[class_0_528]": 50801,
+  "[class_0_529]": 50802,
+  "[class_0_52]": 50325,
+  "[class_0_530]": 50803,
+  "[class_0_531]": 50804,
+  "[class_0_532]": 50805,
+  "[class_0_533]": 50806,
+  "[class_0_534]": 50807,
+  "[class_0_535]": 50808,
+  "[class_0_536]": 50809,
+  "[class_0_537]": 50810,
+  "[class_0_538]": 50811,
+  "[class_0_539]": 50812,
+  "[class_0_53]": 50326,
+  "[class_0_540]": 50813,
+  "[class_0_541]": 50814,
+  "[class_0_542]": 50815,
+  "[class_0_543]": 50816,
+  "[class_0_544]": 50817,
+  "[class_0_545]": 50818,
+  "[class_0_546]": 50819,
+  "[class_0_547]": 50820,
+  "[class_0_548]": 50821,
+  "[class_0_549]": 50822,
+  "[class_0_54]": 50327,
+  "[class_0_550]": 50823,
+  "[class_0_551]": 50824,
+  "[class_0_552]": 50825,
+  "[class_0_553]": 50826,
+  "[class_0_554]": 50827,
+  "[class_0_555]": 50828,
+  "[class_0_556]": 50829,
+  "[class_0_557]": 50830,
+  "[class_0_558]": 50831,
+  "[class_0_559]": 50832,
+  "[class_0_55]": 50328,
+  "[class_0_560]": 50833,
+  "[class_0_561]": 50834,
+  "[class_0_562]": 50835,
+  "[class_0_563]": 50836,
+  "[class_0_564]": 50837,
+  "[class_0_565]": 50838,
+  "[class_0_566]": 50839,
+  "[class_0_567]": 50840,
+  "[class_0_568]": 50841,
+  "[class_0_569]": 50842,
+  "[class_0_56]": 50329,
+  "[class_0_570]": 50843,
+  "[class_0_571]": 50844,
+  "[class_0_572]": 50845,
+  "[class_0_573]": 50846,
+  "[class_0_574]": 50847,
+  "[class_0_575]": 50848,
+  "[class_0_576]": 50849,
+  "[class_0_577]": 50850,
+  "[class_0_578]": 50851,
+  "[class_0_579]": 50852,
+  "[class_0_57]": 50330,
+  "[class_0_580]": 50853,
+  "[class_0_581]": 50854,
+  "[class_0_582]": 50855,
+  "[class_0_583]": 50856,
+  "[class_0_584]": 50857,
+  "[class_0_585]": 50858,
+  "[class_0_586]": 50859,
+  "[class_0_587]": 50860,
+  "[class_0_588]": 50861,
+  "[class_0_589]": 50862,
+  "[class_0_58]": 50331,
+  "[class_0_590]": 50863,
+  "[class_0_591]": 50864,
+  "[class_0_592]": 50865,
+  "[class_0_593]": 50866,
+  "[class_0_594]": 50867,
+  "[class_0_595]": 50868,
+  "[class_0_596]": 50869,
+  "[class_0_597]": 50870,
+  "[class_0_598]": 50871,
+  "[class_0_599]": 50872,
+  "[class_0_59]": 50332,
+  "[class_0_5]": 50278,
+  "[class_0_600]": 50873,
+  "[class_0_601]": 50874,
+  "[class_0_602]": 50875,
+  "[class_0_603]": 50876,
+  "[class_0_604]": 50877,
+  "[class_0_605]": 50878,
+  "[class_0_606]": 50879,
+  "[class_0_607]": 50880,
+  "[class_0_608]": 50881,
+  "[class_0_609]": 50882,
+  "[class_0_60]": 50333,
+  "[class_0_610]": 50883,
+  "[class_0_611]": 50884,
+  "[class_0_612]": 50885,
+  "[class_0_613]": 50886,
+  "[class_0_614]": 50887,
+  "[class_0_615]": 50888,
+  "[class_0_616]": 50889,
+  "[class_0_617]": 50890,
+  "[class_0_618]": 50891,
+  "[class_0_619]": 50892,
+  "[class_0_61]": 50334,
+  "[class_0_620]": 50893,
+  "[class_0_621]": 50894,
+  "[class_0_622]": 50895,
+  "[class_0_623]": 50896,
+  "[class_0_624]": 50897,
+  "[class_0_625]": 50898,
+  "[class_0_626]": 50899,
+  "[class_0_627]": 50900,
+  "[class_0_628]": 50901,
+  "[class_0_629]": 50902,
+  "[class_0_62]": 50335,
+  "[class_0_630]": 50903,
+  "[class_0_631]": 50904,
+  "[class_0_632]": 50905,
+  "[class_0_633]": 50906,
+  "[class_0_634]": 50907,
+  "[class_0_635]": 50908,
+  "[class_0_636]": 50909,
+  "[class_0_637]": 50910,
+  "[class_0_638]": 50911,
+  "[class_0_639]": 50912,
+  "[class_0_63]": 50336,
+  "[class_0_640]": 50913,
+  "[class_0_641]": 50914,
+  "[class_0_642]": 50915,
+  "[class_0_643]": 50916,
+  "[class_0_644]": 50917,
+  "[class_0_645]": 50918,
+  "[class_0_646]": 50919,
+  "[class_0_647]": 50920,
+  "[class_0_648]": 50921,
+  "[class_0_649]": 50922,
+  "[class_0_64]": 50337,
+  "[class_0_650]": 50923,
+  "[class_0_651]": 50924,
+  "[class_0_652]": 50925,
+  "[class_0_653]": 50926,
+  "[class_0_654]": 50927,
+  "[class_0_655]": 50928,
+  "[class_0_656]": 50929,
+  "[class_0_657]": 50930,
+  "[class_0_658]": 50931,
+  "[class_0_659]": 50932,
+  "[class_0_65]": 50338,
+  "[class_0_660]": 50933,
+  "[class_0_661]": 50934,
+  "[class_0_662]": 50935,
+  "[class_0_663]": 50936,
+  "[class_0_664]": 50937,
+  "[class_0_665]": 50938,
+  "[class_0_666]": 50939,
+  "[class_0_667]": 50940,
+  "[class_0_668]": 50941,
+  "[class_0_669]": 50942,
+  "[class_0_66]": 50339,
+  "[class_0_670]": 50943,
+  "[class_0_671]": 50944,
+  "[class_0_672]": 50945,
+  "[class_0_673]": 50946,
+  "[class_0_674]": 50947,
+  "[class_0_675]": 50948,
+  "[class_0_676]": 50949,
+  "[class_0_677]": 50950,
+  "[class_0_678]": 50951,
+  "[class_0_679]": 50952,
+  "[class_0_67]": 50340,
+  "[class_0_680]": 50953,
+  "[class_0_681]": 50954,
+  "[class_0_682]": 50955,
+  "[class_0_683]": 50956,
+  "[class_0_684]": 50957,
+  "[class_0_685]": 50958,
+  "[class_0_686]": 50959,
+  "[class_0_687]": 50960,
+  "[class_0_688]": 50961,
+  "[class_0_689]": 50962,
+  "[class_0_68]": 50341,
+  "[class_0_690]": 50963,
+  "[class_0_691]": 50964,
+  "[class_0_692]": 50965,
+  "[class_0_693]": 50966,
+  "[class_0_694]": 50967,
+  "[class_0_695]": 50968,
+  "[class_0_696]": 50969,
+  "[class_0_697]": 50970,
+  "[class_0_698]": 50971,
+  "[class_0_699]": 50972,
+  "[class_0_69]": 50342,
+  "[class_0_6]": 50279,
+  "[class_0_700]": 50973,
+  "[class_0_701]": 50974,
+  "[class_0_702]": 50975,
+  "[class_0_703]": 50976,
+  "[class_0_704]": 50977,
+  "[class_0_705]": 50978,
+  "[class_0_706]": 50979,
+  "[class_0_707]": 50980,
+  "[class_0_708]": 50981,
+  "[class_0_709]": 50982,
+  "[class_0_70]": 50343,
+  "[class_0_710]": 50983,
+  "[class_0_711]": 50984,
+  "[class_0_712]": 50985,
+  "[class_0_713]": 50986,
+  "[class_0_714]": 50987,
+  "[class_0_715]": 50988,
+  "[class_0_716]": 50989,
+  "[class_0_717]": 50990,
+  "[class_0_718]": 50991,
+  "[class_0_719]": 50992,
+  "[class_0_71]": 50344,
+  "[class_0_720]": 50993,
+  "[class_0_721]": 50994,
+  "[class_0_722]": 50995,
+  "[class_0_723]": 50996,
+  "[class_0_724]": 50997,
+  "[class_0_725]": 50998,
+  "[class_0_726]": 50999,
+  "[class_0_727]": 51000,
+  "[class_0_728]": 51001,
+  "[class_0_729]": 51002,
+  "[class_0_72]": 50345,
+  "[class_0_730]": 51003,
+  "[class_0_731]": 51004,
+  "[class_0_732]": 51005,
+  "[class_0_733]": 51006,
+  "[class_0_734]": 51007,
+  "[class_0_735]": 51008,
+  "[class_0_736]": 51009,
+  "[class_0_737]": 51010,
+  "[class_0_738]": 51011,
+  "[class_0_739]": 51012,
+  "[class_0_73]": 50346,
+  "[class_0_740]": 51013,
+  "[class_0_741]": 51014,
+  "[class_0_742]": 51015,
+  "[class_0_743]": 51016,
+  "[class_0_744]": 51017,
+  "[class_0_745]": 51018,
+  "[class_0_746]": 51019,
+  "[class_0_747]": 51020,
+  "[class_0_748]": 51021,
+  "[class_0_749]": 51022,
+  "[class_0_74]": 50347,
+  "[class_0_750]": 51023,
+  "[class_0_751]": 51024,
+  "[class_0_752]": 51025,
+  "[class_0_753]": 51026,
+  "[class_0_754]": 51027,
+  "[class_0_755]": 51028,
+  "[class_0_756]": 51029,
+  "[class_0_757]": 51030,
+  "[class_0_758]": 51031,
+  "[class_0_759]": 51032,
+  "[class_0_75]": 50348,
+  "[class_0_760]": 51033,
+  "[class_0_761]": 51034,
+  "[class_0_762]": 51035,
+  "[class_0_763]": 51036,
+  "[class_0_764]": 51037,
+  "[class_0_765]": 51038,
+  "[class_0_766]": 51039,
+  "[class_0_767]": 51040,
+  "[class_0_768]": 51041,
+  "[class_0_769]": 51042,
+  "[class_0_76]": 50349,
+  "[class_0_770]": 51043,
+  "[class_0_771]": 51044,
+  "[class_0_772]": 51045,
+  "[class_0_773]": 51046,
+  "[class_0_774]": 51047,
+  "[class_0_775]": 51048,
+  "[class_0_776]": 51049,
+  "[class_0_777]": 51050,
+  "[class_0_778]": 51051,
+  "[class_0_779]": 51052,
+  "[class_0_77]": 50350,
+  "[class_0_780]": 51053,
+  "[class_0_781]": 51054,
+  "[class_0_782]": 51055,
+  "[class_0_783]": 51056,
+  "[class_0_784]": 51057,
+  "[class_0_785]": 51058,
+  "[class_0_786]": 51059,
+  "[class_0_787]": 51060,
+  "[class_0_788]": 51061,
+  "[class_0_789]": 51062,
+  "[class_0_78]": 50351,
+  "[class_0_790]": 51063,
+  "[class_0_791]": 51064,
+  "[class_0_792]": 51065,
+  "[class_0_793]": 51066,
+  "[class_0_794]": 51067,
+  "[class_0_795]": 51068,
+  "[class_0_796]": 51069,
+  "[class_0_797]": 51070,
+  "[class_0_798]": 51071,
+  "[class_0_799]": 51072,
+  "[class_0_79]": 50352,
+  "[class_0_7]": 50280,
+  "[class_0_800]": 51073,
+  "[class_0_801]": 51074,
+  "[class_0_802]": 51075,
+  "[class_0_803]": 51076,
+  "[class_0_804]": 51077,
+  "[class_0_805]": 51078,
+  "[class_0_806]": 51079,
+  "[class_0_807]": 51080,
+  "[class_0_808]": 51081,
+  "[class_0_809]": 51082,
+  "[class_0_80]": 50353,
+  "[class_0_810]": 51083,
+  "[class_0_811]": 51084,
+  "[class_0_812]": 51085,
+  "[class_0_813]": 51086,
+  "[class_0_814]": 51087,
+  "[class_0_815]": 51088,
+  "[class_0_816]": 51089,
+  "[class_0_817]": 51090,
+  "[class_0_818]": 51091,
+  "[class_0_819]": 51092,
+  "[class_0_81]": 50354,
+  "[class_0_820]": 51093,
+  "[class_0_821]": 51094,
+  "[class_0_822]": 51095,
+  "[class_0_823]": 51096,
+  "[class_0_824]": 51097,
+  "[class_0_825]": 51098,
+  "[class_0_826]": 51099,
+  "[class_0_827]": 51100,
+  "[class_0_828]": 51101,
+  "[class_0_829]": 51102,
+  "[class_0_82]": 50355,
+  "[class_0_830]": 51103,
+  "[class_0_831]": 51104,
+  "[class_0_832]": 51105,
+  "[class_0_833]": 51106,
+  "[class_0_834]": 51107,
+  "[class_0_835]": 51108,
+  "[class_0_836]": 51109,
+  "[class_0_837]": 51110,
+  "[class_0_838]": 51111,
+  "[class_0_839]": 51112,
+  "[class_0_83]": 50356,
+  "[class_0_840]": 51113,
+  "[class_0_841]": 51114,
+  "[class_0_842]": 51115,
+  "[class_0_843]": 51116,
+  "[class_0_844]": 51117,
+  "[class_0_845]": 51118,
+  "[class_0_846]": 51119,
+  "[class_0_847]": 51120,
+  "[class_0_848]": 51121,
+  "[class_0_849]": 51122,
+  "[class_0_84]": 50357,
+  "[class_0_850]": 51123,
+  "[class_0_851]": 51124,
+  "[class_0_852]": 51125,
+  "[class_0_853]": 51126,
+  "[class_0_854]": 51127,
+  "[class_0_855]": 51128,
+  "[class_0_856]": 51129,
+  "[class_0_857]": 51130,
+  "[class_0_858]": 51131,
+  "[class_0_859]": 51132,
+  "[class_0_85]": 50358,
+  "[class_0_860]": 51133,
+  "[class_0_861]": 51134,
+  "[class_0_862]": 51135,
+  "[class_0_863]": 51136,
+  "[class_0_864]": 51137,
+  "[class_0_865]": 51138,
+  "[class_0_866]": 51139,
+  "[class_0_867]": 51140,
+  "[class_0_868]": 51141,
+  "[class_0_869]": 51142,
+  "[class_0_86]": 50359,
+  "[class_0_870]": 51143,
+  "[class_0_871]": 51144,
+  "[class_0_872]": 51145,
+  "[class_0_873]": 51146,
+  "[class_0_874]": 51147,
+  "[class_0_875]": 51148,
+  "[class_0_876]": 51149,
+  "[class_0_877]": 51150,
+  "[class_0_878]": 51151,
+  "[class_0_879]": 51152,
+  "[class_0_87]": 50360,
+  "[class_0_880]": 51153,
+  "[class_0_881]": 51154,
+  "[class_0_882]": 51155,
+  "[class_0_883]": 51156,
+  "[class_0_884]": 51157,
+  "[class_0_885]": 51158,
+  "[class_0_886]": 51159,
+  "[class_0_887]": 51160,
+  "[class_0_888]": 51161,
+  "[class_0_889]": 51162,
+  "[class_0_88]": 50361,
+  "[class_0_890]": 51163,
+  "[class_0_891]": 51164,
+  "[class_0_892]": 51165,
+  "[class_0_893]": 51166,
+  "[class_0_894]": 51167,
+  "[class_0_895]": 51168,
+  "[class_0_896]": 51169,
+  "[class_0_897]": 51170,
+  "[class_0_898]": 51171,
+  "[class_0_899]": 51172,
+  "[class_0_89]": 50362,
+  "[class_0_8]": 50281,
+  "[class_0_900]": 51173,
+  "[class_0_901]": 51174,
+  "[class_0_902]": 51175,
+  "[class_0_903]": 51176,
+  "[class_0_904]": 51177,
+  "[class_0_905]": 51178,
+  "[class_0_906]": 51179,
+  "[class_0_907]": 51180,
+  "[class_0_908]": 51181,
+  "[class_0_909]": 51182,
+  "[class_0_90]": 50363,
+  "[class_0_910]": 51183,
+  "[class_0_911]": 51184,
+  "[class_0_912]": 51185,
+  "[class_0_913]": 51186,
+  "[class_0_914]": 51187,
+  "[class_0_915]": 51188,
+  "[class_0_916]": 51189,
+  "[class_0_917]": 51190,
+  "[class_0_918]": 51191,
+  "[class_0_919]": 51192,
+  "[class_0_91]": 50364,
+  "[class_0_920]": 51193,
+  "[class_0_921]": 51194,
+  "[class_0_922]": 51195,
+  "[class_0_923]": 51196,
+  "[class_0_924]": 51197,
+  "[class_0_925]": 51198,
+  "[class_0_926]": 51199,
+  "[class_0_927]": 51200,
+  "[class_0_928]": 51201,
+  "[class_0_929]": 51202,
+  "[class_0_92]": 50365,
+  "[class_0_930]": 51203,
+  "[class_0_931]": 51204,
+  "[class_0_932]": 51205,
+  "[class_0_933]": 51206,
+  "[class_0_934]": 51207,
+  "[class_0_935]": 51208,
+  "[class_0_936]": 51209,
+  "[class_0_937]": 51210,
+  "[class_0_938]": 51211,
+  "[class_0_939]": 51212,
+  "[class_0_93]": 50366,
+  "[class_0_940]": 51213,
+  "[class_0_941]": 51214,
+  "[class_0_942]": 51215,
+  "[class_0_943]": 51216,
+  "[class_0_944]": 51217,
+  "[class_0_945]": 51218,
+  "[class_0_946]": 51219,
+  "[class_0_947]": 51220,
+  "[class_0_948]": 51221,
+  "[class_0_949]": 51222,
+  "[class_0_94]": 50367,
+  "[class_0_950]": 51223,
+  "[class_0_951]": 51224,
+  "[class_0_952]": 51225,
+  "[class_0_953]": 51226,
+  "[class_0_954]": 51227,
+  "[class_0_955]": 51228,
+  "[class_0_956]": 51229,
+  "[class_0_957]": 51230,
+  "[class_0_958]": 51231,
+  "[class_0_959]": 51232,
+  "[class_0_95]": 50368,
+  "[class_0_960]": 51233,
+  "[class_0_961]": 51234,
+  "[class_0_962]": 51235,
+  "[class_0_963]": 51236,
+  "[class_0_964]": 51237,
+  "[class_0_965]": 51238,
+  "[class_0_966]": 51239,
+  "[class_0_967]": 51240,
+  "[class_0_968]": 51241,
+  "[class_0_969]": 51242,
+  "[class_0_96]": 50369,
+  "[class_0_970]": 51243,
+  "[class_0_971]": 51244,
+  "[class_0_972]": 51245,
+  "[class_0_973]": 51246,
+  "[class_0_974]": 51247,
+  "[class_0_975]": 51248,
+  "[class_0_976]": 51249,
+  "[class_0_977]": 51250,
+  "[class_0_978]": 51251,
+  "[class_0_979]": 51252,
+  "[class_0_97]": 50370,
+  "[class_0_980]": 51253,
+  "[class_0_981]": 51254,
+  "[class_0_982]": 51255,
+  "[class_0_983]": 51256,
+  "[class_0_984]": 51257,
+  "[class_0_985]": 51258,
+  "[class_0_986]": 51259,
+  "[class_0_987]": 51260,
+  "[class_0_988]": 51261,
+  "[class_0_989]": 51262,
+  "[class_0_98]": 50371,
+  "[class_0_990]": 51263,
+  "[class_0_991]": 51264,
+  "[class_0_992]": 51265,
+  "[class_0_993]": 51266,
+  "[class_0_994]": 51267,
+  "[class_0_995]": 51268,
+  "[class_0_996]": 51269,
+  "[class_0_997]": 51270,
+  "[class_0_998]": 51271,
+  "[class_0_999]": 51272,
+  "[class_0_99]": 50372,
+  "[class_0_9]": 50282
+}

last-checkpoint/config.json

@@ -25,6 +25,6 @@
   "torch_dtype": "float32",
   "transformers_version": "4.52.4",
   "use_cache": true,
-  "vocab_size": 51291,
+  "vocab_size": 51297,
   "word_embed_proj_dim": 768
 }

last-checkpoint/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:60ef70e916534602727c685bc3e560089cfddeb9a05e3837fc8190396a9a7e7b
-size 504109968
+oid sha256:a7cc629374fc46886b1e24ab16dc92fe214f23fc080c19fa4a7a615068db5c5b
+size 504128400

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bedf0d1b7c3bbf0111aba64f6132b02f55b24345be4304477e408503c9887b6d
-size 1008339066
+oid sha256:8a2c30197a0c52ac2794b295388908da96887c32e5211b307643b94cc3633690
+size 1008375930

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ccceff689fde6fd741090e83495b2f3a7e45ddfe8bbd282dc268cffe901b0a8e
+oid sha256:f46f04b2334750ca6d05fa3a065b5b08c86ce7573e9349b4799fd4232bbfff55
 size 14244

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:188c81e189c658e48078a1ca1cd3c2acf1186f080e97455286a1864a893eb44e
+oid sha256:ae3e354ca816fa1bbc609ddc3283d10d94d7b8a4065a9dd53bdef6ddf1669777
 size 988

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9fd3f0a20426ba838eb0d45cc61bd3a8604c8e9e67dcb30df34f1aa0f2e13603
+oid sha256:cea666a390eccba21e7e10bcf8f4d615cd341461a1c7a63f919ff96000ca6f4b
 size 1064

last-checkpoint/special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

last-checkpoint/trainer_state.json

@@ -1,50 +1,237 @@
 {
-  "best_global_step": 84,
-  "best_metric": 2.854076623916626,
-  "best_model_checkpoint": "./opt_thinker_ckpts/checkpoint-84",
-  "epoch": 3.0,
+  "best_global_step": 300,
+  "best_metric": 2.361220121383667,
+  "best_model_checkpoint": "./opt_thinker_ckpts2/checkpoint-300",
+  "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 84,
+  "global_step": 300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0,
-      "eval_loss": 4.2282938957214355,
-      "eval_runtime": 5.5985,
-      "eval_samples_per_second": 8.574,
-      "eval_steps_per_second": 4.287,
-      "step": 28
+      "epoch": 0.03333333333333333,
+      "grad_norm": 11.137995719909668,
+      "learning_rate": 9.999931461237134e-05,
+      "loss": 7.4006,
+      "step": 10
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 5.385880947113037,
+      "learning_rate": 9.999383162408304e-05,
+      "loss": 4.4949,
+      "step": 20
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5095698833465576,
+      "learning_rate": 9.998286624877786e-05,
+      "loss": 4.0902,
+      "step": 30
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 2.4948348999023438,
+      "learning_rate": 9.996641968893282e-05,
+      "loss": 3.7388,
+      "step": 40
     },
     {
-      "epoch": 1.8073394495412844,
-      "grad_norm": 15.655576705932617,
-      "learning_rate": 9.898376992116179e-05,
-      "loss": 36.5614,
+      "epoch": 0.16666666666666666,
+      "grad_norm": 3.0202476978302,
+      "learning_rate": 9.994449374809851e-05,
+      "loss": 3.2958,
       "step": 50
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 3.1219027042388916,
-      "eval_runtime": 5.6122,
-      "eval_samples_per_second": 8.553,
-      "eval_steps_per_second": 4.276,
-      "step": 56
+      "epoch": 0.2,
+      "grad_norm": 3.4056878089904785,
+      "learning_rate": 9.991709083070143e-05,
+      "loss": 2.9765,
+      "step": 60
+    },
+    {
+      "epoch": 0.23333333333333334,
+      "grad_norm": 3.8623976707458496,
+      "learning_rate": 9.988421394178027e-05,
+      "loss": 2.8075,
+      "step": 70
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 4.10685920715332,
+      "learning_rate": 9.98458666866564e-05,
+      "loss": 2.7051,
+      "step": 80
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.230323791503906,
+      "learning_rate": 9.980205327053848e-05,
+      "loss": 2.5966,
+      "step": 90
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 2.5417559146881104,
+      "learning_rate": 9.975277849806133e-05,
+      "loss": 2.5829,
+      "step": 100
+    },
+    {
+      "epoch": 0.36666666666666664,
+      "grad_norm": 3.1019628047943115,
+      "learning_rate": 9.9698047772759e-05,
+      "loss": 2.5723,
+      "step": 110
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.955507278442383,
+      "learning_rate": 9.963786709647228e-05,
+      "loss": 2.605,
+      "step": 120
+    },
+    {
+      "epoch": 0.43333333333333335,
+      "grad_norm": 3.4431023597717285,
+      "learning_rate": 9.957224306869053e-05,
+      "loss": 2.5316,
+      "step": 130
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 3.066657304763794,
+      "learning_rate": 9.950118288582788e-05,
+      "loss": 2.5645,
+      "step": 140
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6040785312652588,
+      "learning_rate": 9.942469434043418e-05,
+      "loss": 2.5448,
+      "step": 150
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 2.585803508758545,
+      "learning_rate": 9.934278582034037e-05,
+      "loss": 2.5451,
+      "step": 160
+    },
+    {
+      "epoch": 0.5666666666666667,
+      "grad_norm": 1.8818528652191162,
+      "learning_rate": 9.92554663077387e-05,
+      "loss": 2.513,
+      "step": 170
     },
     {
-      "epoch": 3.0,
-      "eval_loss": 2.854076623916626,
-      "eval_runtime": 5.6227,
-      "eval_samples_per_second": 8.537,
-      "eval_steps_per_second": 4.268,
-      "step": 84
+      "epoch": 0.6,
+      "grad_norm": 1.9180946350097656,
+      "learning_rate": 9.916274537819775e-05,
+      "loss": 2.4365,
+      "step": 180
+    },
+    {
+      "epoch": 0.6333333333333333,
+      "grad_norm": 3.170057535171509,
+      "learning_rate": 9.906463319961225e-05,
+      "loss": 2.4846,
+      "step": 190
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 3.2630980014801025,
+      "learning_rate": 9.896114053108829e-05,
+      "loss": 2.4806,
+      "step": 200
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7932326793670654,
+      "learning_rate": 9.88522787217632e-05,
+      "loss": 2.4994,
+      "step": 210
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 2.0861623287200928,
+      "learning_rate": 9.87380597095611e-05,
+      "loss": 2.4128,
+      "step": 220
+    },
+    {
+      "epoch": 0.7666666666666667,
+      "grad_norm": 2.4601433277130127,
+      "learning_rate": 9.861849601988383e-05,
+      "loss": 2.4513,
+      "step": 230
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.431674599647522,
+      "learning_rate": 9.849360076423734e-05,
+      "loss": 2.4656,
+      "step": 240
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.9531445503234863,
+      "learning_rate": 9.836338763879385e-05,
+      "loss": 2.4561,
+      "step": 250
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 2.1023504734039307,
+      "learning_rate": 9.822787092288991e-05,
+      "loss": 2.4226,
+      "step": 260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.7438138723373413,
+      "learning_rate": 9.808706547746057e-05,
+      "loss": 2.4111,
+      "step": 270
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 1.9399806261062622,
+      "learning_rate": 9.794098674340965e-05,
+      "loss": 2.4352,
+      "step": 280
+    },
+    {
+      "epoch": 0.9666666666666667,
+      "grad_norm": 2.0438737869262695,
+      "learning_rate": 9.778965073991651e-05,
+      "loss": 2.3811,
+      "step": 290
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.9183045625686646,
+      "learning_rate": 9.763307406267932e-05,
+      "loss": 2.3174,
+      "step": 300
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 2.361220121383667,
+      "eval_runtime": 10.6371,
+      "eval_samples_per_second": 9.401,
+      "eval_steps_per_second": 4.701,
+      "step": 300
     }
   ],
-  "logging_steps": 50,
-  "max_steps": 700,
+  "logging_steps": 10,
+  "max_steps": 3000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 25,
+  "num_train_epochs": 10,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -58,8 +245,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 1255480655616000.0,
-  "train_batch_size": 2,
+  "total_flos": 940651315200000.0,
+  "train_batch_size": 3,
   "trial_name": null,
   "trial_params": null
 }

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d34b9553bbb2026c2c4963f5ef04a7de90dae66202d435b4b713acbf8cee3894
+oid sha256:ea469926b8e2f70b091c99a67fbe0ac8c6a6d1414b186aec873f352693c2db70
 size 5304