llama : fix Viking pre-tokenizer regex

The order was previously wrong, which caused errors in some tests.
This commit is contained in:
Francis Couture-Harpin 2024-07-08 16:34:39 -04:00
parent f9d42c598b
commit 31a1b0eeaa

View File

@ -15440,8 +15440,8 @@ struct llm_tokenizer_bpe {
break; break;
case LLAMA_VOCAB_PRE_TYPE_VIKING: case LLAMA_VOCAB_PRE_TYPE_VIKING:
regex_exprs = { regex_exprs = {
"\\p{N}",
" ?[^(\\s|.,!?…。,、।۔،)]+", " ?[^(\\s|.,!?…。,、।۔،)]+",
"\\p{N}",
}; };
break; break;
default: default: