MrJackSpade

Appreciate the response. Before I dig too deep into it, I should probably ask for clarification on one point. What you're saying is, by the time it gets to `path_session.clear();`,...

I'm going to close this, because while I cant explain why the model kept forgetting the prompt, after continuing to dig through the code for the last week I can...

I know almost nothing about c++ but I took a wild stab in the dark and tried reverting "Fix for mingw (#1462)" locally, and it resolved the problem.

Apologies. It looks like some information was accidentally removed and some not properly updated while I was attempting to diagnose the issue. > I'm still able to use it through...

> I have the same issue with clang failing to compile a test program. > > ``` > C:\Users\andre\Desktop\development\llama.cpp\build>cmake -G Ninja -DAMDGPU_TARGETS=gfx1100 -DLLAMA_HIPBLAS=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release .. > -- The...

I see a commit to add a definition for Llama4, however I get an error message when trying to actually merge ``` RuntimeError: Tensor vision_model.model.layers.33.attention.v_proj.bias required but not present in...

For what its worth, I've been having issues with this since support for QWEN2 was added. It would appear to gen for a few messages, and then collapse into garbage....

Just disabling MMQ is allowing me to generate without garbage, even without enabling Flash Attention.