mxfp8

FLAMIA 提出了一族面向 NVIDIA Blackwell GPU 的混合精度注意力内核，利用 MXFP8 与 NVFP4/FP8 路径，在避免因果信息泄漏的同时加速 Transformer 训练，在 B200/B300 上实现约 1.06–1.42 倍的前向与反向注意力加速。

Meta Engineering extends FlashAttention-4 with MXFP8 support for NVIDIA Blackwell, achieving up to 2.85 PFLOP/s forward performance and integrating into production training workflows like GEM.

DealignAI releases CRACK-abliterated and MXFP4/MXFP8 quantized versions of Qwen3.6-27B and 35B models, preserving MTP for faster speculative decoding on Apple Silicon.