A saliency-weighted quantization-aware training method lets 4-bit quantized imitation-learning policies match full-precision success rates across robot manipulation, driving, and control benchmarks.
{TVM}: An automated {End- to-End} optimizing compiler for deep learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Saliency-Aware Quantized Imitation Learning for Efficient Robotic Control
A saliency-weighted quantization-aware training method lets 4-bit quantized imitation-learning policies match full-precision success rates across robot manipulation, driving, and control benchmarks.