agents

agents copied to clipboard

Network._build assumes input array spec is bounded

1

When defining a custom network, the `Network` class attempts to build by calling `tensor_spec.sample_spec_nest` to sample an input. Line 134 in `tf_agents/network/network.py`: ``` def _build(self): if not self.built and self.input_tensor_spec...

ericzhao28

Tutorial 8: Networks - faulty link to Keras' network.py

1

Hello, in Tutorial 8: Networks, there's a faulty link to Keras' network.py in the section Defining Networks. I am not sure how this bug came about but my guess is...

jan-gebauer

Failed to run the Train_eval_atari.py in categorical DQN example

2

I have been trying to run the example by following, ``` python tf_agents/agents/categorical_dqn/examples/train_eval_atari.py \ --root_dir=$HOME/atari/pong \ --alsologtostderr ``` However, this error occurs... ``` W tensorflow/core/kernels/data/generator_dataset_op.cc:107] Error occurred when finalizing GeneratorDataset...

okvlam

Dear, I want to change the loss function a little bit, just as this [https://arxiv.org/abs/2012.06 644]. Is this possible to this within the tf-agents? Thank you in advance! Best regards,...

WangMengqi32C

Hi! I'm wondering how best to add a second head to, for example a DQN Agent to train an auxiliary supervised task. So far the only work I've found with...

npqst

Q policies have no way of knowing whether the Q-Network underlying them should be called with the training mode flag = True or False. By default they are implicitly always...

FMalerba

Hi everyone, I am trying to understand the Rubik Cube behaviour however I have found something I think is wrong: 1.) I created an environment (rubik cube) in a specific...

JorgeQuinteroM

I have a project where I would like to know whether the last action applied to the environment came from the agent's policy or from a random action (as a...

ngroves08

Custom OpenAI Gym environment wrapped as a PyEnvironment time_step not matching expected time_step_spec

1

Greetings. This problem is similar to the one here at: https://stackoverflow.com/questions/57259497/py-environment-time-step-doesnt-match-time-step-spec . However, with the minimal documentation on how the PyEnvironment wrapper works, I'm having some trouble resolving. The custom...

mjssimon

Random_TF_Environment Bug

4

Hi there, I think I've found a minor bug in the Random_TF_Enviroment class. I've discovered that it fails when the time_step_spec includes a reward_spec that is a multi-dimensional array (as...

ngroves08