mirror of
https://github.com/FlashML-org/FreeToken.git
synced 2026-11-22 12:00:36 +08:00
* feat(moe): serve NVFP4 and MXFP4 experts fused through the expert banks Resident experts skip the state dict and load through the offload expert-bank loader (HF pieces or FTW banks) into per-layer GPU banks, so fused serves bf16, fp8_block, NVFP4 and MXFP4 alike. Unified-memory GPUs auto-select fused for these formats and ignore offload-only flags with a warning. BREAKING CHANGE: ft checkpoint always writes MoE experts as banks; an FTW converted with --moe-backend fused or triton (dense experts) no longer loads and must be reconverted. * docs(moe): name only GB10 / DGX Spark for unified memory