mTOR_RECAP_DeepSMILES / vocab.json
aakothari's picture
Upload 9 files
68e339d verified
{
"!": 5,
"\"": 6,
"#": 7,
"$": 8,
"%": 9,
"&": 10,
"'": 11,
"(": 12,
")": 13,
"))": 261,
")))": 300,
"))))": 263,
")))))": 289,
"))))))": 269,
")))))))": 280,
"))))))))": 267,
")))))))))": 288,
"))))))))))": 282,
")))))))))))": 299,
"))))))))))))": 307,
")))))))))))))": 331,
"))))))))))))))": 349,
")))))))))))))))": 381,
"))))))))))))))))": 433,
")))))))))))))))))": 471,
"))))))))))))))))))": 611,
")))))))))))))))))))": 679,
")))))))))))))[": 748,
"))))))))))))[": 656,
")))))))))))[": 575,
"))))))))))[": 484,
")))))))))[": 453,
"))))))))[": 419,
")))))))[": 389,
"))))))=": 674,
"))))))[": 375,
")))))[": 486,
"))))[": 516,
")))[": 493,
"))=": 518,
"))[": 393,
")/": 366,
")=": 283,
")=[": 721,
")[": 275,
"*": 14,
"+": 15,
"+]": 274,
"+]%": 708,
"+])": 401,
"+]))": 462,
"+])[": 675,
"+]=": 314,
"+][": 325,
",": 16,
"-": 17,
"-]": 290,
"-])": 360,
"-]))": 317,
"-])))": 545,
"-]))))": 628,
"-]))))))))": 594,
"-])))))))))": 622,
"-]))))))))))": 635,
"-]))[": 623,
"-])=": 562,
"-])[": 547,
".": 18,
"/": 19,
"/[": 532,
"0": 20,
"1": 21,
"10": 292,
"106": 350,
"11": 441,
"12": 572,
"13": 411,
"14": 552,
"2": 22,
"3": 23,
"4": 24,
"5": 25,
"58": 648,
"59": 492,
"6": 26,
"69": 332,
"7": 27,
"8": 28,
"85": 582,
"9": 29,
"95": 398,
"96": 354,
":": 30,
";": 31,
"<": 32,
"</s>": 1,
"<mask>": 3,
"<pad>": 2,
"<s>": 0,
"<unk>": 4,
"=": 33,
"=[": 402,
">": 34,
"?": 35,
"@": 36,
"@@": 271,
"@@]": 334,
"@@]=": 505,
"@]": 328,
"@]=": 563,
"A": 37,
"B": 38,
"Br": 308,
"C": 39,
"CC": 264,
"CCC": 270,
"CCCC": 286,
"CCCCC": 326,
"CCCCCC": 367,
"CCCCCCC": 519,
"CCCCCCCC": 627,
"CCCCCNC": 716,
"CCCCN": 469,
"CCCCNC": 446,
"CCCCO": 529,
"CCCN": 353,
"CCCNC": 335,
"CCCNS": 531,
"CCCO": 337,
"CCCOC": 670,
"CCCOCC": 510,
"CCCS": 499,
"CCCSCC": 707,
"CCCcccccc": 501,
"CCF": 614,
"CCN": 355,
"CCNC": 313,
"CCNCC": 384,
"CCNCCC": 693,
"CCNCCNC": 735,
"CCNCcccccc": 739,
"CCNS": 633,
"CCNcccccc": 596,
"CCO": 305,
"CCOC": 339,
"CCOCC": 448,
"CCOccc": 759,
"CCOcccc": 477,
"CCOccccC": 615,
"CCOccccNC": 598,
"CCOccccc": 763,
"CCOcccccc": 429,
"CCS": 345,
"CCSC": 496,
"CCccC": 671,
"CCccc": 720,
"CCcccc": 487,
"CCccccC": 681,
"CCccccF": 757,
"CCccccNC": 746,
"CCcccccc": 385,
"CCcn": 626,
"CCcnc": 620,
"CCcncC": 753,
"CCcnnC": 535,
"CCl": 509,
"CCn": 576,
"CCnc": 460,
"CCncC": 730,
"CCncc": 522,
"CCnn": 703,
"CCnncC": 667,
"CF": 371,
"CN": 306,
"CNC": 294,
"CNCC": 338,
"CNCCC": 428,
"CNCCCC": 599,
"CNCCNC": 561,
"CNCCO": 610,
"CNCCOCC": 602,
"CNCcccccc": 520,
"CNS": 498,
"CNcccc": 467,
"CNcccccc": 475,
"CO": 279,
"COC": 324,
"COCC": 423,
"COCCC": 715,
"COCCCNC": 687,
"COCCN": 725,
"COCCNC": 568,
"COCCO": 698,
"COccc": 408,
"COcccC": 473,
"COcccOC": 650,
"COcccc": 315,
"COccccBr": 737,
"COccccC": 374,
"COccccCC": 658,
"COccccCNC": 587,
"COccccCl": 546,
"COccccF": 578,
"COccccNC": 438,
"COccccOC": 542,
"COccccS": 604,
"COccccc": 426,
"COcccccC": 456,
"COcccccCNC": 755,
"COcccccNC": 659,
"COcccccc": 342,
"CS": 316,
"CSC": 472,
"CSCC": 581,
"CScccccc": 586,
"CScnnc": 607,
"Cc": 432,
"Ccc": 319,
"CccC": 449,
"CccCl": 676,
"CccF": 641,
"Cccc": 330,
"CcccBr": 696,
"CcccC": 361,
"CcccCl": 588,
"CcccF": 512,
"CcccNC": 476,
"Ccccc": 291,
"CccccBr": 685,
"CccccC": 352,
"CccccCC": 688,
"CccccCNC": 584,
"CccccCl": 461,
"CccccF": 437,
"CccccNC": 392,
"CccccO": 543,
"CccccOC": 654,
"CccccOCC": 728,
"CccccS": 483,
"Cccccc": 356,
"CcccccC": 416,
"CcccccCl": 669,
"CcccccF": 613,
"CcccccNC": 478,
"CcccccO": 734,
"Ccccccc": 310,
"Ccccccn": 617,
"Cccccnc": 489,
"Ccccco": 555,
"Cccccs": 514,
"Ccccn": 701,
"Ccccnc": 683,
"Ccccncc": 636,
"Ccccsc": 569,
"Cccn": 651,
"Cccnc": 580,
"Cccncc": 743,
"Cccnn": 711,
"Cccsc": 474,
"CccscC": 722,
"Ccn": 346,
"Ccnc": 391,
"CcncC": 464,
"Ccncc": 691,
"Ccncccccc": 661,
"Ccnccn": 762,
"Ccnn": 415,
"CcnnC": 564,
"Ccnnc": 605,
"Ccnoc": 624,
"CcnocC": 500,
"Cl": 295,
"Cn": 450,
"Cnc": 455,
"CncC": 717,
"Cncc": 407,
"CnccC": 642,
"Cncccn": 744,
"Cnccnc": 729,
"Cnn": 482,
"Cnnc": 709,
"Cnnccc": 756,
"D": 40,
"E": 41,
"F": 42,
"FCF": 690,
"G": 43,
"H": 44,
"I": 45,
"J": 46,
"K": 47,
"L": 48,
"M": 49,
"N": 50,
"NC": 268,
"NCC": 277,
"NCCC": 301,
"NCCCC": 329,
"NCCCCC": 369,
"NCCCCCC": 425,
"NCCCCCCC": 673,
"NCCCCCCCC": 754,
"NCCCNC": 508,
"NCCCO": 653,
"NCCCcccccc": 732,
"NCCF": 644,
"NCCN": 454,
"NCCNC": 347,
"NCCNCC": 612,
"NCCNCcccccc": 761,
"NCCNS": 597,
"NCCNcccc": 758,
"NCCNcccccc": 579,
"NCCO": 418,
"NCCOC": 513,
"NCCOCC": 357,
"NCCS": 668,
"NCCcc": 751,
"NCCcccccc": 443,
"NCN": 680,
"NCccc": 752,
"NCcccc": 538,
"NCccccC": 571,
"NCccccCl": 704,
"NCccccF": 591,
"NCcccccc": 386,
"NCcccccn": 747,
"NCccccnc": 634,
"NCcccco": 551,
"NCccccs": 705,
"NH": 278,
"NN": 436,
"NNC": 382,
"NO": 719,
"NS": 377,
"NSC": 736,
"Ncc": 517,
"NccC": 537,
"Nccc": 442,
"NcccC": 465,
"NcccCl": 565,
"NcccF": 666,
"Ncccc": 303,
"NccccBr": 590,
"NccccC": 387,
"NccccCC": 694,
"NccccCl": 431,
"NccccF": 395,
"NccccNC": 528,
"NccccOC": 660,
"NccccS": 589,
"Nccccc": 410,
"NcccccC": 414,
"NcccccCF": 699,
"NcccccCl": 485,
"NcccccF": 639,
"NcccccNC": 662,
"Ncccccc": 323,
"Nccccnc": 603,
"Ncccnn": 718,
"Nccn": 723,
"Nccnn": 700,
"Ncn": 424,
"Ncnc": 495,
"NcncC": 706,
"Ncncc": 710,
"Ncncccc": 727,
"Ncnccs": 618,
"Ncncnc": 684,
"Ncnnc": 577,
"Ncsccc": 745,
"O": 51,
"OC": 287,
"OCC": 297,
"OCCC": 600,
"OCCCC": 629,
"OCCCCC": 682,
"OCCCO": 631,
"OCCF": 556,
"OCCO": 405,
"OCF": 466,
"OCO": 394,
"OCcccccc": 567,
"Occcc": 697,
"OccccC": 609,
"OccccCl": 695,
"OccccF": 738,
"Occcccc": 406,
"P": 52,
"Q": 53,
"R": 54,
"S": 55,
"SC": 333,
"SCC": 370,
"Scccccc": 583,
"Scnnc": 616,
"T": 56,
"U": 57,
"V": 58,
"W": 59,
"X": 60,
"Y": 61,
"Z": 62,
"[": 63,
"\\": 64,
"\\[": 649,
"]": 65,
"]%": 412,
"][": 320,
"^": 66,
"_": 67,
"`": 68,
"a": 69,
"b": 70,
"c": 71,
"cBr": 534,
"cC": 281,
"cCC": 657,
"cCF": 741,
"cCl": 327,
"cF": 362,
"cN": 549,
"cNC": 417,
"cO": 533,
"cOC": 409,
"cS": 523,
"cc": 262,
"ccBr": 550,
"ccC": 336,
"ccCl": 427,
"ccF": 400,
"ccN": 678,
"ccNC": 526,
"ccO": 677,
"ccOC": 511,
"ccc": 276,
"cccBr": 494,
"cccC": 309,
"cccCF": 637,
"cccCl": 376,
"cccF": 378,
"cccNC": 451,
"cccOC": 536,
"cccS": 574,
"cccc": 265,
"ccccBr": 390,
"ccccC": 284,
"ccccCC": 548,
"ccccCF": 524,
"ccccCNC": 515,
"ccccCl": 304,
"ccccF": 318,
"ccccN": 601,
"ccccNC": 358,
"ccccNCC": 713,
"ccccO": 435,
"ccccOC": 403,
"ccccOCC": 647,
"ccccOCF": 686,
"ccccS": 447,
"ccccSC": 652,
"ccccc": 296,
"cccccBr": 502,
"cccccC": 363,
"cccccCF": 541,
"cccccCl": 399,
"cccccF": 373,
"cccccNC": 383,
"cccccO": 645,
"cccccOC": 530,
"cccccS": 619,
"cccccc": 272,
"ccccccc": 632,
"cccccccc": 458,
"ccccccccc": 503,
"cccccccccc": 592,
"cccccn": 368,
"ccccn": 640,
"ccccnc": 351,
"cccco": 359,
"ccccs": 344,
"cccn": 463,
"cccnc": 497,
"cccncc": 379,
"cccnn": 664,
"cccoc": 540,
"cccsc": 422,
"ccn": 348,
"ccnC": 608,
"ccnc": 440,
"ccncC": 714,
"ccncc": 430,
"ccnccn": 595,
"ccnn": 404,
"ccnnC": 506,
"ccs": 468,
"ccsc": 444,
"ccscC": 712,
"ccscNC": 646,
"ccscc": 764,
"cn": 273,
"cnc": 302,
"cncC": 372,
"cncCC": 663,
"cncCCC": 665,
"cncNC": 766,
"cncc": 311,
"cnccC": 544,
"cnccc": 672,
"cncccc": 445,
"cncccccc": 365,
"cncccn": 490,
"cnccn": 525,
"cnccs": 420,
"cncn": 573,
"cncnc": 507,
"cncncc": 724,
"cncnn": 731,
"cnn": 322,
"cnnC": 470,
"cnnc": 340,
"cnncC": 504,
"cnncNC": 557,
"cnncSCC": 750,
"cnncccccn": 749,
"cnncn": 479,
"cnnn": 733,
"cnnnn": 553,
"cnoc": 480,
"cnocC": 539,
"co": 488,
"cs": 343,
"csc": 558,
"cscC": 726,
"csccc": 452,
"cscnc": 740,
"d": 72,
"e": 73,
"f": 74,
"g": 75,
"h": 76,
"i": 77,
"j": 78,
"k": 79,
"l": 80,
"m": 81,
"n": 82,
"nC": 397,
"nCC": 570,
"nH": 293,
"nc": 285,
"ncC": 388,
"ncN": 655,
"ncS": 742,
"ncSCC": 689,
"ncc": 312,
"nccC": 606,
"nccc": 702,
"ncccc": 521,
"ncccccc": 421,
"ncccn": 457,
"nccnc": 481,
"ncn": 396,
"ncnc": 527,
"ncncccccc": 765,
"ncncn": 491,
"ncnnc": 643,
"ncnnn": 638,
"nn": 298,
"nnC": 554,
"nnc": 380,
"nncC": 439,
"nnccc": 630,
"nnn": 459,
"nnnnc": 621,
"no": 364,
"noc": 593,
"o": 83,
"oc": 341,
"ocC": 413,
"occc": 625,
"occcccc": 559,
"on": 434,
"onc": 692,
"p": 84,
"q": 85,
"r": 86,
"s": 87,
"sc": 321,
"scC": 560,
"scNC": 760,
"scc": 566,
"sccc": 585,
"t": 88,
"u": 89,
"v": 90,
"w": 91,
"x": 92,
"y": 93,
"z": 94,
"{": 95,
"|": 96,
"}": 97,
"~": 98,
"¡": 99,
"¢": 100,
"£": 101,
"¤": 102,
"¥": 103,
"¦": 104,
"§": 105,
"¨": 106,
"©": 107,
"ª": 108,
"«": 109,
"¬": 110,
"®": 111,
"¯": 112,
"°": 113,
"±": 114,
"²": 115,
"³": 116,
"´": 117,
"µ": 118,
"¶": 119,
"·": 120,
"¸": 121,
"¹": 122,
"º": 123,
"»": 124,
"¼": 125,
"½": 126,
"¾": 127,
"¿": 128,
"À": 129,
"Á": 130,
"Â": 131,
"Ã": 132,
"Ä": 133,
"Å": 134,
"Æ": 135,
"Ç": 136,
"È": 137,
"É": 138,
"Ê": 139,
"Ë": 140,
"Ì": 141,
"Í": 142,
"Î": 143,
"Ï": 144,
"Ð": 145,
"Ñ": 146,
"Ò": 147,
"Ó": 148,
"Ô": 149,
"Õ": 150,
"Ö": 151,
"×": 152,
"Ø": 153,
"Ù": 154,
"Ú": 155,
"Û": 156,
"Ü": 157,
"Ý": 158,
"Þ": 159,
"ß": 160,
"à": 161,
"á": 162,
"â": 163,
"ã": 164,
"ä": 165,
"å": 166,
"æ": 167,
"ç": 168,
"è": 169,
"é": 170,
"ê": 171,
"ë": 172,
"ì": 173,
"í": 174,
"î": 175,
"ï": 176,
"ð": 177,
"ñ": 178,
"ò": 179,
"ó": 180,
"ô": 181,
"õ": 182,
"ö": 183,
"÷": 184,
"ø": 185,
"ù": 186,
"ú": 187,
"û": 188,
"ü": 189,
"ý": 190,
"þ": 191,
"ÿ": 192,
"Ā": 193,
"ā": 194,
"Ă": 195,
"ă": 196,
"Ą": 197,
"ą": 198,
"Ć": 199,
"ć": 200,
"Ĉ": 201,
"ĉ": 202,
"Ċ": 203,
"ċ": 204,
"Č": 205,
"č": 206,
"čĊ": 266,
"Ď": 207,
"ď": 208,
"Đ": 209,
"đ": 210,
"Ē": 211,
"ē": 212,
"Ĕ": 213,
"ĕ": 214,
"Ė": 215,
"ė": 216,
"Ę": 217,
"ę": 218,
"Ě": 219,
"ě": 220,
"Ĝ": 221,
"ĝ": 222,
"Ğ": 223,
"ğ": 224,
"Ġ": 225,
"ġ": 226,
"Ģ": 227,
"ģ": 228,
"Ĥ": 229,
"ĥ": 230,
"Ħ": 231,
"ħ": 232,
"Ĩ": 233,
"ĩ": 234,
"Ī": 235,
"ī": 236,
"Ĭ": 237,
"ĭ": 238,
"Į": 239,
"į": 240,
"İ": 241,
"ı": 242,
"IJ": 243,
"ij": 244,
"Ĵ": 245,
"ĵ": 246,
"Ķ": 247,
"ķ": 248,
"ĸ": 249,
"Ĺ": 250,
"ĺ": 251,
"Ļ": 252,
"ļ": 253,
"Ľ": 254,
"ľ": 255,
"Ŀ": 256,
"ŀ": 257,
"Ł": 258,
"ł": 259,
"Ń": 260
}