EL-Attention Memory Efficient Lossless Attention for Generation

EL-Attention:MemoryEfficientLosslessAttentionforGenerationYuYan1JiushengChen1WeizhenQi2NikhilBhendawade1YeyunGong3NanDuan3RuofeiZhang4Abstractpruninglayer(Fanetal.,2019)ortrainingasmallerstu-dentmo...