Search papers, labs, and topics across Lattice.
1
0
3
Achieving 80% accuracy on VQA v2.0 proves that combining Visual BERT, ViLT, and memory-augmented attention can significantly outperform traditional VQA models.
