FLARE replaces floating-point quantization and softmax in transformers with integer-only eMSB-based methods and low-ENOB analog-to-digital converters, aiming for fast and low-energy attention.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FLARE: FP-Less PTQ and Low-ENOB ADC Based AMS-PiM for Error-Resilient, Fast, and Efficient Transformer Acceleration
FLARE replaces floating-point quantization and softmax in transformers with integer-only eMSB-based methods and low-ENOB analog-to-digital converters, aiming for fast and low-energy attention.