Files
DeepGEMM/deep_gemm/mega
biondizzle fbfeb54c9a Fix fold_global_scale: UE4M3 scales use .to(float32), not shift-by-23
Checkpoint stores float8_e4m3fn (standard NVFP4), not UE8M0.
The shift-by-23 was misinterpreting E4M3 bytes as E8M0 exponents.
2026-05-12 05:52:33 +00:00
..