Unverified50% confidenceBenchmarkExact time
In the experiment, under the naive mean method, the concept distance to the anchor was 0.326 and the similarity to the attack target was +0.413, while under the controlled anchor method, the distance to the anchor was 0.093 and the similarity to the attack target was only +0.018.
1
Sources
50%
Confidence
Long-term
Relevance
8/17/2026
First Seen
Sources
抗投毒概念锚定：防御AI数据污染的新思路
Related Claims
UnverifiedIn the queue admission control task, the RL policy improved from a baseline of −16.2 to 25.6.55% similarUnverifiedOn the AIME-2024 mathematical reasoning benchmark, pass@1 improved from 0.39 to 0.49.54% similarUnverifiedAprilTag 36h11采用6×6数据位矩阵加4层边框，汉明距离为1152% similarUnverifiedPathway reported that a BDH-based system achieved a 97.4% Top-1 accuracy rate on approximately 250,000 Sudoku Extreme puzzles without using CoT or solution backtracking.52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/765128API
curl https://kongchang.com/api/v1/knowledge/claims/765128MCP
get_claim(id=765128)