This is a heavily interactive web application, and JavaScript is required. Simple HTML interfaces are possible, but that is not what this is.
Post
Ferenc Huszár
inference.vc
did:plc:dwqbbgbceprottqihgkziz4x
Can language models transcend the limitations of training data?
We train LMs on a formal grammar, then prompt them OUTSIDE of this grammar. We find that LMs often extrapolate logical rules and apply them OOD, too. Proof of a useful inductive bias.
Check it out at NeurIPS:
https://nips.cc/virtual/2024/poster/95576
2024-12-06T13:31:21.558Z