Stack Overflow
Stack Overflow

Questions tagged [dqn]

DQN is a multi-layered neural network, added target network and experience replay to Q-learning

206 questions
2
votes
1 answer

Atari score vs reward in rllib DQN implementation

I'm trying to replicate DQN scores for Breakout using RLLib. After 5M steps the average reward is 2.0 while the known score for Breakout using DQN is 100+. I'm wondering if this is because of reward clipping and therefore actual reward does not…
Shital Shah
  • 63,284
  • 17
  • 238
  • 185
1
vote
0 answers

ValueError: RolloutWorker has no `input_reader` object! Cannot call `sample()` occurs when try to start training in RLlib

I just started to use Ray-RLlib and I try to restructure my VSL algorithm (using SUMO) to be able to work with RLlib. This is a MARL (4 agents) DQN approach. I restructured the environment based on the documentation and created the training…
komate1995
  • 11
  • 2
1
vote
0 answers

Large Discrete States for DQN when using ReinforcementLearning.jl

I am using the Julia package ReinforcementLearning.jl. I would like to gain from the fact that DQN does not require enumerate and revising the whole state space. So, my question is how to describe state_space for discrete environments with no need…
Alejandro Lamas
  • 11
  • 2
1
vote
1 answer

DDPG training in Tensorflow.js

I'm trying to figure out how to implement DDPG in Tensorflow.js using Python examples such as this one from keras website. I got stuck on the training code: with tf.GradientTape() as tape: target_actions =…
Oleg Khalidov
  • 5,108
  • 1
  • 28
  • 29
1
vote
0 answers

Defining state and action for Q-learning in the code

I am trying to understand the following code for the simulator to avoid collision with the help of Q-learning. The examples and tutorials which I followed had the space divided into blocks such as taxiv3, so it was rather easier to define state and…
itsChibi
  • 61
  • 5
1
vote
0 answers

DQN from scratch not able to learn any environment, no matter parameters using

I've attempted to implement a DQN from scratch, without importing the neural network from a library, and have attempted to get it to work with the Luna Lander environment, however no matter the values for the parameters and even after 600 episodes…
David Hestelow
  • 11
  • 2
1
vote
1 answer

Saving and Loading RL model - DQN algorithm

I am training a RL model using the DQN algorithm. At every iteration, I save the model as follows: agent = dqn.DQNTrainer(env=CustomEnv,config=config) for n in range(100): result = agent.train() agent.save() I want to evluate the…
EngGu
  • 459
  • 3
  • 14
1
vote
1 answer

Open AI gym render function not working(MACbook pro M1 max chip)

Try to run the cartpole environment on my macbook pro and the render function cannot work as there shows no animation about the cart. However, the code seems to work and my DQN can train under the environment. Still no animation after trying some…
Jackmeory_thelatestversion
  • 21
  • 3
1
vote
0 answers

I'm trying to export gym-gazebo GazeboCircuit2TurtlebotLidar as a custom env to my algorithm. I could not import gym-gazebo as typical gym env

I am trying to connect an agent(GazeboCircuit2TurtlebotLidar-v0) from gym-gazebo library(gym-gazebo is an extension of the initial OpenAI gym for robotics using ROS and Gazebo, an advanced 3D modeling and rendering tool) with a deep reinforcement…
Janani
  • 11
  • 2
1
vote
1 answer

Continuous action and space in DQN with Gym.spaces.Box?

I wanna investigate 3 functions in comparison with my cost function in a specific period. my action and observation spaces are continuous. how do I solve this error? import math import numpy as np import gym from gym import spaces from gym.spaces…
Hamed HSNY
  • 11
  • 2
1
vote
1 answer

GPU utilization is low when training Deep Q Network (DQN)

I am try to train a DQN model with the following code. The GPU (cuda) usage is always lower than 25 percent. I know the tensorflow backend is consulting the GPU resources, but the usage is low. Is there any way I can improve the utilization of the…
StackOver
  • 65
  • 4
1
vote
1 answer

TF Agent taking the same action for all test states after training in Reinforcement Learning

I am trying to create a Custom PyEnvironment for making an agent learn the optimum hour to send the notification to the users, based on the rewards received by clicking on the notifications sent in previous 7 days. After the training is complete,…
Sukrit Mehta
  • 11
  • 2
1
vote
0 answers

Deep Reinforcement Learning 1-step TD not converging

Is there some trick to getting 1-step TD (temporal difference) prediction to converge with a neural net? The network is a simple feed forward network using ReLU. I've got the network working for Q-learning in the following way: gamma = 0.9 q0 =…
John Hoeck
  • 11
  • 1
1
vote
1 answer

Why are the mean_q and mae for keras-rl2 DQN agent logged as NaN

Copied the codes over from https://github.com/keras-rl/keras-rl/blob/master/examples/dqn_atari.py but only the rewards and number of steps are logged and the error metrics are all NaN memory = SequentialMemory(limit=1000000,…
Dukey
  • 11
  • 2
1
vote
0 answers

tensorflow, using a `tf.Tensor` as a Python `bool` is not allowed in Graph execution

raise errors.OperatorNotAllowedInGraphError( tensorflow.python.framework.errors_impl.OperatorNotAllowedInGraphError: using a `tf.Tensor` as a Python `bool` is not allowed in Graph execution. Use Eager execution or decorate this function with…
JA_Gamer_
  • 11
  • 1
1 2
3
…
13 14