matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-16 10:47:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

model: support for mistral-small in the ollama runner

Browse Source 
Mistral is a popular research lab making open source models. This updates
the forward pass of llama architecture models to support both llama models
and mistral models by accounting for additional metadata present in mistral
models, and finding the correct dimensions for the output projection.
This commit is contained in:
Bruce MacDonald
2025-03-14 16:56:32 -07:00
committed by Michael Yang
parent 1861fbdeb5
commit 6bd0a983cd
27 changed files with 1116 additions and 350 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
llama/llama.cpp/src/llama-quant.cpp vendored
View File

@@ -738,13 +738,8 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
bool quantize = name.rfind("weight") == name.size() - 6; // ends with 'weight'?
// don't quantize vision stuff
quantize &= name.find("v.blk.") == std::string::npos;
quantize &= name.find("mm.mm_input_projection.weight") == std::string::npos;
quantize &= name.find("mm.mm_soft_emb_norm.weight") == std::string::npos;
quantize &= name.find("v.patch_embedding.weight") == std::string::npos;
quantize &= name.find("v.position_embedding.weight") == std::string::npos;
quantize &= name.find("v.post_layernorm.weight") == std::string::npos;
quantize &= name.find("v.") == std::string::npos;
quantize &= name.find("mm.") == std::string::npos;
// quantize only 2D and 3D tensors (experts)
quantize &= (ggml_n_dims(tensor) >= 2);