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

fix nec_agent

Browse Source
This commit is contained in:
Zach Dwiel
2018-02-16 13:36:14 -05:00
parent ee6e0bdc3b
commit 943e41ba58
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
agents/nec_agent.py
View File

@@ -18,6 +18,7 @@ import numpy as np
from agents.value_optimization_agent import ValueOptimizationAgent from agents.value_optimization_agent import ValueOptimizationAgent
from logger import screen from logger import screen
from utils import RunPhase
# Neural Episodic Control - https://arxiv.org/pdf/1703.01988.pdf # Neural Episodic Control - https://arxiv.org/pdf/1703.01988.pdf
@@ -40,7 +41,7 @@ class NECAgent(ValueOptimizationAgent):
screen.log_title("Finished collecting initial entries in DND. Starting to train network...") screen.log_title("Finished collecting initial entries in DND. Starting to train network...")
current_states, next_states, actions, rewards, game_overs, total_return = self.extract_batch(batch) current_states, next_states, actions, rewards, game_overs, total_return = self.extract_batch(batch)
result = self.main_network.train_and_sync_networks([current_states, actions], total_return) result = self.main_network.train_and_sync_networks(current_states, total_return)
total_loss = result[0] total_loss = result[0]
return total_loss return total_loss