I don’t love the lack of quantitative comparison to Alpaca but a commercial model (which sounds like it’s in the works) would finally move the needle on democra

I don’t love the lack of quantitative comparison to Alpaca but a commercial model (which sounds like it’s in the works) would finally move the needle on democratizing access to LLMs. Will also commend the authors for not falling into the “LLMs can’t perform without 200B params!” fallacy. For anyone reading, 6B params is enough to train on a 3090. A PC rig for training or running inference with this would put you back maybe 4k$. The end game here is likely getting the model to perform well in millions of parameters on specific tasks. Most business uses of ChatGPT are pretty closed domain tasks, it wouldn’t be a huge step to distill this model on a specific task and get it down to 150-350M params (which is roughly BART size and can run on AWS Lambda even).