We can also get 1.3 better pass rate over the ChatGPT Code

Current deep networks are very data-hungry and benefit from training on large-scale datasets, which are often time-consuming to collect and annotate.

The idea is illustrated in Figure 1.

Procedural Content Generation (PCG) is a technique to generate complex and diverse environments in an automated way.

Existing solutions often resort to down-sampling operations, such as pooling, to reduce computational cost. Unfortunately, such operations are non-invertible and can result in information loss.