MLLM

LLM-as-a-Judge is a technique in natural language processing in which a large language model (LLM) is used to assess the quality, relevance, or correctness of a text output, typically one generated by another model, against predefined criteria. It is used as a scalable, lower-cost alternative to hum

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.