News for hackers. Stuff that matters.
Re:DyLoRA: Parameter Efficient Tuning of Pre-Trained Models by Ldorigo on Tuesday April 11, 2023 @03:05AM (#35521798)
Why is the only advantage at training time? I might misunderstand something but with this method you can train once, and then deploy models that use arbitrary rank (according to end-users compute requirements) and expect to have a model that performs best for that specific rank.

There are no comments on this comment yet.