add tokenizer
Browse files
runs/Apr04_22-14-02_TUF/1649135705.5762658/events.out.tfevents.1649135705.TUF.19268.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bb7837525eacf2cd323bef1cabdfffb23712b8607cc845cdd089251fa1315f89
|
3 |
+
size 4779
|
runs/Apr04_22-14-02_TUF/events.out.tfevents.1649135705.TUF.19268.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ce3339f570d11eb7dad4d5c58aa89b59b943066cd13f9e0fc4c7e2ca7b5b5e69
|
3 |
+
size 4996
|
vocab.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"stick": 680, "tear": 707, "except": 214, "in": 340, "the": 711, "winter": 824, "when": 807, "ooze": 488, "or": 490, "snow": 661, "ice": 336, "prevents": 535, "pat": 510, "up": 768, "meat": 424, "know": 366, "he": 311, "slowly": 654, "takes": 703, "short": 624, "walk": 786, "open": 489, "air": 12, "each": 190, "day": 158, "swarm": 695, "double": 181, "no": 465, "storm": 685, "usually": 775, "minus": 434, "several": 609, "buttons": 104, "dug": 189, "you": 853, "wished": 827, "to": 735, "all": 15, "about": 2, "my": 449, "grandfather": 287, "knee": 362, "feet": 236, "train": 748, "but": 103, "always": 22, "answers": 28, "banana": 48, "oil": 482, "beat": 59, "quick": 548, "brown": 98, "fox": 260, "jumps": 354, "over": 496, "lazy": 374, "dog": 175, "she": 613, "had": 295, "your": 855, "dark": 154, "suit": 690, "greasy": 289, "wash": 791, "water": 793, "year": 847, "car": 110, "giving": 275, "those": 723, "who": 813, "observe": 475, "him": 321, "pronounced": 538, "feeling": 235, "of": 477, "utmost": 776, "respect": 576, "we": 796, "have": 310, "often": 480, "urged": 770, "more": 439, "and": 26, "smoke": 657, "less": 385, "rake": 557, "swore": 701, "rave": 562, "for": 253, "sip": 643, "grow": 293, "bit": 75, "chair": 120, "one": 486, "fair": 224, "two": 761, "beta": 69, "rage": 554, "spark": 669, "race": 552, "form": 256, "farm": 226, "raid": 555, "fee": 232, "weed": 799, "bat": 53, "horn": 328, "jagged": 348, "sheet": 615, "bubble": 99, "warm": 789, "ship": 620, "hair": 296, "share": 612, "rate": 560, "goat": 280, "trace": 744, "go": 279, "alpha": 18, "left": 380, "sticks": 681, "trouble": 758, "dagger": 152, "read": 565, "long": 404, "flowing": 251, "beard": 58, "clings": 129, "his": 324, "chin": 123, "slip": 653, "glow": 277, "trade": 745, "range": 558, "stubble": 687, "floor": 248, "bug": 100, "lead": 375, "yet": 852, "still": 682, "thinks": 718, "as": 36, "swiftly": 699, "ever": 209, "jacket": 347, "born": 86, "corn": 140, "right": 583, "knew": 363, "well": 803, "is": 343, "nearly": 454, "ninety": 463, "three": 726, "years": 848, "old": 484, "feed": 233, "twice": 760, "plays": 526, "skillfully": 649, "with": 830, "zest": 857, "upon": 769, "our": 494, "small": 656, "organ": 491, "rain": 556, "park": 507, "tip": 734, "down": 182, "gadget": 267, "store": 684, "dresses": 186, "himself": 322, "an": 24, "ancient": 25, "black": 79, "frock": 262, "coat": 130, "likes": 392, "be": 55, "modern": 437, "language": 369, "torn": 741, "speaks": 672, "voice": 783, "just": 356, "cracked": 144, "quivers": 551, "trifle": 756, "trait": 749, "yes": 851, "ninetythree": 464, "little": 398, "ozone": 501, "bag": 47, "gnaw": 278, "don": 177, "ask": 37, "me": 423, "carry": 113, "oily": 483, "rag": 553, "like": 391, "that": 710, "select": 604, "bad": 46, "pile": 518, "shear": 614, "looked": 406, "noticed": 471, "men": 427, "re": 564, "back": 45, "steer": 679, "jane": 349, "may": 422, "earn": 192, "money": 438, "by": 105, "working": 837, "hard": 300, "side": 631, "got": 284, "wet": 806, "ate": 40, "scrubbed": 600, "floors": 249, "thoroughly": 722, "fruit": 263, "sigh": 633, "troop": 757, "nothing": 470, "offensive": 479, "innocence": 342, "approached": 30, "depot": 166, "light": 390, "this": 720, "was": 790, "easy": 193, "us": 772, "wish": 826, "shoot": 622, "prior": 536, "witch": 829, "feel": 234, "can": 109, "play": 523, "weekend": 801, "lip": 395, "bear": 57, "coupe": 141, "mut": 448, "fate": 228, "boot": 85, "thinner": 719, "than": 709, "am": 23, "their": 712, "house": 333, "grey": 291, "white": 811, "toot": 739, "big": 72, "baseball": 51, "players": 525, "get": 273, "traded": 746, "write": 841, "mere": 429, "whoop": 815, "why": 816, "yell": 849, "worry": 838, "silly": 636, "items": 346, "used": 774, "being": 65, "on": 485, "field": 237, "dot": 180, "error": 204, "tile": 731, "will": 819, "allow": 16, "rare": 559, "lie": 388, "pad": 503, "sleep": 651, "are": 32, "grades": 286, "higher": 319, "lower": 413, "nancy": 451, "cycle": 150, "humidity": 335, "overwhelming": 498, "there": 715, "sister": 644, "made": 415, "flowered": 250, "curtains": 148, "explore": 221, "hear": 313, "validated": 777, "acts": 7, "school": 598, "districts": 171, "galore": 269, "gathered": 271, "shells": 617, "beach": 56, "menu": 428, "try": 759, "do": 173, "best": 68, "conscious": 136, "time": 732, "they": 716, "carried": 112, "off": 478, "stretcher": 686, "sway": 696, "urgent": 771, "tie": 730, "rocks": 589, "fear": 229, "though": 724, "droop": 187, "loop": 407, "if": 337, "destroy": 167, "confidence": 135, "banks": 49, "something": 665, "economy": 196, "said": 595, "better": 70, "taking": 704, "cold": 131, "shower": 626, "cheer": 122, "rock": 588, "peer": 513, "deer": 161, "robin": 587, "wear": 798, "yellow": 850, "lily": 393, "islands": 344, "sparsely": 671, "populated": 530, "bright": 95, "sunshine": 692, "shimmers": 619, "ocean": 476, "jungle": 355, "single": 641, "xxx": 844, "briar": 94, "delta": 165, "where": 808, "were": 805, "while": 809, "away": 44, "everything": 211, "went": 804, "real": 566, "smooth": 658, "sheriff": 618, "pot": 531, "other": 493, "cases": 115, "also": 19, "under": 763, "advisement": 9, "pay": 511, "charlie": 121, "thigh": 717, "both": 87, "figures": 238, "would": 839, "later": 370, "much": 443, "hate": 309, "tried": 755, "tell": 708, "people": 514, "community": 133, "near": 452, "knot": 365, "sparkle": 670, "sprinkle": 675, "group": 292, "slay": 650, "root": 593, "snoop": 660, "forward": 258, "four": 259, "mother": 442, "sews": 611, "gingham": 274, "aprons": 31, "here": 317, "losing": 408, "replace": 574, "it": 345, "immediately": 339, "cart": 114, "take": 702, "seven": 608, "after": 10, "some": 663, "new": 459, "episode": 202, "reported": 575, "leak": 377, "fat": 227, "lick": 387, "spain": 668, "shy": 629, "echo": 195, "blow": 82, "rode": 590, "horseback": 329, "sergeant": 607, "rich": 580, "quebec": 547, "written": 842, "thought": 725, "schoolhouse": 599, "stood": 683, "empty": 199, "seeds": 603, "hem": 316, "expect": 216, "ll": 401, "bounce": 89, "week": 800, "injuries": 341, "same": 596, "leg": 381, "part": 508, "loyal": 414, "oscar": 492, "box": 90, "contained": 137, "sweaters": 697, "golf": 281, "hum": 334, "seed": 602, "museum": 445, "hires": 323, "musicians": 447, "every": 210, "evening": 207, "roll": 591, "wire": 825, "lay": 373, "wall": 787, "alimony": 14, "harms": 304, "divorced": 172, "man": 417, "wealth": 797, "mike": 430, "city": 126, "sitting": 647, "witty": 831, "sign": 635, "kitten": 361, "hark": 301, "brawn": 92, "league": 376, "five": 244, "eight": 197, "liar": 386, "lair": 367, "has": 306, "been": 62, "done": 178, "advantage": 8, "enabling": 200, "legislation": 382, "fitting": 243, "thorn": 721, "bell": 66, "rung": 594, "spit": 674, "reek": 570, "nine": 462, "six": 648, "dolphins": 176, "swam": 694, "around": 35, "boat": 83, "treehouse": 753, "above": 3, "eventually": 208, "became": 60, "rather": 561, "proud": 544, "claim": 127, "mitten": 436, "barn": 50, "pit": 519, "start": 677, "hitting": 325, "before": 63, "thursday": 728, "exam": 212, "review": 579, "formula": 257, "students": 688, "watched": 792, "out": 495, "books": 84, "very": 781, "expensive": 218, "harp": 305, "foxtrot": 261, "misguided": 435, "souls": 667, "lost": 410, "way": 795, "knitting": 364, "carl": 111, "lives": 400, "lively": 399, "home": 327, "vat": 779, "fill": 239, "good": 282, "win": 821, "juliet": 352, "zero": 856, "pretty": 534, "swing": 700, "arm": 34, "high": 318, "not": 468, "program": 537, "socialized": 662, "medicine": 426, "trespassers": 754, "prosecuted": 541, "fined": 240, "sitter": 646, "shop": 623, "nice": 460, "value": 778, "themselves": 713, "life": 389, "winners": 823, "dread": 183, "travel": 751, "only": 487, "leisure": 383, "wicker": 818, "night": 461, "geese": 272, "blend": 80, "sierra": 632, "muck": 444, "sight": 634, "doesn": 174, "hanging": 299, "chip": 124, "whitecapped": 812, "waves": 794, "covered": 143, "entire": 201, "bay": 54, "willing": 820, "experiment": 219, "first": 241, "hold": 326, "fish": 242, "across": 5, "pond": 529, "quart": 546, "dress": 184, "bread": 93, "bird": 74, "hotels": 332, "available": 43, "nearby": 453, "most": 441, "young": 854, "rise": 585, "early": 191, "morning": 440, "mill": 431, "asked": 38, "loan": 402, "fed": 231, "provides": 545, "funds": 265, "clear": 128, "slums": 655, "help": 315, "colleges": 132, "build": 101, "dormitories": 179, "fact": 222, "whole": 814, "defensive": 162, "unit": 765, "did": 169, "job": 351, "calico": 107, "cat": 117, "keep": 357, "court": 142, "took": 738, "results": 578, "disappointing": 170, "place": 521, "volunteered": 784, "jump": 353, "tread": 752, "fork": 255, "cork": 139, "rid": 581, "decision": 160, "eat": 194, "kilo": 359, "tort": 742, "bed": 61, "at": 39, "glitter": 276, "hat": 308, "bitter": 78, "snake": 659, "sew": 610, "chop": 125, "harmonize": 303, "marvelously": 421, "neat": 455, "lee": 379, "dead": 159, "aluminum": 21, "silverware": 637, "flimsy": 247, "red": 569, "zip": 858, "papa": 505, "heart": 314, "hotel": 331, "owner": 499, "shrugged": 628, "ve": 780, "kept": 358, "since": 639, "dice": 168, "ride": 582, "stagecoach": 676, "zulu": 859, "attitude": 42, "unbeatable": 762, "shred": 627, "cash": 116, "rip": 584, "rest": 577, "definitely": 163, "notch": 469, "sort": 666, "pair": 504, "shoes": 621, "nobody": 466, "really": 567, "expects": 217, "evacuate": 206, "bitten": 77, "selected": 605, "places": 522, "families": 225, "welcome": 802, "sink": 642, "meats": 425, "prospects": 543, "cupping": 147, "prospect": 542, "cutting": 149, "spending": 673, "unpleasant": 766, "any": 29, "governor": 285, "etiquette": 205, "mandates": 418, "compliance": 134, "existing": 215, "regulations": 572, "participate": 509, "athletic": 41, "activities": 6, "critical": 145, "equipment": 203, "needs": 457, "proper": 539, "maintenance": 416, "slicker": 652, "whisky": 810, "victor": 782, "bill": 73, "ray": 563, "lock": 403, "further": 266, "proposed": 540, "grants": 288, "unspecified": 767, "sum": 691, "experimental": 220, "hospitals": 330, "nut": 474, "lend": 384, "brought": 97, "yankee": 845, "received": 568, "annoying": 27, "phone": 516, "calls": 108, "sing": 640, "dart": 156, "sin": 638, "see": 601, "tango": 705, "hill": 320, "wait": 785, "work": 836, "want": 788, "catch": 118, "played": 524, "basketball": 52, "toward": 743, "law": 371, "degree": 164, "prefer": 533, "traditional": 747, "steel": 678, "kind": 360, "bought": 88, "goose": 283, "wrapped": 840, "package": 502, "hastily": 307, "wore": 835, "fleecy": 245, "wooden": 832, "wool": 833, "oh": 481, "woolen": 834, "overalls": 497, "foe": 252, "able": 1, "dance": 153, "too": 737, "greg": 290, "pick": 517, "peck": 512, "potatoes": 532, "wicked": 817, "uniform": 764, "lake": 368, "nest": 458, "sell": 606, "celebrate": 119, "brother": 96, "success": 689, "dressed": 185, "area": 33, "sweet": 698, "november": 472, "alfalfa": 13, "healthy": 312, "mush": 446, "darn": 155, "else": 198, "fails": 223, "use": 773, "force": 254, "lovely": 412, "pitting": 520, "bloat": 81, "cake": 106, "period": 515, "transition": 750, "bran": 91, "many": 419, "benefits": 67, "top": 740, "lima": 394, "bite": 76, "harm": 302, "dad": 151, "academic": 4, "bidding": 71, "shell": 616, "ticker": 729, "tin": 733, "crucified": 146, "nailed": 450, "yard": 846, "loss": 409, "owners": 500, "regularly": 571, "look": 405, "plumbing": 528, "leaks": 378, "xray": 843, "sand": 597, "ham": 297, "although": 20, "alone": 17, "survive": 693, "excellent": 213, "team": 706, "please": 527, "window": 822, "quickly": 550, "jaw": 350, "cope": 138, "drop": 188, "shout": 625, "through": 727, "ignorance": 338, "relax": 573, "hand": 298, "now": 473, "lawyers": 372, "love": 411, "millionaires": 432, "wit": 828, "sit": 645, "nod": 467, "beg": 64, "guard": 294, "gallon": 268, "gas": 270, "ned": 456, "min": 433, "paper": 506, "today": 736, "date": 157, "february": 230, "13th": 0, "litter": 397, "romeo": 592, "again": 11, "then": 714, "someplace": 664, "liquor": 396, "sicker": 630, "flicker": 246, "quicker": 549, "fun": 264, "roast": 586, "marshmallows": 420, "burner": 102, "[UNK]": 860, "[PAD]": 861}
|
|
|
1 |
+
{"t": 0, "i": 1, "m": 2, "q": 3, " ": 4, "a": 5, "s": 6, "e": 7, "p": 8, "h": 9, "w": 10, "c": 11, "3": 12, "r": 13, "k": 14, "n": 15, "z": 16, "y": 17, "l": 18, "b": 19, "o": 20, "d": 21, "f": 22, "1": 23, "g": 24, "j": 25, "v": 26, "'": 27, "x": 28, "u": 29, "[UNK]": 30, "[PAD]": 31}
|