Unverified50% confidenceFactExact time
The benchmark consists of 60 minimal-pair cases (30 minimal pairs spanning six categories) and 12 adversarial control cases.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn few-shot scenarios (e.g., approximately 20 samples per class), multi-class subdivision should be abandoned, all defects should be collapsed into a binary classification problem (defective/non-defective), and specific classification should be left to manual post-processing.68% similarUnverifiedAmong 27 evaluation settings, distillation narrowed the group gap in 21 of them.68% similarUnverifiedEach model was paired with four prompt types—zero-shot, few-shot, chain-of-thought, and few-shot plus chain-of-thought—each in two groups, producing approximately 52,000 labels in total.63% similarUnverifiedThe study's evaluation covers 11 agent decision points, with data drawn from 18 public sources, comprising a total of 7,283 base cases plus 6,640 robustness variants.62% similarUnverifiedAmong the 500 divergence positions, 19 had immediately executable lethal combinations; Jev selected lethal 16 times, the neural network selected it 0 times, and in the remaining 3 cases neither chose it.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1003395API
curl https://kongchang.com/api/v1/knowledge/claims/1003395MCP
get_claim(id=1003395)