frames_judge
frames_judge
¶
FRAMES scorer — LLM-as-judge for multi-hop factual retrieval.
Adapted from IPW's frames.py evaluation handler.
Classes¶
FRAMESScorer
¶
FRAMESScorer(
judge_backend: InferenceBackend, judge_model: str
)
Bases: LLMJudgeScorer
LLM-as-judge evaluation for FRAMES multi-hop factual retrieval.