Skip to content

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention #10169

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention

[CUDA] Support FP8 (E4M3) KV Cache for Group Query Attention #10169

Annotations

4 warnings

Build Linux x64 Debug (ASan)  /  build_test_pipeline

succeeded Feb 14, 2026 in 32m 57s