conftrace_

reduced precision

3 papers

Explore in graph

Co-occurring keywords · papers mentioning both

8-bit floating point (1) reduced precision training (1) batch training (1) model parallel (1) stochastic rounding (1) large batch training (1) transformer model (1) activation quantization (1) translation quality (1) compute efficiency (1)

Papers

Scaling Neural Machine Translation EMNLP 2018 Training Deep Neural Networks with 8-bit Floating Point Numbers NeurIPS 2018 WRPN: Wide Reduced-Precision Networks ICLR 2018