mirror of
https://github.com/vale981/ray
synced 2025-03-06 10:31:39 -05:00
60d4d5e1aa
* Remove all __future__ imports from RLlib. * Remove (object) again from tf_run_builder.py::TFRunBuilder. * Fix 2xLINT warnings. * Fix broken appo_policy import (must be appo_tf_policy) * Remove future imports from all other ray files (not just RLlib). * Remove future imports from all other ray files (not just RLlib). * Remove future import blocks that contain `unicode_literals` as well. Revert appo_tf_policy.py to appo_policy.py (belongs to another PR). * Add two empty lines before Schedule class. * Put back __future__ imports into determine_tests_to_run.py. Fails otherwise on a py2/print related error.
56 lines
1.9 KiB
Python
56 lines
1.9 KiB
Python
from ray.rllib.models.model import Model
|
|
from ray.rllib.models.tf.misc import normc_initializer, get_activation_fn
|
|
from ray.rllib.utils.annotations import override
|
|
from ray.rllib.utils import try_import_tf
|
|
|
|
tf = try_import_tf()
|
|
|
|
|
|
# Deprecated: see as an alternative models/tf/fcnet_v2.py
|
|
class FullyConnectedNetwork(Model):
|
|
"""Generic fully connected network."""
|
|
|
|
@override(Model)
|
|
def _build_layers(self, inputs, num_outputs, options):
|
|
"""Process the flattened inputs.
|
|
|
|
Note that dict inputs will be flattened into a vector. To define a
|
|
model that processes the components separately, use _build_layers_v2().
|
|
"""
|
|
|
|
hiddens = options.get("fcnet_hiddens")
|
|
activation = get_activation_fn(options.get("fcnet_activation"))
|
|
|
|
if len(inputs.shape) > 2:
|
|
inputs = tf.layers.flatten(inputs)
|
|
|
|
with tf.name_scope("fc_net"):
|
|
i = 1
|
|
last_layer = inputs
|
|
for size in hiddens:
|
|
# skip final linear layer
|
|
if options.get("no_final_linear") and i == len(hiddens):
|
|
output = tf.layers.dense(
|
|
last_layer,
|
|
num_outputs,
|
|
kernel_initializer=normc_initializer(1.0),
|
|
activation=activation,
|
|
name="fc_out")
|
|
return output, output
|
|
|
|
label = "fc{}".format(i)
|
|
last_layer = tf.layers.dense(
|
|
last_layer,
|
|
size,
|
|
kernel_initializer=normc_initializer(1.0),
|
|
activation=activation,
|
|
name=label)
|
|
i += 1
|
|
|
|
output = tf.layers.dense(
|
|
last_layer,
|
|
num_outputs,
|
|
kernel_initializer=normc_initializer(0.01),
|
|
activation=None,
|
|
name="fc_out")
|
|
return output, last_layer
|