Unverified50% confidenceBenchmarkExact time
In SWE-bench trajectory evaluation, when using information from the first 8 steps, OnTrack's ability to rank failed trajectories is better than content similarity methods, with an AUROC higher by 0.057.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
VerifiedIn class-imbalanced scenarios, Accuracy is misleading, and Precision-Recall curves and AUC-ROC are more reliable performance measurement tools.67% similarUnverifiedA model's utilization of context follows a U-shaped curve: accuracy is highest at the beginning (primacy bias), second highest at the end (recency bias), with a notable dip in the middle.66% similarUnverifiedOpenAmer's self-improvement loop reads the outcome ledger, and improving the pass rate means overcoming the dominant failure records66% similarUnverifiedThe scale recovery accuracy of monocular VIO is highly dependent on IMU calibration quality and initialization motion excitation; if only slow translation without rotation occurs during the startup phase, the scale estimate may be inaccurate.65% similarUnverifiedThe noticeably higher error in the classification stage compared to the detection stage in the end-to-end scenario reflects the effect of error cascading in multi-stage pipeline systems.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1013315API
curl https://kongchang.com/api/v1/knowledge/claims/1013315MCP
get_claim(id=1013315)