llama : enforce the same K and V cache types for DeepSeek V4; enable FA if V cache is quantized - #25871

llama : enforce the same K and V cache types for DeepSeek V4; enable FA if V cache is quantized#25871

ggerganov merged 3 commits into

ggml-org:masterggml-org/llama.cpp:masterfrom

fairydreaming:force-fa-quant-kv-cachefairydreaming/llama.cpp:force-fa-quant-kv-cacheCopy head branch name to clipboard