LLM Metrics
Properties
created
19.10.2024, 15:22
modified
19.07.2026, 18:31
published
Empty
sources
AI Metrics that Matter
topics
Evaluation, LLM Metrics
authors
Empty
ai-assisted
No
- List of metrics that are used to evaluate performance and outputs of LLM powered applications.
# Metrics Categories
- Qualitative Metrics
- Quantitative metrics are objective, numerical measures used to evaluate specific attributes of a system or process.
- LLM-as-a-Judge
- Use of LLM to evaluate LLM outputs
- Agent-as-a-Judge
- Agent-as-a-Judge: Evaluate Agents with Agents
- 2024-10-16
- 104 citations
- A Survey on Agent-as-a-Judge
- 2026-01-08
- 0 citations
- Agent-as-a-Judge: Evaluate Agents with Agents
- Grounded Fact Verification
- Specialized classifiers (MiniCheck, FaithLens, NLI models) that score whether a claim is supported by a grounding document
- Rubrics based criteria scoring
- Other
- Deterministic Assertion - check that the output does not contain specified substring
- Cosine Similarity - between answer and expected answer