Shawn Tan
commited on
Commit
·
f5eaf34
1
Parent(s):
bf62d95
Cleanup config.
Browse files- config.json +40 -80
config.json
CHANGED
|
@@ -308,8 +308,7 @@
|
|
| 308 |
"attention_multiplier": 0.0078125,
|
| 309 |
"dropout": 0,
|
| 310 |
"num_key_value_heads": 4,
|
| 311 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 312 |
-
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"add_bias": false,
|
|
@@ -317,8 +316,7 @@
|
|
| 317 |
"attention_multiplier": 0.0078125,
|
| 318 |
"dropout": 0,
|
| 319 |
"num_key_value_heads": 4,
|
| 320 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 321 |
-
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"add_bias": false,
|
|
@@ -326,8 +324,7 @@
|
|
| 326 |
"attention_multiplier": 0.0078125,
|
| 327 |
"dropout": 0,
|
| 328 |
"num_key_value_heads": 4,
|
| 329 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 330 |
-
|
| 331 |
},
|
| 332 |
{
|
| 333 |
"add_bias": false,
|
|
@@ -335,8 +332,7 @@
|
|
| 335 |
"attention_multiplier": 0.0078125,
|
| 336 |
"dropout": 0,
|
| 337 |
"num_key_value_heads": 4,
|
| 338 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 339 |
-
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"add_bias": false,
|
|
@@ -344,8 +340,7 @@
|
|
| 344 |
"attention_multiplier": 0.0078125,
|
| 345 |
"dropout": 0,
|
| 346 |
"num_key_value_heads": 4,
|
| 347 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 348 |
-
|
| 349 |
},
|
| 350 |
{
|
| 351 |
"add_bias": false,
|
|
@@ -353,8 +348,7 @@
|
|
| 353 |
"attention_multiplier": 0.0078125,
|
| 354 |
"dropout": 0,
|
| 355 |
"num_key_value_heads": 4,
|
| 356 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 357 |
-
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"add_bias": false,
|
|
@@ -362,8 +356,7 @@
|
|
| 362 |
"attention_multiplier": 0.0078125,
|
| 363 |
"dropout": 0,
|
| 364 |
"num_key_value_heads": 4,
|
| 365 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 366 |
-
|
| 367 |
},
|
| 368 |
{
|
| 369 |
"add_bias": false,
|
|
@@ -371,8 +364,7 @@
|
|
| 371 |
"attention_multiplier": 0.0078125,
|
| 372 |
"dropout": 0,
|
| 373 |
"num_key_value_heads": 4,
|
| 374 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 375 |
-
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"add_bias": false,
|
|
@@ -380,8 +372,7 @@
|
|
| 380 |
"attention_multiplier": 0.0078125,
|
| 381 |
"dropout": 0,
|
| 382 |
"num_key_value_heads": 4,
|
| 383 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 384 |
-
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"add_bias": false,
|
|
@@ -389,8 +380,7 @@
|
|
| 389 |
"attention_multiplier": 0.0078125,
|
| 390 |
"dropout": 0,
|
| 391 |
"num_key_value_heads": 4,
|
| 392 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 393 |
-
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"add_bias": false,
|
|
@@ -398,8 +388,7 @@
|
|
| 398 |
"attention_multiplier": 0.0078125,
|
| 399 |
"dropout": 0,
|
| 400 |
"num_key_value_heads": 4,
|
| 401 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 402 |
-
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"add_bias": false,
|
|
@@ -407,8 +396,7 @@
|
|
| 407 |
"attention_multiplier": 0.0078125,
|
| 408 |
"dropout": 0,
|
| 409 |
"num_key_value_heads": 4,
|
| 410 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 411 |
-
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"add_bias": false,
|
|
@@ -416,8 +404,7 @@
|
|
| 416 |
"attention_multiplier": 0.0078125,
|
| 417 |
"dropout": 0,
|
| 418 |
"num_key_value_heads": 4,
|
| 419 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 420 |
-
|
| 421 |
},
|
| 422 |
{
|
| 423 |
"add_bias": false,
|
|
@@ -425,8 +412,7 @@
|
|
| 425 |
"attention_multiplier": 0.0078125,
|
| 426 |
"dropout": 0,
|
| 427 |
"num_key_value_heads": 4,
|
| 428 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 429 |
-
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"add_bias": false,
|
|
@@ -434,8 +420,7 @@
|
|
| 434 |
"attention_multiplier": 0.0078125,
|
| 435 |
"dropout": 0,
|
| 436 |
"num_key_value_heads": 4,
|
| 437 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 438 |
-
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"add_bias": false,
|
|
@@ -443,8 +428,7 @@
|
|
| 443 |
"attention_multiplier": 0.0078125,
|
| 444 |
"dropout": 0,
|
| 445 |
"num_key_value_heads": 4,
|
| 446 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 447 |
-
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"add_bias": false,
|
|
@@ -452,8 +436,7 @@
|
|
| 452 |
"attention_multiplier": 0.0078125,
|
| 453 |
"dropout": 0,
|
| 454 |
"num_key_value_heads": 4,
|
| 455 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 456 |
-
|
| 457 |
},
|
| 458 |
{
|
| 459 |
"add_bias": false,
|
|
@@ -461,8 +444,7 @@
|
|
| 461 |
"attention_multiplier": 0.0078125,
|
| 462 |
"dropout": 0,
|
| 463 |
"num_key_value_heads": 4,
|
| 464 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 465 |
-
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"add_bias": false,
|
|
@@ -470,8 +452,7 @@
|
|
| 470 |
"attention_multiplier": 0.0078125,
|
| 471 |
"dropout": 0,
|
| 472 |
"num_key_value_heads": 4,
|
| 473 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 474 |
-
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"add_bias": false,
|
|
@@ -479,8 +460,7 @@
|
|
| 479 |
"attention_multiplier": 0.0078125,
|
| 480 |
"dropout": 0,
|
| 481 |
"num_key_value_heads": 4,
|
| 482 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 483 |
-
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"add_bias": false,
|
|
@@ -488,8 +468,7 @@
|
|
| 488 |
"attention_multiplier": 0.0078125,
|
| 489 |
"dropout": 0,
|
| 490 |
"num_key_value_heads": 4,
|
| 491 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 492 |
-
|
| 493 |
},
|
| 494 |
{
|
| 495 |
"add_bias": false,
|
|
@@ -497,8 +476,7 @@
|
|
| 497 |
"attention_multiplier": 0.0078125,
|
| 498 |
"dropout": 0,
|
| 499 |
"num_key_value_heads": 4,
|
| 500 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 501 |
-
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"add_bias": false,
|
|
@@ -506,8 +484,7 @@
|
|
| 506 |
"attention_multiplier": 0.0078125,
|
| 507 |
"dropout": 0,
|
| 508 |
"num_key_value_heads": 4,
|
| 509 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 510 |
-
|
| 511 |
},
|
| 512 |
{
|
| 513 |
"add_bias": false,
|
|
@@ -515,8 +492,7 @@
|
|
| 515 |
"attention_multiplier": 0.0078125,
|
| 516 |
"dropout": 0,
|
| 517 |
"num_key_value_heads": 4,
|
| 518 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 519 |
-
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"add_bias": false,
|
|
@@ -524,8 +500,7 @@
|
|
| 524 |
"attention_multiplier": 0.0078125,
|
| 525 |
"dropout": 0,
|
| 526 |
"num_key_value_heads": 4,
|
| 527 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 528 |
-
|
| 529 |
},
|
| 530 |
{
|
| 531 |
"add_bias": false,
|
|
@@ -533,8 +508,7 @@
|
|
| 533 |
"attention_multiplier": 0.0078125,
|
| 534 |
"dropout": 0,
|
| 535 |
"num_key_value_heads": 4,
|
| 536 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 537 |
-
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"add_bias": false,
|
|
@@ -542,8 +516,7 @@
|
|
| 542 |
"attention_multiplier": 0.0078125,
|
| 543 |
"dropout": 0,
|
| 544 |
"num_key_value_heads": 4,
|
| 545 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 546 |
-
|
| 547 |
},
|
| 548 |
{
|
| 549 |
"add_bias": false,
|
|
@@ -551,8 +524,7 @@
|
|
| 551 |
"attention_multiplier": 0.0078125,
|
| 552 |
"dropout": 0,
|
| 553 |
"num_key_value_heads": 4,
|
| 554 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 555 |
-
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"add_bias": false,
|
|
@@ -560,8 +532,7 @@
|
|
| 560 |
"attention_multiplier": 0.0078125,
|
| 561 |
"dropout": 0,
|
| 562 |
"num_key_value_heads": 4,
|
| 563 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 564 |
-
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"add_bias": false,
|
|
@@ -569,8 +540,7 @@
|
|
| 569 |
"attention_multiplier": 0.0078125,
|
| 570 |
"dropout": 0,
|
| 571 |
"num_key_value_heads": 4,
|
| 572 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 573 |
-
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"add_bias": false,
|
|
@@ -578,8 +548,7 @@
|
|
| 578 |
"attention_multiplier": 0.0078125,
|
| 579 |
"dropout": 0,
|
| 580 |
"num_key_value_heads": 4,
|
| 581 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 582 |
-
|
| 583 |
},
|
| 584 |
{
|
| 585 |
"add_bias": false,
|
|
@@ -587,8 +556,7 @@
|
|
| 587 |
"attention_multiplier": 0.0078125,
|
| 588 |
"dropout": 0,
|
| 589 |
"num_key_value_heads": 4,
|
| 590 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 591 |
-
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"add_bias": false,
|
|
@@ -596,8 +564,7 @@
|
|
| 596 |
"attention_multiplier": 0.0078125,
|
| 597 |
"dropout": 0,
|
| 598 |
"num_key_value_heads": 4,
|
| 599 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 600 |
-
|
| 601 |
},
|
| 602 |
{
|
| 603 |
"add_bias": false,
|
|
@@ -605,8 +572,7 @@
|
|
| 605 |
"attention_multiplier": 0.0078125,
|
| 606 |
"dropout": 0,
|
| 607 |
"num_key_value_heads": 4,
|
| 608 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 609 |
-
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"add_bias": false,
|
|
@@ -614,8 +580,7 @@
|
|
| 614 |
"attention_multiplier": 0.0078125,
|
| 615 |
"dropout": 0,
|
| 616 |
"num_key_value_heads": 4,
|
| 617 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 618 |
-
|
| 619 |
},
|
| 620 |
{
|
| 621 |
"add_bias": false,
|
|
@@ -623,8 +588,7 @@
|
|
| 623 |
"attention_multiplier": 0.0078125,
|
| 624 |
"dropout": 0,
|
| 625 |
"num_key_value_heads": 4,
|
| 626 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 627 |
-
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"add_bias": false,
|
|
@@ -632,8 +596,7 @@
|
|
| 632 |
"attention_multiplier": 0.0078125,
|
| 633 |
"dropout": 0,
|
| 634 |
"num_key_value_heads": 4,
|
| 635 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 636 |
-
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"add_bias": false,
|
|
@@ -641,8 +604,7 @@
|
|
| 641 |
"attention_multiplier": 0.0078125,
|
| 642 |
"dropout": 0,
|
| 643 |
"num_key_value_heads": 4,
|
| 644 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 645 |
-
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"add_bias": false,
|
|
@@ -650,8 +612,7 @@
|
|
| 650 |
"attention_multiplier": 0.0078125,
|
| 651 |
"dropout": 0,
|
| 652 |
"num_key_value_heads": 4,
|
| 653 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 654 |
-
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"add_bias": false,
|
|
@@ -659,8 +620,7 @@
|
|
| 659 |
"attention_multiplier": 0.0078125,
|
| 660 |
"dropout": 0,
|
| 661 |
"num_key_value_heads": 4,
|
| 662 |
-
"sequence_mixer_type": "stickbreaking_attention"
|
| 663 |
-
|
| 664 |
}
|
| 665 |
],
|
| 666 |
"transformers_version": "4.49.0.dev0",
|
|
|
|
| 308 |
"attention_multiplier": 0.0078125,
|
| 309 |
"dropout": 0,
|
| 310 |
"num_key_value_heads": 4,
|
| 311 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"add_bias": false,
|
|
|
|
| 316 |
"attention_multiplier": 0.0078125,
|
| 317 |
"dropout": 0,
|
| 318 |
"num_key_value_heads": 4,
|
| 319 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"add_bias": false,
|
|
|
|
| 324 |
"attention_multiplier": 0.0078125,
|
| 325 |
"dropout": 0,
|
| 326 |
"num_key_value_heads": 4,
|
| 327 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"add_bias": false,
|
|
|
|
| 332 |
"attention_multiplier": 0.0078125,
|
| 333 |
"dropout": 0,
|
| 334 |
"num_key_value_heads": 4,
|
| 335 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 336 |
},
|
| 337 |
{
|
| 338 |
"add_bias": false,
|
|
|
|
| 340 |
"attention_multiplier": 0.0078125,
|
| 341 |
"dropout": 0,
|
| 342 |
"num_key_value_heads": 4,
|
| 343 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 344 |
},
|
| 345 |
{
|
| 346 |
"add_bias": false,
|
|
|
|
| 348 |
"attention_multiplier": 0.0078125,
|
| 349 |
"dropout": 0,
|
| 350 |
"num_key_value_heads": 4,
|
| 351 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"add_bias": false,
|
|
|
|
| 356 |
"attention_multiplier": 0.0078125,
|
| 357 |
"dropout": 0,
|
| 358 |
"num_key_value_heads": 4,
|
| 359 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"add_bias": false,
|
|
|
|
| 364 |
"attention_multiplier": 0.0078125,
|
| 365 |
"dropout": 0,
|
| 366 |
"num_key_value_heads": 4,
|
| 367 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"add_bias": false,
|
|
|
|
| 372 |
"attention_multiplier": 0.0078125,
|
| 373 |
"dropout": 0,
|
| 374 |
"num_key_value_heads": 4,
|
| 375 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"add_bias": false,
|
|
|
|
| 380 |
"attention_multiplier": 0.0078125,
|
| 381 |
"dropout": 0,
|
| 382 |
"num_key_value_heads": 4,
|
| 383 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 384 |
},
|
| 385 |
{
|
| 386 |
"add_bias": false,
|
|
|
|
| 388 |
"attention_multiplier": 0.0078125,
|
| 389 |
"dropout": 0,
|
| 390 |
"num_key_value_heads": 4,
|
| 391 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 392 |
},
|
| 393 |
{
|
| 394 |
"add_bias": false,
|
|
|
|
| 396 |
"attention_multiplier": 0.0078125,
|
| 397 |
"dropout": 0,
|
| 398 |
"num_key_value_heads": 4,
|
| 399 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"add_bias": false,
|
|
|
|
| 404 |
"attention_multiplier": 0.0078125,
|
| 405 |
"dropout": 0,
|
| 406 |
"num_key_value_heads": 4,
|
| 407 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 408 |
},
|
| 409 |
{
|
| 410 |
"add_bias": false,
|
|
|
|
| 412 |
"attention_multiplier": 0.0078125,
|
| 413 |
"dropout": 0,
|
| 414 |
"num_key_value_heads": 4,
|
| 415 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 416 |
},
|
| 417 |
{
|
| 418 |
"add_bias": false,
|
|
|
|
| 420 |
"attention_multiplier": 0.0078125,
|
| 421 |
"dropout": 0,
|
| 422 |
"num_key_value_heads": 4,
|
| 423 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"add_bias": false,
|
|
|
|
| 428 |
"attention_multiplier": 0.0078125,
|
| 429 |
"dropout": 0,
|
| 430 |
"num_key_value_heads": 4,
|
| 431 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"add_bias": false,
|
|
|
|
| 436 |
"attention_multiplier": 0.0078125,
|
| 437 |
"dropout": 0,
|
| 438 |
"num_key_value_heads": 4,
|
| 439 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 440 |
},
|
| 441 |
{
|
| 442 |
"add_bias": false,
|
|
|
|
| 444 |
"attention_multiplier": 0.0078125,
|
| 445 |
"dropout": 0,
|
| 446 |
"num_key_value_heads": 4,
|
| 447 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"add_bias": false,
|
|
|
|
| 452 |
"attention_multiplier": 0.0078125,
|
| 453 |
"dropout": 0,
|
| 454 |
"num_key_value_heads": 4,
|
| 455 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"add_bias": false,
|
|
|
|
| 460 |
"attention_multiplier": 0.0078125,
|
| 461 |
"dropout": 0,
|
| 462 |
"num_key_value_heads": 4,
|
| 463 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 464 |
},
|
| 465 |
{
|
| 466 |
"add_bias": false,
|
|
|
|
| 468 |
"attention_multiplier": 0.0078125,
|
| 469 |
"dropout": 0,
|
| 470 |
"num_key_value_heads": 4,
|
| 471 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"add_bias": false,
|
|
|
|
| 476 |
"attention_multiplier": 0.0078125,
|
| 477 |
"dropout": 0,
|
| 478 |
"num_key_value_heads": 4,
|
| 479 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"add_bias": false,
|
|
|
|
| 484 |
"attention_multiplier": 0.0078125,
|
| 485 |
"dropout": 0,
|
| 486 |
"num_key_value_heads": 4,
|
| 487 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"add_bias": false,
|
|
|
|
| 492 |
"attention_multiplier": 0.0078125,
|
| 493 |
"dropout": 0,
|
| 494 |
"num_key_value_heads": 4,
|
| 495 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"add_bias": false,
|
|
|
|
| 500 |
"attention_multiplier": 0.0078125,
|
| 501 |
"dropout": 0,
|
| 502 |
"num_key_value_heads": 4,
|
| 503 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 504 |
},
|
| 505 |
{
|
| 506 |
"add_bias": false,
|
|
|
|
| 508 |
"attention_multiplier": 0.0078125,
|
| 509 |
"dropout": 0,
|
| 510 |
"num_key_value_heads": 4,
|
| 511 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 512 |
},
|
| 513 |
{
|
| 514 |
"add_bias": false,
|
|
|
|
| 516 |
"attention_multiplier": 0.0078125,
|
| 517 |
"dropout": 0,
|
| 518 |
"num_key_value_heads": 4,
|
| 519 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"add_bias": false,
|
|
|
|
| 524 |
"attention_multiplier": 0.0078125,
|
| 525 |
"dropout": 0,
|
| 526 |
"num_key_value_heads": 4,
|
| 527 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"add_bias": false,
|
|
|
|
| 532 |
"attention_multiplier": 0.0078125,
|
| 533 |
"dropout": 0,
|
| 534 |
"num_key_value_heads": 4,
|
| 535 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"add_bias": false,
|
|
|
|
| 540 |
"attention_multiplier": 0.0078125,
|
| 541 |
"dropout": 0,
|
| 542 |
"num_key_value_heads": 4,
|
| 543 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"add_bias": false,
|
|
|
|
| 548 |
"attention_multiplier": 0.0078125,
|
| 549 |
"dropout": 0,
|
| 550 |
"num_key_value_heads": 4,
|
| 551 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"add_bias": false,
|
|
|
|
| 556 |
"attention_multiplier": 0.0078125,
|
| 557 |
"dropout": 0,
|
| 558 |
"num_key_value_heads": 4,
|
| 559 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 560 |
},
|
| 561 |
{
|
| 562 |
"add_bias": false,
|
|
|
|
| 564 |
"attention_multiplier": 0.0078125,
|
| 565 |
"dropout": 0,
|
| 566 |
"num_key_value_heads": 4,
|
| 567 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"add_bias": false,
|
|
|
|
| 572 |
"attention_multiplier": 0.0078125,
|
| 573 |
"dropout": 0,
|
| 574 |
"num_key_value_heads": 4,
|
| 575 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"add_bias": false,
|
|
|
|
| 580 |
"attention_multiplier": 0.0078125,
|
| 581 |
"dropout": 0,
|
| 582 |
"num_key_value_heads": 4,
|
| 583 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 584 |
},
|
| 585 |
{
|
| 586 |
"add_bias": false,
|
|
|
|
| 588 |
"attention_multiplier": 0.0078125,
|
| 589 |
"dropout": 0,
|
| 590 |
"num_key_value_heads": 4,
|
| 591 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"add_bias": false,
|
|
|
|
| 596 |
"attention_multiplier": 0.0078125,
|
| 597 |
"dropout": 0,
|
| 598 |
"num_key_value_heads": 4,
|
| 599 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"add_bias": false,
|
|
|
|
| 604 |
"attention_multiplier": 0.0078125,
|
| 605 |
"dropout": 0,
|
| 606 |
"num_key_value_heads": 4,
|
| 607 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 608 |
},
|
| 609 |
{
|
| 610 |
"add_bias": false,
|
|
|
|
| 612 |
"attention_multiplier": 0.0078125,
|
| 613 |
"dropout": 0,
|
| 614 |
"num_key_value_heads": 4,
|
| 615 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"add_bias": false,
|
|
|
|
| 620 |
"attention_multiplier": 0.0078125,
|
| 621 |
"dropout": 0,
|
| 622 |
"num_key_value_heads": 4,
|
| 623 |
+
"sequence_mixer_type": "stickbreaking_attention"
|
|
|
|
| 624 |
}
|
| 625 |
],
|
| 626 |
"transformers_version": "4.49.0.dev0",
|