3LA workshop paper

Deep Learning

From DSLs to Accelerator-rich Platform: Addressing the Mapping Gap will be at LATTE 2021 (ASPLOS)

3LA workshop paper

Deep Learning

Dynamic Tensor Rematerialization will be at ICLR 2021

Our paper Dynamic Tensor Rematerialization (Kirisame et al.) is accepted to ICLR 2021. Cheers!

Research

From Models to Computation Graphs (Part I)

Note on Converting a PyTorch model to a computation graph. The second part will be about compiling the model back to a python source file.

PyTorch

Machine Learning

Proof of Commutativity (+) in Haskell

Using GADTs and dummy dependent-type in Haskell

Haskell

Theorem Proving