Indras-Mirror/llama.cpp-turboq-mtp

Qwen4-Exp (Qwen3.8-Flash-Next) SWA + MTP + PLE-on-iswa — bounded deep-context decode w/ long-range recall, TBQ4 4.25 bpv KV, DSpark + RotorQuant. Main (master) = Qwen4-Exp build; the Qwen35 SWA hybrid lives on the qwen35 branch. RTX 4090.

语言：C++ · GitHub 星标：95 · Fork：13

话题：cuda · flash-attention · fwht · kv-cache · llama-cpp · mtp · multi-token-prediction · quantization

在 GitHub 上查看原仓库