Unverified95% confidenceFactTime unknown
MEME基准测试中表现最好的记忆系统MD-Flat整体准确率仅为0.42
1
Sources
95%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
MEME基准测试揭示LLM记忆系统致命缺陷:依赖推理准确率不足50%
bilibili熊二等兵
Related Entities
Related Claims
UnverifiedQwen3 27B在256K长上下文记忆检索测试(大海捞针)中取得100%通过率,15次运行全部准确66% similarUnverified标称0.1g精度但5kg大量程的秤要警惕,量程越大分度值越大时误差越明显,真正0.1g稳定读数的一般量程不超过3kg62% similarVerifiedCoT提示将PaLM 540B在GSM8K数学基准测试中的准确率从17.9%提升至58.1%61% similarUnverifiedPathway报告称基于BDH的系统在约25万道Sudoku Extreme题目上达到97.4%的Top-1准确率,且未使用CoT、未进行解答回溯60% similarUnverifiedFP8量化相比FP16原始精度,在主流benchmark上的性能下降通常不超过1-2%60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/21542API
curl https://kongchang.com/api/v1/knowledge/claims/21542MCP
get_claim(id=21542)