Apply for access

Set up an account; an admin will review and approve. You'll receive an email when the application has been reviewed.

Most of these questions are optional, but the more you tell us the easier it is to evaluate your application — if we don't have enough signal we'll err on the side of rejecting.

Fields marked * are required.

Email Name (a full name is preferred; a publicly used pseudonym is acceptable) Organization (or “Independent”) Profile link Link to a profile showing your work, e.g. Google Scholar, X/Twitter, LessWrong, LinkedIn, etc. Planned usage What kinds of experiments or explorations would you run on the platform? Please include specifics if you can. Results What would be the ideal results of your explorations, and why would they be interesting? Where will you publish the results (arXiv, LessWrong, X/Twitter, blog, …)? Prior work Relevant prior work or thinking in LLM psychology, interpretability, cyborgism etc. Link us to it — X/Twitter, LessWrong, blog, Google Scholar, papers. If you don't have a good link: what have you previously noticed or found interesting about base models? How did you hear about us? E.g. X/Twitter, LessWrong, a Slack/Discord community, a friend, a talk — ideally with specifics (which account, which community, who).

Platform rules

Your access is personal. Do not share your account or API keys. If someone else needs access, point them to this signup form.
Do not use the platform for illegal, harmful, or clearly unethical purposes.
This platform is in research-preview stage: expect rough edges, and tell us when something breaks.
If the platform significantly helps you gain insights, please consider linking to https://infra.acsresearch.org in the paper or post presenting them.

I have read these rules and agree to follow them.

Already have an account? Sign in.