vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : add Minerva 7B model support (#10673)

Browse source 
* Support for Minerva 7B

* Update convert_hf_to_gguf_update.py
This commit is contained in:
Riccardo Orlando 2024-12-05 19:30:59 +01:00 • committed by GitHub
parent 0cd182ebcc
commit 6fe6247831
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 9 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
src/llama-vocab.cpp
View file

@ -418,6 +418,7 @@ struct llm_tokenizer_bpe : llm_tokenizer {
case LLAMA_VOCAB_PRE_TYPE_SMOLLM:
case LLAMA_VOCAB_PRE_TYPE_CODESHELL:
case LLAMA_VOCAB_PRE_TYPE_EXAONE:
case LLAMA_VOCAB_PRE_TYPE_MINERVA:
regex_exprs = {
"\\p{N}",
"'s|'t|'re|'ve|'m|'ll|'d| ?\\p{L}+| ?\\p{N}+| ?[^\\s\\p{L}\\p{N}]+|\\s+(?!\\S)",

3
src/llama.cpp
View file

@ -6479,6 +6479,9 @@ static void llm_load_vocab(
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_CHAMELEON;
vocab.tokenizer_add_bos = true;
vocab.tokenizer_clean_spaces = false;
} else if (
tokenizer_pre == "minerva-7b") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_MINERVA;
} else {
throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
}