diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index bc27a4af3928647add254248b384b0fab7041fca..6733cb4aef9458e927831c76ae76382d7e50b595 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -23,10 +23,8 @@ tests:
stage: tests
script:
- apt update
- - apt install -y libgl1-mesa-glx xvfb
+ - apt install -y libgl1-mesa-glx xvfb graphviz xdg-utils libcairo2-dev libjpeg-dev libgif-dev
- pip install tox
- - apt install -y graphviz xdg-utils
- - apt install -y libcairo2-dev libjpeg-dev libgif-dev
- xvfb-run tox -v --recreate
build_and_deploy_docs:
@@ -37,13 +35,13 @@ build_and_deploy_docs:
dependencies:
- tests
before_script:
- - pip install awscli
- apt update
- - apt install -y graphviz libgl1-mesa-glx xvfb xdg-utils
- - apt install -y libcairo2-dev libjpeg-dev libgif-dev
- - echo "Bucket $BUCKET_NAME ${BUCKET_NAME}"
+ - apt install -y graphviz libgl1-mesa-glx xvfb xdg-utils libcairo2-dev libjpeg-dev libgif-dev
+ - pip install awscli
script:
- - echo "Bucket $BUCKET_NAME ${BUCKET_NAME}"
+ - echo "Bucket=${BUCKET_NAME}"
+ - echo "AWS_DEFAULT_REGION=${AWS_DEFAULT_REGION}"
+ - echo "CI_COMMIT_REF_SLUG=${CI_COMMIT_REF_SLUG}"
- pip install -r requirements_dev.txt -r requirements_continuous_integration.txt
- python setup.py install
- xvfb-run make docs
diff --git a/benchmarks/play_model_benchmark.py b/benchmarks/play_model_benchmark.py
index a3c087e8ab8ed33a24bd447cd3203b85eb1a28f2..41c15ef0ef1cbbeebcf547ffa967dc9c31fecbe6 100644
--- a/benchmarks/play_model_benchmark.py
+++ b/benchmarks/play_model_benchmark.py
@@ -1,16 +1,13 @@
import random
-import time
-from collections import deque
import numpy as np
from benchmarker import Benchmarker
from flatland.envs.generators import complex_rail_generator
from flatland.envs.rail_env import RailEnv
-from flatland.utils.rendertools import RenderTool
-def main(render=True, delay=0.0):
+def main():
random.seed(1)
np.random.seed(1)
@@ -19,18 +16,8 @@ def main(render=True, delay=0.0):
rail_generator=complex_rail_generator(nr_start_goal=5, nr_extra=20, min_dist=12),
number_of_agents=5)
- if render:
- env_renderer = RenderTool(env, gl="QTSVG")
-
n_trials = 20
- eps = 1.
- eps_end = 0.005
- eps_decay = 0.998
action_dict = dict()
- scores_window = deque(maxlen=100)
- done_window = deque(maxlen=100)
- scores = []
- dones_list = []
action_prob = [0] * 4
def max_lt(seq, val):
@@ -46,82 +33,32 @@ def main(render=True, delay=0.0):
idx -= 1
return None
- iFrame = 0
- tStart = time.time()
for trials in range(1, n_trials + 1):
# Reset environment
obs = env.reset()
- if render:
- env_renderer.set_new_rail()
for a in range(env.get_num_agents()):
norm = max(1, max_lt(obs[a], np.inf))
obs[a] = np.clip(np.array(obs[a]) / norm, -1, 1)
- # env.obs_builder.util_print_obs_subtree(tree=obs[0], num_elements_per_node=5)
-
- score = 0
- env_done = 0
-
# Run episode
for step in range(100):
- # if trials > 114:
- # env_renderer.renderEnv(show=True)
- # print(step)
# Action
for a in range(env.get_num_agents()):
action = np.random.randint(0, 4)
action_prob[action] += 1
action_dict.update({a: action})
- if render:
- env_renderer.renderEnv(show=True, frames=True, iEpisode=trials, iStep=step, action_dict=action_dict)
- if delay > 0:
- time.sleep(delay)
-
- iFrame += 1
-
# Environment step
next_obs, all_rewards, done, _ = env.step(action_dict)
for a in range(env.get_num_agents()):
norm = max(1, max_lt(next_obs[a], np.inf))
next_obs[a] = np.clip(np.array(next_obs[a]) / norm, -1, 1)
- # Update replay buffer and train agent
- for a in range(env.get_num_agents()):
- # agent.step(obs[a], action_dict[a], all_rewards[a], next_obs[a], done[a])
- score += all_rewards[a]
- obs = next_obs.copy()
if done['__all__']:
- env_done = 1
break
- # Epsilon decay
- eps = max(eps_end, eps_decay * eps) # decrease epsilon
-
- done_window.append(env_done)
- scores_window.append(score) # save most recent score
- scores.append(np.mean(scores_window))
- dones_list.append((np.mean(done_window)))
-
- print(('\rTraining {} Agents.\tEpisode {}\tAverage Score: {:.0f}\tDones: {:.2f}%' +
- '\tEpsilon: {:.2f} \t Action Probabilities: \t {}').format(
- env.get_num_agents(),
- trials,
- np.mean(scores_window),
- 100 * np.mean(done_window),
- eps, action_prob / np.sum(action_prob)),
- end=" ")
if trials % 100 == 0:
- tNow = time.time()
- rFps = iFrame / (tNow - tStart)
- print(('\rTraining {} Agents.\tEpisode {}\tAverage Score: {:.0f}\tDones: {:.2f}%' +
- '\tEpsilon: {:.2f} fps: {:.2f} \t Action Probabilities: \t {}').format(
- env.get_num_agents(),
- trials,
- np.mean(scores_window),
- 100 * np.mean(done_window),
- eps, rFps, action_prob / np.sum(action_prob)))
action_prob = [1] * 4
@@ -129,4 +66,4 @@ if __name__ == "__main__":
with Benchmarker(cycle=20, extra=1) as bench:
@bench("Everything")
def _(bm):
- main(render=False, delay=0)
+ main()
diff --git a/setup.py b/setup.py
index ce7b232fd005d696bb2803733d455904c75352df..649a4ac69bdf11c8800a571bb067e1ed16141129 100644
--- a/setup.py
+++ b/setup.py
@@ -21,13 +21,12 @@ with open('HISTORY.rst') as history_file:
-# install pycairo
+# install pycairo on Windows
if os.name == 'nt':
p = platform.architecture()
is64bit = p[0] == '64bit'
if sys.version[0:3] == '3.5':
if is64bit:
-
url = 'https://download.lfd.uci.edu/pythonlibs/t4jqbe6o/pycairo-1.18.0-cp35-cp35m-win_amd64.whl'
else:
url = 'https://download.lfd.uci.edu/pythonlibs/t4jqbe6o/pycairo-1.18.0-cp35-cp35m-win32.whl'
@@ -47,7 +46,7 @@ if os.name == 'nt':
try:
import pycairo
except:
- call_cmd = "pip install " + url
+ call_cmd = "pip install --user " + url
os.system(call_cmd)
import site