Bagel在UniGenBench上的效果

作者你好，

我们测试了Bagel在我们最新提出的[UniGenBench](https://github.com/CodeGoat24/UniGenBench) (Long/Short English/Chinese)四个榜单上的表现，无论是中/英文还是长/短prompt的效果在开源模型中都非常惊艳！

我们已经更新到了我们的leaderboard中，欢迎关注～

1. 短英文leaderboard：[CodeGoat24/UniGenBench_Leaderboard](https://huggingface.co/spaces/CodeGoat24/UniGenBench_Leaderboard)

2. 长英文leaderboard：[CodeGoat24/UniGenBench_Leaderboard_English_Long](https://huggingface.co/spaces/CodeGoat24/UniGenBench_Leaderboard_English_Long)

3. 短中文leaderboard：[CodeGoat24/UniGenBench_Leaderboard_Chinese](https://huggingface.co/spaces/CodeGoat24/UniGenBench_Leaderboard_Chinese)

4. 长中文leaderboard：[CodeGoat24/UniGenBench_Leaderboard_Chinese_Long](https://huggingface.co/spaces/CodeGoat24/UniGenBench_Leaderboard_Chinese_Long)

我们还开源了所有模型用于评测的图像数据，可以在[CodeGoat24/UniGenBench-Eval-Images](https://huggingface.co/datasets/CodeGoat24/UniGenBench-Eval-Images)中下载～

Start by reviewing the linked UniGenBench leaderboards and evaluation-image dataset to understand the benchmark result being reported. The issue does not name a file, test, requested change, or completion criterion, so no implementation path or definition of done is provided.

Written by the indexing model from the issue text.