llama.cpp b9905 Release Fixes Quantized KV-Cache for dsv4

Core summary

One sentence to understand this update

The llama.cpp b9905 release addresses a bug by fixing the quantized KV-cache for dsv4 and lists various supported platforms including macOS, iOS, and Linux.

Impact & opportunity

What this could mean

Developers using llama.cpp can expect improved stability and performance for dsv4 models, especially on platforms like Apple Silicon and Linux.