llama-cpp-python
fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf
#2180

Open

fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf #2180

avion23 wants to merge 1 commit into abetlen:main from avion23:fix/perf-and-iswa

avion23 force pushed from 5935c64d to 939fa72d 77 days ago

avion23 force pushed from 939fa72d to 9609c824 77 days ago

avion23 changed the title ~~perf: vectorize hot-path operations + fix SWA/ISWA KV cache corruption (Gemma-4)~~ fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf 77 days ago

avion23 marked this pull request as ready for review 77 days ago

avion23 force pushed from c9bbd6d8 to 3538232d 77 days ago

perf: vectorize hot-path ops, reduce Python overhead, fix SWA/ISWA KV…

e54fe594

avion23 force pushed from 3538232d to e54fe594 76 days ago

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

llama-cpp-python fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf #2180 Open

fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf #2180

llama-cpp-python
fix: prevent KV cache corruption on SWA/ISWA models + hot-path perf
#2180

Open