Sebastian Raschka
Sebastian Raschka
Thanks @vmirly! Would you mind adding this to the Errata document here? https://github.com/rasbt/machine-learning-book/tree/main/ERRATA
Hi there, and sorry for the late response, it's been a super intense week. Regarding the tokenizer, usually there shouldn't be a modification necessary as it loads the tokenizer from...
Oh sure, that would be nice!
Thanks for the feedback! Regarding your question, yeah they way they motivate DoRA and the way it's implemented is a bit misleading. So in their Eq 1 they formulate the...
Whoa, thanks for doing this! This is unexpected, I agree. Perhaps it could be an artifact of the relatively small DistilBERT model, but there could also be something else going...
Ah yes, that was incorrect. I updated it and think it should be correct now (need to think about it a bit more before actually implementing).
Thanks for the update. I am currently wrapping up another project but hope to get back to that soon and integrate it into Lit-GPT for experiments on other LLMs like...
Thanks for your interest in this. I think it would require a few adjustments to use it. We would have to make those carefully to ensure everything works correctly. I...
> will DoRA be integrated into lit-gpt? Sorry for the late response. I currently have some other work projects I have to get to first, but yeah, ultimately, I am...
This looks awesome! Thanks for sharing. Just saved & sent it to my e-reader and plan to reading it carefully in the upcoming days!