Related papers: Eliminating Hidden Serialization in Multi-Node Meg…

Mixture of Experts (MoE) models have emerged as the de facto architecture for scaling up language models without significantly increasing the computational cost. Recent MoE models demonstrate a clear trend towards high expert granularity…

Mixture-of-Experts (MoE) model architecture has emerged as a promising solution for scaling transformer models efficiently, offering sparse activation that reduces computational costs while increasing model capacity. However, as MoE models…

Mixture-of-Experts (MoE) has become a dominant architecture in large language models (LLMs) due to its ability to scale model capacity via sparse expert activation. Meanwhile, serverless computing, with its elasticity and pay-per-use…

Mixture-of-Experts-based (MoE-based) diffusion models demonstrate remarkable scalability in high-fidelity image generation, yet their reliance on expert parallelism introduces critical communication bottlenecks. State-of-the-art methods…

We propose Chain-of-Experts (CoE), a new Mixture-of-Experts (MoE) architecture that introduces sequential expert communication within each layer. Unlike traditional MoE models, where experts operate independently in parallel, CoE processes…

The Mixture-of-Experts (MoE) model has emerged as a prominent architecture in the field of Large Language Models (LLMs), providing a better balance between model performance and computational efficiency. However the General Matrix Multiply…

Sparse Mixture-of-Experts (MoE) layers route tokens through a handful of experts, and learning-free compression of these layers reduces inference cost without retraining. A subtle obstruction blocks every existing compressor in this family:…

Online analytical processing of queries on datasets in the many-terabyte range is only possible with costly distributed computing systems. To decrease the cost and increase the throughput, systems can leverage accelerators such as GPUs,…

The sparsely gated mixture of experts (MoE) architecture sends different inputs to different subnetworks, i.e., experts, through trainable routers. MoE reduces the training computation significantly for large models, but its deployment can…

The mixture-of-experts (MoE) architecture scales model size with sublinear computational increase but suffers from memory-intensive inference due to KV caches and sparse expert activation. Recent disaggregated expert parallelism (DEP)…

Mixture-of-Experts (MoE) models offer computational efficiency during inference by activating only a subset of specialized experts for a given input. This enables efficient model scaling on multi-GPU systems that use expert parallelism…

Despite the high computational throughput of GPUs, limited memory capacity and bandwidth-limited CPU-GPU communication via PCIe links remain significant bottlenecks for accelerating large-scale data analytics workloads. This paper…

A recent discovery of a new class of microarchitectural attacks called Spectre picked up the attention of the security community as these attacks can circumvent many traditional mechanisms of defense. One of the attacks---Bounds Check…

Multi-party computing (MPC) has been gaining popularity as a secure computing model over the past few years. However, prior works have demonstrated that MPC protocols still pay substantial performance penalties compared to plaintext,…

Diffusion Transformers (DiTs) are increasingly adopted in scientific computing, yet growing model sizes and resolutions make distributed multi-GPU inference essential. Ulysses sequence parallelism scales DiT inference but introduces…

Mixture-of-Expert (MoE) models outperform conventional models by selectively activating different subnets, named experts, on a per-token basis. This gated computation generates dynamic communications that cannot be determined beforehand,…

Signal processing is crucial for satisfying the high data rate requirements of future sixth-generation (6G) wireless networks. However, the rapid growth of wireless networks has brought about massive data traffic, which hinders the…

The optimization of the transpose convolution layer for deep learning applications is achieved with the kernel segregation mechanism. However, kernel segregation has disadvantages, such as computing extra elements to obtain the output…

The Mixture of Experts (MoE) architecture has emerged as a key technique for scaling Large Language Models by activating only a subset of experts per query. Deploying MoE on consumer-grade edge hardware, however, is constrained by limited…

Fully Homomorphic Encryption (FHE) enables privacy-preserving Transformer inference, but long-sequence encrypted Transformers quickly exceed single-GPU memory capacity because encoded weights are already large and encrypted activations grow…