mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-05-08 10:04:10 +00:00
* server : fix speculative decoding with context shift ggml-ci * server : take into account speculative limits ggml-ci * server : add tests