swa full used unless ctx shift and fast forward disabled

This commit is contained in:
Concedo 2025-05-21 22:47:45 +08:00
commit 9f976e9c65
16 changed files with 1429 additions and 654 deletions

View file

@ -319,6 +319,7 @@ struct common_params {
bool flash_attn = false; // flash attention
bool no_perf = false; // disable performance metrics
bool ctx_shift = true; // context shift on inifinite text generation
bool swa_full = false; // use full-size SWA cache (https://github.com/ggml-org/llama.cpp/pull/13194#issuecomment-2868343055)
bool input_prefix_bos = false; // prefix BOS to user inputs, preceding input_prefix
bool use_mmap = true; // use mmap for faster loads