pom
commited on
Commit
•
39ea5e0
1
Parent(s):
629f866
Update conf files
Browse files- config.json +3 -3
- generation_config.json +12 -0
- tokenizer.json +269 -13
config.json
CHANGED
@@ -14,14 +14,14 @@
|
|
14 |
"initializer_range": 0.02,
|
15 |
"intermediate_size": 13824,
|
16 |
"max_position_embeddings": 8192,
|
|
|
17 |
"model_type": "xverse",
|
18 |
"num_attention_heads": 40,
|
19 |
"num_hidden_layers": 40,
|
20 |
"rms_norm_eps": 1e-06,
|
21 |
"tie_word_embeddings": false,
|
22 |
-
"torch_dtype": "
|
23 |
"transformers_version": "4.28.1",
|
24 |
"use_cache": true,
|
25 |
-
"vocab_size":
|
26 |
}
|
27 |
-
|
|
|
14 |
"initializer_range": 0.02,
|
15 |
"intermediate_size": 13824,
|
16 |
"max_position_embeddings": 8192,
|
17 |
+
"max_tokenizer_truncation": 6144,
|
18 |
"model_type": "xverse",
|
19 |
"num_attention_heads": 40,
|
20 |
"num_hidden_layers": 40,
|
21 |
"rms_norm_eps": 1e-06,
|
22 |
"tie_word_embeddings": false,
|
23 |
+
"torch_dtype": "bfloat16",
|
24 |
"transformers_version": "4.28.1",
|
25 |
"use_cache": true,
|
26 |
+
"vocab_size": 100534
|
27 |
}
|
|
generation_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"pad_token_id": 1,
|
3 |
+
"bos_token_id": 2,
|
4 |
+
"eos_token_id": 3,
|
5 |
+
"max_new_tokens": 2048,
|
6 |
+
"temperature": 0.5,
|
7 |
+
"top_k": 30,
|
8 |
+
"top_p": 0.85,
|
9 |
+
"repetition_penalty": 1.1,
|
10 |
+
"do_sample": true,
|
11 |
+
"transformers_version": "4.29.1"
|
12 |
+
}
|
tokenizer.json
CHANGED
@@ -58,14 +58,6 @@
|
|
58 |
"special": true
|
59 |
}
|
60 |
],
|
61 |
-
"normalizer": {
|
62 |
-
"type": "Sequence",
|
63 |
-
"normalizers": [
|
64 |
-
{
|
65 |
-
"type": "NFKC"
|
66 |
-
}
|
67 |
-
]
|
68 |
-
},
|
69 |
"pre_tokenizer": {
|
70 |
"type": "Sequence",
|
71 |
"pretokenizers": [
|
@@ -86,9 +78,17 @@
|
|
86 |
},
|
87 |
"post_processor": null,
|
88 |
"decoder": {
|
89 |
-
"type": "
|
90 |
-
"
|
91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
},
|
93 |
"model": {
|
94 |
"type": "BPE",
|
@@ -100376,7 +100376,263 @@
|
|
100376 |
"nj": 100274,
|
100377 |
"iful": 100275,
|
100378 |
"▁solution": 100276,
|
100379 |
-
"\n": 100277
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
100380 |
},
|
100381 |
"merges": [
|
100382 |
"▁ t",
|
@@ -104090,4 +104346,4 @@
|
|
104090 |
"▁sol ution"
|
104091 |
]
|
104092 |
}
|
104093 |
-
}
|
|
|
58 |
"special": true
|
59 |
}
|
60 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
"pre_tokenizer": {
|
62 |
"type": "Sequence",
|
63 |
"pretokenizers": [
|
|
|
78 |
},
|
79 |
"post_processor": null,
|
80 |
"decoder": {
|
81 |
+
"type": "Sequence",
|
82 |
+
"decoders": [
|
83 |
+
{
|
84 |
+
"type": "Metaspace",
|
85 |
+
"replacement": "▁",
|
86 |
+
"add_prefix_space": false
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"type": "ByteFallback"
|
90 |
+
}
|
91 |
+
]
|
92 |
},
|
93 |
"model": {
|
94 |
"type": "BPE",
|
|
|
100376 |
"nj": 100274,
|
100377 |
"iful": 100275,
|
100378 |
"▁solution": 100276,
|
100379 |
+
"\n": 100277,
|
100380 |
+
"<0x00>": 100278,
|
100381 |
+
"<0x01>": 100279,
|
100382 |
+
"<0x02>": 100280,
|
100383 |
+
"<0x03>": 100281,
|
100384 |
+
"<0x04>": 100282,
|
100385 |
+
"<0x05>": 100283,
|
100386 |
+
"<0x06>": 100284,
|
100387 |
+
"<0x07>": 100285,
|
100388 |
+
"<0x08>": 100286,
|
100389 |
+
"<0x09>": 100287,
|
100390 |
+
"<0x0A>": 100288,
|
100391 |
+
"<0x0B>": 100289,
|
100392 |
+
"<0x0C>": 100290,
|
100393 |
+
"<0x0D>": 100291,
|
100394 |
+
"<0x0E>": 100292,
|
100395 |
+
"<0x0F>": 100293,
|
100396 |
+
"<0x10>": 100294,
|
100397 |
+
"<0x11>": 100295,
|
100398 |
+
"<0x12>": 100296,
|
100399 |
+
"<0x13>": 100297,
|
100400 |
+
"<0x14>": 100298,
|
100401 |
+
"<0x15>": 100299,
|
100402 |
+
"<0x16>": 100300,
|
100403 |
+
"<0x17>": 100301,
|
100404 |
+
"<0x18>": 100302,
|
100405 |
+
"<0x19>": 100303,
|
100406 |
+
"<0x1A>": 100304,
|
100407 |
+
"<0x1B>": 100305,
|
100408 |
+
"<0x1C>": 100306,
|
100409 |
+
"<0x1D>": 100307,
|
100410 |
+
"<0x1E>": 100308,
|
100411 |
+
"<0x1F>": 100309,
|
100412 |
+
"<0x20>": 100310,
|
100413 |
+
"<0x21>": 100311,
|
100414 |
+
"<0x22>": 100312,
|
100415 |
+
"<0x23>": 100313,
|
100416 |
+
"<0x24>": 100314,
|
100417 |
+
"<0x25>": 100315,
|
100418 |
+
"<0x26>": 100316,
|
100419 |
+
"<0x27>": 100317,
|
100420 |
+
"<0x28>": 100318,
|
100421 |
+
"<0x29>": 100319,
|
100422 |
+
"<0x2A>": 100320,
|
100423 |
+
"<0x2B>": 100321,
|
100424 |
+
"<0x2C>": 100322,
|
100425 |
+
"<0x2D>": 100323,
|
100426 |
+
"<0x2E>": 100324,
|
100427 |
+
"<0x2F>": 100325,
|
100428 |
+
"<0x30>": 100326,
|
100429 |
+
"<0x31>": 100327,
|
100430 |
+
"<0x32>": 100328,
|
100431 |
+
"<0x33>": 100329,
|
100432 |
+
"<0x34>": 100330,
|
100433 |
+
"<0x35>": 100331,
|
100434 |
+
"<0x36>": 100332,
|
100435 |
+
"<0x37>": 100333,
|
100436 |
+
"<0x38>": 100334,
|
100437 |
+
"<0x39>": 100335,
|
100438 |
+
"<0x3A>": 100336,
|
100439 |
+
"<0x3B>": 100337,
|
100440 |
+
"<0x3C>": 100338,
|
100441 |
+
"<0x3D>": 100339,
|
100442 |
+
"<0x3E>": 100340,
|
100443 |
+
"<0x3F>": 100341,
|
100444 |
+
"<0x40>": 100342,
|
100445 |
+
"<0x41>": 100343,
|
100446 |
+
"<0x42>": 100344,
|
100447 |
+
"<0x43>": 100345,
|
100448 |
+
"<0x44>": 100346,
|
100449 |
+
"<0x45>": 100347,
|
100450 |
+
"<0x46>": 100348,
|
100451 |
+
"<0x47>": 100349,
|
100452 |
+
"<0x48>": 100350,
|
100453 |
+
"<0x49>": 100351,
|
100454 |
+
"<0x4A>": 100352,
|
100455 |
+
"<0x4B>": 100353,
|
100456 |
+
"<0x4C>": 100354,
|
100457 |
+
"<0x4D>": 100355,
|
100458 |
+
"<0x4E>": 100356,
|
100459 |
+
"<0x4F>": 100357,
|
100460 |
+
"<0x50>": 100358,
|
100461 |
+
"<0x51>": 100359,
|
100462 |
+
"<0x52>": 100360,
|
100463 |
+
"<0x53>": 100361,
|
100464 |
+
"<0x54>": 100362,
|
100465 |
+
"<0x55>": 100363,
|
100466 |
+
"<0x56>": 100364,
|
100467 |
+
"<0x57>": 100365,
|
100468 |
+
"<0x58>": 100366,
|
100469 |
+
"<0x59>": 100367,
|
100470 |
+
"<0x5A>": 100368,
|
100471 |
+
"<0x5B>": 100369,
|
100472 |
+
"<0x5C>": 100370,
|
100473 |
+
"<0x5D>": 100371,
|
100474 |
+
"<0x5E>": 100372,
|
100475 |
+
"<0x5F>": 100373,
|
100476 |
+
"<0x60>": 100374,
|
100477 |
+
"<0x61>": 100375,
|
100478 |
+
"<0x62>": 100376,
|
100479 |
+
"<0x63>": 100377,
|
100480 |
+
"<0x64>": 100378,
|
100481 |
+
"<0x65>": 100379,
|
100482 |
+
"<0x66>": 100380,
|
100483 |
+
"<0x67>": 100381,
|
100484 |
+
"<0x68>": 100382,
|
100485 |
+
"<0x69>": 100383,
|
100486 |
+
"<0x6A>": 100384,
|
100487 |
+
"<0x6B>": 100385,
|
100488 |
+
"<0x6C>": 100386,
|
100489 |
+
"<0x6D>": 100387,
|
100490 |
+
"<0x6E>": 100388,
|
100491 |
+
"<0x6F>": 100389,
|
100492 |
+
"<0x70>": 100390,
|
100493 |
+
"<0x71>": 100391,
|
100494 |
+
"<0x72>": 100392,
|
100495 |
+
"<0x73>": 100393,
|
100496 |
+
"<0x74>": 100394,
|
100497 |
+
"<0x75>": 100395,
|
100498 |
+
"<0x76>": 100396,
|
100499 |
+
"<0x77>": 100397,
|
100500 |
+
"<0x78>": 100398,
|
100501 |
+
"<0x79>": 100399,
|
100502 |
+
"<0x7A>": 100400,
|
100503 |
+
"<0x7B>": 100401,
|
100504 |
+
"<0x7C>": 100402,
|
100505 |
+
"<0x7D>": 100403,
|
100506 |
+
"<0x7E>": 100404,
|
100507 |
+
"<0x7F>": 100405,
|
100508 |
+
"<0x80>": 100406,
|
100509 |
+
"<0x81>": 100407,
|
100510 |
+
"<0x82>": 100408,
|
100511 |
+
"<0x83>": 100409,
|
100512 |
+
"<0x84>": 100410,
|
100513 |
+
"<0x85>": 100411,
|
100514 |
+
"<0x86>": 100412,
|
100515 |
+
"<0x87>": 100413,
|
100516 |
+
"<0x88>": 100414,
|
100517 |
+
"<0x89>": 100415,
|
100518 |
+
"<0x8A>": 100416,
|
100519 |
+
"<0x8B>": 100417,
|
100520 |
+
"<0x8C>": 100418,
|
100521 |
+
"<0x8D>": 100419,
|
100522 |
+
"<0x8E>": 100420,
|
100523 |
+
"<0x8F>": 100421,
|
100524 |
+
"<0x90>": 100422,
|
100525 |
+
"<0x91>": 100423,
|
100526 |
+
"<0x92>": 100424,
|
100527 |
+
"<0x93>": 100425,
|
100528 |
+
"<0x94>": 100426,
|
100529 |
+
"<0x95>": 100427,
|
100530 |
+
"<0x96>": 100428,
|
100531 |
+
"<0x97>": 100429,
|
100532 |
+
"<0x98>": 100430,
|
100533 |
+
"<0x99>": 100431,
|
100534 |
+
"<0x9A>": 100432,
|
100535 |
+
"<0x9B>": 100433,
|
100536 |
+
"<0x9C>": 100434,
|
100537 |
+
"<0x9D>": 100435,
|
100538 |
+
"<0x9E>": 100436,
|
100539 |
+
"<0x9F>": 100437,
|
100540 |
+
"<0xA0>": 100438,
|
100541 |
+
"<0xA1>": 100439,
|
100542 |
+
"<0xA2>": 100440,
|
100543 |
+
"<0xA3>": 100441,
|
100544 |
+
"<0xA4>": 100442,
|
100545 |
+
"<0xA5>": 100443,
|
100546 |
+
"<0xA6>": 100444,
|
100547 |
+
"<0xA7>": 100445,
|
100548 |
+
"<0xA8>": 100446,
|
100549 |
+
"<0xA9>": 100447,
|
100550 |
+
"<0xAA>": 100448,
|
100551 |
+
"<0xAB>": 100449,
|
100552 |
+
"<0xAC>": 100450,
|
100553 |
+
"<0xAD>": 100451,
|
100554 |
+
"<0xAE>": 100452,
|
100555 |
+
"<0xAF>": 100453,
|
100556 |
+
"<0xB0>": 100454,
|
100557 |
+
"<0xB1>": 100455,
|
100558 |
+
"<0xB2>": 100456,
|
100559 |
+
"<0xB3>": 100457,
|
100560 |
+
"<0xB4>": 100458,
|
100561 |
+
"<0xB5>": 100459,
|
100562 |
+
"<0xB6>": 100460,
|
100563 |
+
"<0xB7>": 100461,
|
100564 |
+
"<0xB8>": 100462,
|
100565 |
+
"<0xB9>": 100463,
|
100566 |
+
"<0xBA>": 100464,
|
100567 |
+
"<0xBB>": 100465,
|
100568 |
+
"<0xBC>": 100466,
|
100569 |
+
"<0xBD>": 100467,
|
100570 |
+
"<0xBE>": 100468,
|
100571 |
+
"<0xBF>": 100469,
|
100572 |
+
"<0xC0>": 100470,
|
100573 |
+
"<0xC1>": 100471,
|
100574 |
+
"<0xC2>": 100472,
|
100575 |
+
"<0xC3>": 100473,
|
100576 |
+
"<0xC4>": 100474,
|
100577 |
+
"<0xC5>": 100475,
|
100578 |
+
"<0xC6>": 100476,
|
100579 |
+
"<0xC7>": 100477,
|
100580 |
+
"<0xC8>": 100478,
|
100581 |
+
"<0xC9>": 100479,
|
100582 |
+
"<0xCA>": 100480,
|
100583 |
+
"<0xCB>": 100481,
|
100584 |
+
"<0xCC>": 100482,
|
100585 |
+
"<0xCD>": 100483,
|
100586 |
+
"<0xCE>": 100484,
|
100587 |
+
"<0xCF>": 100485,
|
100588 |
+
"<0xD0>": 100486,
|
100589 |
+
"<0xD1>": 100487,
|
100590 |
+
"<0xD2>": 100488,
|
100591 |
+
"<0xD3>": 100489,
|
100592 |
+
"<0xD4>": 100490,
|
100593 |
+
"<0xD5>": 100491,
|
100594 |
+
"<0xD6>": 100492,
|
100595 |
+
"<0xD7>": 100493,
|
100596 |
+
"<0xD8>": 100494,
|
100597 |
+
"<0xD9>": 100495,
|
100598 |
+
"<0xDA>": 100496,
|
100599 |
+
"<0xDB>": 100497,
|
100600 |
+
"<0xDC>": 100498,
|
100601 |
+
"<0xDD>": 100499,
|
100602 |
+
"<0xDE>": 100500,
|
100603 |
+
"<0xDF>": 100501,
|
100604 |
+
"<0xE0>": 100502,
|
100605 |
+
"<0xE1>": 100503,
|
100606 |
+
"<0xE2>": 100504,
|
100607 |
+
"<0xE3>": 100505,
|
100608 |
+
"<0xE4>": 100506,
|
100609 |
+
"<0xE5>": 100507,
|
100610 |
+
"<0xE6>": 100508,
|
100611 |
+
"<0xE7>": 100509,
|
100612 |
+
"<0xE8>": 100510,
|
100613 |
+
"<0xE9>": 100511,
|
100614 |
+
"<0xEA>": 100512,
|
100615 |
+
"<0xEB>": 100513,
|
100616 |
+
"<0xEC>": 100514,
|
100617 |
+
"<0xED>": 100515,
|
100618 |
+
"<0xEE>": 100516,
|
100619 |
+
"<0xEF>": 100517,
|
100620 |
+
"<0xF0>": 100518,
|
100621 |
+
"<0xF1>": 100519,
|
100622 |
+
"<0xF2>": 100520,
|
100623 |
+
"<0xF3>": 100521,
|
100624 |
+
"<0xF4>": 100522,
|
100625 |
+
"<0xF5>": 100523,
|
100626 |
+
"<0xF6>": 100524,
|
100627 |
+
"<0xF7>": 100525,
|
100628 |
+
"<0xF8>": 100526,
|
100629 |
+
"<0xF9>": 100527,
|
100630 |
+
"<0xFA>": 100528,
|
100631 |
+
"<0xFB>": 100529,
|
100632 |
+
"<0xFC>": 100530,
|
100633 |
+
"<0xFD>": 100531,
|
100634 |
+
"<0xFE>": 100532,
|
100635 |
+
"<0xFF>": 100533
|
100636 |
},
|
100637 |
"merges": [
|
100638 |
"▁ t",
|
|
|
104346 |
"▁sol ution"
|
104347 |
]
|
104348 |
}
|
104349 |
+
}
|