I'm using Tensorboard to see the progress of the PettingZoo environment that my agents are playing. I can see the reward go up with time, which is good, but I'd like to add other metrics that are specific to my environment. i.e. I'd like TensorBoard to show me more charts with my metrics and how they improve over time.
The only way I could figure out how to do that was by inserting a few lines into the learn method of OnPolicyAlgorithm that's part of SB3. This works and I got the charts I wanted:
(The two bottom charts are the ones I added.)
But obviously editing library code isn't a good practice. I should make the modifications in my own code, not in the libraries. Is there currently a more elegant way to add a metric from my PettingZoo environment into TensorBoard?
You can add a callback to add your own logs. See the below example. In this case the call back is called every step. There are other callbacks that you case use depending on your use case.
import numpy as np
from stable_baselines3 import SAC
from stable_baselines3.common.callbacks import BaseCallback
model = SAC("MlpPolicy", "Pendulum-v1", tensorboard_log="/tmp/sac/", verbose=1)
class TensorboardCallback(BaseCallback):
"""
Custom callback for plotting additional values in tensorboard.
"""
def __init__(self, verbose=0):
super(TensorboardCallback, self).__init__(verbose)
def _on_step(self) -> bool:
# Log scalar value (here a random variable)
value = np.random.random()
self.logger.record('random_value', value)
return True
model.learn(50000, callback=TensorboardCallback())
Related
I have a folder where I will upload one file every month. The file will have the same format in every month.
First problem
The idea is to concatenate all the files in this folder into one file. Currently I am hardcoding the filenames (filename[0], filename[1], filename[2]..) but imagine later I will have 50 files, should I explicitly add them to the transform_df decorator? Is there any other method to handle this?
Second problem:
Currently I have let's say 4 files (2021_07, 2021_08, 2021_09, 2021_10) and I want whenever I add the file presenting 2021_12 data to avoid changing the code.
If I add input_5 = Input(path_to_2021_12_do_not_exists) the code will not be run and give an error.
How can I implement the code for future files and let the code ignore the input if it does not exist without manually each month add a new value to my code?
Thank you
# from pyspark.sql import functions as F
from transforms.api import transform_df, Input, Output
from pyspark.sql.functions import to_date, year, col
from pyspark.sql.types import StringType
from myproject.datasets import utils
from pyspark.sql import DataFrame
from functools import reduce
input_dir = '/Company/Project_name/'
prefix_filename = 'DataInput1_'
suffixes = ['2021_07', '2021_08', '2021_09', '2021_10', '2021_11', '2021_12']
filenames = [input_dir + prefix_filename + suffixe for suffixe in suffixes]
#transform_df(
Output("/Company/Project_name/Data/clean/File_concat"),
input_1=Input(filenames[0]),
input_2=Input(filenames[1]),
input_3=Input(filenames[2]),
input_4=Input(filenames[3]),
)
def compute(input_1, input_2, input_3, input_4):
input_dfs = [input_1, input_2, input_3, input_4]
dfs = []
def transformation_input(df):
# some transformation
return df
for input_df in input_dfs:
dfs.append(transformation_input(input_df))
dfs = reduce(DataFrame.unionByName, dfs)
return dfs
This question comes up a lot, the simple answer is that you don't. Defining datasets and executing a build on them are two different steps executed at different stages.
Whenever you commit your code and run the checks, your overall python code is executed during the renderSchrinkwrap stage, except for the compute part. This allows Foundry to discover what datasets exist and publish.
Publishing involves creating your dataset and putting whatever is inside your compute function is published into the jobspec of the dataset, so foundry knows what code to execute whenever you run a build.
Once you hit build on the dataset, Foundry will only pick up whatever is on the jobspec and execute it. Any other code has already run during your checks, and it has run just once.
So any dynamic input/output would require you to re-run checks on your repo, which means that some code change would have had to happen since the Checks is part of the CI process, not part of the build.
Taking a step back, assuming each of your input files has the same schema, Foundry would expect you to have all of those files in the same dataset as append transactions.
This might not be possible though, if for instance, the only indication of the "year" of the data is embedded in the filename, but your sample code would indicate that you expect all these datasets to have the same schema and easily union together.
You can do this manually through the Dataset Preview - just use the Upload File button or drag-and-drop the new file into the Preview window - or, if it's an "end user" workflow, with a File Upload Widget in a Workshop app. You may need to coordinate with your Foundry support team if this widget isn't available.
Bit late to the post although for anyone who is interested in an answer to most of the question. Dynamically determining file names from within a folder is not doable although having some level of dynamic input is possible as follows:
# from pyspark.sql import functions as F
from transforms.api import transform, Input, Output
from pyspark.sql.functions import to_date, year, col
from pyspark.sql.types import StringType
from myproject.datasets import utils
from pyspark.sql import DataFrame
# from functools import reduce
from transforms.verbs.dataframes import union_many # use this instead of reduce
input_dir = '/Company/Project_name/'
prefix_filename = 'DataInput1_'
suffixes = ['2021_07', '2021_08', '2021_09', '2021_10', '2021_11', '2021_12']
filenames = [input_dir + prefix_filename + suffixe for suffixe in suffixes]
inputs = {('input{}'.format(index)): Input(filename) for (index, filename) in enumerate(filenames))}
#transform(
output=Output("/Company/Project_name/Data/clean/File_concat"),
**inputs
)
def compute(output, **kwargs):
# Extract dataframes from input datasets
input_dfs = [dataset_df.dataframe() for dataset_name, dataset_df in kwargs.items()]
dfs = []
def transformation_input(df):
# some transformation
return df
for input_df in input_dfs:
dfs.append(transformation_input(input_df))
# dfs = reduce(DataFrame.unionByName, dfs)
unioned_dfs = union_many(*dfs)
return unioned_dfs
Couple points:
Created dynamic input dict.
That dict is read into the transform using **kwargs.
Using transform decorator not transform_df, we can extract the dataframes.
(not in question) Combine multiple dataframes using union_many function from transforms_verbs library.
Let's say I train an autoencoder.
I want to freeze the parameters of the encoder for the training, so only the decoder trains.
I can do this using:
# assuming it's a single layer called 'encoder'
model.encoder.weights.data.requers_grad = False
Or I can pass only the decoder's parameters to the optimizer. Is there a difference?
The most practical way is to iterate through all parameters of the module you want to freeze and set required_grad to False. This gives you the flexibility to switch your modules on and off without having to initialize a new optimizer each time. You can do this using the parameters generator available on all nn.Modules:
for param in module.parameters():
param.requires_grad = False
This method is model agnostic since you don't have to worry whether your module contains multiple layers or sub-modules.
Alternatively, you can call the function nn.Module.requires_grad_ once as:
module.requires_grad_(False)
Due to the way pytest works, it is not possible (or recommended) to import other modules in a pytest module. Instead, one should properly edit it's conftest.py file.
Several times, I am put in a situation where I need to share constants/functions to several tests modules. And fixture fails to be as practical as functions. Even if they can be indirectly parametrized with the indirect parameter, they are still situations where it's not possible, or simple, to use this approach.
For constants, I am in the following situation, here is an extract of my conftest.py:
TARGET_NAME_1 = 'MY_OP4510'
TARGET_NAME_2 = 'MY_ML605'
TARGET_NAME_3 = 'TARGET_WITH_CHILD'
CONFIG_FILE_NAME = 'config.ini'
#pytest.fixture()
def target_name_1():
"""This fixture returns a target name"""
return TARGET_NAME_1
#pytest.fixture()
def target_name_2():
"""This fixture returns a target name"""
return TARGET_NAME_2
#pytest.fixture()
def target_name_3():
"""This fixture returns a target name"""
return TARGET_NAME_3
#pytest.fixture()
def target_config_path():
"""This fixture returns the config path"""
return CONFIG_FILE_NAME
Every time I have to add a constant, I have to add a fixture. Also, this increase the number of parameters the tests functions will receive (if in this case, I could use the autouse parameter, for some other fixtures that actually execute code, I do not necessary want to auto-use them as they could prevent other test cases from working).
I am looking for a way to simplify this code, would you have a good pattern/implementation to suggest ?
We are working on a Top-Down-RPG-like Multiplayer game for learning purposes (and fun!) with some friends. We already have some Entities in the Game and Inputs are working, but the network implementation gives us headache :D
The Issues
When trying to convert with dict some values will still contain the pygame.Surface, which I dont want to transfer and it causes errors when trying to jsonfy them. Other objects I would like to transfer in a simplyfied way like Rectangle cannot be converted automatically.
Already functional
Client-Server connection
Transfering JSON objects in both directions
Async networking and synchronized putting into a Queue
Situation
A new player connects to the server and wants to get the current game state with all objects.
Data-Structure
We use a "Entity-Component" based architecture, so we separated the game logic very strictly into "systems", while the data is stored in the "components" of each Entity. The Entity is a very simple container and has nothing more than a ID and a list of components. Example Entity (shorten for better readability):
Entity
|-- Component (Moveable)
|-- Component (Graphic)
| |- complex datatypes like pygame.SURFACE
| `- (...)
`- Component (Inventory)
We tried different approaches, but all seems not to fit very well or feel "hacky".
pickle
Very Python near, so not easy to implement other clients in future. And I´ve read about some security risks when creating items from network in this dynamic way how pickle it offers. It does not even solve the Surface/Rectangle issue.
__dict__
Still contains the reference to the old objects, so a "cleanup" or "filter" for unwanted datatypes happens also in the origin. A deepcopy throws Exception.
...\Python\Python36\lib\copy.py", line 169, in deepcopy
rv = reductor(4)
TypeError: can't pickle pygame.Surface objects
Show some code
The method of the "EnitityManager" Class which should generate the Snapshot of all Entities, including their components. This Snapshot should be converted to JSON without any errors - and if possible without much configuration in this core-class.
class EnitityManager:
def generate_world_snapshot(self):
""" Returns a dictionary with all Entities and their components to send
this to the client. This function will probably generate a lot of data,
but, its to send the whole current game state when a new player
connects or when a complete refresh is required """
# It should be possible to add more objects to the snapshot, so we
# create our own Snapshot-Datastructure
result = {'entities': {}}
entities = self.get_all_entities()
for e in entities:
result['entities'][e.id] = deepcopy(e.__dict__)
# Components are Objects, but dictionary is required for transfer
cmp_obj_list = result['entities'][e.id]['components']
# Empty the current list of components, its going to be filled with
# dictionaries of each cmp which are cleaned for the dump, because
# of the errors directly coverting the whole datastructure to JSON
result['entities'][e.id]['components'] = {}
for cmp in cmp_obj_list:
cmp_copy = deepcopy(cmp)
cmp_dict = cmp_copy.__dict__
# Only list, dict, int, str, float and None will stay, while
# other Types are being simply deleted including their key
# Lists and directories will be cleaned ob recursive as well
cmp_dict = self.clean_complex_recursive(cmp_dict)
result['entities'][e.id]['components'][type(cmp_copy).__name__] \
= cmp_dict
logging.debug("EntityMgr: Entity#3: %s" % result['entities'][3])
return result
Expectation and actual results
We can find a way to manually override elements which we dont want. But as the list of components will increase we have to put all the filter logic into this core class, which should not contain any components specializations.
Do we really have to put all the logic into the EntityManager for filtering the right objects? This does not feel good, as I would like to have all convertion to JSON done without any hardcoded configuration.
How to convert all this complex data in a most generic approach?
Thanks for reading so far and thank you very much for your help in advance!
Interesting articles which we were already working threw and maybe helpful for others with similar issues
https://gafferongames.com/post/what_every_programmer_needs_to_know_about_game_networking/
http://code.activestate.com/recipes/408859/
https://docs.python.org/3/library/pickle.html
UPDATE: Solution - thx 2 sloth
We used a combination of the following architecture, which works really great so far and is also good to maintain!
Entity Manager now calls the get_state() function of the entity.
class EntitiyManager:
def generate_world_snapshot(self):
""" Returns a dictionary with all Entities and their components to send
this to the client. This function will probably generate a lot of data,
but, its to send the whole current game state when a new player
connects or when a complete refresh is required """
# It should be possible to add more objects to the snapshot, so we
# create our own Snapshot-Datastructure
result = {'entities': {}}
entities = self.get_all_entities()
for e in entities:
result['entities'][e.id] = e.get_state()
return result
The Entity has only some basic attributes to add to the state and forwards the get_state() call to all the Components:
class Entity:
def get_state(self):
state = {'name': self.name, 'id': self.id, 'components': {}}
for cmp in self.components:
state['components'][type(cmp).__name__] = cmp.get_state()
return state
The components itself now inherit their get_state() method from their new superclass components, which simply cares about all simple datatypes:
class Component:
def __init__(self):
logging.debug('generic component created')
def get_state(self):
state = {}
for attr, value in self.__dict__.items():
if value is None or isinstance(value, (str, int, float, bool)):
state[attr] = value
elif isinstance(value, (list, dict)):
# logging.warn("Generating state: not supporting lists yet")
pass
return state
class GraphicComponent(Component):
# (...)
Now every developer has the opportunity to overlay this function to create a more detailed get_state() function for complex types directly in the Component Classes (like Graphic, Movement, Inventory, etc.) if it is required to safe the state in a more accurate way - which is a huge thing for maintaining the code in future, to have these code pieces in one Class.
Next step is to implement the static method for creating the items from the state in the same Class. This makes this working really smooth.
Thank you so much sloth for your help.
Do we really have to put all the logic into the EntityManager for filtering the right objects?
No, you should use polymorphism.
You need a way to represent your game state in a form that can be shared between different systems; so maybe give your components a method that will return all of their state, and a factory method that allows you create the component instances out of that very state.
(Python already has the __repr__ magic method, but you don't have to use it)
So instead of doing all the filtering in the entity manager, just let him call this new method on all components and let each component decide that the result will look like.
Something like this:
...
result = {'entities': {}}
entities = self.get_all_entities()
for e in entities:
result['entities'][e.id] = {'components': {}}
for cmp in e.components:
result['entities'][e.id]['components'][type(cmp).__name__] = cmp.get_state()
...
And a component could implement it like this:
class GraphicComponent:
def __init__(self, pos=...):
self.image = ...
self.rect = ...
self.whatever = ...
def get_state(self):
return { 'pos_x': self.rect.x, 'pos_y': self.rect.y, 'image': 'name_of_image.jpg' }
#staticmethod
def from_state(state):
return GraphicComponent(pos=(state.pos_x, state.pos_y), ...)
And a client's EntityManager that recieves the state from the server would iterate for the component list of each entity and call from_state to create the instances.
I am trying to run the seq2seq translate example present on Tensorflow website and getting following error. I am using tensorflow-gpu==1.1.0
ValueError: Attempt to reuse RNNCell
with a different variable scope than its first use.
First use of cell was with scope
'embedding_attention_seq2seq/embedding_attention_decoder/attention_decoder/gru_cell',
this attempt is with scope 'embedding_attention_seq2seq/rnn/gru_cell'.
Please create a new instance of the cell if you would like it to use a
different set of weights. If before you were using:
MultiRNNCell([GRUCell(...)] * num_layers), change to:
MultiRNNCell([GRUCell(...) for _ in range(num_layers)]). If before
you were using the same cell instance as both the forward and reverse
cell of a bidirectional RNN, simply create two instances (one for
forward, one for reverse). In May 2017, we will start transitioning
this cell's behavior to use existing stored weights, if any, when it
is called with scope=None (which can lead to silent model degradation,
so this error will remain until then.)
On github people were suggesting to change the add the reuse argument to cell as follows :
def single_cell():
return tf.contrib.rnn.GRUCell(size, reuse = tf.get_variable_scope().reuse)
if use_lstm:
def single_cell():
return tf.contrib.rnn.BasicLSTMCell(size, reuse = tf.get_variable_scope().reuse)
cell = single_cell()
if num_layers > 1:
cell = tf.contrib.rnn.MultiRNNCell([single_cell() for _ in range(num_layers)])
But still I am getting the same error. What's the issue and how to resolve it ?
Any help is highly appreciated.
P.S: A similar post was there on stackoverflow,but that solution didn't work for me and since the version of TF is different, I created a new post.