mirror of
https://github.com/vale981/ray
synced 2025-03-08 19:41:38 -05:00
428516056a
* Policy-classes cleanup and torch/tf unification. - Make Policy abstract. - Add `action_dist` to call to `extra_action_out_fn` (necessary for PPO torch). - Move some methods and vars to base Policy (from TFPolicy): num_state_tensors, ACTION_PROB, ACTION_LOGP and some more. * Fix `clip_action` import from Policy (should probably be moved into utils altogether). * - Move `is_recurrent()` and `num_state_tensors()` into TFPolicy (from DynamicTFPolicy). - Add config to all Policy c'tor calls (as 3rd arg after obs and action spaces). * Add `config` to c'tor call to TFPolicy. * Add missing `config` to c'tor call to TFPolicy in marvil_policy.py. * Fix test_rollout_worker.py::MockPolicy and BadPolicy classes (Policy base class is now abstract). * Fix LINT errors in Policy classes. * Implement StatefulPolicy abstract methods in test cases: test_multi_agent_env.py. * policy.py LINT errors. * Create a simple TestPolicy to sub-class from when testing Policies (reduces code in some test cases). * policy.py - Remove abstractmethod from `apply_gradients` and `compute_gradients` (these are not required iff `learn_on_batch` implemented). - Fix docstring of `num_state_tensors`. * Make QMIX torch Policy a child of TorchPolicy (instead of Policy). * QMixPolicy add empty implementations of abstract Policy methods. * Store Policy's config in self.config in base Policy c'tor. * - Make only compute_actions in base Policy's an abstractmethod and provide pass implementation to all other methods if not defined. - Fix state_batches=None (most Policies don't have internal states). * Cartpole tf learning. * Cartpole tf AND torch learning (in ~ same ts). * Cartpole tf AND torch learning (in ~ same ts). 2 * Cartpole tf (torch syntax-broken) learning (in ~ same ts). 3 * Cartpole tf AND torch learning (in ~ same ts). 4 * Cartpole tf AND torch learning (in ~ same ts). 5 * Cartpole tf AND torch learning (in ~ same ts). 6 * Cartpole tf AND torch learning (in ~ same ts). Pendulum tf learning. * WIP. * WIP. * SAC torch learning Pendulum. * WIP. * SAC torch and tf learning Pendulum and Cartpole after cleanup. * WIP. * LINT. * LINT. * SAC: Move policy.target_model to policy.device as well. * Fixes and cleanup. * Fix data-format of tf keras Conv2d layers (broken for some tf-versions which have data_format="channels_first" as default). * Fixes and LINT. * Fixes and LINT. * Fix and LINT. * WIP. * Test fixes and LINT. * Fixes and LINT. Co-authored-by: Sven Mika <sven@Svens-MacBook-Pro.local>
85 lines
2.8 KiB
Python
85 lines
2.8 KiB
Python
from ray.rllib.models.model import Model
|
|
from ray.rllib.models.tf.misc import flatten
|
|
from ray.rllib.utils.annotations import override
|
|
from ray.rllib.utils.framework import get_activation_fn, try_import_tf
|
|
|
|
tf = try_import_tf()
|
|
|
|
|
|
# Deprecated: see as an alternative models/tf/visionnet_v2.py
|
|
class VisionNetwork(Model):
|
|
"""Generic vision network."""
|
|
|
|
@override(Model)
|
|
def _build_layers_v2(self, input_dict, num_outputs, options):
|
|
inputs = input_dict["obs"]
|
|
filters = options.get("conv_filters")
|
|
if not filters:
|
|
filters = _get_filter_config(inputs.shape.as_list()[1:])
|
|
|
|
activation = get_activation_fn(options.get("conv_activation"))
|
|
|
|
with tf.name_scope("vision_net"):
|
|
for i, (out_size, kernel, stride) in enumerate(filters[:-1], 1):
|
|
inputs = tf.layers.conv2d(
|
|
inputs,
|
|
out_size,
|
|
kernel,
|
|
stride,
|
|
activation=activation,
|
|
padding="same",
|
|
name="conv{}".format(i))
|
|
out_size, kernel, stride = filters[-1]
|
|
|
|
# skip final linear layer
|
|
if options.get("no_final_linear"):
|
|
fc_out = tf.layers.conv2d(
|
|
inputs,
|
|
num_outputs,
|
|
kernel,
|
|
stride,
|
|
activation=activation,
|
|
padding="valid",
|
|
name="fc_out")
|
|
return flatten(fc_out), flatten(fc_out)
|
|
|
|
fc1 = tf.layers.conv2d(
|
|
inputs,
|
|
out_size,
|
|
kernel,
|
|
stride,
|
|
activation=activation,
|
|
padding="valid",
|
|
name="fc1")
|
|
fc2 = tf.layers.conv2d(
|
|
fc1,
|
|
num_outputs, [1, 1],
|
|
activation=None,
|
|
padding="same",
|
|
name="fc2")
|
|
return flatten(fc2), flatten(fc1)
|
|
|
|
|
|
def _get_filter_config(shape):
|
|
shape = list(shape)
|
|
filters_84x84 = [
|
|
[16, [8, 8], 4],
|
|
[32, [4, 4], 2],
|
|
[256, [11, 11], 1],
|
|
]
|
|
filters_42x42 = [
|
|
[16, [4, 4], 2],
|
|
[32, [4, 4], 2],
|
|
[256, [11, 11], 1],
|
|
]
|
|
if len(shape) == 3 and shape[:2] == [84, 84]:
|
|
return filters_84x84
|
|
elif len(shape) == 3 and shape[:2] == [42, 42]:
|
|
return filters_42x42
|
|
else:
|
|
raise ValueError(
|
|
"No default configuration for obs shape {}".format(shape) +
|
|
", you must specify `conv_filters` manually as a model option. "
|
|
"Default configurations are only available for inputs of shape "
|
|
"[42, 42, K] and [84, 84, K]. You may alternatively want "
|
|
"to use a custom model or preprocessor.")
|