Any data sets produced before 2022 will be very valuable compared to anything after. Maybe the only way we avoid this is to stick to training LLMs on older data and prompt inject anything newer, rather than training for it.

You can run your own models at home if you don’t mind some tech struggle: https://github.com/ggerganov/llama.cpp lets you run the LLaMA/Alpaca/Vicuna/Wizard family of LLMs.

I might try Sea of Thieves just for this.

Pretty hyped for this one. Please don’t be another Redfall.

I paid $1100 for a 3070 during the pandemic with a newegg bundle deal (trash stuff they couldn’t sell). I already had a 2070 and it was a complete waste of money.