Unverified50% confidenceFactExact time
AlpacaEval 2.0 is an instruction-following benchmark containing 805 diverse user instructions that uses an LLM as an automatic judge and introduces length-controlled (LC) win rate to control for length bias.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEleutherAI 的 lm-evaluation-harness 是 NLP 领域广泛使用的公开评测工具64% similarUnverifiedThe evaluation adopted two approaches: automatic scoring based on large language models (LLMs) and human evaluation by 11 annotators.63% similarUnverifiedThe LFORLA evaluation mechanism contains four rules: fixed prompts, a fixed judging model, scoring split by dimension, and public verbatim answer records.63% similarUnverifiedLLM-as-a-judge is the most commonly used automated scoring paradigm in current safety evaluation, using a language model to label target model responses according to predefined categories.62% similarUnverifiedEvaluation uses automatic metrics (ROUGE and BLEU) and answer accuracy on multiple-choice generation, supplemented by a five-criteria human evaluation protocol.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134223API
curl https://kongchang.com/api/v1/knowledge/claims/1134223MCP
get_claim(id=1134223)