Neural Symbolic Simplifier

This repo provides the code to replicate the experiments in the paper

Hui Shi, Yang Zhang, Xinyun Chen, Yuandong Tian, Jishen Zhao Deep Symbolic Superoptimization Without Human Knowledge. in ICLR 2020

A short video introduction [YouTube]

Dependency

python >= 3.5

pytorch >= 1.1.0

To reproduce the results, you may first download the data to the folder Data/simplify, as instructed here. Then run:

python run_optimizer.py

The hyperparameters are specified in configs/halide.json.

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
Data		Data
Models		Models
Optimizer		Optimizer
configs		configs
LICENSE		LICENSE
README.md		README.md
run_optimizer.py		run_optimizer.py