Uni-Fold supports online evaluation with the UnifoldDataset created by the task, which uses the eval_*.json configuration files to specify the evaluation metrics.
Notably, for large evaluation sets, we don't recommend online evaluations, because it takes long. Alternatively, evaluate the dumped ckpts with additional resources (different gpus).
Uni-Fold supports online evaluation with the UnifoldDataset created by the task, which uses the eval_*.json configuration files to specify the evaluation metrics.
Notably, for large evaluation sets, we don't recommend online evaluations, because it takes long. Alternatively, evaluate the dumped ckpts with additional resources (different gpus).