From 90593dcbc2dca1aa9b55ffda00edf731d403855d Mon Sep 17 00:00:00 2001
From: Guillaume Mollard <guillaume@iccluster028.iccluster.epfl.ch>
Date: Thu, 18 Jul 2019 22:15:58 +0200
Subject: [PATCH] done metric fix
---
RLLib_training/experiment_configs/config_example/config.gin | 4 ++--
RLLib_training/train_experiment.py | 6 +++---
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/RLLib_training/experiment_configs/config_example/config.gin b/RLLib_training/experiment_configs/config_example/config.gin
index 7a4c7c1..1fc64cb 100644
--- a/RLLib_training/experiment_configs/config_example/config.gin
+++ b/RLLib_training/experiment_configs/config_example/config.gin
@@ -5,10 +5,10 @@ run_experiment.hidden_sizes = [32, 32]
run_experiment.map_width = 20
run_experiment.map_height = 20
-run_experiment.n_agents = 7#{"grid_search": [3, 4, 5, 6, 7, 8]}
+run_experiment.n_agents = {"grid_search": [3, 4, 5, 6, 7, 8]}
run_experiment.rail_generator = "complex_rail_generator"
run_experiment.nr_extra = 5
-run_experiment.policy_folder_name = "ppo_policy_two_obs_with_predictions_n_agents_{config[n_agents]}_"#_map_size_{config[map_width]}"
+run_experiment.policy_folder_name = "ppo_policy_two_obs_with_predictions_n_agents_{config[n_agents]}_map_size_{config[map_width]}_"
run_experiment.seed = 123
diff --git a/RLLib_training/train_experiment.py b/RLLib_training/train_experiment.py
index 1bac614..44fa26d 100644
--- a/RLLib_training/train_experiment.py
+++ b/RLLib_training/train_experiment.py
@@ -51,9 +51,9 @@ def on_episode_end(info):
for k, v in episode._agent_reward_history.items():
score += np.sum(v)
score /= (len(episode._agent_reward_history) * episode.horizon)
- done = 1
- if len(episode._agent_reward_history) == episode.horizon:
- done = 0
+ done = 0
+ if len(episode._agent_reward_history[0]) <= episode.horizon-5:
+ done = 1
episode.custom_metrics["score"] = score
episode.custom_metrics["proportion_episode_solved"] = done
--
GitLab