A Bitcoin-native LLM: dataset, architecture and open questions

[0xB10C]

@Tsua00021

Hi @0xB10C Your post sounds pretty self-promoting. Do you want to add something precisely on the discussion?

I’m not sure I understand your problem with my comment.

@Tsua00021

• Bitcoin mailing list and Delving threads — years of high-signal technical discussion
• Bitcoin Core source annotations and functional tests — ground truth for consensus behavior

your are looking for datasets on technical Bitcoin discussion, I link to available GitHub issue comment and PR review datasets containing these kinds of discussions. While I happen to provide these to the community for free (which seems to be well received e.g. GitLab Backups for Bitcoin Core repository - #10 by TheCharlatan) and paying for the infrastructure to run them out of my pocket, I don’t see the problem with suggesting to use these GitHub comments and reviews as a dataset for LLM training. But feel free to ignore and don’t use them!