Open ChengLeiYu-Maker opened 1 year ago
Hello, if I'm understanding the question correctly, you're interested in running parallel predictions in a single batch, which is possible in principle. However, because the refinement step for each prediction isn't run on GPU, there would likely be little effect on overall runtime for full-atom structure prediction. If you just need an embedding or an approximate distance-matrix, batch mode should be faster.
How can I use batch mode for multiply sequence as input? Can you give an example? Thanks
Can we use put multiple antibody sequences into the model ?