center-for-threat-informed-defense / tram

TRAM is an open-source platform designed to advance research into automating the mapping of cyber threat intelligence reports to MITRE ATT&CK®.

Home Page:https://ctid.mitre-engenuity.org/our-work/tram/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Need Guidance: How can i generate config.json and pytorch_model.bin for fine tuned SciBERT?

abhishekdhiman25 opened this issue · comments

Hi Reader
Greetings,

I have installed tram using developer's setup guide on my windows 10 OS.
I also performed the fine tuning using "fine_tune_multi_label.ipynb" notebook (path: tram\user_notebooks\fine_tune_multi_label.ipynb).
I know that to use this fine tuned Sci-BERT, I have to change corresponding config.json, pytorch_model.bin and classes.txt in bert_model directory (path: tram\data\ml-models\bert_model).
I have generated classes.txt with my data.
I analysed config.json file in both directories scibert_multi_label_model directory (under user notebooks generated after running fine tuning notebook) and bert_model directory (path: tram\data\ml-models\bert_model), Both JSON files are similar.
Even after running fine tuning notebook with my data the config.json was not updated under scibert_multi_label_model directory.

Question: To use fine tuned Sci-BERT for report analysis, Do I need to generate config.json and pytorch model.bin separately, If Yes how to generate them. If No how these files will be updated automatically.
Please correct me if I am wrong somewhere.

Thanks for your help in advance.

I believe this is answered in #214. Please re-open if you feel that answer is insufficient.