mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-06-29 12:35:16 +00:00
* server : fix speculative decoding with context shift ggml-ci * server : take into account speculative limits ggml-ci * server : add tests