mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-05 04:21:03 +02:00
ae4b922614
This allows for setting the --no-context-shift value in llama-imatrix which is required for models like DeepSeek