Datasets:
XAI
/

Modalities:
Image
Text
Formats:
parquet
Languages:
English
ArXiv:
DOI:
Libraries:
Datasets
pandas
License:

请问大家有在qwen2.5-vl上测试过这个数据集么

#2
by Curitis7 - opened

想知道大家测试下来精度是多少

@Curitis7

感谢您的来信!

我们在论文修订版第 4.7 节 (https://arxiv.org/pdf/2407.06581) 中对 Qwen-2.0 VL 进行了 BlindTest 测试,结果显示其表现与 Gemini 1.5 Pro 不相上下.

Thank you for reaching out!

We have tested Qwen-2.0VL on BlindTest (section 4.7 in https://arxiv.org/pdf/2407.06581), and it achieves 54.13% accuracy. We find that to be almost on par with Gemini 1.5 Pro (58.48%).

Sign up or log in to comment