PrismServe: Efficient LLM Serving on Heterogeneous Accelerator Clusters
cs.LG, cs.AI, cs.CV
Submitted: 2025-08-09
Updated: 2026-10-31
Code: https://github.com/paper-submission-goii/MoQE
Terminology
Sources
- Learning Factored Representations in a Deep Mixture of Experts
- GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers
- Improving Post Training Neural Quantization: Layer-wise Calibration and Integer Programming
- BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction
- MultiWay-Adapater: Adapting large-scale multi-modal models for scalable image-text retrieval
- Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis