TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

H-Matched Tracker: Now with MathVista, PubMedQA, and Interactive Charts

1 点作者 mrconter114 个月前

1 comment

mrconter114 个月前
Hi! Quick update to H-Matched, the website tracking AI's progress toward human-level performance. Since last post, I've added 6 new benchmarks (now 20 total) and made the visualization interactive! The site shows how AI's 'catch-up time' has dramatically shrunk - from 6+ years with ImageNet to just months now. Explore the timeline with release dates, solve dates, and links to papers. Would love to hear your thoughts and if there are any benchmarks I've missed!