metadata
language: en
license: apache-2.0
library_name: pytorch
tags:
- deep-reinforcement-learning
- reinforcement-learning
- DI-engine
- Walker2d-v3
benchmark_name: OpenAI/Gym/MuJoCo
task_name: Walker2d-v3
pipeline_tag: reinforcement-learning
model-index:
- name: TD3
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: OpenAI/Gym/MuJoCo-Walker2d-v3
type: OpenAI/Gym/MuJoCo-Walker2d-v3
metrics:
- type: mean_reward
value: 4331.88 +/- 12.08
name: mean_reward
Play Walker2d-v3 with TD3 Policy
Model Description
This is a simple TD3 implementation to OpenAI/Gym/MuJoCo Walker2d-v3 using the DI-engine library and the DI-zoo.
DI-engine is a python library for solving general decision intelligence problems, which is based on implementations of reinforcement learning framework using PyTorch or JAX. This library aims to standardize the reinforcement learning framework across different algorithms, benchmarks, environments, and to support both academic researches and prototype applications. Besides, self-customized training pipelines and applications are supported by reusing different abstraction levels of DI-engine reinforcement learning framework.
Model Usage
Install the Dependencies
(Click for Details)
# install huggingface_ding
git clone https://github.com/opendilab/huggingface_ding.git
pip3 install -e ./huggingface_ding/
# install environment dependencies if needed
sudo apt update -y && sudo apt install -y build-essential libgl1-mesa-dev libgl1-mesa-glx libglew-dev libosmesa6-dev libglfw3 libglfw3-dev libsdl2-dev libsdl2-image-dev libglm-dev libfreetype6-dev patchelf
mkdir -p ~/.mujoco
wget https://mujoco.org/download/mujoco210-linux-x86_64.tar.gz -O mujoco.tar.gz
tar -xf mujoco.tar.gz -C ~/.mujoco
echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.mujoco/mjpro210/bin:~/.mujoco/mujoco210/bin" >> ~/.bashrc
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.mujoco/mjpro210/bin:~/.mujoco/mujoco210/bin
pip3 install "cython<3"
pip3 install DI-engine[common_env]
Git Clone from Huggingface and Run the Model
(Click for Details)
# running with trained model
python3 -u run.py
run.py
from ding.bonus import SACAgent
from ding.config import Config
from easydict import EasyDict
import torch
# Pull model from files which are git cloned from huggingface
policy_state_dict = torch.load("pytorch_model.bin", map_location=torch.device("cpu"))
cfg = EasyDict(Config.file_to_dict("policy_config.py").cfg_dict)
# Instantiate the agent
agent = SACAgent(env_id="Walker2d-v3", exp_name="Walker2d-v3-TD3", cfg=cfg.exp_config, policy_state_dict=policy_state_dict)
# Continue training
agent.train(step=5000)
# Render the new agent performance
agent.deploy(enable_save_replay=True)
Run Model by Using Huggingface_ding
(Click for Details)
# running with trained model
python3 -u run.py
run.py
from ding.bonus import TD3Agent
from huggingface_ding import pull_model_from_hub
# Pull model from Hugggingface hub
policy_state_dict, cfg = pull_model_from_hub(repo_id="OpenDILabCommunity/Walker2d-v3-TD3")
# Instantiate the agent
agent = TD3Agent(env_id="Walker2d-v3", exp_name="Walker2d-v3-TD3", cfg=cfg.exp_config, policy_state_dict=policy_state_dict)
# Continue training
agent.train(step=5000)
# Render the new agent performance
agent.deploy(enable_save_replay=True)
Model Training
Train the Model and Push to Huggingface_hub
(Click for Details)
#Training Your Own Agent
python3 -u train.py
train.py
from ding.bonus import TD3Agent
from huggingface_ding import push_model_to_hub
# Instantiate the agent
agent = TD3Agent(env_id="Walker2d-v3", exp_name="Walker2d-v3-TD3")
# Train the agent
return_ = agent.train(step=int(5000000))
# Push model to huggingface hub
push_model_to_hub(
agent=agent.best,
env_name="OpenAI/Gym/MuJoCo",
task_name="Walker2d-v3",
algo_name="TD3",
wandb_url=return_.wandb_url,
github_repo_url="https://github.com/opendilab/DI-engine",
github_doc_model_url="https://di-engine-docs.readthedocs.io/en/latest/12_policies/td3.html",
github_doc_env_url="https://di-engine-docs.readthedocs.io/en/latest/13_envs/mujoco.html",
installation_guide='''
sudo apt update -y \
&& sudo apt install -y \
build-essential \
libgl1-mesa-dev \
libgl1-mesa-glx \
libglew-dev \
libosmesa6-dev \
libglfw3 \
libglfw3-dev \
libsdl2-dev \
libsdl2-image-dev \
libglm-dev \
libfreetype6-dev \
patchelf
mkdir -p ~/.mujoco
wget https://mujoco.org/download/mujoco210-linux-x86_64.tar.gz -O mujoco.tar.gz
tar -xf mujoco.tar.gz -C ~/.mujoco
echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.mujoco/mjpro210/bin:~/.mujoco/mujoco210/bin" >> ~/.bashrc
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.mujoco/mjpro210/bin:~/.mujoco/mujoco210/bin
pip3 install "cython<3"
pip3 install DI-engine[common_env]
''',
usage_file_by_git_clone="./td3/walker2d_td3_deploy.py",
usage_file_by_huggingface_ding="./td3/walker2d_td3_download.py",
train_file="./td3/walker2d_td3.py",
repo_id="OpenDILabCommunity/Walker2d-v3-TD3",
create_repo=False
)
Configuration
(Click for Details)
exp_config = {
'env': {
'manager': {
'episode_num': float("inf"),
'max_retry': 1,
'retry_type': 'reset',
'auto_reset': True,
'step_timeout': None,
'reset_timeout': None,
'retry_waiting_time': 0.1,
'cfg_type': 'BaseEnvManagerDict'
},
'stop_value': 6000,
'n_evaluator_episode': 8,
'env_id': 'Walker2d-v3',
'norm_obs': {
'use_norm': False
},
'norm_reward': {
'use_norm': False
},
'collector_env_num': 1,
'evaluator_env_num': 8,
'env_wrapper': 'mujoco_default'
},
'policy': {
'model': {
'twin_critic': True,
'obs_shape': 17,
'action_shape': 6,
'actor_head_hidden_size': 256,
'critic_head_hidden_size': 256,
'action_space': 'regression'
},
'learn': {
'learner': {
'train_iterations': 1000000000,
'dataloader': {
'num_workers': 0
},
'log_policy': True,
'hook': {
'load_ckpt_before_run': '',
'log_show_after_iter': 100,
'save_ckpt_after_iter': 10000,
'save_ckpt_after_run': True
},
'cfg_type': 'BaseLearnerDict'
},
'update_per_collect': 1,
'batch_size': 256,
'learning_rate_actor': 0.001,
'learning_rate_critic': 0.001,
'ignore_done': False,
'target_theta': 0.005,
'discount_factor': 0.99,
'actor_update_freq': 2,
'noise': True,
'noise_sigma': 0.2,
'noise_range': {
'min': -0.5,
'max': 0.5
}
},
'collect': {
'collector': {},
'unroll_len': 1,
'noise_sigma': 0.1,
'n_sample': 1
},
'eval': {
'evaluator': {
'eval_freq': 5000,
'render': {
'render_freq': -1,
'mode': 'train_iter'
},
'figure_path': None,
'cfg_type': 'InteractionSerialEvaluatorDict',
'stop_value': 6000,
'n_episode': 8
}
},
'other': {
'replay_buffer': {
'replay_buffer_size': 1000000
}
},
'on_policy': False,
'cuda': True,
'multi_gpu': False,
'bp_update_sync': True,
'traj_len_inf': False,
'type': 'td3',
'priority': False,
'priority_IS_weight': False,
'random_collect_size': 25000,
'transition_with_policy_data': False,
'action_space': 'continuous',
'reward_batch_norm': False,
'multi_agent': False,
'cfg_type': 'TD3PolicyDict'
},
'exp_name': 'Walker2d-v3-TD3',
'seed': 0,
'wandb_logger': {
'gradient_logger': True,
'video_logger': True,
'plot_logger': True,
'action_logger': True,
'return_logger': False
}
}
Training Procedure
- Weights & Biases (wandb): monitor link
Model Information
- Github Repository: repo link
- Doc: DI-engine-docs Algorithm link
- Configuration: config link
- Demo: video
- Parameters total size: 1690.06 KB
- Last Update Date: 2023-09-22
Environments
- Benchmark: OpenAI/Gym/MuJoCo
- Task: Walker2d-v3
- Gym version: 0.25.1
- DI-engine version: v0.4.9
- PyTorch version: 2.0.1+cu117
- Doc: DI-engine-docs Environments link