sunxm2357/DIME-FM

Inquiry about the Reproducibility of Results and Required GPU for Open-Source Code

Opened this issue · 2 comments

I am highly interested in your project and am planning to replicate the results mentioned in the associated paper. I am currently reading the relevant research paper and would like to inquire whether the open-source code provided in your project is capable of reproducing the results reported in the paper.

Furthermore, I would appreciate it if you could provide information regarding the recommended GPU specifications necessary to run the code efficiently. Understanding the GPU requirements will help me ensure that I have access to the appropriate hardware resources for successful replication.

Thank you for your attention to this matter. I look forward to your response.

@liuxy1103 Did you reproduce the results successfully?

For ViT-B/32 experiments, we use 12288 as our batch size and 24 A100 GPUs for the parallel training. Hopefully, this information is helpful.