The thing that annoys me most is that there have been studies done on LLMs where, when trained on subsets of output, it produces increasingly noisier output.

Sources (unordered):

• What is model collapse?
• AI models collapse when trained on recursively generated data
• Large Language Models Suffer From Their Own Output: An Analysis of the Self-Consuming Training Loop
• Collapse of Self-trained Language Models

Whatever nonsense Muskrat is spewing, it is factually incorrect. He won’t be able to successfully retrain any model on generated content. At least, not an LLM if he wants a successful product. If anything, he will be producing a model that is heavily trained on censored datasets.

This is way too opinionated imo. Many of these options I disagree with. There are github repos with much larger content suggestions and they all benefit from offering multiple options. There is often no such thing as a singular “best”.

So, my opinion of this list is that, unfortunately, while it is informative by listing names of software, it offers zero description as to why you’ve deemed it best. As I said, often times software has trade-offs, and that allows a competitor can be a better choice when those weaker areas are more important to a user. I suggest thinking about that and improving to fit that into your site.

But again, repos exist that already cover this in depth. So try to be different. Perhaps do reviews, add screenshots, and do more than those repo readmes can.

I want to use Thunderbird but my university won’t let me log into my email outside of Outlook… So dumb.