Training in progress, step 5

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1 @@
+checkpoint-*/

added_tokens.json

@@ -0,0 +1,1054 @@
+{
+  "abundant": 250922,
+  "achachi": 250100,
+  "achachila": 250101,
+  "achachilan tatapa": 250102,
+  "achacu": 250103,
+  "achaku": 250104,
+  "achupalla": 250105,
+  "achuqallu": 250106,
+  "aeroplane": 250866,
+  "afflict": 250976,
+  "afflicted": 251054,
+  "afternoon": 250943,
+  "ahuayo": 250107,
+  "ahuicha": 250108,
+  "ahuichan mamapa": 250109,
+  "ajanu": 250110,
+  "ajarankhu": 250111,
+  "ajhanu": 250112,
+  "ajusa": 250113,
+  "akhan tuyuña walikiti": 250114,
+  "alajgpacha": 250115,
+  "alajha": 250116,
+  "alajhpacha": 250117,
+  "alga": 250985,
+  "aliqa": 250118,
+  "aljheri": 250119,
+  "allchhi": 250120,
+  "allpachu": 250121,
+  "allpi": 250122,
+  "allqa": 250123,
+  "allqamari": 250124,
+  "allu": 250125,
+  "almost": 251013,
+  "alpaca": 250852,
+  "although": 250932,
+  "amaya": 250126,
+  "ampar moqo": 250127,
+  "ampar quta": 250128,
+  "ampara": 250129,
+  "amulet": 251007,
+  "amuyu": 250130,
+  "amwulansya thaqanim": 250131,
+  "anaqa": 250132,
+  "anata": 250133,
+  "anqa": 250134,
+  "apachi": 250135,
+  "apnaqeri": 250136,
+  "apples": 251000,
+  "armadillo": 250969,
+  "armpit": 250899,
+  "aru apiri": 250137,
+  "aruma": 250138,
+  "arumanti": 250139,
+  "arumi": 250140,
+  "arumita": 250141,
+  "asiru": 250142,
+  "aski": 250143,
+  "askiru churata tatay": 250144,
+  "aspirin": 250864,
+  "aspirina": 250145,
+  "ass": 250894,
+  "asu": 250146,
+  "asukara": 250147,
+  "aucch'i": 250148,
+  "auqui": 250149,
+  "autipacha": 250150,
+  "awti": 250151,
+  "awtu": 250152,
+  "awyuna": 250153,
+  "aycha kanka": 250154,
+  "aychathejhe": 250155,
+  "aymara": 250156,
+  "aynacha": 250157,
+  "añuthaya": 250158,
+  "barley": 251087,
+  "beefsteak": 250867,
+  "belly": 251028,
+  "biscuits": 251004,
+  "bitter": 250941,
+  "black and white": 250854,
+  "blacken": 251066,
+  "blurry": 250898,
+  "braid": 250957,
+  "breathing": 251081,
+  "broom": 251036,
+  "brother": 250945,
+  "brother-in-law": 250986,
+  "bunch": 251139,
+  "butcher": 251002,
+  "buttock": 250883,
+  "callachi": 250159,
+  "callchi": 250160,
+  "canary": 250878,
+  "caratacu": 250161,
+  "carrot": 251082,
+  "catari": 250162,
+  "catjata": 250163,
+  "cattle": 251132,
+  "catuña": 250164,
+  "cauquitaastasa?": 250165,
+  "cayu": 250166,
+  "cayumoqo": 250167,
+  "ch'aca": 250168,
+  "ch'aka": 250169,
+  "ch'aku": 250170,
+  "ch'alla": 250171,
+  "ch'amaka": 250172,
+  "ch'apaqa": 250173,
+  "ch'api": 250174,
+  "ch'araña": 250175,
+  "ch'armiri": 250176,
+  "ch'arqi": 250177,
+  "ch'axwa": 250178,
+  "ch'ayña": 250179,
+  "ch'añari": 250180,
+  "ch'ejhe": 250181,
+  "ch'epa ampara": 250182,
+  "ch'eqa ampara": 250183,
+  "ch'ihu": 250184,
+  "ch'iji": 250185,
+  "ch'ijma": 250186,
+  "ch'iki": 250187,
+  "ch'ina": 250188,
+  "ch'ipoqo": 250189,
+  "ch'ipuqu": 250190,
+  "ch'iqa": 250191,
+  "ch'islli": 250192,
+  "ch'iyar sajuna": 250193,
+  "ch'iyara": 250194,
+  "ch'ojhña": 250195,
+  "ch'oqe": 250196,
+  "ch'uju": 250197,
+  "ch'ulla": 250198,
+  "ch'umphi": 250199,
+  "ch'umpi": 250200,
+  "ch'uqi": 250201,
+  "ch'urahui": 250202,
+  "ch'uru": 250203,
+  "ch'usa": 250204,
+  "ch'uspi": 250205,
+  "ch'utu": 250206,
+  "ch'uxña": 250207,
+  "chaca": 250208,
+  "chacha": 250209,
+  "chacha yaq'aña": 250210,
+  "challa": 250211,
+  "challhua": 250212,
+  "challwa": 250213,
+  "chaullacaturi": 250214,
+  "chawlla": 250215,
+  "chayanta": 250216,
+  "chchiara": 250217,
+  "cheek": 250968,
+  "cheers!": 251123,
+  "cheese": 250978,
+  "chest": 251151,
+  "chhaqachhaqa": 250218,
+  "chharpu": 250219,
+  "chhaxlla": 250220,
+  "chheqanqara": 250221,
+  "chhichhillanca": 250222,
+  "chhichhillanka": 250223,
+  "chhijchi": 250224,
+  "chhipchi": 250225,
+  "chhiqanqara": 250226,
+  "chhojhchiphorqe": 250227,
+  "chhullchu": 250228,
+  "chhullunkhaya": 250229,
+  "chhuxchi": 250230,
+  "chhuxlla": 250231,
+  "chhuxu": 250232,
+  "chica": 250233,
+  "chica uru": 250234,
+  "chicäruma": 250235,
+  "chinqi": 250236,
+  "chips": 251020,
+  "chiqa": 250237,
+  "chira": 250238,
+  "chiuchi": 250239,
+  "chiwchi": 250240,
+  "chiñi": 250241,
+  "chopsticks": 250998,
+  "chore": 250995,
+  "chuhua": 250242,
+  "chuima": 250243,
+  "chujhaña": 250244,
+  "chullpa": 250245,
+  "chullunkiya": 250246,
+  "chupa": 250247,
+  "chupika": 250248,
+  "chuqa": 250249,
+  "chuqi": 250250,
+  "chuqupa": 250251,
+  "chuxu": 250252,
+  "chuyma": 250253,
+  "chuymani": 250254,
+  "cloudy": 250890,
+  "condor": 250911,
+  "convulsion": 250966,
+  "corpse": 250856,
+  "cough": 251131,
+  "coward": 250901,
+  "credit card": 250910,
+  "credit tarjita": 250255,
+  "cub": 251056,
+  "cucumber": 251037,
+  "cullaca": 250256,
+  "cunca": 250257,
+  "cunturi": 250258,
+  "cup and saucer": 251039,
+  "cupi ampara": 250259,
+  "curmi": 250260,
+  "curse": 251063,
+  "cusicusi": 250261,
+  "cyclone": 251120,
+  "dancer": 251097,
+  "dandruff": 251050,
+  "dark blue": 250886,
+  "darkness": 250873,
+  "daughter-in-law": 251148,
+  "daugther": 251034,
+  "deer": 251092,
+  "desert": 250916,
+  "diligent": 251047,
+  "dirt": 250885,
+  "disarranged": 250902,
+  "distant": 250942,
+  "donkey": 250895,
+  "dough": 251038,
+  "dozen": 251112,
+  "dried beef": 250876,
+  "drunken": 251122,
+  "dungeon": 250853,
+  "ear": 250947,
+  "eight": 251071,
+  "eighteen": 251115,
+  "eighty": 251072,
+  "elbow": 251005,
+  "elegant": 250958,
+  "eternal": 251142,
+  "excrement": 250930,
+  "excuse me": 251001,
+  "eyebrow": 250884,
+  "eyelash": 250900,
+  "farmer": 251147,
+  "father-in-law": 250865,
+  "fertilizing": 250949,
+  "fetus": 251090,
+  "fib": 251033,
+  "fifth": 251031,
+  "fifty": 251030,
+  "fingernail": 251085,
+  "firefly": 251011,
+  "fisherman": 250897,
+  "flamingo": 251024,
+  "flatulence": 251086,
+  "flea": 250960,
+  "flour": 250929,
+  "forefront": 251022,
+  "forehead": 251023,
+  "fork": 251103,
+  "forty": 251045,
+  "four hundred": 251043,
+  "fourteen": 251114,
+  "fourth": 251044,
+  "full moon": 251125,
+  "garlic": 250848,
+  "gents' toilet": 250896,
+  "get an ambulance!": 250859,
+  "glowworm": 251012,
+  "glue": 251121,
+  "godfather": 250921,
+  "godmother": 250920,
+  "good afternoon": 251141,
+  "good evening": 251144,
+  "good morning": 250863,
+  "good-bye": 251083,
+  "grandfather": 250841,
+  "grandmother": 250845,
+  "grandson": 250851,
+  "grapefruit": 251119,
+  "grapes": 251129,
+  "great grandfather": 250842,
+  "great grandmother": 250846,
+  "grippe": 250956,
+  "guanaco": 251134,
+  "guinea pig": 251135,
+  "hamacura": 250262,
+  "hanchi": 250263,
+  "handful": 251048,
+  "hankko": 250264,
+  "have you got a pen?": 251040,
+  "help!": 251146,
+  "herd": 250860,
+  "hiwisa": 250265,
+  "hollow": 251017,
+  "honeybee": 251041,
+  "how much is this?": 251059,
+  "huahuita": 250266,
+  "huallpa": 250267,
+  "huampara": 250268,
+  "huank'u": 250269,
+  "huarahuara": 250270,
+  "huaranqa": 250271,
+  "huaranqanhuaranqapa": 250272,
+  "huarmi": 250273,
+  "huasara": 250274,
+  "huasuru": 250275,
+  "huat'a": 250276,
+  "huayna": 250277,
+  "huayronqo": 250278,
+  "hucumari": 250279,
+  "huila": 250280,
+  "huit'u": 250281,
+  "hunttuña": 250282,
+  "hupa llamayu": 250283,
+  "i don't eat meat": 250933,
+  "i don't know": 250937,
+  "i don't smoke": 250936,
+  "i don't understand": 250934,
+  "i feel ill": 251127,
+  "i need a doctor": 251107,
+  "i'd like a window seat, please": 251143,
+  "ichumama": 250284,
+  "ichutata": 250285,
+  "identical": 250977,
+  "idleness": 251075,
+  "ijma": 250286,
+  "ikiña": 250287,
+  "illapu": 250288,
+  "illness": 251126,
+  "illpa": 250289,
+  "imilla": 250290,
+  "in front": 250967,
+  "inkilla": 250291,
+  "insect": 251019,
+  "insipid": 250874,
+  "intelligent": 250882,
+  "intestine": 250948,
+  "intijalanta": 250292,
+  "intijalsu": 250293,
+  "iquilla": 250294,
+  "iquiña": 250295,
+  "is it safe to swim here?": 250849,
+  "isk'aqota": 250296,
+  "ispa": 250297,
+  "ispillo": 250298,
+  "itapallu": 250299,
+  "jach'a luk'ana": 250300,
+  "jach'a uta": 250301,
+  "jach'a wasu": 250302,
+  "jach'aqollo": 250303,
+  "jach'aqota": 250304,
+  "jacha": 250305,
+  "jahuari": 250306,
+  "jahuira": 250307,
+  "jak'u": 250308,
+  "jallu": 250309,
+  "jamach'i": 250310,
+  "jamp'atu": 250311,
+  "jamwurjisa": 250312,
+  "janchi": 250313,
+  "janis": 250314,
+  "janiw aych manq'rïktti": 250315,
+  "janiw intintktti": 250316,
+  "janiw pitañ luwarakiti": 250317,
+  "janiw pitirïktti": 250318,
+  "janiw yatktti": 250319,
+  "janiwa": 250320,
+  "janq'o": 250321,
+  "janq'u": 250322,
+  "janq'ulla": 250323,
+  "jant'aku": 250324,
+  "jaqi": 250325,
+  "jararankha": 250326,
+  "jararankhu": 250327,
+  "jaru": 250328,
+  "jatha": 250329,
+  "jathi": 250330,
+  "jawasa": 250331,
+  "jaxu": 250332,
+  "jay": 250333,
+  "jayp'u": 250334,
+  "jayra": 250335,
+  "jayri": 250336,
+  "jayu": 250337,
+  "jichhüru": 250338,
+  "jikhani": 250339,
+  "jilata": 250340,
+  "jilatata": 250341,
+  "jincho": 250342,
+  "jinchu": 250343,
+  "jinq'i": 250344,
+  "jiphilla": 250345,
+  "jiru": 250346,
+  "jisa": 250347,
+  "jiski": 250348,
+  "jiwaka": 250349,
+  "jiwaki": 250350,
+  "jiwata": 250351,
+  "jiwq'i": 250352,
+  "juch'usa": 250353,
+  "jucha": 250354,
+  "jucu": 250355,
+  "jucumari": 250356,
+  "jump'i": 250357,
+  "junt'u": 250358,
+  "junt'upacha": 250359,
+  "jupa": 250360,
+  "jupuqu": 250361,
+  "juq'ullu": 250362,
+  "jurma": 250363,
+  "juykhu": 250364,
+  "juyphipacha": 250365,
+  "juyra": 250366,
+  "k'achatak arst'am": 250367,
+  "k'achi": 250368,
+  "k'aira": 250369,
+  "k'aja": 250370,
+  "k'ana": 250371,
+  "k'anchilla": 250372,
+  "k'ank'a": 250373,
+  "k'ari": 250374,
+  "k'auna": 250375,
+  "k'awna": 250376,
+  "k'axilu": 250377,
+  "k'ayra": 250378,
+  "k'ita": 250379,
+  "k'iwcha": 250380,
+  "k'uchi": 250381,
+  "k'ulli": 250382,
+  "k'ullu": 250383,
+  "k'uma": 250384,
+  "k'umara": 250385,
+  "k'umu": 250386,
+  "k'usillo": 250387,
+  "k'uti": 250388,
+  "kallachi": 250389,
+  "kamisaraki": 250390,
+  "kanqiri": 250391,
+  "kaphiya": 250392,
+  "kawallu": 250393,
+  "kawkhans taksi utji": 250394,
+  "kawkharus awt istasyunirista": 250395,
+  "kawki": 250396,
+  "kawkins awt alkili": 250397,
+  "kawkits pasaj alirista": 250398,
+  "khallu": 250399,
+  "khankha": 250400,
+  "kharkhati": 250401,
+  "kharu": 250402,
+  "khaycatana": 250403,
+  "khañu": 250404,
+  "khirkhi": 250405,
+  "khuchi": 250406,
+  "khuchi li'ki": 250407,
+  "khuchuña": 250408,
+  "khullu": 250409,
+  "khuna": 250410,
+  "khunu": 250411,
+  "khunuqollonaca": 250412,
+  "khusa": 250413,
+  "kichu": 250414,
+  "kikipa": 250415,
+  "kimsa": 250416,
+  "kiru": 250417,
+  "kisu": 250418,
+  "kkunkkuri": 250419,
+  "knife": 250973,
+  "knives": 250974,
+  "kucha": 250420,
+  "kuchara": 250421,
+  "kuchillu": 250422,
+  "kukakula": 250423,
+  "kuliktiwu": 250424,
+  "kullaka": 250425,
+  "kumri": 250426,
+  "kuna uraxisa": 250427,
+  "kunata": 250428,
+  "kunjata": 250429,
+  "kunka": 250430,
+  "kuns umañ munta": 250431,
+  "kuntura": 250432,
+  "kupi": 250433,
+  "kusikusi": 250434,
+  "laca": 250435,
+  "lacach'aca": 250436,
+  "ladies' toilet": 251137,
+  "lagoon": 250925,
+  "lajhra": 250437,
+  "laka chaca": 250438,
+  "lakhampu": 250439,
+  "lamana": 250440,
+  "lamb": 250914,
+  "landmark": 251084,
+  "lankhu": 250441,
+  "lap'a": 250442,
+  "laphi": 250443,
+  "laq'a": 250444,
+  "laq'u": 250445,
+  "laqa": 250446,
+  "laqatu": 250447,
+  "laqha": 250448,
+  "laqu": 250449,
+  "larama": 250450,
+  "laranja": 250451,
+  "larankha": 250452,
+  "latanu": 250453,
+  "laxra": 250454,
+  "laxwi": 250455,
+  "layqa": 250456,
+  "left hand": 250880,
+  "lichi": 250457,
+  "light blue": 251062,
+  "lightning": 250991,
+  "liju": 250458,
+  "lik'i": 250459,
+  "limp": 250909,
+  "limuna": 250460,
+  "lip'ichi": 250461,
+  "liqi liqi": 250462,
+  "lirpu": 250463,
+  "lizard": 250939,
+  "llaki": 250464,
+  "llallawa": 250465,
+  "llama": 251061,
+  "llaphi": 250466,
+  "llatunctunca": 250467,
+  "llawi": 250468,
+  "lliju lliju": 250469,
+  "llijulliju": 250470,
+  "llilli": 250471,
+  "llink'i": 250472,
+  "llumpaqa": 250473,
+  "llätunca": 250474,
+  "lodging": 251080,
+  "louse": 250984,
+  "luk'ana": 250475,
+  "lukkana": 250476,
+  "lunthata": 250477,
+  "luraña": 250478,
+  "luxru": 250479,
+  "magician": 250987,
+  "magnificent": 250988,
+  "maize": 251118,
+  "makaña": 250480,
+  "malitanakasti": 250481,
+  "mallku": 250482,
+  "mallq'a": 250483,
+  "mamani": 250484,
+  "manku": 250485,
+  "manq'a jayt'uña": 250486,
+  "manq'añ sirk'i": 250487,
+  "manqha": 250488,
+  "mansana": 250489,
+  "mantikilla": 250490,
+  "mascot": 250906,
+  "mat'aqi": 250491,
+  "may i use your phone?": 251100,
+  "mayay jiskt'asïma": 250492,
+  "mañaso": 250493,
+  "mijilla": 250494,
+  "mik'i": 250495,
+  "mikha": 250496,
+  "millu": 250497,
+  "mirq'i": 250498,
+  "misk'i": 250499,
+  "misk'i t'ant'alla": 250500,
+  "mosquito": 250893,
+  "mother-in-law": 251094,
+  "mountain range": 250975,
+  "moustache": 251091,
+  "mucus": 250953,
+  "mug": 250928,
+  "mujlli": 250501,
+  "muk'i": 250502,
+  "mullu": 250503,
+  "muraru": 250504,
+  "muruqu": 250505,
+  "mushroom": 251046,
+  "mushrooms": 250999,
+  "mute": 250858,
+  "muxsa": 250506,
+  "muxsa chhuxllu": 250507,
+  "muyta": 250508,
+  "naira": 250509,
+  "nalra": 250510,
+  "nappy": 250913,
+  "nayra": 250511,
+  "near": 251014,
+  "nearly": 251015,
+  "neat": 250993,
+  "neighbor": 251128,
+  "nerve": 250861,
+  "nettle": 250927,
+  "new moon": 250944,
+  "ninalaq'o": 250512,
+  "ninanqu": 250513,
+  "nineteen": 251110,
+  "ninety": 250990,
+  "non-smoking please": 250935,
+  "noon": 250903,
+  "nose": 251010,
+  "one hundred": 251026,
+  "one thousand": 250915,
+  "oraqe": 250514,
+  "orphan": 251133,
+  "outside": 250862,
+  "outstanding": 250961,
+  "p'eke": 250515,
+  "p'iqi": 250516,
+  "p'isaqa": 250517,
+  "p'iya": 250518,
+  "p'ujru chuwa": 250519,
+  "paca": 250520,
+  "pachpa": 250521,
+  "palatu": 250522,
+  "palaya": 250523,
+  "palaya kawkhankisa": 250524,
+  "pampa": 250525,
+  "pancake": 251096,
+  "pankataya": 250526,
+  "panqara": 250527,
+  "papa phirita": 250528,
+  "papataca": 250529,
+  "paqallqo": 250530,
+  "paqallqtunca": 250531,
+  "paqallqu": 250532,
+  "pariwana": 250533,
+  "parrot": 251124,
+  "partridge": 251016,
+  "pasapurt lunthatapxitu": 250534,
+  "pasapurti": 250535,
+  "pataca": 250536,
+  "patacmara": 250537,
+  "pataka": 250538,
+  "patunca": 250539,
+  "paya": 250540,
+  "peanut": 250908,
+  "peninsula": 250919,
+  "perqa": 250541,
+  "phajhsi": 250542,
+  "phalahuiru": 250543,
+  "phantta": 250544,
+  "pharmasya kawkhankisa": 250545,
+  "phatanka": 250546,
+  "phaxcha": 250547,
+  "phaxi": 250548,
+  "phesqa": 250549,
+  "phesqatunca": 250550,
+  "phesqëri": 250551,
+  "phichitanqa": 250552,
+  "phisi": 250553,
+  "phisna": 250554,
+  "phisqa": 250555,
+  "phucha": 250556,
+  "phucu": 250557,
+  "phuju": 250558,
+  "phurutilla": 250559,
+  "phuyu": 250560,
+  "pichana": 250561,
+  "pichaña": 250562,
+  "pichu": 250563,
+  "pigeon": 251074,
+  "pillow": 250881,
+  "pilpinto": 250564,
+  "pilpintu": 250565,
+  "pineapple": 250843,
+  "pipinu": 250566,
+  "pirasa": 250567,
+  "pirtunitayya": 250568,
+  "pisi": 250569,
+  "pitu": 250570,
+  "platillu": 250571,
+  "please speak more slowly": 250955,
+  "plentiful": 250923,
+  "poqota": 250572,
+  "pork": 250970,
+  "porker": 250971,
+  "potato": 250887,
+  "potatoes": 250889,
+  "ppekke": 250573,
+  "provisions": 250954,
+  "puncu": 250574,
+  "puntawul lanktati": 250575,
+  "pupa": 250576,
+  "puquta": 250577,
+  "puraca": 250578,
+  "puraka": 250579,
+  "pusi": 250580,
+  "pusipataca": 250581,
+  "pusiri": 250582,
+  "pusitunca": 250583,
+  "päpataca": 250584,
+  "pätunca": 250585,
+  "q'allampa": 250586,
+  "q'anu": 250587,
+  "q'apa": 250588,
+  "q'api": 250589,
+  "q'ara": 250590,
+  "q'asa": 250591,
+  "q'atawi": 250592,
+  "q'ayma": 250593,
+  "q'añu": 250594,
+  "q'ejhoq'ejho": 250595,
+  "q'ellu": 250596,
+  "q'ili": 250597,
+  "q'illu": 250598,
+  "q'ita": 250599,
+  "q'iwi": 250600,
+  "q'uruta": 250601,
+  "q'uya": 250602,
+  "qachu": 250603,
+  "qallu": 250604,
+  "qallu luk'ana": 250605,
+  "qamaqe": 250606,
+  "qamaqi": 250607,
+  "qampu": 250608,
+  "qanjili": 250609,
+  "qanjlli": 250610,
+  "qapa": 250611,
+  "qaqa": 250612,
+  "qaqawara": 250613,
+  "qarqa": 250614,
+  "qasiwi": 250615,
+  "qasïwi": 250616,
+  "qawqhas akaxa": 250617,
+  "qawqhaxisa": 250618,
+  "qawra": 250619,
+  "qaxsi": 250620,
+  "qellqeri": 250621,
+  "qenaya": 250622,
+  "qhala": 250623,
+  "qhana": 250624,
+  "qhana sajuna": 250625,
+  "qhatu": 250626,
+  "qhepa": 250627,
+  "qheri": 250628,
+  "qhespi": 250629,
+  "qhilla": 250630,
+  "qhincha": 250631,
+  "qhini": 250632,
+  "qhiphürkama": 250633,
+  "qhisti": 250634,
+  "qhontiri": 250635,
+  "qhuna": 250636,
+  "qilla": 250637,
+  "qimara": 250638,
+  "qinaya": 250639,
+  "qisimi": 250640,
+  "qismi": 250641,
+  "qolli": 250642,
+  "qollqe": 250643,
+  "qonqori": 250644,
+  "qont'asina": 250645,
+  "qont'asiña": 250646,
+  "qoqa": 250647,
+  "qoqaqoqa": 250648,
+  "qora": 250649,
+  "qota": 250650,
+  "quimsa": 250651,
+  "quimsapataca": 250652,
+  "quimsaqallqo": 250653,
+  "quimsaqallqtunca": 250654,
+  "quimsatunca": 250655,
+  "quimsiri": 250656,
+  "quimsïri": 250657,
+  "quitula": 250658,
+  "qulayu": 250659,
+  "qulla": 250660,
+  "qullana": 250661,
+  "qullpa": 250662,
+  "qullqi": 250663,
+  "qullu": 250664,
+  "quna": 250665,
+  "qunchu": 250666,
+  "qunquri": 250667,
+  "qupa": 250668,
+  "quqa": 250669,
+  "ququ": 250670,
+  "quri": 250671,
+  "qurpa": 250672,
+  "quta": 250673,
+  "quña": 250674,
+  "quñchu": 250675,
+  "raspberries": 251008,
+  "remedy": 251076,
+  "right hand": 250912,
+  "rusara": 250676,
+  "sacred": 251077,
+  "sajuna": 250677,
+  "salesman": 250850,
+  "saliva": 251098,
+  "saltpeter": 251078,
+  "samana": 250678,
+  "sanawrya": 250679,
+  "sand": 250872,
+  "sapha": 250680,
+  "saru": 250681,
+  "sarxäya": 250682,
+  "sausage": 250868,
+  "saxra": 250683,
+  "saywa": 250684,
+  "scare": 251006,
+  "scorpion": 250847,
+  "scurf": 251051,
+  "seashell": 250891,
+  "see you tomorrow": 251065,
+  "seed": 250905,
+  "serpent": 250870,
+  "seventeen": 251111,
+  "sheep": 251130,
+  "shower": 251106,
+  "sillq'i": 250685,
+  "sillu": 250686,
+  "siqusa": 250687,
+  "sira": 250688,
+  "sirwisa": 250689,
+  "sirwuyla": 250690,
+  "siwara": 250691,
+  "siwi": 250692,
+  "siwulla": 250693,
+  "sixteen": 251117,
+  "sixth": 251089,
+  "sixty": 251088,
+  "skunk": 250869,
+  "slack": 251067,
+  "smooth": 251079,
+  "snail": 250892,
+  "sojhta": 250694,
+  "sojhtatunca": 250695,
+  "sojhtïri": 250696,
+  "solitary": 250888,
+  "someone stole my passport": 251025,
+  "son-in-law": 251105,
+  "soot": 251068,
+  "sorrowful": 251055,
+  "soup": 250996,
+  "sour": 250940,
+  "sparrow": 251032,
+  "sparrow hawk": 250855,
+  "squirrel": 251145,
+  "stomach": 251042,
+  "strawberries": 251035,
+  "stupid": 250924,
+  "sullcatata": 250697,
+  "sullu": 250698,
+  "suma": 250699,
+  "sunkha": 250700,
+  "suti": 250701,
+  "suxta": 250702,
+  "swallow": 250879,
+  "sweat": 250951,
+  "sweet corn": 251009,
+  "sweet potato": 251064,
+  "swimmer": 250983,
+  "swine": 250972,
+  "t'ant'a": 250703,
+  "t'anta": 250704,
+  "t'awra": 250705,
+  "t'inkha": 250706,
+  "t'ixni": 250707,
+  "t'ojho": 250708,
+  "t'una": 250709,
+  "tadpole": 250952,
+  "taksi": 250710,
+  "taqi": 250711,
+  "tarantula": 251058,
+  "taruka": 250712,
+  "tarwi": 250713,
+  "tasteless": 251049,
+  "tatter": 251095,
+  "tawaqu": 250714,
+  "taxa": 250715,
+  "tayca": 250716,
+  "taycch'i": 250717,
+  "tayka": 250718,
+  "tea": 251104,
+  "teeth": 250982,
+  "ten thousand": 251109,
+  "tepid": 250989,
+  "testicle": 251053,
+  "thaki": 250719,
+  "thank you": 251149,
+  "thank you very much": 251150,
+  "thantaha": 250720,
+  "thantha": 250721,
+  "thaxti": 250722,
+  "thaya": 250723,
+  "the bill, please": 251060,
+  "thievish": 250994,
+  "thirteen": 251116,
+  "thirty": 251073,
+  "thoqori": 250724,
+  "thorny": 250875,
+  "three hundred": 251070,
+  "thujsa": 250725,
+  "thuri": 250726,
+  "thusa": 250727,
+  "thutha": 250728,
+  "tiliphun numirusti": 250729,
+  "tiliphunu": 250730,
+  "tiliphunuy mayt'ita": 250731,
+  "tinta kawkhankisa": 250732,
+  "tirin istasyuna kawkinkisa": 250733,
+  "tirina": 250734,
+  "tirinchi": 250735,
+  "tiwu": 250736,
+  "tixi": 250737,
+  "tiyi": 250738,
+  "to disturb": 250877,
+  "toad": 250931,
+  "toe": 251057,
+  "tollqa": 250739,
+  "tomatoes": 251108,
+  "tongue": 250981,
+  "tonqo": 250740,
+  "tucha": 250741,
+  "tuktur nisisitta": 250742,
+  "tullpa": 250743,
+  "tullqa": 250744,
+  "tumati": 250745,
+  "tunca": 250746,
+  "tuncahuaranqa": 250747,
+  "tuncallätuncani": 250748,
+  "tuncamayani": 250749,
+  "tuncapaqallquni": 250750,
+  "tuncapayajalja": 250751,
+  "tuncapayani": 250752,
+  "tuncaphesqani": 250753,
+  "tuncapusini": 250754,
+  "tuncaqimsaqallquni": 250755,
+  "tuncaquimsani": 250756,
+  "tuncasojhtani": 250757,
+  "tunka": 250758,
+  "tunqu": 250759,
+  "tunu": 250760,
+  "turunja": 250761,
+  "tustat t'ant'a": 250762,
+  "tutuca": 250763,
+  "twelve": 251113,
+  "twin": 250926,
+  "twisted": 251052,
+  "two hundred": 251021,
+  "tï": 250764,
+  "uichinca": 250765,
+  "umata": 250766,
+  "umaña": 250767,
+  "umt'añani": 250768,
+  "uncle": 250946,
+  "untamed": 250959,
+  "uqi": 250769,
+  "uritu": 250770,
+  "urpu": 250771,
+  "urqu": 250772,
+  "urt'a": 250773,
+  "uspitala": 250774,
+  "usutjamätwa": 250775,
+  "usuyiri": 250776,
+  "utajak'a": 250777,
+  "utajanana": 250778,
+  "utajanaña": 250779,
+  "uturunqu": 250780,
+  "uturuqu": 250781,
+  "uwasa": 250782,
+  "uwija": 250783,
+  "uyhua": 250784,
+  "uywa": 250785,
+  "vagina": 250904,
+  "vinegar": 251140,
+  "waiter": 250918,
+  "wajcha": 250786,
+  "wajlli": 250787,
+  "wallpa": 250788,
+  "wallq'i": 250789,
+  "wanaku": 250790,
+  "wank'u": 250791,
+  "wanku kawkinkisa": 250792,
+  "warawara": 250793,
+  "warku": 250794,
+  "warmi": 250795,
+  "warmi yaq'aña": 250796,
+  "waruru": 250797,
+  "wasara": 250798,
+  "wasita": 250799,
+  "waterfall": 251029,
+  "waxra": 250800,
+  "wayk'a": 250801,
+  "wayu": 250802,
+  "waña": 250803,
+  "weasel": 250844,
+  "what is the telephone number?": 251099,
+  "what time is it?": 250979,
+  "what would you like to drink?": 250980,
+  "where are the suitcases?": 250997,
+  "where are you from?": 250871,
+  "where can i buy a ticket?": 250965,
+  "where can i get a taxi?": 250962,
+  "where can i hire a car?": 250964,
+  "where can i park the car?": 250963,
+  "where is the bank?": 251136,
+  "where is the beach?": 251018,
+  "where is the chemist?": 251027,
+  "where is the shop?": 251101,
+  "where is the train station?": 251102,
+  "whitish": 250938,
+  "wich'inka": 250804,
+  "widow": 250907,
+  "wila": 250805,
+  "wila-chipica": 250806,
+  "wild olive": 251069,
+  "wilderness": 251138,
+  "willulli": 250807,
+  "winagri": 250808,
+  "winas taris": 250809,
+  "winas täris": 250810,
+  "winaya": 250811,
+  "wintantuqiruw qunt'añ munta": 250812,
+  "winu": 250813,
+  "winus nuchis": 250814,
+  "winus tartis": 250815,
+  "winus tiyas": 250816,
+  "wiryu wasu": 250817,
+  "wisiklita": 250818,
+  "wisk'achu": 250819,
+  "wiñaya": 250820,
+  "worn out": 251003,
+  "wound": 250992,
+  "wrist": 250857,
+  "yanapt'ita!": 250821,
+  "yanqha": 250822,
+  "yant'a": 250823,
+  "yapu": 250824,
+  "yapuchiri": 250825,
+  "yatichiri": 250826,
+  "yawri": 250827,
+  "yes": 250950,
+  "yesterday": 250917,
+  "yojhch'a": 250828,
+  "yoqa": 250829,
+  "yoqalla": 250830,
+  "young woman": 251093,
+  "yunca": 250831,
+  "yuqa": 250832,
+  "yuqalla": 250833,
+  "yuspagara": 250834,
+  "yuspajarkätam": 250835,
+  "yuspara": 250836,
+  "yuxch'a": 250837,
+  "ñicuta": 250838,
+  "ñik'uta": 250839,
+  "ñuñu": 250840
+}

config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "alvations/mt5-aym-lex",
+  "architectures": [
+    "MT5ForConditionalGeneration"
+  ],
+  "d_ff": 2048,
+  "d_kv": 64,
+  "d_model": 768,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "mt5",
+  "num_decoder_layers": 12,
+  "num_heads": 12,
+  "num_layers": 12,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "T5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.29.1",
+  "use_cache": true,
+  "vocab_size": 251152
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a608c585403de1b3b7608a0b9073217e66ce1da90c334bcb687131d655693ec8
+size 2336092213

runs/May16_09-23-39_8f80d7c5c961/1684229048.7773266/events.out.tfevents.1684229048.8f80d7c5c961.1279.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbad7328412acfabe813095ddf577a7afe666480ef57cca5d925a444c6577425
+size 6197

runs/May16_09-23-39_8f80d7c5c961/1684229053.1926608/events.out.tfevents.1684229053.8f80d7c5c961.1279.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fba5acb664ea3e7acd65e2f4700b9dad1c7b33b4c9cfd20b2067a81b74785917
+size 6197

runs/May16_09-23-39_8f80d7c5c961/events.out.tfevents.1684229048.8f80d7c5c961.1279.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36856c1582c429d62cfe724f2702ec9dcd87ef07d22ae7384df81e0f19402d7e
+size 8700

runs/May16_09-25-06_8f80d7c5c961/1684229119.5403051/events.out.tfevents.1684229119.8f80d7c5c961.1279.4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:372a969107476b3f7d7239419706407a966793a82c4ae4cc3f05f7df73a0b75c
+size 6197

runs/May16_09-25-06_8f80d7c5c961/1684229121.4736967/events.out.tfevents.1684229121.8f80d7c5c961.1279.5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa52c76e67f9338f52693bf595540a954b63007913f3a64fe14942119731fabf
+size 6197

runs/May16_09-25-06_8f80d7c5c961/events.out.tfevents.1684229119.8f80d7c5c961.1279.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:502b8beb15ff6fec1a78507b6af8ed24b83a8db3d5b591a2d3fe69bcf1236e43
+size 8700

special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef78f86560d809067d12bac6c09f19a462cb3af3f54d2b8acbba26e1433125d6
+size 4309802

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:812d584bfa2fbbd089f09abea7584511ff23db66c044460c061b10d00c3d3791
+size 16592845

tokenizer_config.json

@@ -0,0 +1,11 @@
+{
+  "additional_special_tokens": null,
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "</s>",
+  "extra_ids": 0,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "T5Tokenizer",
+  "unk_token": "<unk>"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9401d24fa242cc6b49ad27b796a580622af9892e5d9566108d919bccfa4acaae
+size 4027