ELDR

ELDR is an AI model that uses expert signatures from prefill to route decode requests, reducing median TPOT by 5.9–13.9% in vLLM at scale. Developed by an unspecified team, it enhances inference efficiency through selective computation.

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.