Unverified50% confidenceFactExact time
Jev's scoring primitive supports up to 11 rating levels (0 to 10).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/20/2026
First Seen
Valid until: 12/19/2026
Sources
Related Entities
Related Claims
UnverifiedFree-text generation in Med-VQA evaluation exhibits answer position bias of up to 10 percentage points.63% similarUnverified系统用轻量小模型对每篇论文摘要进行 1-10 分的相关性评分，评分依据是用户编写的描述研究兴趣的 Markdown 文件59% similarUnverifiedThe diarization error rate (DER) on standard evaluation sets has dropped below 10%.59% similarUnverifiedThe core of evaluation is scoring, typically expressing the quality of results using a continuous value from 0 to 1.59% similarUnverifiedInter-annotator agreement among different human annotators in open-ended text quality assessment is typically only 0.6–0.8.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/936347API
curl https://kongchang.com/api/v1/knowledge/claims/936347MCP
get_claim(id=936347)