Checkpointer and PolicySaver

View on TensorFlow.org Run in Google Colab View source on GitHub Download notebook

Introduction

tf_agents.utils.common.Checkpointer is a utility to save/load the training state, policy state, and replay_buffer state to/from a local storage.

tf_agents.policies.policy_saver.PolicySaver is a tool to save/load only the policy, and is lighter than Checkpointer. You can use PolicySaver to deploy the model as well without any knowledge of the code that created the policy.

In this tutorial, we will use DQN to train a model, then use Checkpointer and PolicySaver to show how we can store and load the states and model in an interactive way. Note that we will use TF2.0's new saved_model tooling and format for PolicySaver.

Setup

If you haven't installed the following dependencies, run:

sudo apt-get update
sudo apt-get install -y xvfb ffmpeg python-opengl
pip install pyglet
pip install 'imageio==2.4.0'
pip install 'xvfbwrapper==0.2.9'
pip install tf-agents[reverb]
pip install tf-keras
 
import os
# Keep using keras-2 (tf-keras) rather than keras-3 (keras).
os.environ['TF_USE_LEGACY_KERAS'] = '1'

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import base64
import imageio
import io
import matplotlib
import matplotlib.pyplot as plt
import os
import shutil
import tempfile
import tensorflow as tf
import zipfile
import IPython

try:
  from google.colab import files
except ImportError:
  files = None
from tf_agents.agents.dqn import dqn_agent
from tf_agents.drivers import dynamic_step_driver
from tf_agents.environments import suite_gym
from tf_agents.environments import tf_py_environment
from tf_agents.eval import metric_utils
from tf_agents.metrics import tf_metrics
from tf_agents.networks import q_network
from tf_agents.policies import policy_saver
from tf_agents.policies import py_tf_eager_policy
from tf_agents.policies import random_tf_policy
from tf_agents.replay_buffers import tf_uniform_replay_buffer
from tf_agents.trajectories import trajectory
from tf_agents.utils import common

tempdir = os.getenv("TEST_TMPDIR", tempfile.gettempdir())

2023-12-22 12:17:45.769390: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2023-12-22 12:17:45.769434: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2023-12-22 12:17:45.771084: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered

# Set up a virtual display for rendering OpenAI gym environments.
import xvfbwrapper
xvfbwrapper.Xvfb(1400, 900, 24).start()

DQN agent

We are going to set up DQN agent, just like in the previous colab. The details are hidden by default as they are not core part of this colab, but you can click on 'SHOW CODE' to see the details.

Hyperparameters

env_name = "CartPole-v1"

collect_steps_per_iteration = 100
replay_buffer_capacity = 100000

fc_layer_params = (100,)

batch_size = 64
learning_rate = 1e-3
log_interval = 5

num_eval_episodes = 10
eval_interval = 1000

Environment

train_py_env = suite_gym.load(env_name)
eval_py_env = suite_gym.load(env_name)

train_env = tf_py_environment.TFPyEnvironment(train_py_env)
eval_env = tf_py_environment.TFPyEnvironment(eval_py_env)

Agent

q_net = q_network.QNetwork(
    train_env.observation_spec(),
    train_env.action_spec(),
    fc_layer_params=fc_layer_params)

optimizer = tf.compat.v1.train.AdamOptimizer(learning_rate=learning_rate)

global_step = tf.compat.v1.train.get_or_create_global_step()

agent = dqn_agent.DqnAgent(
    train_env.time_step_spec(),
    train_env.action_spec(),
    q_network=q_net,
    optimizer=optimizer,
    td_errors_loss_fn=common.element_wise_squared_loss,
    train_step_counter=global_step)
agent.initialize()