Summary by i_am_dom, 100 messages

[high dust]

Summary from #💬off-topic message to #💬off-topic message. This summary contains 121 messages. The conversation starts with users discussing different AI models and their potential. A user mentions the desire for the release of Trenkt GPT and another user suggests using GPT-J instead. There is a mention of Google potentially keeping the transformer to themselves, but Microsoft being excluded from the conversation. Some users engage in a playful banter about names and offensive remarks. The leaked parameters of GPT-4 are mentioned, causing some surprise. A user wonders about the use of GPT-3 in AutoModelForCasualLM, which leads to someone reporting them. The conversation then shifts to discussing the potential of models learning from themselves, the involvement of different companies in the AI race, and a joke about a random GIF posted. Some users express their skepticism about models learning independently, while others mention missed opportunities and the motivation behind AI development. There is a mention of a podcast, and a user shares a funny GIF. The conversation ends with discussions about the repo, GPT training data, the possibility of models learning from themselves, and the comparison between different models.