I find that bart large 410m [0] parameters does a fine job at summarizing. In Summer AI I alternate between using a copy of that bart large getting hyper-traine

I find that bart large 410m [0] parameters does a fine job at summarizing. In Summer AI I alternate between using a copy of that bart large getting hyper-trained on feedback and Chat GPT 3.3 and honestly I don't have a preference between the results. However, thanks to this article I might revisit the summarization techniques used a try a fine tuned 3.5. It would be great to see these techniques compared to Chat GPT 4 Turbo. [0]: https://huggingface.co/facebook/bart-large-cnn https://huggingface.co/facebook/bart-large-cnn