Have you tried fine tuning this? I am attempting on a single 40gb A100 but hitting OOM. If each sample of my training data is small it seems OK (false positive I think) but, as soon as I increase the token size of my training data I hit OOM errors.
· Sign up or log in to comment
