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

Exclude embeddings and output tensor

This commit is contained in:
Ed Addario 2025-08-21 09:48:29 +01:00
parent 35ad0fc4ad
commit 5ef493ea1a
No known key found for this signature in database
GPG Key ID: E7875815A3230993
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/llama-quant.cpp
View File

@ -697,8 +697,10 @@ static std::unordered_map<std::string, ggml_type> target_bpw_type(
q &= name.find("time_mix_decay_w2.weight") == std::string::npos;
q &= name.find("time_mix_lerp_fused.weight") == std::string::npos;
q &= name.find("attn_rel_b.weight") == std::string::npos;
q &= params->quantize_output_tensor || name != "output.weight";
q &= !params->only_copy;
// TODO: Exclude embeddings and output tensors?
q &= params->quantize_output_tensor || name != "output.weight";
q &= name != name_tn(LLM_TENSOR_TOKEN_EMBD, "weight");
return q;
};