Skip to content

Commit

Permalink
convert : add Poro-34B-chat tokenizer support (#7713)
Browse files Browse the repository at this point in the history
* support for Poro chat pre-tokenizer

* add support for Poro pre-tokenizer

* Update convert-hf-to-gguf-update.py

Co-authored-by: Georgi Gerganov <[email protected]>

* Change Poro-34B-chat to poro-chat

* Change Poro-34B-chat to poro-chat

* Update convert-hf-to-gguf-update.py

* Update llama.cpp

---------

Co-authored-by: Georgi Gerganov <[email protected]>
  • Loading branch information
ezosa and ggerganov authored Jun 14, 2024
1 parent 172c825 commit 41b9260
Show file tree
Hide file tree
Showing 4 changed files with 13 additions and 0 deletions.
1 change: 1 addition & 0 deletions convert-hf-to-gguf-update.py
Original file line number Diff line number Diff line change
Expand Up @@ -83,6 +83,7 @@ class TOKENIZER_TYPE(IntEnum):
{"name": "jina-v2-es", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-es", },
{"name": "jina-v2-de", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-de", },
{"name": "smaug-bpe", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/abacusai/Smaug-Llama-3-70B-Instruct", },
{"name": "poro-chat", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LumiOpen/Poro-34B-chat", },
{"name": "jina-v2-code", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/jinaai/jina-embeddings-v2-base-code", },
]

Expand Down
3 changes: 3 additions & 0 deletions convert-hf-to-gguf.py
Original file line number Diff line number Diff line change
Expand Up @@ -477,6 +477,9 @@ def get_vocab_base_pre(self, tokenizer) -> str:
if chkhsh == "c136ed14d01c2745d4f60a9596ae66800e2b61fa45643e72436041855ad4089d":
# ref: https://huggingface.co/abacusai/Smaug-Llama-3-70B-Instruct
res = "smaug-bpe"
if chkhsh == "c7ea5862a53e4272c035c8238367063e2b270d51faa48c0f09e9d5b54746c360":
# ref: https://huggingface.co/LumiOpen/Poro-34B-chat
res = "poro-chat"
if chkhsh == "7967bfa498ade6b757b064f31e964dddbb80f8f9a4d68d4ba7998fcf281c531a":
# ref: https://huggingface.co/jinaai/jina-embeddings-v2-base-code
res = "jina-v2-code"
Expand Down
8 changes: 8 additions & 0 deletions llama.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -4713,6 +4713,9 @@ static void llm_load_vocab(
} else if (
tokenizer_pre == "smaug-bpe") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_SMAUG;
} else if (
tokenizer_pre == "poro-chat") {
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_PORO;
} else {
throw std::runtime_error(format("unknown pre-tokenizer type: '%s'", tokenizer_pre.c_str()));
}
Expand Down Expand Up @@ -13028,6 +13031,11 @@ struct llm_tokenizer_bpe {
"(?:'[sS]|'[tT]|'[rR][eE]|'[vV][eE]|'[mM]|'[lL][lL]|'[dD])|[^\\r\\n\\p{L}\\p{N}]?\\p{L}+|\\p{N}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
});
break;
case LLAMA_VOCAB_PRE_TYPE_PORO:
word_collection = unicode_regex_split(text, {
" ?[^(\\s|.,!?…。,、।۔،)]+",
});
break;
default:
// default regex for BPE tokenization pre-processing
word_collection = unicode_regex_split(text, {
Expand Down
1 change: 1 addition & 0 deletions llama.h
Original file line number Diff line number Diff line change
Expand Up @@ -86,6 +86,7 @@ extern "C" {
LLAMA_VOCAB_PRE_TYPE_OLMO = 12,
LLAMA_VOCAB_PRE_TYPE_DBRX = 13,
LLAMA_VOCAB_PRE_TYPE_SMAUG = 14,
LLAMA_VOCAB_PRE_TYPE_PORO = 15,
};

// note: these values should be synchronized with ggml_rope
Expand Down

0 comments on commit 41b9260

Please sign in to comment.