Closed Novestars closed 5 months ago
In the paper, wd was 0, while in the code base wd is set to default value which is 0.02
@Novestars Appreciate you bringing this to our attention. The wd is set to 0 during the training of both EVA-CLIP-18B and EVA-CLIP-8B models.
In the paper, wd was 0, while in the code base wd is set to default value which is 0.02