Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

synthesis quality #10

Closed
13 tasks done
KiaraGrouwstra opened this issue Jan 21, 2020 · 0 comments
Closed
13 tasks done

synthesis quality #10

KiaraGrouwstra opened this issue Jan 21, 2020 · 0 comments
Labels
nice-to-have synthesis features needed for program synthesis

Comments

@KiaraGrouwstra
Copy link
Owner

KiaraGrouwstra commented Jan 21, 2020

experiments

  • confirm accuracy/learning
  • early stopping
  • lr decay
  • hyper-params
    • hyperparameter optimization:
      • implement grid search, including distinguishing validation vs training sets
      • implement evolutionary search (see 1, 2)
    • [ ] consider LearnedInitialization for LSTMs
  • let a learner pick which hole to fill
  • add explicit type signature for this i/o so programs won't fail type check over uninferred types (which can be a problem for e.g. Either, or for Maybe in case of Nothing)
  • try incorporating expansion embeddings in node embeddings thru either pass?
  • can I meaningfully encode anything more than just 'expression' in symbols?
    • hole status?
    • fn arities, as seemingly done in NSPS?
    • block (before variants)?
    • types?
@KiaraGrouwstra KiaraGrouwstra added nice-to-have synthesis features needed for program synthesis labels Jan 21, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
nice-to-have synthesis features needed for program synthesis
Projects
None yet
Development

No branches or pull requests

1 participant