All Questions
Tagged with openai-gym tensorflow
106 questions
1
vote
0
answers
17
views
"Install TensorFlow and Gym with Conda for usage with Jupyter Notebook in macOS Sonoma, Intel processor
I am trying to set up a virtual environment using Conda to code a lab activity regarding RL, but it is proving to be quite a nightmare due to incompatibilities among different library versions. The ...
- 191
0
votes
0
answers
35
views
I'm getting a "invalid type "ResourceVariable" must be a string or Tensor." on the rL compile function when trying to train a DQM with Keras
I was trying to train a model on a gym environment when I encountered the following error:
TypeError: Argument `fetch` = <tf.Variable 'dense/kernel:0' shape=(4, 24) dtype=float32> has invalid ...
1
vote
0
answers
132
views
Building a Reinforcement Learning Model by combining outputs of many different ML models
Compared to many people, I am an amateur about coding and I need some guidance.
After careful writing and tweaking to build a stock price predictor (hopefull a trading bot after that), I built 3 ...
- 13
1
vote
0
answers
486
views
Keras Sequential Unhashable type: 'list error at import
I'm having problems with running a deep q-learning model with Keras-RL and OpenAI Gym in Python. In particular I get an error when loading the Sequential model from the keras package.
The code is the ...
- 187
0
votes
0
answers
105
views
Error when creating DQNAgent with Keras-RL: "Keras symbolic inputs/outputs do not implement __len__"
I'm trying to create a DQNAgent using Keras-RL to train a CartPole-v0 environment. However, I'm encountering an issue related to the Keras symbolic inputs/outputs not implementing len. Here's the ...
0
votes
0
answers
93
views
OpenAI Gym 'CarRacing-v2' (discrete!) ValueError wrapping in TensorFlow Py-Environment
I first instantiate the discrete version of the 'CarRacing v2' environment and then want to wrap it in TensorFlow(TF):
env = gym.make("CarRacing-v2", continuous = False) #discrete version ...
0
votes
1
answer
565
views
Deep Reinforcement Learning on Raspberry Pi
I am trying to run a deep reinforcement problem on a Raspberry Pi 4. The code successfully runs on Colab but shows the following error on my Pi.
/home/pi/.local/lib/python3.9/site-packages/...
- 1
0
votes
1
answer
249
views
cannot import name '__version__' issue
i would like to implement Deep neural network+RL in python, here is my code :
import random
import gym
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense,...
- 215
0
votes
0
answers
138
views
tf_agents reset environment using actor
I'm trying to understand how to use Actor class in tf_agents. I am using DDPG (actor-critic, although this doesn't really matter per say). I also am learning off of gym package, although again this ...
- 367
0
votes
1
answer
101
views
Multiple errors in deep reinforcement learning project with keras and openai gym
I have copied code from a source about deep q-learning to try to learn from it but it is an older source so there are many things that are going wrong with both keras and openai gym. I have tried for ...
0
votes
1
answer
551
views
A2C and stable_baselines3
I´m trying to use this code from a repo in GitHub (https://github.com/nicknochnack/Reinforcement-Learning-for-Trading-Custom-Signals/blob/main/Custom%20Signals.ipynb) in Point 3:
model = A2C('...
- 1
1
vote
2
answers
590
views
DQN, TF, nested spcaes.Dict : How to deal with variable size observation space?
I am very new to RL and DQN. And trying to code an agent for my problem statement. I am using Replay Buffer concept and trying to learn to code this agent manually. My observation space is a ...
- 49
0
votes
1
answer
557
views
Attribute error in PPO algorithm for Cartpole gym environment
I'm trying to run the code from here (Github link on this page): https://keras.io/examples/rl/ppo_cartpole/
I'm getting an attribute error in the training section from observation = observation....
- 13
0
votes
2
answers
337
views
Im having problems adding observation space to a custom Gym enviroment
I am using this code. I have modified it to work with a car (0 Left, 1 Straight, 2 Right).
I would like to add some observations, such as Destination (XY), Car Location (XY) bearing (angle), ...
- 13
0
votes
1
answer
2k
views
cannot import name 'load_model' from 'keras.engine.saving'
Issue :
I am trying to use the gym connect4 env.
When running this sample code :
import gym
from gym_connect_four import RandomPlayer, ConnectFourEnv
env: ConnectFourEnv = gym.make("ConnectFour-...
- 1
0
votes
0
answers
80
views
KerasRl ValueError: Error when checking input: expected input_3 to have 3 dimensions, but got array with shape (1, 1, 9, 9)
I made an env with Gym for Sudoku puzzle and I want to train an AI on it using KerasRL (I've removed the step reset and render method of the environment to not have too much code for StackOverflow).
I ...
2
votes
0
answers
201
views
I need to install tensorflow 1.x and the code works until last week on google colab. Now it does not work
I need to install TensorFlow 1.x for solving my problem. When I run the below codes a week back on google colab (python version 3.8). It installed successfully, and since then I haven't changed ...
- 351
1
vote
1
answer
2k
views
How do you use OpenAI Gym 'wrappers' with a custom Gym environment in Ray Tune?
How do you use OpenAI Gym 'wrappers' with a custom Gym environment in Ray Tune?
Let's say I built a Python class called CustomEnv (similar to the 'CartPoleEnv' class used to create the OpenAI Gym &...
- 81
2
votes
2
answers
383
views
Tf-agent Actor/Learner: TFUniform ReplayBuffer dimensionality issue - invalid shape of Replay Buffer vs. Actor update
I try to adapt the this tf-agents actor<->learner DQN Atari Pong example to my windows machine using a TFUniformReplayBuffer instead of the ReverbReplayBuffer which only works on linux machine ...
- 41
0
votes
1
answer
282
views
TypeError: Only integers, slices (`:`), ellipsis (`...`), tf.newaxis (`None`) and scalar tf.int32/tf.int64 tensors are valid indices
Before I start, I know there are a lot of questions with the same error but none of them solved the issue for me.
I have a PPO implementation for playing the CarRacing-v2 environment from gym (gym==0....
- 171
0
votes
1
answer
351
views
tf_agents and reverb produce incompatible tensor
I'm trying to implement a DDPG using tf_agents and reverb but I can't figure out how both libraries to work together. For this, I'm trying to use the code from the DQL-Tutorial from tf_agents with my ...
- 322
0
votes
1
answer
267
views
How to take two arrays as output from Gym.Env to fit to DQN NN
Can't figure out how to make the gym.Env put out two separate arrays. It just seems to combine them into 1 array containing 2 arrays. But fitting to DQN NN expects two arrays.
I'm hoping to put the ...
0
votes
1
answer
129
views
ValueError: Error when checking input: expected flatten_input to have shape... but got the shape
When trying to implement a DQN with Tensorflow/Keras, on an openai-gym environment, I'm encountering this error:
ValueError: Error when checking input: expected flatten_input to have shape (1, 4) but ...
- 81
3
votes
6
answers
4k
views
ValueError: Error when checking input: expected flatten_input to have shape (1, 4) but got array with shape (1, 2)
I'm fairly new to RL and i can't really understand why I'm getting this error.
import random
import numpy as np
import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras....
1
vote
1
answer
463
views
TensorFlow with custom gym environment: Layer "dense_6" expects 1 input(s), but it received 2 input tensors
I am trying to use TF to solve a custom gym environment, all within Google Colab.
The main script is the TF "DQN Tutorial" available here.
In place of env_name = "CartPole-v0" I ...
- 2,387
1
vote
2
answers
268
views
Why does my model not learn? Very high loss
I built a simulation model where trucks collect garbage containers based on their fill level.
I used OpenAi Gym and Tensorflow/keras to create my Deep Reinforcement Learning model...
But my training ...
- 27
2
votes
0
answers
966
views
Omnet++ with Reinforcement Learning Tools [ML]
I am currently failing into find an easy and modular framework to link openAI gym or tensorflow or keras with omnet++ in such a way I can produce communication between each tool and have online ...
- 29
0
votes
1
answer
785
views
real tme keras rl DQN predictions
hello everyone I followed that tutorial
https://www.youtube.com/watch?v=hCeJeq8U0lo&list=PLgNJO2hghbmjlE6cuKMws2ejC54BTAaWV&index=2
to train a DQN Agent
everything works
env = gym.make('...
- 33
0
votes
0
answers
413
views
Error:DQN expects a model that has one dimension for each action, in this case (1, 2, 1, 0)
i am building an RL agent for which the model is defined:
def build_model(states, actions):
azioni = list(actions)
model = Sequential()
model.add(Dense(4, activation='relu', input_shape=[len(azioni)]))...
- 15
1
vote
0
answers
1k
views
TypeError: 'Sequential' object is not subscriptable when using the model to predict something
i've tried to create a model and a function that plays one step in reinforcement learning in lunar lander.
import gym
env = gym.make("LunarLander-v2")
this is the environment
keras.backend....
- 21
2
votes
1
answer
715
views
ValueError: Error when checking input: expected dense_input to have 2 dimensions, but got array with shape (1, 1, 2)
Edit: Problem Solved. Solution below.
Attempting to build a RL model to handle a task.
There are two inputs: x and y, both are measured on an int scale of 1 to 100.
Based on these two inputs there ...
- 35
0
votes
0
answers
131
views
Keras-RL error when training agent for tic tac toe game: "expected dense_16_input to have 2 dimensions, but got array with shape (1, 1, 3, 3)"
I just recently tried using Keras-RL to train an agent in a tictactoe game I made to practice making custom environments for my final third year project which involves doing this but on a much larger ...
0
votes
1
answer
308
views
Bipedalwalker from openai gym solving using tensorflow
i am trying to solve the Bipedalwalker from openai. The Problem is that i always get the error:
The shape of the output should be 4 values between -1 and 1(like: [ 0.45099565 -0.7659952 -0.01972992 ...
5
votes
0
answers
661
views
Alternatives of Stable Baselines3
can you suggest some alternative of stable baselines that I can use to train my agent in reinforcement learning.
P.s. I'm using gym mini-grid environment so tell me those who work in this environment.
- 51
2
votes
2
answers
2k
views
Error: DQN expects a model that has one dimension for each action
I am building an RL agent for which the model is defined:
def build_model(height,width,channels,actions):
model = Sequential()
model.add(Conv2D(32,(8,8),strides=(4,4),activation='relu',...
- 179
0
votes
1
answer
246
views
I get horrible results with my DDPG model TF2
Hello my DDPG model that I have implemented in TF 2 get's horrible results at every env on openai-gym that has continuous actions I need help to find what's the problem. I run this on my GPU. On env ...
- 41
0
votes
1
answer
629
views
How do I get Target Q-values in Bipedalwalker-v3 in openai-gym, reinforcement learning?
I am new to reinforcement learning and I was trying to solve the BipedalWalker-v3 using Deep Q learning. However I found out that the env.action_space.sample() = numpy array with 4 elements and I am ...
- 247
0
votes
1
answer
7k
views
ValueError: Error when checking input: expected dense_input to have 2 dimensions, but got array with shape (1, 1, 15)
I am trying to make a custom Gym Environment so that I can use it in a Keras Network. But there is a problem that is happening to me when I try to fit de neural network.
ValueError: Error when ...
0
votes
1
answer
496
views
How to use a trained RL model to make a prediction?
I would like to use my trained RL model for a discrete test prediction.
This is how the model is built:
model = Sequential()
model.add(Dense(60, activation='relu', input_shape=states))
model.add(Dense(...
- 2,821
3
votes
1
answer
968
views
How to build a DQN that outputs 1 discrete and 1 continuous value as a pair?
I am building a DQN for an Open Gym environment. My observation space is only 1 discrete value but my actions are:
self.action_space = (Discrete(3), Box(-100, 100, (1,)))
ex: [1,56], [0,24], [2,-78].....
- 2,821
2
votes
1
answer
1k
views
Simple DQN to slow to train
I have been trying to solve the OpenAI lunar lander game with a DQN taken from this paper
https://arxiv.org/pdf/2006.04938v2.pdf
The issue is that it takes 12 hours to train 50 episodes so something ...
- 16.5k
0
votes
1
answer
249
views
Python neural network with Keras runs on CPU, but crashes on the GPU
I implemented a neural network that learns to play PacMan using gym,box2d and gym[atari] with Keras models. The training was very slow so I tried to make in run on my GTX 1060 Max-Q.
I installed the ...
- 15
3
votes
0
answers
3k
views
Reinforced learning agent build DQNAgent causes unknown problem
I wanted to get into reinforced learning a bit, so I started with the fairly simple example "Cartpole" by following a hands-on tutorial.
Github link of the tutorial source code (identical ...
- 31
0
votes
2
answers
248
views
Deep Reinforcement Learning Motion in Observation
I am trying to implement a DRL (Deep Reinforcement Learning) Agent for self-driving vehicles. I am currently teaching my agent not to bump on other cars, using a simple camera. There are many ways to ...
- 476
0
votes
1
answer
450
views
saving RL agent by pickle, cannot save because of pickle.thread_RLock -- what is the source of this error?
I am trying to save my reinforcement learning agent class after training for further training later on by pickling it.
The script used is:
with open('agent.pickle','wb') as agent_file:
pickle....
user9317212
0
votes
1
answer
978
views
How to use own environment for DDPG without gym
I'm using Keras to build a ddpg model,I followed the official instruction from here enter link description here
But I want to my own environment, not gym,here is my own environment:
class Environment1:...
- 4,020
1
vote
0
answers
134
views
proximal policy gradient tensorflow pendulum issue
import gym
import numpy as np
import tensorflow as tf
class Memory(object):
def __init__(self):
self.ep_obs, self.ep_act, self.ep_rwd, self.ep_neglogp = [], [], [], []
def ...
1
vote
1
answer
1k
views
Why DQN for cartpole game has a ascending reward while loss is not descending?
I wrote a DQN to play the OpenAI gym cart pole game with TensorFlow and tf_agents. The code looks like the following:
def compute_avg_return(environment, policy, num_episodes=10):
total_return = 0....
- 822
1
vote
0
answers
238
views
Failed to find data adapter that can handle input: <class 'numpy.uint8'>, <class 'NoneType'> - Keras Reinforcement Learning in AI Gym
I have been trying to train an agent to play Atari games from the Open Ai Gym environment using the Deep Q Networks but I am running in to an error when trying to use the current state given by the ...
- 453
0
votes
0
answers
918
views
How to solve "KeyError: '/conv2d_1/kernel:0'"
I am trying to use the colab to run the gym package with pacman, since the spec in colab is more powerful than my notebook. This program is successful simulate in Jupyter in my notebook, which using ...
- 29