Unverified50% confidenceBenchmarkExact time
The 7 models that passed the surrogate test had a change rate of 15.9%, while the 6 models that did not pass had a rate as high as 25.8%, indicating that better-performing models are less susceptible to interference but still more so than the effect of prompt modification.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
Unverified靶向数据工程中，针对性数据占总微调数据的 5%-15% 通常能取得良好效果，过高比例可能导致模型在其他维度退化73% similarUnverifiedAmong the 7 models that passed the test, adding aligned images led to a 15.9% label change, adding misleading images led to a 14.7% change, while the control group that only modified the prompt had just 8.8%.72% similarUnverifiedAfter matched supervised training, model performance improved to 99.8%, while removing the relational links caused accuracy to plummet to 23.4%, even below the best shortcut's 27.0%.72% similarUnverifiedMeta AI research shows that having a model iteratively optimize its own prompts can improve performance by 15-30% on certain standard benchmarks, but carries approximately a 20% probability of falling into a degenerative loop.72% similarUnverifiedOn performance model code construction tasks, the average pass rate of all model configurations other than GPT-5.6 Sol was below 15%, with extremely large fluctuations across multiple runs.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/977925API
curl https://kongchang.com/api/v1/knowledge/claims/977925MCP
get_claim(id=977925)