News for hackers. Stuff that matters.
Re: by jph00 on Thursday July 27, 2023 @06:29PM (#36900996)
I haven't seen any recent papers that show that fine-tuning is obsolete - I've only seen papers showing the opposite. I'd be very interested to see any papers that have demonstrated applications where fine-tuning is not effective nowadays, if you have any links.

Here's an example of a paper that shows good results from fine-tuning for instance: https://arxiv.org/abs/2305.02301

Re: by treprinum on Thursday July 27, 2023 @06:31PM (#36901015)
This Stanford seminar video can provide some references:

https://www.youtube.com/watch?v=tVtOevLrt5U

Re: by jph00 on Thursday July 27, 2023 @09:16PM (#36902107)
That video shows examples of new models with one-shot performance better than old models with fine-tuning. It doesn't compare new models with fine-tuning to new models without fine-tuning.
Re: by treprinum on Friday July 28, 2023 @07:41AM (#36905461)
Right, the page 28 of this article mentions "On some benchmarks, task-general models (not explicitly trained to perform a task) surpass prior state-of-the-art performance held by a task-specific model" but they don't provide data for fine-tuning of their very large models:

https://arxiv.org/pdf/2206.07682.pdf

Re: by ssivark on Thursday July 27, 2023 @07:09PM (#36901299)
What’s your take on the argument that fine tuning might be destroying model calibration and causing it to overfit to aspects of the finetuning dataset eg: causing it to appear more confident, rather than becoming more correct?

See https://twitter.com/animaanandkumar/status/16816906540601876...

Re: by jph00 on Thursday July 27, 2023 @09:12PM (#36902087)
That tweet is about a specific type of fine tuning: instruction tuning and RLHF. It causes to get better at the thing that it's being fine tuned for, which is making users feel better about the answer. Whether that's actually desirable in all cases is the question -- but that's not really related to whether fine tuning is a useful tool in general.