Commit c1e46f6

committed

Achieve pre-chunk-attention parity; remove most of the LLM generated crap

1 parent c87e8d5 commit c1e46f6Copy full SHA for c1e46f6

2 files changed

+202

-60

lines changed

ggml/src/ggml-cpu
- ops.cpp
src/models
- llm_build_qwen3next.cpp

2 files changed

+202

-60

lines changed

Comments

(0)