Sparse models, including sparse Mixture-of-Experts (MoE) models, have emerged as an effective approach for scaling Transformer models. However, they often suffer from computational inefficiency since a significant number of parameters are unnecessarily involved in computations by multiplying values by zero or low activation values. To address this issue, we present XMoE, a novel MoE designed to enhance both the efficacy and efficiency of sparse MoE models. XMoE leverages small experts and a threshold-based router to enable tokens to selectively engage only essential parameters. Our extensive experiments on language modeling and machine translation tasks demonstrate that enhances model performance and can decrease the computation load at MoE layers by over 50% without sacrificing performance. Furthermore, we present the versatility of by applying it to dense models, enabling sparse computation during inference. We provide a comprehensive analysis and make our code available at https://anonymous.4open.science/r/XMoE.

Transformer-based models have achieved great success on sentence pair modeling tasks, such as answer selection and natural language inference (NLI). These models generally perform cross-attention over input pairs, leading to prohibitive computational cost. Recent studies propose dual-encoder and late interaction architectures for faster computation. However, the balance between the expressive of cross-attention and computation speedup still needs better coordinated. To this end, this paper introduces a novel paradigm TopicAns for efficient sentence pair modeling. TopicAns involves a lightweight cross-attention mechanism. It conducts query encoding only once while modeling the query-candidate interaction in parallel. Extensive experiments conducted on four tasks demonstrate that our TopicAnscan speed up sentence pairing by over 113x while achieving comparable performance as the more expensive cross-attention models.