mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-08-31 17:41:26 +02:00
873e5d8e39
* model: use ggml_rope_set_offset() * partially apply to deepseek2