LLM Output Evaluation

高级testing最低上下文：32K

Defines repeatable quality evaluation for LLM outputs using representative datasets, scoring rubrics, model-graded checks, human review sampling, and regression thresholds.

llmevaluationqualityregression-testing

使用场景

  • LLM quality gates
  • Prompt benchmarking
  • Model migration validation

示例提示词

Create an evaluation plan for this LLM feature, including representative cases, a scored rubric, automated checks, human-review sampling, and release thresholds.

推荐模型

claude-opus-4-6gpt-5gemini-2-5-pro

兼容工具

claude-codecursorkiroany

模态

输入: text, file
→
输出: text, code

相关 Skills

llm-eval-harness-builderprompt-engineering

作者

OpenModels Community

@openmodelsrun