#moe
4 briefs tagged #moe.
Research
New studies probe serving systems for agentic AI
Four papers examine latency, caching, traces and edge deployment as LLM workloads become more agentic.
arXiv cs.AI+1 outlet·1d ago
Research
Researchers target MoE inference bottlenecks
New papers propose MoE designs for faster vision encoding, decoding and load balancing.
HF Daily Papers+1 outlet·2d ago
Research
Motif 3 report details 314B-parameter MoE model
The technical report describes a sparse decoder-only model with 13.2B active parameters per token.
HF Daily Papers+1 outlet·Aug 10
Models
LG AI Research releases K-EXAONE 2.0
The open-weight multilingual MoE model has 750B total parameters and a 256K-token context window.
HF Daily Papers+1 outlet·Aug 5