mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-11-01 09:01:57 +00:00
* speculative : update default params * speculative : do not discard the last drafted token
9.1 KiB
9.1 KiB