This is the repository for the Analysis of Fine-tuning RoBERTa for Content Moderation on Reddit. For this, we propose a RoBERTa model fine-tuned on a labeled dataset of flagged comments from a subreddit to create an automatic comment-flagging model specific to the rules of a given subreddit.
- The glue.py file describes the changes made to the Transformers repository for our project at the following path: https://github.com/huggingface/transformers/tree/master/src/transformers/data/processors.
- The init.py file describes the changes made to the Transformers repository for our project at the following path: https://github.com/huggingface/transformers/tree/master/src/transformers/data/metrics."