atomic-llama-cpp-turboquant
llama.cpp fork with TurboQuant WHT-rotated KV cache & weight compression + Gemma 4 MTP and Qwen 3.6 NextN speculative decoding (+30-50% throughput).
Community
Post about atomic-llama-cpp-turboquant...PostLoading...
Post about atomic-llama-cpp-turboquant...Post