Counting 3,463 Big Data & Machine Learning Frameworks, Toolsets, and Examples...
Suggestion? Feedback? Tweet @stkim1

Last Commit
Jan. 13, 2019
Apr. 7, 2018


Dependently-typed tensor computation.


  • Parametric polymorphism
  • Compile time type checking
  • Dependently typed tensors
  • Multiple targets(Tensorflow, PyTorch, more to come!)
  • Pipes operator

Pipes operator

Pipes operator is a syntax sugar for chained function calls inspired by F#, Elixir and R. For example,

x |> lin1 |> leaky_relu(p=0.2) |> sigmoid

compiles to

x = lin1(x)
x = leaky_relu(x, p=0.2)
x = sigmoid(x)


Build Status

The language is not usable in production or development.


  1. implement module pattern matching
  2. type level computation (resolved tensor dimension)
  3. BUG: dimension mismatch for mnist example need to create fresh type variables for different static forward functions
  4. BUG: non-determinism
  5. BUG: impl Hash, Eq for Type
  6. set up examples and tests
  7. set up commandline
  8. more examples
  9. better errors in parser
  10. code gen: PyTorch
  11. add more examples
  12. lift dim and tsr to top level
  13. add dim level computation dim1 * dim1
  14. use Linear as L; aliasing
  15. add binary ops (+, -, *, /, %)
  16. add if else expression
  17. add let binding
  18. add more tests