search

clinicalml / TabLLM

MIT License
265 stars 42 forks source link

Could you please upload the entire source codes? #11

Open A11en0 opened 1 year ago

A11en0 commented 1 year ago

Thanks for your interesting work! I clone your repo and try to reproduce your results in the paper, however, I find that there are too many missing modules in these codes, including in the procedure of data processing, models, etc. So, could you please clean your codes again and add these missing parts, I think it would significantly benefit and push forward a large step in the field of tabular learning.

stefanhgm commented 11 months ago

Hello @A11en0,

thanks for reaching out and using our code!

We agree that it would definitely be simpler to provide the entire code. However, as stated in the readme, we relied on the codebase of the t-few project for our main experiments. We did not want to copy over their code to benefit from future improvements of t-few and credit their work. We added all necessary files for using t-few in the folder t-few.

Could you already complete all steps in the readme? It would be helpful if you could provide some more details about your specific problem, e.g. error messages or unclear instructions in the readme.

stefanhgm commented 10 months ago

Hello @A11en0 ,

just as a heads-up: based on the feedback in the issues, we updated the readme now with all steps to reproduce a performance entry from the paper. Maybe that is also helpful for you!