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

mtmd: fix "v.patch_embd" quant and unsupported im2col ops on Metal for deepseek-ocr (#21027) 

* mtmd: fix "v.patch_embd" quant and unsupported im2col ops on Metal for deepseek-ocr

* Update src/llama-quant.cpp

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

---------

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
Saba Fallah 2026-03-27 00:07:55 +01:00 committed by GitHub
parent 7ca0c9cca7
commit 1743d98057
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
convert_hf_to_gguf.py
View File

@ -7150,6 +7150,8 @@ class DeepseekOCRVisionModel(MmprojModel):
return gguf.GGMLQuantizationType.F32
if ".rel_pos_h" in name or '.rel_pos_w' in name:
return gguf.GGMLQuantizationType.F32
if ".neck." in name or ".net_" in name:
return gguf.GGMLQuantizationType.F32
return super().tensor_force_quant(name, new_name, bid, n_dims)
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:

5
src/llama-quant.cpp
View File

@ -345,9 +345,12 @@ static bool tensor_allows_quantization(const llama_model_quantize_params * param
// do not quantize specific multimodal tensors
quantize &= name.find(".position_embd") == std::string::npos;
quantize &= name.find("sam.patch_embd") == std::string::npos;
quantize &= name.find("sam.pos_embd") == std::string::npos;
quantize &= name.find("sam.neck.") == std::string::npos;
quantize &= name.find("sam.net_") == std::string::npos;
quantize &= name.find(".rel_pos") == std::string::npos;
quantize &= name.find(".patch_embd") == std::string::npos;
quantize &= name.find(".patch_merger") == std::string::npos;
return quantize;
}