search

zengyan-97 / X-VLM

X-VLM: Multi-Grained Vision Language Pre-Training (ICML 2022)
BSD 3-Clause "New" or "Revised" License
442 stars 51 forks source link

The code saves the best testing results on Image-Text Retrieval #22

Open yangbang18 opened 1 year ago

yangbang18 commented 1 year ago

Hi, the following code is inappropriate:

https://github.com/zengyan-97/X-VLM/blob/e7b960256d194952321b5adad39770c03e6ce9c2/Retrieval.py#L310-L333