Thank you for organizing the challenge.
I have a question regarding the training setup.
Would it be permissible to create an additional dataset for fine-tuning that is similar to the Phase 2 test dataset?
Specifically, I am wondering whether it would be allowed to construct a dataset that includes the 9 potential causes in the Phase 2 test dataset but not included in the original Telelogs training dataset, or some background knowledge that may be helpful to solve general questions in phase 2 test data.
Thank you.
Yes. You can create sinthetic data in the format you prefer for fine-tuning your model
Thank you for your kind reply!
Are we constrained to use same qwen models to create this synthetic data?
I've asked this in another thread and as far as I recall you said to use same qwen model as the task entails. Is this still true?