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。

Show HN: A simple benchmark for AIGC that is actually difficult

1 点作者 tianshuo大约 1 年前
I've created a benchmark of 10 text-to-image tasks and matched 10 pictures for each task, which are pictures that you can easily find on Google. It is extremely easy to check that for every task, all the current AIGC platforms are totally incapable of creating pictures that can fulfill them, so this is a benchmark that can be used to check the intelligence of future AIGC frameworks.

1 comment

tianshuo大约 1 年前
I greatly respect how image benchmarks like MNIST and Imagenet have contributed to AI advancements, so this is my attempt at developing a benchmark. Please leave feedback.