Unverified50% confidenceFactExact time
When the user's own historical background was fed to the evaluation model (user context-aware evaluation), the high-risk proportion rose by another 13 to 17 percentage points compared with context-blind evaluation.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCausal probes achieved a 17%–118% improvement in steering capability across different models, with only a 3% decrease in concept detection capability.68% similarUnverifiedIn the ablation study, when no data is used and only parametric memory is relied on, even the best model performs poorly; after providing early-quarter data, performance improves by about 25 percentage points; but there is still a 12-16 percentage point gap from using the latest data.67% similarUnverified早期模型在某些领域的事实幻觉率高达15-25%，尤其在长尾知识领域幻觉率更高65% similarUnverified研究表明即使是轻微的数据污染也可能导致评测分数虚高5-15个百分点65% similarUnverifiedA gap exists between high scores on internal evals and real-world performance; the Vercel team achieved 30% on evals but received poor real-user feedback.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023653API
curl https://kongchang.com/api/v1/knowledge/claims/1023653MCP
get_claim(id=1023653)