Oscar Mañas
Oscar Mañas
Hi, Have you tried running DeepDraw with other net architectures? For example, I've been able to run it with the ResNet-50 targeting the Softmax layer, but I'm not getting useful...
The goal of a machine learning algorithm is to **minimize** the expected validation/generalization error, not **maximize** it [[1]](#1). [1] Goodfellow, Ian, Yoshua Bengio, and Aaron Courville. Deep learning. MIT press,...
I was trying to reproduce results with BLIP on VQAv2 test-dev and I observed a non-negligible difference between the VQA accuracy obtained using the [published checkpoint](https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_vqa_capfilt_large.pth) (**77.41%**) and the number...