Popular repositories Loading

1. llama-cpp-mtp-turboquant-sm120-blackwell-windows llama-cpp-mtp-turboquant-sm120-blackwell-windows Public

   Windows prebuilt of llama.cpp combining Multi-Token Prediction (MTP) + TurboQuant KV cache compression + native sm_120 (Blackwell consumer GPU, FP4 tensor cores). For RTX 5060 Ti / 5070 / 5080 / 5090.

   15 4

2. beellama.cpp beellama.cpp Public

   Forked from Anbeeld/beellama.cpp

   DFlash & TurboQuant in llama.cpp with up to 3x faster generation and 7.5x more KV cache in same VRAM

   C++ 3

3. weatherland weatherland Public

   Weatherland — live desktop wallpaper for Windows. Product page and privacy policy.

   HTML

4. strata-flint strata-flint Public

   Strata Fork with Flint 40B MoE (256 experts) support, Tiered KV Context Streaming (256K context), and modular inference backend

   C++