Apto: Self-Adaptive CPU-GPU Hybrid MoE Inference on Shared PCs
cs.DC
Submitted: 2026-11-01
Updated: 2026-11-01
Code: https://github.com/facebook/zstd
Terminology
Sources
- OptiMA: A Transaction-Based Framework with Throughput Optimization for Very Complex Multi-Agent Systems
- DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models
- DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
- Fast Inference of Mixture-of-Experts Language Models with Offloading
- Magentic-One: A Generalist Multi-Agent System for Solving Complex Tasks
- Mixtral of Experts
- Fiddler: CPU-GPU Orchestration for Fast Inference of Mixture-of-Experts Models
- Multi-agent Collaboration with State Management
- ProMoE: Fast MoE-based LLM Serving using Proactive Caching
- PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU
- Qwen3 Technical Report
- On-Device Language Models: A Comprehensive Review
- MoE-Infinity: Efficient MoE Inference on Personal Machines with Sparsity-Aware Expert Cache
- Qwen2 Technical Report
- SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
- EdgeMoE: Empowering Sparse Large Language Models on Mobile Devices
- Optimizing Sequential Multi-Step Tasks with Parallel LLM Agents
- AutoCodeRover: Autonomous Program Improvement
- SMoE: An Algorithm-System Co-Design for Pushing MoE to the Edge via Expert Substitution
Related papers
- Programmable Probabilistic Computer with 1,000,000 p-bits
- BatchGen: An Architecture for Scalable and Efficient Batch Inference
- Seamless Reconfiguration for DAG-BFT (Technical Report)
- ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving
- iScheduler: Reinforcement Learning-Driven Continual Optimization for Large-Scale Resource Investment Problems
- TIRx: A Unified Tile Primitive ML Compiler Across Heterogeneous Backends