gaspersic/baselines
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #22 from ngc92/doc_fixes

Browse Source 
docstring and comment fixes
This commit is contained in:
Szymon Sidor
2017-06-04 00:41:34 -07:00
committed by GitHub
parent fba0ac30ca 02919483f2
commit 184440ffd3
2 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
baselines/deepq/build_graph.py
View File

@@ -27,7 +27,7 @@ The functions in this file can are used to create the following functions:
Function that takes a transition (s,a,r,s') and optimizes Bellman equation's error: Function that takes a transition (s,a,r,s') and optimizes Bellman equation's error:
td_error = Q(s,a) - (r + gamma * max_a' Q(s', a')) td_error = Q(s,a) - (r + gamma * max_a' Q(s', a'))
loss = hauber_loss[td_error] loss = huber_loss[td_error]
Parameters Parameters
---------- ----------
@@ -127,12 +127,12 @@ def build_act(make_obs_ph, q_func, num_actions, scope="deepq", reuse=None):
def build_train(make_obs_ph, q_func, num_actions, optimizer, grad_norm_clipping=None, gamma=1.0, double_q=True, scope="deepq", reuse=None): def build_train(make_obs_ph, q_func, num_actions, optimizer, grad_norm_clipping=None, gamma=1.0, double_q=True, scope="deepq", reuse=None):
"""Creates the act function: """Creates the train function:
Parameters Parameters
---------- ----------
make_obs_ph: str -> tf.placeholder or TfInput make_obs_ph: str -> tf.placeholder or TfInput
a function that take a name and creates a placeholder of input with that name a function that takes a name and creates a placeholder of input with that name
q_func: (tf.Variable, int, str, bool) -> tf.Variable q_func: (tf.Variable, int, str, bool) -> tf.Variable
the model that takes the following inputs: the model that takes the following inputs:
observation_in: object observation_in: object
@@ -150,7 +150,7 @@ def build_train(make_obs_ph, q_func, num_actions, optimizer, grad_norm_clipping=
optimizer: tf.train.Optimizer optimizer: tf.train.Optimizer
optimizer to use for the Q-learning objective. optimizer to use for the Q-learning objective.
grad_norm_clipping: float or None grad_norm_clipping: float or None
clip graident norms to this value. If None no clipping is performed. clip gradient norms to this value. If None no clipping is performed.
gamma: float gamma: float
discount rate. discount rate.
double_q: bool double_q: bool

2
baselines/deepq/simple.py
View File

@@ -123,11 +123,11 @@ def learn(env,
final value of random action probability final value of random action probability
train_freq: int train_freq: int
update the model every `train_freq` steps. update the model every `train_freq` steps.
set to None to disable printing
batch_size: int batch_size: int
size of a batched sampled from replay buffer for training size of a batched sampled from replay buffer for training
print_freq: int print_freq: int
how often to print out training progress how often to print out training progress
set to None to disable printing
checkpoint_freq: int checkpoint_freq: int
how often to save the model. This is so that the best version is restored how often to save the model. This is so that the best version is restored
at the end of the training. If you do not wish to restore the best version at at the end of the training. If you do not wish to restore the best version at