phantom-kv removes LLM refusals via 18MB KV cache injection

Developer lordx64 released phantom-kv, an open-source method that removes LLM refusal behavior without touching model weights, by injecting a small 18MB offline-trained KV tensor library into the model's KV cache.

2026-09-22 ~ 2026-09-22 · 3 related posts

• phantom-kv: uncensor LLMs per-request with an 18MB trained KV-cache, no weight edits — Anony6666 · 2026-09-22
• phantom-kv removes LLM refusals via hot-swappable KV cache, no weight edits needed — cyb3rops · 2026-09-22

1 near-duplicate retellings: Anony6666