tdw419 commited on
Commit
f49218d
·
verified ·
1 Parent(s): 8690148

Upload folder using huggingface_hub

Browse files
bilingual_tokenizer_v4/asm_tokenizer.json ADDED
@@ -0,0 +1,543 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "token2id": {
3
+ "<PAD>": 0,
4
+ "<BOS>": 1,
5
+ "<EOS>": 2,
6
+ "<NL>": 3,
7
+ "<COMMENT>": 4,
8
+ "<NUM>": 5,
9
+ "<LABEL>": 6,
10
+ "<STR>": 7,
11
+ ",": 8,
12
+ ":": 9,
13
+ "ABS": 10,
14
+ "ADD": 11,
15
+ "ADDI": 12,
16
+ "AI_AGENT": 13,
17
+ "AI_INJECT": 14,
18
+ "ALARM_CLR": 15,
19
+ "ALARM_SET": 16,
20
+ "AND": 17,
21
+ "ANDI": 18,
22
+ "ASM": 19,
23
+ "ASMSELF": 20,
24
+ "ASM_RAM": 21,
25
+ "AUDIO_PLAY": 22,
26
+ "AUDIO_STATUS": 23,
27
+ "AUDIO_STOP": 24,
28
+ "BCLR": 25,
29
+ "BEEP": 26,
30
+ "BFE": 27,
31
+ "BFI": 28,
32
+ "BGE": 29,
33
+ "BITCLR": 30,
34
+ "BITSET": 31,
35
+ "BITTEST": 32,
36
+ "BLEND": 33,
37
+ "BLENDR": 34,
38
+ "BLT": 35,
39
+ "BNOT": 36,
40
+ "BREAKPOINT": 37,
41
+ "BSET": 38,
42
+ "BTST": 39,
43
+ "CALL": 40,
44
+ "CHDIR": 41,
45
+ "CIRCLE": 42,
46
+ "CLAMP": 43,
47
+ "CLIPCLR": 44,
48
+ "CLIPSET": 45,
49
+ "CLIP_COPY": 46,
50
+ "CLIP_HISTORY": 47,
51
+ "CLIP_PASTE": 48,
52
+ "CLIP_TEXT": 49,
53
+ "CLOSE": 50,
54
+ "CMOV": 51,
55
+ "CMP": 52,
56
+ "CMPI": 53,
57
+ "CONNECT": 54,
58
+ "COPY": 55,
59
+ "CSEL": 56,
60
+ "DISCONNECT": 57,
61
+ "DIV": 58,
62
+ "DRAWTEXT": 59,
63
+ "EXEC": 60,
64
+ "EXECP": 61,
65
+ "EXIT": 62,
66
+ "FCOPY": 63,
67
+ "FILL": 64,
68
+ "FLOOD": 65,
69
+ "FMKDIR": 66,
70
+ "FONT_SELECT": 67,
71
+ "FORK": 68,
72
+ "FORMULA": 69,
73
+ "FORMULACLEAR": 70,
74
+ "FORMULAREM": 71,
75
+ "FRAME": 72,
76
+ "FSCLOSE": 73,
77
+ "FSLS": 74,
78
+ "FSOPEN": 75,
79
+ "FSREAD": 76,
80
+ "FSTAT": 77,
81
+ "FSWRITE": 78,
82
+ "FUNLINK": 79,
83
+ "GETCWD": 80,
84
+ "GETENV": 81,
85
+ "GETPID": 82,
86
+ "HALT": 83,
87
+ "HASHGET": 84,
88
+ "HASHINIT": 85,
89
+ "HASHSET": 86,
90
+ "HERMES": 87,
91
+ "HITCLR": 88,
92
+ "HITQ": 89,
93
+ "HITSET": 90,
94
+ "HTPARSE": 91,
95
+ "HTTPGET": 92,
96
+ "HYPERVISOR": 93,
97
+ "IKEY": 94,
98
+ "IMOUSE": 95,
99
+ "INV": 96,
100
+ "IOCTL": 97,
101
+ "JMP": 98,
102
+ "JNZ": 99,
103
+ "JZ": 100,
104
+ "KILL": 101,
105
+ "LDI": 102,
106
+ "LINE": 103,
107
+ "LLM": 104,
108
+ "LOAD": 105,
109
+ "LOADPNG": 106,
110
+ "LOADS": 107,
111
+ "LOADSRCIMG": 108,
112
+ "LS": 109,
113
+ "MATMUL": 110,
114
+ "MATVEC": 111,
115
+ "MAX": 112,
116
+ "MEDTEXT": 113,
117
+ "MEMCPY": 114,
118
+ "MEMSET": 115,
119
+ "MIN": 116,
120
+ "MOD": 117,
121
+ "MOUSEB": 118,
122
+ "MOUSECLICK": 119,
123
+ "MOUSEQ": 120,
124
+ "MOUSEX": 121,
125
+ "MOUSEY": 122,
126
+ "MOV": 123,
127
+ "MSGRCV": 124,
128
+ "MSGSND": 125,
129
+ "MUL": 126,
130
+ "NEG": 127,
131
+ "NET_RECV": 128,
132
+ "NET_SEND": 129,
133
+ "NOP": 130,
134
+ "NOT": 131,
135
+ "NOTE": 132,
136
+ "NPROC": 133,
137
+ "OPEN": 134,
138
+ "OR": 135,
139
+ "ORI": 136,
140
+ "PATCH": 137,
141
+ "PATCHW": 138,
142
+ "PEEK": 139,
143
+ "PIPE": 140,
144
+ "PIXEL_HISTORY": 141,
145
+ "POP": 142,
146
+ "PROCINFO": 143,
147
+ "PROCLS": 144,
148
+ "PROFILE": 145,
149
+ "PSET": 146,
150
+ "PSETI": 147,
151
+ "PTYCLOSE": 148,
152
+ "PTYOPEN": 149,
153
+ "PTYREAD": 150,
154
+ "PTYSIZE": 151,
155
+ "PTYWRITE": 152,
156
+ "PUSH": 153,
157
+ "RAND": 154,
158
+ "READ": 155,
159
+ "READLN": 156,
160
+ "RECT": 157,
161
+ "RECTF": 158,
162
+ "RELU": 159,
163
+ "REPLAY": 160,
164
+ "RET": 161,
165
+ "RETK": 162,
166
+ "ROTATE": 163,
167
+ "RUNNEXT": 164,
168
+ "SAR": 165,
169
+ "SARI": 166,
170
+ "SAVEPNG": 167,
171
+ "SCALE": 168,
172
+ "SCREENA": 169,
173
+ "SCREENP": 170,
174
+ "SCROLL": 171,
175
+ "SCRSHOT": 172,
176
+ "SEEK": 173,
177
+ "SETCAPS": 174,
178
+ "SETENV": 175,
179
+ "SETPRIORITY": 176,
180
+ "SHL": 177,
181
+ "SHLI": 178,
182
+ "SHR": 179,
183
+ "SHRI": 180,
184
+ "SHUTDOWN": 181,
185
+ "SIGNAL": 182,
186
+ "SIGSET": 183,
187
+ "SLEEP": 184,
188
+ "SMALLTEXT": 185,
189
+ "SNAP_TRACE": 186,
190
+ "SOCKRECV": 187,
191
+ "SOCKSEND": 188,
192
+ "SPAWN": 189,
193
+ "SPAWNC": 190,
194
+ "SPRANIM": 191,
195
+ "SPRBLT": 192,
196
+ "SPRFRAME": 193,
197
+ "SPRITE": 194,
198
+ "SPRITEANIM": 195,
199
+ "SPRITE_FRAME": 196,
200
+ "SPRITE_LOAD": 197,
201
+ "SPRLOAD": 198,
202
+ "STORE": 199,
203
+ "STORES": 200,
204
+ "STRCMP": 201,
205
+ "STRCPY": 202,
206
+ "STRLEN": 203,
207
+ "STRO": 204,
208
+ "SUB": 205,
209
+ "SUBI": 206,
210
+ "SYSCALL": 207,
211
+ "TEXT": 208,
212
+ "TEXTI": 209,
213
+ "TILEMAP": 210,
214
+ "TMR_GET": 211,
215
+ "TMR_WAIT": 212,
216
+ "TRACE_READ": 213,
217
+ "UNLINK": 214,
218
+ "VM_KILL": 215,
219
+ "VM_LIST": 216,
220
+ "VM_LIVE_KILL": 217,
221
+ "VM_LIVE_SPAWN": 218,
222
+ "VM_LIVE_STEP": 219,
223
+ "VM_PAUSE": 220,
224
+ "VM_RESUME": 221,
225
+ "VM_SET_BUDGET": 222,
226
+ "VM_SPAWN": 223,
227
+ "VM_STATUS": 224,
228
+ "VSTAT": 225,
229
+ "VWTXT": 226,
230
+ "WAITPID": 227,
231
+ "WINSYS": 228,
232
+ "WPIXEL": 229,
233
+ "WREAD": 230,
234
+ "WRITE": 231,
235
+ "WRITESTR": 232,
236
+ "XOR": 233,
237
+ "XORI": 234,
238
+ "YIELD": 235,
239
+ "r0": 236,
240
+ "r1": 237,
241
+ "r2": 238,
242
+ "r3": 239,
243
+ "r4": 240,
244
+ "r5": 241,
245
+ "r6": 242,
246
+ "r7": 243,
247
+ "r8": 244,
248
+ "r9": 245,
249
+ "r10": 246,
250
+ "r11": 247,
251
+ "r12": 248,
252
+ "r13": 249,
253
+ "r14": 250,
254
+ "r15": 251,
255
+ "r16": 252,
256
+ "r17": 253,
257
+ "r18": 254,
258
+ "r19": 255,
259
+ "r20": 256,
260
+ "r21": 257,
261
+ "r22": 258,
262
+ "r23": 259,
263
+ "r24": 260,
264
+ "r25": 261,
265
+ "r26": 262,
266
+ "r27": 263,
267
+ "r28": 264,
268
+ "r29": 265,
269
+ "r30": 266,
270
+ "r31": 267
271
+ },
272
+ "id2token": {
273
+ "0": "<PAD>",
274
+ "1": "<BOS>",
275
+ "2": "<EOS>",
276
+ "3": "<NL>",
277
+ "4": "<COMMENT>",
278
+ "5": "<NUM>",
279
+ "6": "<LABEL>",
280
+ "7": "<STR>",
281
+ "8": ",",
282
+ "9": ":",
283
+ "10": "ABS",
284
+ "11": "ADD",
285
+ "12": "ADDI",
286
+ "13": "AI_AGENT",
287
+ "14": "AI_INJECT",
288
+ "15": "ALARM_CLR",
289
+ "16": "ALARM_SET",
290
+ "17": "AND",
291
+ "18": "ANDI",
292
+ "19": "ASM",
293
+ "20": "ASMSELF",
294
+ "21": "ASM_RAM",
295
+ "22": "AUDIO_PLAY",
296
+ "23": "AUDIO_STATUS",
297
+ "24": "AUDIO_STOP",
298
+ "25": "BCLR",
299
+ "26": "BEEP",
300
+ "27": "BFE",
301
+ "28": "BFI",
302
+ "29": "BGE",
303
+ "30": "BITCLR",
304
+ "31": "BITSET",
305
+ "32": "BITTEST",
306
+ "33": "BLEND",
307
+ "34": "BLENDR",
308
+ "35": "BLT",
309
+ "36": "BNOT",
310
+ "37": "BREAKPOINT",
311
+ "38": "BSET",
312
+ "39": "BTST",
313
+ "40": "CALL",
314
+ "41": "CHDIR",
315
+ "42": "CIRCLE",
316
+ "43": "CLAMP",
317
+ "44": "CLIPCLR",
318
+ "45": "CLIPSET",
319
+ "46": "CLIP_COPY",
320
+ "47": "CLIP_HISTORY",
321
+ "48": "CLIP_PASTE",
322
+ "49": "CLIP_TEXT",
323
+ "50": "CLOSE",
324
+ "51": "CMOV",
325
+ "52": "CMP",
326
+ "53": "CMPI",
327
+ "54": "CONNECT",
328
+ "55": "COPY",
329
+ "56": "CSEL",
330
+ "57": "DISCONNECT",
331
+ "58": "DIV",
332
+ "59": "DRAWTEXT",
333
+ "60": "EXEC",
334
+ "61": "EXECP",
335
+ "62": "EXIT",
336
+ "63": "FCOPY",
337
+ "64": "FILL",
338
+ "65": "FLOOD",
339
+ "66": "FMKDIR",
340
+ "67": "FONT_SELECT",
341
+ "68": "FORK",
342
+ "69": "FORMULA",
343
+ "70": "FORMULACLEAR",
344
+ "71": "FORMULAREM",
345
+ "72": "FRAME",
346
+ "73": "FSCLOSE",
347
+ "74": "FSLS",
348
+ "75": "FSOPEN",
349
+ "76": "FSREAD",
350
+ "77": "FSTAT",
351
+ "78": "FSWRITE",
352
+ "79": "FUNLINK",
353
+ "80": "GETCWD",
354
+ "81": "GETENV",
355
+ "82": "GETPID",
356
+ "83": "HALT",
357
+ "84": "HASHGET",
358
+ "85": "HASHINIT",
359
+ "86": "HASHSET",
360
+ "87": "HERMES",
361
+ "88": "HITCLR",
362
+ "89": "HITQ",
363
+ "90": "HITSET",
364
+ "91": "HTPARSE",
365
+ "92": "HTTPGET",
366
+ "93": "HYPERVISOR",
367
+ "94": "IKEY",
368
+ "95": "IMOUSE",
369
+ "96": "INV",
370
+ "97": "IOCTL",
371
+ "98": "JMP",
372
+ "99": "JNZ",
373
+ "100": "JZ",
374
+ "101": "KILL",
375
+ "102": "LDI",
376
+ "103": "LINE",
377
+ "104": "LLM",
378
+ "105": "LOAD",
379
+ "106": "LOADPNG",
380
+ "107": "LOADS",
381
+ "108": "LOADSRCIMG",
382
+ "109": "LS",
383
+ "110": "MATMUL",
384
+ "111": "MATVEC",
385
+ "112": "MAX",
386
+ "113": "MEDTEXT",
387
+ "114": "MEMCPY",
388
+ "115": "MEMSET",
389
+ "116": "MIN",
390
+ "117": "MOD",
391
+ "118": "MOUSEB",
392
+ "119": "MOUSECLICK",
393
+ "120": "MOUSEQ",
394
+ "121": "MOUSEX",
395
+ "122": "MOUSEY",
396
+ "123": "MOV",
397
+ "124": "MSGRCV",
398
+ "125": "MSGSND",
399
+ "126": "MUL",
400
+ "127": "NEG",
401
+ "128": "NET_RECV",
402
+ "129": "NET_SEND",
403
+ "130": "NOP",
404
+ "131": "NOT",
405
+ "132": "NOTE",
406
+ "133": "NPROC",
407
+ "134": "OPEN",
408
+ "135": "OR",
409
+ "136": "ORI",
410
+ "137": "PATCH",
411
+ "138": "PATCHW",
412
+ "139": "PEEK",
413
+ "140": "PIPE",
414
+ "141": "PIXEL_HISTORY",
415
+ "142": "POP",
416
+ "143": "PROCINFO",
417
+ "144": "PROCLS",
418
+ "145": "PROFILE",
419
+ "146": "PSET",
420
+ "147": "PSETI",
421
+ "148": "PTYCLOSE",
422
+ "149": "PTYOPEN",
423
+ "150": "PTYREAD",
424
+ "151": "PTYSIZE",
425
+ "152": "PTYWRITE",
426
+ "153": "PUSH",
427
+ "154": "RAND",
428
+ "155": "READ",
429
+ "156": "READLN",
430
+ "157": "RECT",
431
+ "158": "RECTF",
432
+ "159": "RELU",
433
+ "160": "REPLAY",
434
+ "161": "RET",
435
+ "162": "RETK",
436
+ "163": "ROTATE",
437
+ "164": "RUNNEXT",
438
+ "165": "SAR",
439
+ "166": "SARI",
440
+ "167": "SAVEPNG",
441
+ "168": "SCALE",
442
+ "169": "SCREENA",
443
+ "170": "SCREENP",
444
+ "171": "SCROLL",
445
+ "172": "SCRSHOT",
446
+ "173": "SEEK",
447
+ "174": "SETCAPS",
448
+ "175": "SETENV",
449
+ "176": "SETPRIORITY",
450
+ "177": "SHL",
451
+ "178": "SHLI",
452
+ "179": "SHR",
453
+ "180": "SHRI",
454
+ "181": "SHUTDOWN",
455
+ "182": "SIGNAL",
456
+ "183": "SIGSET",
457
+ "184": "SLEEP",
458
+ "185": "SMALLTEXT",
459
+ "186": "SNAP_TRACE",
460
+ "187": "SOCKRECV",
461
+ "188": "SOCKSEND",
462
+ "189": "SPAWN",
463
+ "190": "SPAWNC",
464
+ "191": "SPRANIM",
465
+ "192": "SPRBLT",
466
+ "193": "SPRFRAME",
467
+ "194": "SPRITE",
468
+ "195": "SPRITEANIM",
469
+ "196": "SPRITE_FRAME",
470
+ "197": "SPRITE_LOAD",
471
+ "198": "SPRLOAD",
472
+ "199": "STORE",
473
+ "200": "STORES",
474
+ "201": "STRCMP",
475
+ "202": "STRCPY",
476
+ "203": "STRLEN",
477
+ "204": "STRO",
478
+ "205": "SUB",
479
+ "206": "SUBI",
480
+ "207": "SYSCALL",
481
+ "208": "TEXT",
482
+ "209": "TEXTI",
483
+ "210": "TILEMAP",
484
+ "211": "TMR_GET",
485
+ "212": "TMR_WAIT",
486
+ "213": "TRACE_READ",
487
+ "214": "UNLINK",
488
+ "215": "VM_KILL",
489
+ "216": "VM_LIST",
490
+ "217": "VM_LIVE_KILL",
491
+ "218": "VM_LIVE_SPAWN",
492
+ "219": "VM_LIVE_STEP",
493
+ "220": "VM_PAUSE",
494
+ "221": "VM_RESUME",
495
+ "222": "VM_SET_BUDGET",
496
+ "223": "VM_SPAWN",
497
+ "224": "VM_STATUS",
498
+ "225": "VSTAT",
499
+ "226": "VWTXT",
500
+ "227": "WAITPID",
501
+ "228": "WINSYS",
502
+ "229": "WPIXEL",
503
+ "230": "WREAD",
504
+ "231": "WRITE",
505
+ "232": "WRITESTR",
506
+ "233": "XOR",
507
+ "234": "XORI",
508
+ "235": "YIELD",
509
+ "236": "r0",
510
+ "237": "r1",
511
+ "238": "r2",
512
+ "239": "r3",
513
+ "240": "r4",
514
+ "241": "r5",
515
+ "242": "r6",
516
+ "243": "r7",
517
+ "244": "r8",
518
+ "245": "r9",
519
+ "246": "r10",
520
+ "247": "r11",
521
+ "248": "r12",
522
+ "249": "r13",
523
+ "250": "r14",
524
+ "251": "r15",
525
+ "252": "r16",
526
+ "253": "r17",
527
+ "254": "r18",
528
+ "255": "r19",
529
+ "256": "r20",
530
+ "257": "r21",
531
+ "258": "r22",
532
+ "259": "r23",
533
+ "260": "r24",
534
+ "261": "r25",
535
+ "262": "r26",
536
+ "263": "r27",
537
+ "264": "r28",
538
+ "265": "r29",
539
+ "266": "r30",
540
+ "267": "r31"
541
+ },
542
+ "vocab_size": 268
543
+ }
bilingual_tokenizer_v4/config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"text_vocab_size": 3000, "offset": 356, "char_range_end": 351}
bilingual_tokenizer_v4/text_tokenizer.json ADDED
The diff for this file is too large to render. See raw diff