IMP-MoE-L focusing on video tasks that achieves new state-of-the-art in zero-shot video classification: 77.0% on Kinetics-400, 76.8% on Kinetics-600, and 68.3% on

Most widely adopted MoE models are discontinuous with respect to their parameters - often referred to as sparse .

Implicit neural representations (INRs) have proven effective in various tasks including image, shape, audio, and video reconstruction. These INRs typically learn the implicit field from sampled input points.

MTL: InMTL, deep learning-based architectures that perform soft-parameter sharing, i.e., share model parameters partially, are proving to be effective at exploiting both the commonalities and differences among tasks [6]. Ourwork is also related to [5] who introduced "routers" (similar to gates) that can choose which layers or components of layers to activate per-task. The routers in the latter work are not differentiable and requirereinforcementlearning. To construct α, there are two cases to consider: (i)s = k and (ii) s < k. If s = k, then set αi = log(w ti) for i [k]. Our base case is fort = 1.

This motivates us to consider a challenging classification problem with intrinsic cluster structures.

The proliferation of large language models (LLMs) has led to the adoption of Mixture-of-Experts (MoE) architectures that dynamically leverage specialized subnetworks for improved efficiency and performance. Despite their benefits, MoE models face significant challenges during inference, including inefficient memory management and suboptimal batching, due to misaligned design choices between the model architecture and the system policies. Furthermore, the conventional approach of training MoEs from scratch is increasingly prohibitive in terms of cost. In this paper, we propose a novel framework $\textit{Read-ME}$ that transforms pre-trained dense LLMs into smaller MoE models (in contrast to ``upcycling generalist MoEs), avoiding the high costs of ground-up training. Our approach employs activation sparsity to extract experts. To compose experts, we examine the widely-adopted layer-wise router design and show its redundancy, and thus we introduce the pre-gating router decoupled from the MoE backbone that facilitates system-friendly pre-computing and lookahead scheduling, enhancing expert-aware batching and caching.Our codesign therefore addresses critical gaps on both the algorithmic and system fronts, establishing a scalable and efficient alternative for LLM inference in resource-constrained settings.$\textit{Read-ME}$