Really? They spend $16.5M (Yep, I did the calculation myself) and release a SOTA model with open weights, and this is TechCrunch's headline. What the actual fuck dude?
@migtissera It sure would be nice to have tech journalists who love and appreciate tech!
@migtissera (what were the key assumptions for your calculation?) we dont know if the 2 months we’re continuous usage
@migtissera Well, to the average ChatGPT user that's probably what matters.
@migtissera centralists don't want open source AI to blossom. Journalist job security benefits from heavy restrictions on LLMs and a weak open source LLM ecosystem. This should not happen, in their eyes.
@migtissera Seems headlines are GPT-4 generated, would explain the degradation in quality 😂
@migtissera They could’ve also mentioned its 1/10 the size of gpt4 with similar performance, without any RLHF or DPO for better performance (not that safety bs open ai does making their models dumber but ‘safer’).