Unverified50% confidenceOpinionExact time
众包式盲测评测被认为比传统学术基准测试（如MMLU、HumanEval）更能反映模型在开放式对话中的综合表现
1
Sources
50%
Confidence
Long-term
Relevance
8/24/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTop-tier models have approached the level of human judges on many evaluation tasks, but exhibit biases such as a tendency to give higher scores to longer responses69% similarUnverifiedResearch finds that when a prompt assigns the model the role of a rigorous academic researcher, both its citation density and authenticity are significantly higher than in a general conversational role.69% similarUnverifiedHuman annotators exhibit a cognitive bias toward more information being better when evaluating answers, causing reward models to implicitly link length with quality, which is one of the causes of model verbosity.67% similarUnverifiedCurrent mainstream LLM evaluation benchmarks such as MMLU, HumanEval, and MT-Bench focus primarily on knowledge capabilities and instruction-following abilities, and systematic evaluation of sycophantic behavior has yet to form an industry standard.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/795684API
curl https://kongchang.com/api/v1/knowledge/claims/795684MCP
get_claim(id=795684)