Generates random English words for pure text benchmarking.
| 325 | |
| 326 | |
| 327 | class RandomTextDataset(BenchmarkDataset): |
| 328 | """ |
| 329 | Generates random English words for pure text benchmarking. |
| 330 | """ |
| 331 | |
| 332 | # Common English words vocabulary |
| 333 | COMMON_WORDS = [ |
| 334 | "the", |
| 335 | "be", |
| 336 | "to", |
| 337 | "of", |
| 338 | "and", |
| 339 | "a", |
| 340 | "in", |
| 341 | "that", |
| 342 | "have", |
| 343 | "i", |
| 344 | "it", |
| 345 | "for", |
| 346 | "not", |
| 347 | "on", |
| 348 | "with", |
| 349 | "he", |
| 350 | "as", |
| 351 | "you", |
| 352 | "do", |
| 353 | "at", |
| 354 | "this", |
| 355 | "but", |
| 356 | "his", |
| 357 | "by", |
| 358 | "from", |
| 359 | "they", |
| 360 | "we", |
| 361 | "say", |
| 362 | "her", |
| 363 | "she", |
| 364 | "or", |
| 365 | "an", |
| 366 | "will", |
| 367 | "my", |
| 368 | "one", |
| 369 | "all", |
| 370 | "would", |
| 371 | "there", |
| 372 | "their", |
| 373 | "what", |
| 374 | "so", |
| 375 | "up", |
| 376 | "out", |
| 377 | "if", |
| 378 | "about", |
| 379 | "who", |
| 380 | "get", |
| 381 | "which", |
| 382 | "go", |
| 383 | "me", |
| 384 | "when", |