gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-17 19:20:19 +01:00
Code Issues Projects Releases Wiki Activity

Multiple improvements and bug fixes (#66)

Browse Source 
* Multiple improvements and bug fixes:

    * Using lazy stacking to save on memory when using a replay buffer
    * Remove step counting for evaluation episodes
    * Reset game between heatup and training
    * Major bug fixes in NEC (is reproducing the paper results for pong now)
    * Image input rescaling to 0-1 is now optional
    * Change the terminal title to be the experiment name
    * Observation cropping for atari is now optional
    * Added random number of noop actions for gym to match the dqn paper
    * Fixed a bug where the evaluation episodes won't start with the max possible ale lives
    * Added a script for plotting the results of an experiment over all the atari games
This commit is contained in:
Itai Caspi
2018-02-26 12:29:07 +02:00
committed by GitHub
parent 4fe9cba445
commit a7206ed702
20 changed files with 465 additions and 158 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
architectures/network_wrapper.py
View File

@@ -125,14 +125,15 @@ class NetworkWrapper(object):
"""
self.online_network.apply_gradients(self.online_network.accumulated_gradients)
def train_and_sync_networks(self, inputs, targets):
def train_and_sync_networks(self, inputs, targets, additional_fetches=[]):
"""
A generic training function that enables multi-threading training using a global network if necessary.
:param inputs: The inputs for the network.
:param targets: The targets corresponding to the given inputs
:param additional_fetches: Any additional tensor the user wants to fetch
:return: The loss of the training iteration
"""
result = self.online_network.accumulate_gradients(inputs, targets)
result = self.online_network.accumulate_gradients(inputs, targets, additional_fetches=additional_fetches)
self.apply_gradients_and_sync_networks()
return result