Left: a Deep Q-Network-based agent trained on a Macbook for several hours plays Snake. Right: cell-wise attributions of the policy network obtained by taking the gradient of the network's output with respect to the inputs.
| Name | Name | Last commit date | ||
|---|---|---|---|---|
Left: a Deep Q-Network-based agent trained on a Macbook for several hours plays Snake. Right: cell-wise attributions of the policy network obtained by taking the gradient of the network's output with respect to the inputs.