Posts tagged #Links

A paper on pruning transformer attention heads during inference without retraining.