Reading the paper it's apparent what a huge contribution HuggingFace made by doing the BLOOM work in the open. There's so much knowledge there that anyone else

Reading the paper it's apparent what a huge contribution HuggingFace made by doing the BLOOM work in the open. There's so much knowledge there that anyone else trying to train a LLM can use. Also it looks like public filings are a large dataset that isn't currently being used by other LLMs.