EnvironmentLogger
EnvLogger is a standard dm_env.Environment class wrapper that records interactions between a real environment and an agent. These interactions are saved on disk as trajectories and can be retrieved in whole, by individual timesteps or by specific episodes.
Metadata
To better categorize your logged data, you may want to add some tags in the metadata when you construct the logger wrapper. The metadata is written once per EnvLogger instance.
env = envlogger.EnvLogger(
env,
data_directory='/tmp/experiment_logs',
metadata={
'environment_type': 'dm_control',
'agent_type': 'D4PG'
})
Examples
Most of the time, it is just a one-liner wrapper, e.g.
import envlogger
from envlogger.testing import catch_env
import numpy as np
env = catch_env.Catch()
with envlogger.EnvLogger(
env, data_directory='/tmp/experiment_logs') as env:
env.reset()
for step in range(100):
action = np.random.randint(low=0, high=3)
timestep = env.step(action)
Reading stored trajectories
Reader can read stored trajectories. Example:
from envlogger import reader
with reader.Reader(
data_directory='/tmp/experiment_logs') as r:
for episode in r.episodes:
for step in episode:
# step is a step_data.StepData.
# Use step.timestep.observation, step.timestep.reward, step.action etc...
Getting Started
EnvLogger currently only supports Linux based OSes and Python 3.
You can install EnvLogger via pip:
pip install envlogger
Compiling from source
For this option you will need to install Bazel and GMP (libgmp-dev on Debian-based systems). Please note that Bazel versions >4.0 are not supported. Our recommended version is 3.7.2. Then:
git clone https://github.com/deepmind/envlogger/
cd envlogger
bazel test --test_output=errors envlogger/...
Running inside Docker
We provide a Docker image that can be used to run tests in a more controlled environment. You can run it as follows:
sh docker/build.sh
docker run -it envlogger bash
bazel test --test_output=errors envlogger/...
Benchmarks
Wrapping your environment with EnvLogger adds an approximately 2 millisecond overhead per environment step. See the following difference in distributions in the case of random agent on a 100 steps per second Catch environment (measured in milliseconds).
| Percentiles | 50th | 95th | 99th | 99.9th | mean (± std) |
|---|---|---|---|---|---|
| w/o EnvLogger | 10.15 | 10.23 | 11.51 | 14.70 | 10.19 (± 1.06) |
| w/ EnvLogger | 12.65 | 14.39 | 15.94 | 19.43 | 12.88 (± 0.34) |
Acknowledgements
We greatly appreciate all the support from the TF-Agents team in setting up building and testing for EnvLogger.
5 Aug 31, 2022
4k Jan 3, 2023
2 Feb 12, 2022
122 Nov 17, 2022
71 Nov 20, 2022
1.7k Jan 4, 2023
1.2k Dec 29, 2022
3.4k Dec 31, 2022
2.2k Jan 3, 2023
682 Dec 28, 2022
2.7k Dec 31, 2022
2.8k Jan 3, 2023
4.8k Jan 5, 2023
68 May 31, 2022
3.2k Jan 9, 2023
678 Dec 22, 2022
3.2k Jan 4, 2023
698 Dec 26, 2022
3 Oct 13, 2022
516 Dec 30, 2022
11 Dec 23, 2022
135 Jan 8, 2023
1.4k Jan 8, 2023
392 Dec 31, 2022
483 Dec 31, 2022
5 Nov 21, 2021
9 Jun 7, 2022