wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-08-17 11:25:34 +08:00
Code Issues Packages Projects Releases Wiki Activity

[rllib] Fix atari reward calculations, add LR annealing, explained var stat for A2C / impala (#2700)

Browse Source 
Changes needed to reproduce Atari plots in IMPALA / A2C: https://github.com/ray-project/rl-experiments
This commit is contained in:
Eric Liang
2018-08-23 17:49:10 -07:00
committed by GitHub
parent 1b3de31ff1
commit aa014af85b
35 changed files with 483 additions and 148 deletions
Download Patch File Download Diff File Expand all files Collapse all files
doc/source/rllib-algorithms.rst
+13 -6
View File
@@ -16,11 +16,14 @@ Tuned examples: `PongNoFrameskip-v4 <https://github.com/ray-project/ray/blob/mas
Advantage Actor-Critic (A2C, A3C)
---------------------------------
`[paper] <https://arxiv.org/abs/1602.01783>`__ `[implementation] <https://github.com/ray-project/ray/blob/master/python/ray/rllib/agents/a3c/a3c.py>`__
RLlib's A3C uses the AsyncGradientsOptimizer to apply gradients computed remotely on policy evaluation actors. It scales to up to 16-32 worker processes, depending on the environment. Both a TensorFlow (LSTM), and PyTorch version are available.
RLlib implements A2C and A3C using SyncSamplesOptimizer and AsyncGradientsOptimizer respectively for policy optimization. These algorithms scale to up to 16-32 worker processes depending on the environment. Both a TensorFlow (LSTM), and PyTorch version are available.
Note that if you have a GPU, `IMPALA <#importance-weighted-actor-learner-architecture>`__ probably will perform better than A3C. You can also use the synchronous variant of A3C, `A2C <https://github.com/ray-project/ray/blob/master/python/ray/rllib/agents/a3c/a2c.py>`__.
.. note::
In most cases, `IMPALA <#importance-weighted-actor-learner-architecture-impala>`__ will outperform A2C / A3C. In our `benchmarks <https://github.com/ray-project/rl-experiments>`__, IMPALA is almost 10x faster than A2C in wallclock time, with similar sample efficiency.
Tuned examples: `PongDeterministic-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-a3c.yaml>`__, `A2C variant <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-a2c.yaml>`__, `PyTorch version <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-a3c-pytorch.yaml>`__
Tuned examples: `PongDeterministic-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-a3c.yaml>`__, `PyTorch version <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-a3c-pytorch.yaml>`__, `{BeamRider,Breakout,Qbert,SpaceInvaders}NoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/atari-a2c.yaml>`__
See also our `Atari results <https://github.com/ray-project/rl-experiments>`__.
Deep Deterministic Policy Gradients (DDPG)
------------------------------------------
@@ -56,12 +59,14 @@ Importance Weighted Actor-Learner Architecture (IMPALA)
`[implementation] <https://github.com/ray-project/ray/blob/master/python/ray/rllib/agents/impala/impala.py>`__
In IMPALA, a central learner runs SGD in a tight loop while asynchronously pulling sample batches from many actor processes. RLlib's IMPALA implementation uses DeepMind's reference `V-trace code <https://github.com/deepmind/scalable_agent/blob/master/vtrace.py>`__. Note that we do not provide a deep residual network out of the box, but one can be plugged in as a `custom model <rllib-models.html#custom-models>`__.
Tuned examples: `PongNoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-impala.yaml>`__, `vectorized configuration <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-impala-vectorized.yaml>`__
Tuned examples: `PongNoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-impala.yaml>`__, `vectorized configuration <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-impala-vectorized.yaml>`__, `{BeamRider,Breakout,Qbert,SpaceInvaders}NoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/atari-impala.yaml>`__
See also our `Atari results <https://github.com/ray-project/rl-experiments>`__.
.. figure:: impala.png
:align: center
RLlib's IMPALA implementation scales from 16 to 128 workers on PongNoFrameskip-v4. With vectorization, similar learning performance to 128 workers can be achieved with only 32 workers. This is about an order of magnitude faster than A3C (not shown here), with similar sample efficiency.
IMPALA solves Atari about 10x faster than A2C / A3C, with similar sample efficiency. Here IMPALA scales from 16 to 128 workers on PongNoFrameskip-v4.
Policy Gradients
----------------
@@ -74,7 +79,9 @@ Proximal Policy Optimization (PPO)
`[paper] <https://arxiv.org/abs/1707.06347>`__ `[implementation] <https://github.com/ray-project/ray/blob/master/python/ray/rllib/agents/ppo/ppo.py>`__
PPO's clipped objective supports multiple SGD passes over the same batch of experiences. RLlib's multi-GPU optimizer pins that data in GPU memory to avoid unnecessary transfers from host memory, substantially improving performance over a naive implementation. RLlib's PPO scales out using multiple workers for experience collection, and also with multiple GPUs for SGD.
Tuned examples: `Humanoid-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/humanoid-ppo-gae.yaml>`__, `Hopper-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/hopper-ppo.yaml>`__, `Pendulum-v0 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pendulum-ppo.yaml>`__, `PongDeterministic-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-ppo.yaml>`__, `Walker2d-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/walker2d-ppo.yaml>`__
Tuned examples: `Humanoid-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/humanoid-ppo-gae.yaml>`__, `Hopper-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/hopper-ppo.yaml>`__, `Pendulum-v0 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pendulum-ppo.yaml>`__, `PongDeterministic-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/pong-ppo.yaml>`__, `Walker2d-v1 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/walker2d-ppo.yaml>`__, `{BeamRider,Breakout,Qbert,SpaceInvaders}NoFrameskip-v4 <https://github.com/ray-project/ray/blob/master/python/ray/rllib/tuned_examples/atari-ppo.yaml>`__
See also our `Atari results <https://github.com/ray-project/rl-experiments>`__.
.. figure:: ppo.png
:width: 500px