avisingh599 / imitation-dagger

[Reimplementation Ross et al 2011] An implementation of DAGGER using ConvNets for driving from pixels.
MIT License
71 stars 20 forks source link

organized dagger code; added comments; improved convergence properties of neural net #4

Closed james-mchugh closed 3 months ago

james-mchugh commented 6 years ago

I organized the code and added comments to improve its readability.

Also, the current set up of the neural network makes it very unlikely to converge. To fix this, I increased the learning rate, changed the kernel initialization, and changed the output activation to linear since it is a regression problem.