mini / vocab.json
StrongDev2024's picture
Add Mini checkpoint
d3166c0 verified
Raw History Blame Contribute Delete
8.32 kB
{
"<pad>": 0,
"<unk>": 1,
"<user>": 2,
"<bot>": 3,
"<end>": 4,
"hello": 5,
"hi": 6,
",": 7,
"i": 8,
"am": 9,
"mini": 10,
".": 11,
"what": 12,
"do": 13,
"you": 14,
"want": 15,
"to": 16,
"talk": 17,
"about": 18,
"?": 19,
"hey": 20,
"there": 21,
"good": 22,
"morning": 23,
"how": 24,
"can": 25,
"help": 26,
"evening": 27,
"is": 28,
"your": 29,
"name": 30,
"my": 31,
"who": 32,
"are": 33,
"a": 34,
"small": 35,
"chatbot": 36,
"trained": 37,
"from": 38,
"scratch": 39,
"tell": 40,
"me": 41,
"doing": 42,
"well": 43,
"today": 44,
"ok": 45,
"glad": 46,
"hear": 47,
"that": 48,
"fine": 49,
"great": 50,
"not": 51,
"bad": 52,
"chat": 53,
"simple": 54,
"things": 55,
"like": 56,
"greetings": 57,
"names": 58,
"and": 59,
"questions": 60,
"short": 61,
"question": 62,
"will": 63,
"try": 64,
"answer": 65,
"need": 66,
"thank": 67,
"welcome": 68,
"thanks": 69,
"bye": 70,
"goodbye": 71,
"later": 72,
"see": 73,
"language": 74,
"speak": 75,
"english": 76,
"yes": 77,
"made": 78,
"was": 79,
"on": 80,
"dialogues": 81,
"where": 82,
"come": 83,
"1": 84,
"+": 85,
"2": 86,
"4": 87,
"3": 88,
"6": 89,
"color": 90,
"the": 91,
"sky": 92,
"blue": 93,
"grass": 94,
"green": 95,
"snow": 96,
"white": 97,
"pizza": 98,
"food": 99,
"cats": 100,
"dogs": 101,
"favorite": 102,
"joke": 103,
"why": 104,
"did": 105,
"robot": 106,
"sit": 107,
"down": 108,
"it": 109,
"had": 110,
"byte": 111,
"eat": 112,
"say": 113,
"something": 114,
"fun": 115,
"capital": 116,
"of": 117,
"france": 118,
"know": 119,
"yet": 120,
"ask": 121,
"quantum": 122,
"physics": 123,
"history": 124,
"write": 125,
"code": 126,
"soon": 127,
"nice": 128,
"meet": 129,
"too": 130,
"night": 131,
"sleep": 132,
"old": 133,
"new": 134,
"so": 135,
"very": 136,
"young": 137,
"live": 138,
"inside": 139,
"this": 140,
"program": 141,
"computer": 142,
"now": 143,
"understand": 144,
"animal": 145,
"cat": 146,
"sun": 147,
"yellow": 148,
"banana": 149,
"many": 150,
"legs": 151,
"does": 152,
"dog": 153,
"have": 154,
"has": 155,
"four": 156,
"8": 157,
"5": 158,
"10": 159,
"20": 160,
"count": 161,
"three": 162,
"one": 163,
"two": 164,
"weather": 165,
"cannot": 166,
"outside": 167,
"raining": 168,
"sad": 169,
"sorry": 170,
"here": 171,
"happy": 172,
"wonderful": 173,
"tired": 174,
"should": 175,
"rest": 176,
"water": 177,
"clear": 178,
"drink": 179,
"hot": 180,
"day": 181,
"no": 182,
"but": 183,
"12": 184,
"7": 185,
"14": 186,
"an": 187,
"apple": 188,
"red": 189,
"days": 190,
"in": 191,
"week": 192,
"seven": 193,
"opposite": 194,
"cold": 195,
"up": 196,
"number": 197,
"story": 198,
"sam": 199,
"kite": 200,
"wind": 201,
"took": 202,
"into": 203,
"tree": 204,
"helped": 205,
"climb": 206,
"free": 207,
"they": 208,
"carried": 209,
"home": 210,
"before": 211,
"named": 212,
"pip": 213,
"walk": 214,
"walked": 215,
"river": 216,
"happened": 217,
"next": 218,
"mended": 219,
"bent": 220,
"corner": 221,
"gentle": 222,
"flew": 223,
"stayed": 224,
"watched": 225,
"window": 226,
"another": 227,
"went": 228,
"market": 229,
"bought": 230,
"apples": 231,
"desk": 232,
"back": 233,
"noon": 234,
"he": 235,
"held": 236,
"string": 237,
"with": 238,
"both": 239,
"hands": 240,
"ran": 241,
"across": 242,
"strong": 243,
"climbed": 244,
"over": 245,
"fence": 246,
"stood": 247,
"by": 248,
"gate": 249,
"dance": 250,
"pulled": 251,
"harder": 252,
"slipped": 253,
"'": 254,
"s": 255,
"hand": 256,
"at": 257,
"end": 258,
"field": 259,
"caught": 260,
"high": 261,
"branch": 262,
"hung": 263,
"thin": 264,
"rope": 265,
"stopped": 266,
"running": 267,
"looked": 268,
"reach": 269,
"lowest": 270,
"close": 271,
"ground": 272,
"still": 273,
"while": 274,
"moved": 275,
"slowly": 276,
"leaves": 277,
"brushed": 278,
"his": 279,
"face": 280,
"bird": 281,
"hopped": 282,
"away": 283,
"reached": 284,
"lifted": 285,
"paper": 286,
"together": 287,
"against": 288,
"chest": 289,
"wound": 290,
"neat": 291,
"loop": 292,
"sat": 293,
"ate": 294,
"each": 295,
"said": 296,
"would": 297,
"fly": 298,
"again": 299,
"after": 300,
"grew": 301,
"quiet": 302,
"along": 303,
"path": 304,
"were": 305,
"left": 306,
"barked": 307,
"once": 308,
"yard": 309,
"then": 310,
"lay": 311,
"lights": 312,
"came": 313,
"windows": 314,
"put": 315,
"chair": 316,
"door": 317,
"piece": 318,
"same": 319,
"time": 320,
"let": 321,
"go": 322,
"rose": 323,
"steadied": 324,
"air": 325,
"laughed": 326,
"counted": 327,
"turns": 328,
"she": 329,
"likes": 330,
"warm": 331,
"places": 332,
"jumps": 333,
"onto": 334,
"sits": 335,
"beside": 336,
"folds": 337,
"her": 338,
"paws": 339,
"shuts": 340,
"eyes": 341,
"works": 342,
"types": 343,
"lines": 344,
"read": 345,
"them": 346,
"listens": 347,
"sound": 348,
"keys": 349,
"pours": 350,
"bowl": 351,
"drinks": 352,
"looks": 353,
"birds": 354,
"wire": 355,
"ears": 356,
"turn": 357,
"toward": 358,
"stays": 359,
"knows": 360,
"shut": 361,
"safe": 362,
"drinking": 363,
"washes": 364,
"paw": 365,
"afternoon": 366,
"moves": 367,
"floor": 368,
"follows": 369,
"patch": 370,
"light": 371,
"sleeps": 372,
"until": 373,
"reaches": 374,
"wall": 375,
"walks": 376,
"kitchen": 377,
"even": 378,
"when": 379,
"full": 380,
"says": 381,
"answers": 382,
"move": 383,
"set": 384,
"climbs": 385,
"watches": 386,
"sometimes": 387,
"hiss": 388,
"only": 389,
"tail": 390,
"house": 391,
"returns": 392,
"curls": 393,
"off": 394,
"room": 395,
"goes": 396,
"dark": 397,
"sunday": 398,
"narrow": 399,
"lined": 400,
"tall": 401,
"stones": 402,
"as": 403,
"found": 404,
"stone": 405,
"gray": 406,
"stripe": 407,
"flat": 408,
"for": 409,
"skipping": 410,
"kept": 411,
"pocket": 412,
"slow": 413,
"shone": 414,
"touched": 415,
"threw": 416,
"skipped": 417,
"twice": 418,
"sank": 419,
"tried": 420,
"times": 421,
"splash": 422,
"duck": 423,
"swam": 424,
"past": 425,
"care": 426,
"log": 427,
"shared": 428,
"bread": 429,
"asked": 430,
"moving": 431,
"sea": 432,
"drank": 433,
"bottle": 434,
"clouds": 435,
"gathered": 436,
"rain": 437,
"fell": 438,
"faded": 439,
"striped": 440,
"nothing": 441,
"memory": 442,
"skips": 443,
"saw": 444,
"stuck": 445,
"empty": 446,
"watching": 447,
"opened": 448,
"jumped": 449,
"rubbed": 450,
"leg": 451,
"table": 452,
"told": 453,
"everyone": 454,
"monday": 455,
"road": 456,
"bright": 457,
"cool": 458,
"bag": 459,
"list": 460,
"square": 461,
"row": 462,
"tables": 463,
"under": 464,
"long": 465,
"cloth": 466,
"roof": 467,
"woman": 468,
"sold": 469,
"wooden": 470,
"box": 471,
"picked": 472,
"paid": 473,
"coins": 474,
"loaves": 475,
"round": 476,
"brown": 477,
"loaf": 478,
"share": 479,
"way": 480,
"boy": 481,
"smaller": 482,
"than": 483,
"look": 484,
"nodded": 485,
"weak": 486,
"man": 487,
"smooth": 488,
"buy": 489,
"already": 490,
"slept": 491,
"wake": 492,
"enough": 493,
"turned": 494,
"sweet": 495,
"gates": 496,
"passed": 497,
"their": 498,
"own": 499,
"just": 500,
"board": 501,
"its": 502,
"could": 503,
"be": 504,
"seen": 505,
"smelled": 506
}