agents
agents copied to clipboard
tensorflow
TF-Agents: A reliable, scalable and easy to use TensorFlow library for Contextual Bandits and Reinforcement Learning.
I need to extend the max episode per each step in the CartPole env under the Gymnasium library. ``` env = suite_gym.load('CartPole-v1', max_episode_steps = 10000) print(env._max_episode_steps) env = tf_py_environment.TFPyEnvironment(env) ```...
I used the following to create part of reinforcement learning environment: from tf_agents.trajectories import trajectory, time_step def _reset(self): self._state = np.zeros(num_service+1, dtype=np.int32) self._episode_ended = False self._task_queue = 0 np.random.seed(123) return...
suite_gym.load('CartPole-v1', max_episode_steps=10000) does not modify the default max_episode_steps
When training a simple CartPole environment trying to extend the default max_episode_steps as following ``` env = suite_gym.load('CartPole-v1', max_episode_steps=10000) env = tf_py_environment.TFPyEnvironment(env) ``` according to https://www.tensorflow.org/agents/api_docs/python/tf_agents/environments/suite_gym/load#args ``` tf_agents.environments.suite_gym.load( environment_name: Text,...
Hi! Would it be possible to a changelog or some kind of release notes? I can't seem to find it anywhere and it would be very useful to track changes....
It seems it is not possible to define a variable decay epsilon during training for dqn_agent.DqnAgent. I have some direct and indirect evidence of this: 1. I tried to define...
When trying to load an environment based on gymnasium instead of gym, the wrapper function outputs the error: ``` File "/home/xxx/anaconda3/envs/rllib/lib/python3.9/site-packages/tf_agents/environments/gym_wrapper.py", line 140, in spec_from_gym_space raise ValueError( ValueError: The gym...
I am trying to train a DDQN agent (a self driving car in GTAV) with some state observations and discrete actions. I have done an initial training of 1 million...
I tried to use tf-agents, the latest version. However, if I run a simple class which only extends PyEnvironment but nothing else, I receive an with a message like **module...
Hi, I'm not an expert in using TF agents, I've started to learn how to use this library quite recently, so I don't know if I'm just ignoring some implementation...
I save the model during certain timesteps so at the end of the run i would have multiple checkpoints, the files are saved as ckpt-1000, ckpt-2000 ... If for example,...
Metadata
Owner
Metadata
TF-Agents: A reliable, scalable and easy to use TensorFlow library for Contextual Bandits and Reinforcement Learning.