happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

conversion for modern-bert model supported rather than just granite-small

This commit is contained in:
ryan-mangeno 2025-10-10 15:12:49 -04:00
parent 9715c2afb5
commit 1d01245182
1 changed files with 10 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
convert_hf_to_gguf.py
View File

@ -8976,6 +8976,16 @@ class ModernBertModel(BertModel):
self.gguf_writer.add_vocab_size(self.hparams["vocab_size"])
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
# rename custom "head" layers to standard bert "cls.predictions" names for compatibility
if name == "head.norm.weight":
name = "cls.predictions.transform.LayerNorm.weight"
elif name == "head.norm.bias":
name = "cls.predictions.transform.LayerNorm.bias"
elif name == "head.dense.weight":
name = "cls.predictions.transform.dense.weight"
elif name == "head.dense.bias":
name = "cls.predictions.transform.dense.bias"
# These layers act as MLM head, so we don't need them
if name.startswith("decoder."):
return []