ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

llama : add Trillion 7B model support (#12556)

Browse source 
* Support Trillion 7B

* Update llama.h

* Update llama.h

* Update llama-vocab.cpp for Trillion

* Update llama-vocab.cpp
This commit is contained in:
Juyoung Suk 2025-03-31 03:38:33 +09:00 • committed by GitHub
parent 7242dd9675
commit b3de7cac73
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 11 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

5
src/llama-vocab.cpp
View file

@ -342,6 +342,7 @@ struct llm_tokenizer_bpe : llm_tokenizer {
case LLAMA_VOCAB_PRE_TYPE_MPT:
case LLAMA_VOCAB_PRE_TYPE_OLMO:
case LLAMA_VOCAB_PRE_TYPE_JAIS:
case LLAMA_VOCAB_PRE_TYPE_TRILLION:
regex_exprs = {
"'s|'t|'re|'ve|'m|'ll|'d| ?\\p{L}+| ?\\p{N}+| ?[^\\s\\p{L}\\p{N}]+|\\s+(?!\\S)",
};
@ -1614,6 +1615,10 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
tokenizer_pre == "superbpe") {
pre_type = LLAMA_VOCAB_PRE_TYPE_SUPERBPE;
clean_spaces = false;
} else if (
tokenizer_pre == "trillion") {
pre_type = LLAMA_VOCAB_PRE_TYPE_TRILLION;
clean_spaces = false;
} else {
throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
}