Unverified50% confidenceBenchmarkExact time
Experimental results show that answer changes produced by different-category substitutions are 49.7 to 60.9 percentage points higher than those from same-category substitutions.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/28/2026
First Seen
Valid until: 1/26/2027
Sources
Related Entities
Related Claims
UnverifiedAblation analysis shows that fine-tuning on the segment paradigm yields a +5.29-point improvement on the test set, while replacing the first 4,500 transcript words with 2,000 retrieved words yields only a +1.55-point improvement.78% similarUnverifiedIn a test of 119 pairs of "near-identical questions", at the 0.81 threshold 62% of modified questions incorrectly hit; even raised to 0.90, 25% still hit; the hit rate for genuine synonymous rewrites dropped from 39% to 4%.72% similarUnverifiedIn earlier tests, a rewrite that raised a tool's score from 31 to 74 instead made both models' selection performance worse.71% similarUnverifiedThe correction method tested by the researchers—averaging each answer with the complement of its paired answer—restores complementary consistency by construction and lowers Brier loss averaged over the two phrasings.71% similarUnverifiedRewriting one criterion in a prompt causes the pass rate of other criteria within the same prompt to fluctuate by 5 to 10 percentage points.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1105811API
curl https://kongchang.com/api/v1/knowledge/claims/1105811MCP
get_claim(id=1105811)