Could you kindly add samples for LM Fine-tuning?

Open dkfjdkfjdkfj opened this issue 5 years ago • 5 comments

I failed in LM fine-tuning with such an error: RuntimeError: CUDA error: device-side assert triggered I don't know how to solve it even though I made many attempts.

It is appreciated if the sample for fine-tuning could be provided.

Feb 06 '20 10:02 dkfjdkfjdkfj

I'm having the same issue and can't seem to resolve.

Feb 07 '20 21:02 aubreycondor

any updates?

Mar 09 '20 17:03 jkhalsa-arabesque

I used transformers language model training to do this. this page

Mar 10 '20 06:03 dkfjdkfjdkfj

I used transformers language model training to do this.

How does this help?

Mar 18 '20 00:03 Shane-Neeley

So this issue has actually been resolved elsewhere https://github.com/kaushaltrivedi/fast-bert/issues/179

It is to do with the version of the transformers library being used. If you downgrade to 2.3.0 it works fine.

Mar 18 '20 09:03 jkhalsa-arabesque