Unverified50% confidenceBenchmarkExact time
GPT-5.4-mini's estimates were within 2 percentage points of zero, showing no clear component language assignment effect.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027
Sources
Related Entities
Related Claims
UnverifiedIn the controlled GPT-2 experiments, although projection updates were clearly higher in lexical hits, the strict answer score of all experimental groups remained zero.74% similarUnverifiedIn the controlled GPT-2 experiments, the unanchored projection update group achieved a target string hit rate of 28.7% under natural language questioning, which dropped to 22.0% after adding module residual anchoring.71% similarUnverifiedIn 96% of levels, GPT-6 required fewer attempts to find the answer than the average person.70% similarUnverifiedThe GPT-4o-mini few-shot configuration had only one confidence value before and after calibration, and its correctness AUROC was only 0.500, indicating that the model has almost no ability to distinguish correct from incorrect predictions.69% similarUnverifiedThe Spearman correlation coefficient between tool description scores and correct model selection was only 0.04, with almost no statistical association.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114730API
curl https://kongchang.com/api/v1/knowledge/claims/1114730MCP
get_claim(id=1114730)