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

convert : remove unused vocab attributes

Browse source
This commit is contained in:
Jared Van Bortel 2024-03-27 12:14:40 -04:00
parent dd1a60c536
commit 72e95e33a9
1 changed files with 6 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

13
convert.py
View file

@ -336,11 +336,11 @@ class BpeVocab:
name = "bpe" name = "bpe"
def __init__(self, fname_tokenizer: Path, fname_added_tokens: Path | None): def __init__(self, fname_tokenizer: Path, fname_added_tokens: Path | None):
self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read()) bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
if isinstance(self.bpe_tokenizer.get('model'), dict): if isinstance(bpe_tokenizer.get('model'), dict):
self.vocab = self.bpe_tokenizer["model"]["vocab"] self.vocab = bpe_tokenizer["model"]["vocab"]
else: else:
self.vocab = self.bpe_tokenizer self.vocab = bpe_tokenizer
added_tokens: dict[str, int] added_tokens: dict[str, int]
if fname_added_tokens is not None: if fname_added_tokens is not None:
# FIXME: Verify that added tokens here _cannot_ overlap with the main vocab. # FIXME: Verify that added tokens here _cannot_ overlap with the main vocab.
@ -356,7 +356,7 @@ class BpeVocab:
(item['content'], item['id']) (item['content'], item['id'])
for item in tokenizer_json.get('added_tokens', []) for item in tokenizer_json.get('added_tokens', [])
# Added tokens here can be duplicates of the main vocabulary. # Added tokens here can be duplicates of the main vocabulary.
if item['content'] not in self.bpe_tokenizer) if item['content'] not in bpe_tokenizer)
vocab_size = len(self.vocab) vocab_size = len(self.vocab)
expected_ids = list(range(vocab_size, vocab_size + len(added_tokens))) expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
@ -371,7 +371,6 @@ class BpeVocab:
self.vocab_size_base = vocab_size self.vocab_size_base = vocab_size
self.vocab_size = self.vocab_size_base + len(self.added_tokens_list) self.vocab_size = self.vocab_size_base + len(self.added_tokens_list)
self.fname_tokenizer = fname_tokenizer self.fname_tokenizer = fname_tokenizer
self.fname_added_tokens = fname_added_tokens
def bpe_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]: def bpe_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
reverse_vocab = {id: encoded_tok for encoded_tok, id in self.vocab.items()} reverse_vocab = {id: encoded_tok for encoded_tok, id in self.vocab.items()}
@ -419,7 +418,6 @@ class SentencePieceVocab:
self.vocab_size_base = vocab_size self.vocab_size_base = vocab_size
self.vocab_size = self.vocab_size_base + len(self.added_tokens_list) self.vocab_size = self.vocab_size_base + len(self.added_tokens_list)
self.fname_tokenizer = fname_tokenizer self.fname_tokenizer = fname_tokenizer
self.fname_added_tokens = fname_added_tokens
def sentencepiece_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]: def sentencepiece_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
tokenizer = self.sentencepiece_tokenizer tokenizer = self.sentencepiece_tokenizer
@ -507,7 +505,6 @@ class HfVocab:
self.vocab_size = self.vocab_size_base + len(self.added_tokens_list) self.vocab_size = self.vocab_size_base + len(self.added_tokens_list)
self.fname_tokenizer = fname_tokenizer self.fname_tokenizer = fname_tokenizer
self.fname_added_tokens = fname_added_tokens
def hf_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]: def hf_tokens(self) -> Iterable[tuple[bytes, float, gguf.TokenType]]:
reverse_vocab = { reverse_vocab = {