Unverified50% confidenceBenchmarkExact time
On RCAEval under default settings, the graph structure's ranking is on par with the strong baseline, with wins and losses across different systems, and lags by 4 percentage points on the held-out Train Ticket.
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAfter re-evaluating with non-overlapping anchors, the net advantage instead grew from +4.1pp to +5.5pp, but significance dropped from 2.4 standard errors to 1.4 standard errors; the inflation factor is about 2.4x, exactly equal to sqrt(912/163).56% similarUnverifiedNeural 3's three-way routing mode is: confidence ≥ 0.7 executes directly, the middle range asks the user for confirmation, and no return results in a reply that the task cannot be completed; the engine's default confidence lower limit is 0.5.56% similarUnverifiedThe combination of EfficientNet-B3 and SVM achieved the strongest baseline performance, with an accuracy of 96.31% and a macro F1 score of 96.36%.56% similarUnverifiedAt 16× compression, AdaMem's substring match improves by up to 3.2 percentage points (5.5%) over the uniform allocation baseline, with an average relative gain of 3.4%.55% similarUnverifiedAcross a range of question sizes spanning 256x, the error judgment rate of a single run is almost unchanged: 0.531 at T=16 and 0.534 at T=4096.54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1054867API
curl https://kongchang.com/api/v1/knowledge/claims/1054867MCP
get_claim(id=1054867)