b
Discover
Models
Search
About
Softmax Output Approximation for Activation Memory-Efficient Training of Attention-based Networks
2023
·
NeurIPS