Local Ai
b9886
Release b9886 of llama.cpp addresses a bug fix for K/V rotation input handling in attention mechanisms, specifically when buffers are unallocated during DFlash speculative decoding's KV-injection pass
Release b9886 of llama.cpp addresses a bug fix for K/V rotation input handling in attention mechanisms, specifically when buffers are unallocated during DFlash speculative decoding's KV-injection pass. The fix adds buffer validation guards to prevent crashes in the set_input_k_rot and set_input_v_rot functions.
Source: llama.cpp Releases | 2026-07-06