Skip to content
Lucas Lessa
Voltar para as notícias
DatabricksBetaImpacto: Médio

Use OpenJev as evaluator for LLM-as-a-judge service policies

Select the OpenJev model as the evaluator for custom LLM-as-a-judge service policies in Unity Gateway.

O que mudou

Use cases

Imagine you need to classify AI request content without generating additional output tokens, ensuring faster responses compared to a traditional chat evaluator. With this feature, you can:

  • Select the OpenJev model (Qwen3.5 4B) as the evaluator service for custom policies.
  • Classify content efficiently for AI governance.

Limitations and requirements

  • This feature is in Beta.
  • Account admins can control access to this feature from the account console Previews page.
Ver nota oficial do Databricks

Use OpenJev as the evaluator for LLM-as-a-judge service policies (Beta)

Resumo e explicação gerados por IA a partir da release note oficial. Em caso de dúvida, a documentação oficial da plataforma prevalece.

Relacionadas