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

convert : fix nomic-bert-moe mask token (#13757)

Browse source
This commit is contained in:
Sigbjørn Skjæret 2025-06-01 18:07:21 +02:00 • committed by GitHub
parent c496fe0b1d
commit 5e1c3aed40
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 13 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

6
convert_hf_to_gguf.py
View file

@ -3889,6 +3889,12 @@ class BertModel(TextModel):
SentencePieceTokenTypes.UNKNOWN, SentencePieceTokenTypes.UNKNOWN,
] + toktypes[3:-1] ] + toktypes[3:-1]
if self.model_arch == gguf.MODEL_ARCH.NOMIC_BERT_MOE:
# Add mask token missing from sentencepiece.bpe.model
tokens[250001] = b'<mask>'
scores[250001] = 0.0
toktypes[250001] = SentencePieceTokenTypes.CONTROL
self.gguf_writer.add_tokenizer_model("t5") self.gguf_writer.add_tokenizer_model("t5")
self.gguf_writer.add_tokenizer_pre("default") self.gguf_writer.add_tokenizer_pre("default")
self.gguf_writer.add_token_list(tokens) self.gguf_writer.add_token_list(tokens)

9
src/llama-vocab.cpp
View file

@ -2080,9 +2080,11 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
std::string model_name; std::string model_name;
std::string tokenizer_pre; std::string tokenizer_pre;
std::string general_arch;
ml.get_key(LLM_KV_GENERAL_NAME, model_name, false); ml.get_key(LLM_KV_GENERAL_NAME, model_name, false);
ml.get_key(LLM_KV_TOKENIZER_PRE, tokenizer_pre, false); ml.get_key(LLM_KV_TOKENIZER_PRE, tokenizer_pre, false);
ml.get_key(LLM_KV_GENERAL_ARCHITECTURE, general_arch, false);
// model name to lowercase // model name to lowercase
std::transform(model_name.begin(), model_name.end(), model_name.begin(), std::transform(model_name.begin(), model_name.end(), model_name.begin(),
@ -2091,8 +2093,11 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
} }
); );
// set attributes by model/tokenizer name // set attributes by model/tokenizer/architecture name
if (_contains_any(tokenizer_pre, {"jina-v2-de", "jina-v2-es", "jina-v2-code"})) { if (false
|| _contains_any(tokenizer_pre, {"jina-v2-de", "jina-v2-es", "jina-v2-code"})
|| _contains_any(general_arch, {"nomic-bert-moe"})
) {
_set_token_attr("<mask>", LLAMA_TOKEN_ATTR_LSTRIP, true); _set_token_attr("<mask>", LLAMA_TOKEN_ATTR_LSTRIP, true);
} else if (_contains_any(model_name, {"phi-3", "phi3"})) { } else if (_contains_any(model_name, {"phi-3", "phi3"})) {
for (auto id : cache_special_tokens) { for (auto id : cache_special_tokens) {