chainer / chainerrl

ChainerRL is a deep reinforcement learning library built on top of Chainer.
MIT License
1.18k stars 224 forks source link

Fixes Rainbow Score to use correct Reporting Protocol #561

Closed prabhatnagarajan closed 5 years ago

prabhatnagarajan commented 5 years ago

Currently, the Rainbow scores show the scores of the final agent, as opposed to a re-evaluation of the best intermediate network.

The new README looks like this: https://github.com/prabhatnagarajan/chainerrl/tree/fix_rainbow_scores/examples/atari/reproduction/rainbow