keras-rl / keras-rl

Deep Reinforcement Learning for Keras.

Home Page:http://keras-rl.readthedocs.io/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

image_dim_ordering, used in example dqn agent, removed in Keras 2.2.5

travisp opened this issue · comments

The example DQN Atari implementation uses image_dim_ordering to decide how to permute the incoming data:
https://github.com/keras-rl/keras-rl/blob/0707d6da3d5bc2fae26dada0c91aa9e2538bbb92/examples/dqn_atari.py

This was removed in Keras 2.2.5 (keras-team/keras#12649) and can be replaced with the more robust image_data_format method that will return channels_last or channels_first

Removing since I missed that this was reported #329