Unverified50% confidenceBenchmarkExact time
After the 1-billion-parameter model was adapted to a fraud detection scenario, detection accuracy improved by 24% compared to the previous solution.
1
Sources
50%
Confidence
Long-term
Relevance
10/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn a fraud detection task with a positive-to-negative sample ratio of 1:99, a model that predicts all samples as 'non-fraud' can still achieve 99% accuracy.73% similarUnverifiedAfter introducing a gating mechanism based on the measurement metrics, accuracy improved from 20.7% to 36.9%, while the system still answered half of all comparisons.69% similarUnverified阿里千问新一代模型参数达2.4万亿，是该系列首个突破万亿的多模态大模型，代码生成准确率据称达到94.7%，较上一代提升约11个百分点68% similarUnverified参数量在1亿以下的轻量模型经领域数据微调后可达95%以上意图分类准确率，单次推理延迟控制在5-20ms以内68% similarUnverifiedA sample size of 30 is relatively small, and this accuracy rate is better suited as a feasibility validation rather than a rigorous performance benchmark.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/999172API
curl https://kongchang.com/api/v1/knowledge/claims/999172MCP
get_claim(id=999172)