Slow training on CPU and GPU in a small network (tensorflow)

问题

Here is the original script I'm trying to run on both CPU and GPU, I'm expecting a much faster training on GPU however it's taking almost the same time. I made the following modification to main()(the first 4 lines) because the original script does not activate / use the GPU. Suggestions ... ?

def main():
    physical_devices = tf.config.experimental.list_physical_devices('GPU')
    if len(physical_devices) > 0:
        tf.config.experimental.set_memory_growth(physical_devices[0], True)
        print('GPU activated')
    env = gym.make('CartPole-v1')
    agent = Agent(env)
    agent.train(max_episodes=1000)

Update:

wandb's report shows 0% GPU utilization which confirms that there is a problem

Full code in question which is not mine and belongs to this repository:

import wandb
import tensorflow as tf
from tensorflow.keras.layers import Input, Dense
from tensorflow.keras.optimizers import Adam

import gym
import argparse
import numpy as np
from collections import deque
import random

tf.keras.backend.set_floatx('float64')
wandb.init(name='DQN', project="deep-rl-tf2")

parser = argparse.ArgumentParser()
parser.add_argument('--gamma', type=float, default=0.95)
parser.add_argument('--lr', type=float, default=0.005)
parser.add_argument('--batch_size', type=int, default=32)
parser.add_argument('--eps', type=float, default=1.0)
parser.add_argument('--eps_decay', type=float, default=0.995)
parser.add_argument('--eps_min', type=float, default=0.01)

args = parser.parse_args()

class ReplayBuffer:
    def __init__(self, capacity=10000):
        self.buffer = deque(maxlen=capacity)
    
    def put(self, state, action, reward, next_state, done):
        self.buffer.append([state, action, reward, next_state, done])
    
    def sample(self):
        sample = random.sample(self.buffer, args.batch_size)
        states, actions, rewards, next_states, done = map(np.asarray, zip(*sample))
        states = np.array(states).reshape(args.batch_size, -1)
        next_states = np.array(next_states).reshape(args.batch_size, -1)
        return states, actions, rewards, next_states, done
    
    def size(self):
        return len(self.buffer)

class ActionStateModel:
    def __init__(self, state_dim, aciton_dim):
        self.state_dim  = state_dim
        self.action_dim = aciton_dim
        self.epsilon = args.eps
        
        self.model = self.create_model()
    
    def create_model(self):
        model = tf.keras.Sequential([
            Input((self.state_dim,)),
            Dense(32, activation='relu'),
            Dense(16, activation='relu'),
            Dense(self.action_dim)
        ])
        model.compile(loss='mse', optimizer=Adam(args.lr))
        return model
    
    def predict(self, state):
        return self.model.predict(state)
    
    def get_action(self, state):
        state = np.reshape(state, [1, self.state_dim])
        self.epsilon *= args.eps_decay
        self.epsilon = max(self.epsilon, args.eps_min)
        q_value = self.predict(state)[0]
        if np.random.random() < self.epsilon:
            return random.randint(0, self.action_dim-1)
        return np.argmax(q_value)

    def train(self, states, targets):
        self.model.fit(states, targets, epochs=1, verbose=0)
    

class Agent:
    def __init__(self, env):
        self.env = env
        self.state_dim = self.env.observation_space.shape[0]
        self.action_dim = self.env.action_space.n

        self.model = ActionStateModel(self.state_dim, self.action_dim)
        self.target_model = ActionStateModel(self.state_dim, self.action_dim)
        self.target_update()

        self.buffer = ReplayBuffer()

    def target_update(self):
        weights = self.model.model.get_weights()
        self.target_model.model.set_weights(weights)
    
    def replay(self):
        for _ in range(10):
            states, actions, rewards, next_states, done = self.buffer.sample()
            targets = self.target_model.predict(states)
            next_q_values = self.target_model.predict(next_states).max(axis=1)
            targets[range(args.batch_size), actions] = rewards + (1-done) * next_q_values * args.gamma
            self.model.train(states, targets)
    
    def train(self, max_episodes=1000):
        for ep in range(max_episodes):
            done, total_reward = False, 0
            state = self.env.reset()
            while not done:
                action = self.model.get_action(state)
                next_state, reward, done, _ = self.env.step(action)
                self.buffer.put(state, action, reward*0.01, next_state, done)
                total_reward += reward
                state = next_state
            if self.buffer.size() >= args.batch_size:
                self.replay()
            self.target_update()
            print('EP{} EpisodeReward={}'.format(ep, total_reward))
            wandb.log({'Reward': total_reward})


def main():
    env = gym.make('CartPole-v1')
    agent = Agent(env)
    agent.train(max_episodes=1000)

if __name__ == "__main__":
    main()

回答1:

I very much suspect that I/O operations are taking up almost everything (particularly with your self-implemented replay-buffer). To check this I suggest using the TF-profiler; to do this, try one of these approaches. There are also some very useful videos on Youtube on how to use the profiler if you have any further problems.

As to possible optimizations to speed up your code I would higly recommend moving to the TF-Agents framework where agents, replay buffers, ecc. are already pre-implemented in an efficient way. It's a bit of a learning curve to get to know it, but it's well worth it for RL

来源：https://stackoverflow.com/questions/65196943/slow-training-on-cpu-and-gpu-in-a-small-network-tensorflow