Closed abhishekdhiman25 closed 4 months ago
Hi @abhishekdhiman25. You can download your model files (pytorch_model.bin
and config.json
) into the tram/data/ml-models/bert_model/
directory. If you are training on a different set of techniques, you may also need to adjust classes.txt
to match.
Hi @abhishekdhiman25. You can download your model files (
pytorch_model.bin
andconfig.json
) into thetram/data/ml-models/bert_model/
directory. If you are training on a different set of techniques, you may also need to adjustclasses.txt
to match.
Hi @mehaase Thanks for your response. I have analysed both config.json files the original used in directory "tram/data/ml-models/bert_model/" and the one in fine tuned, they data inside looks same in both. Question: Can you please confirm if it is required to change data manually in config.json in fine tuning case. ? I am using the 100 ATT&CK labels for fine tuning, do I need to change the JSON data in config.json for 100 label.
Thanks for your help again.
By way of correction, the classes.txt
file needs to contain class labels in whatever order the model is expecting them. Changing the classes.txt file may break the model. Refer to my message in #214.
Closing this due to inactivity. Please re-open if your question has not been answered.
Hi Reader
I have installed TRAM using the developer's setup documnetation in my Windows 10 operating system. I have fine tuned Sci-BERT using the notebook "fine_tune_multi_label.ipynb " (path to notebook: tram >> user_notebooks >>fine_tune_multi_label.ipynb ). I used the my own JSON data for fine tuning which has same format as "multi_label.json". Now I want to know how can I use this fine tuned Sci-BERT for report analysis so that if I run the pipeline using command "tram pipeline run --model bert" the fine tuned Sci-BERT should run. I want to do this for some experiments and testing on my system.
Thanks for your help in advance.