LLM Digest

Story

arxiv_cs_lg · Oct 27, 2026 · paper

Source brief

CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers

arxiv.orgOct 27, 2026
original source linked

In brief

Mixture-of-Experts (MoE) inference on memory-constrained edge platforms is often dominated by expert-weight movement rather than computation. Although each token activates only a small subset of experts, routing is in...

Read the original at arxiv.org →Open in live feed Read that day’s brief

CRISP: Co-designing Pruning and Scheduling for Efficient MoE Inference on Edge Servers

Earlier in this thread 4 items

DeepSeek-V4.1-Flash Packs 552B Parameters With Efficient MoE Inference

EmbeddingGemma 2: Unified and Efficient Multimodal Representations

Uber Moves Beyond CPU Cores with Performance-Aware Scheduling for Stateful Workloads

SEMA-Auto: Sighted and Efficient Multi-Turn Attack with Autonomy