MX-SAFE proposes a versatile MXFP format with on-the-fly bit allocation and tile-based design that reports small accuracy gains over prior MX formats and an accelerator using 24.9% less energy than BF16 while matching its accuracy.
A Framework for Few-shot Language Model Evaluation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MX-SAFE: Versatile Inference- and Training-Proof Microscaling Format with On-the-Fly Exponent and Mantissa Bit Allocation
MX-SAFE proposes a versatile MXFP format with on-the-fly bit allocation and tile-based design that reports small accuracy gains over prior MX formats and an accelerator using 24.9% less energy than BF16 while matching its accuracy.