nasheedyasin / synthetic-squad Goto Github PK

Fix utility for task1, downsample rows for every algorithm

• Error analysis in the report
• Conclusion in the report
• Future research/room for improvement in the report
• Ensure that the tables, diagrams, confusion matrices, etc. are correct and properly labeled
• Make sure all code and files have been pushed to the GitHub repository
• Look over the PowerPoint
• Decide the division of work/whose turn is whose for the presentation

• Write about motivation for Tf-idf in the report (I started a little bit of this already)
• Create diagram for Tf-idf architecture
• Add results to the report (tables, confusion matrices)
• Decide how to summarize results in the PowerPoint

• Write about motivations for perplexity in the report (I have already started on this a little bit)
• Create diagram for perplexity architecture
• Upload perplexity results to the report (tables, confusion matrices)
• Decide how to summarize these results in the PowerPoint

• Check the data and make sure the labels are correct
• Retrain RoBERTa
• Get results and upload to report (confusion matrices, tables)
• Write about results in the PowerPoint

• Retrain DistilBERT (most likely)
• Remake diagram for BERT models
• Add results to tables and confusion matrices in the report
• Finish linguistic analysis
• Make sure all sources are cited
• Create section in the PowerPoint to discuss results

• add overall and granular F1 metric into the evaluation schema.