LLM contributions to Yosys are on the rise and we are considering adding LLM-specific rules to contribution guidelines. We’ve had brief discussions of the applicability of LLM coding tools internally, and I would like to gather feedback from contributors and potential contributors as well. Submissions are not anonymous but will be anonymized and aggregated before sharing with anybody beyond myself, nobody has g drive access to responses other than me.
2 Likes
I’ve submitted my response, but would you consider opening this up as a general forum for discussion? A lot of these questions were quite free-form and I think it might be good for the community to discuss if it doesn’t get too heated (although I am not actually aware of any YosysHQ maintainers/contributors actually in favour of using LLMs extensively?)
Thank you for your response! Please refrain from discussing LLMs on this thread, and ideally, this forum, before we have some internal discussions among maintainers.
There are two sorts of setting where I would trust things not to get heated:
- settings where everybody agrees already entirely (I think you’re supposing this is the case) including but not limited to algorithmic echochambers
- in social groups where every pair of two people has a long-standing friendship that is backing genuine good faith required to discus differing views
The answers so far are diverse across respondents. There is currently no consensus to what is permissible or which associated problems are important. You can also see that currently other open source projects are running the entire gamut of possible policies and approaches