From a8631db63b1e8fdf6d4df746c00af614a1b902a2 Mon Sep 17 00:00:00 2001
From: "Egli Adrian (IT-SCI-API-PFI)" <adrian.egli@sbb.ch>
Date: Sun, 15 Nov 2020 20:04:05 +0100
Subject: [PATCH] refactored file name
---
reinforcement_learning/multi_agent_training.py | 4 ++--
run.py | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/reinforcement_learning/multi_agent_training.py b/reinforcement_learning/multi_agent_training.py
index 4bdbe4e..b83bd91 100755
--- a/reinforcement_learning/multi_agent_training.py
+++ b/reinforcement_learning/multi_agent_training.py
@@ -172,7 +172,7 @@ def train_agent(train_params, train_env_params, eval_env_params, obs_params):
completion_window = deque(maxlen=checkpoint_interval)
# Double Dueling DQN policy
- USE_SINGLE_AGENT_TRAINING = True
+ USE_SINGLE_AGENT_TRAINING = False
UPDATE_POLICY2_N_EPISODE = 1000
policy = DDDQNPolicy(state_size, action_size, train_params)
# policy = PPOAgent(state_size, action_size, n_agents)
@@ -521,7 +521,7 @@ if __name__ == "__main__":
parser.add_argument("--learning_rate", help="learning rate", default=0.5e-4, type=float)
parser.add_argument("--hidden_size", help="hidden size (2 fc layers)", default=128, type=int)
parser.add_argument("--update_every", help="how often to update the network", default=10, type=int)
- parser.add_argument("--use_gpu", help="use GPU if available", default=True, type=bool)
+ parser.add_argument("--use_gpu", help="use GPU if available", default=False, type=bool)
parser.add_argument("--num_threads", help="number of threads PyTorch can use", default=4, type=int)
parser.add_argument("--render", help="render 1 episode in 100", action='store_true')
parser.add_argument("--load_policy", help="policy filename (reference) to load", default="", type=str)
diff --git a/run.py b/run.py
index ed3afa0..143f44a 100644
--- a/run.py
+++ b/run.py
@@ -27,7 +27,7 @@ VERBOSE = True
# Checkpoint to use (remember to push it!)
# checkpoint = "./checkpoints/201112143850-5400.pth" # 21.220418678677177 DEPTH=2 AGENTS=10
-checkpoint = "./checkpoints/201113211844-6100.pth" # 19.690047767961005 DEPTH=2 AGENTS=20
+checkpoint = "./checkpoints/201113211844-6200.pth" # 19.690047767961005 DEPTH=2 AGENTS=20
# Use last action cache
--
GitLab