torchgfn
torchgfn copied to clipboard
API Change -- Always Recalculate Everything
- Right now, to do on-policy, or off policy, or "true" off policy sampling efficiently, you need to set "recalculate" flags appropriately.
- If you do this incorrectly, you end up with bugs.
- We propose to change things such that everything is reclaulated by default. Therefore the user always gets the expected output, and will only need to learn the API to speed up their code.
- This will need to be accompanied by documentation.