We introduce Voyager, the first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention. Voyager consists of three key components: 1) an automatic curriculum that maximizes exploration, 2) an ever-growing skill library of executable code for storing and retrieving complex behaviors, and 3) a new iterative prompting mechanism that incorporates environment feedback, execution errors, and self-verification for program improvement. Voyager interacts with GPT-4 via blackbox queries, which bypasses the need for model parameter fine-tuning. The skills developed by Voyager are temporally extended, interpretable, and compositional, which compounds the agent’s abilities rapidly and alleviates catastrophic forgetting. Empirically, Voyager shows strong in-context lifelong learning capability and exhibits exceptional proficiency in playing Minecraft. It obtains 3.3× more unique items, travels 2.3× longer distances, and unlocks key tech tree milestones up to 15.3× faster than prior SOTA. Voyager is able to utilize the learned skill library in a new Minecraft world to solve novel tasks from scratch, while other techniques struggle to generalize.
In this repo, we provide Voyager code. This codebase is under MIT License.
Voyager requires Python 3.10 and Node.js ≥ 16.13.0.
If you don't have them already, please download and install the following:
Modrinth is recommended for setting up your Minecraft client. You can also use any means you're comfortable with to install the necessary mods.
In Modrinth, create a new profile and call it "Voyager" (or some other name meaningful to you). Use Minecraft version 1.19.4. Install the following content:
- CompleteConfig 2.3.1
- Fabric API 0.87.2+1.19.4
- Mod Menu 6.3.1
- Multiplayer Server Pause 1.3.0
- iChunUtil 1.0.0
Modrinth will automatically select compatible versions: you don't have to specify them manually.
In the Command Prompt, navigate to the Voyager repo and type the following:
voyager install
You will see the following on the screen:
>voyager install
Starting installation...
Checking Python version...
Found Python version 3.10.11
Enter your OpenAI API key:
Copy and paste your OpenAI API Key and hit enter. It will be invisible when you paste it.
Next, you will see the following:
Available OpenAI models:
1. gpt-4
2. gpt-4o
3. gpt-4o-mini
Select a model by entering the corresponding number:
Select the OpenAI model you want to use. GPT-4 is the most capable, but also very expensive. Consider starting with GPT-4o for the most balanced experience. After this, the installation should complete after a few minutes.
Should you ever want to change these options, you can edit the .env file.
You should see the following:
✅ .env file created successfully!
Your configuration has been saved in '.env'.
Checking Node.js version...
Found Node.js version 22.12.0
Removing existing virtual environment...
Creating new virtual environment...
Activating virtual environment...
Installing Python dependencies...
Setting up Mineflayer environment...
Installing Node.js dependencies...
Setting up mineflayer-collectblock...
Installation completed successfully!
Run 'voyager.bat' without parameters to start the application.
After the installation process, you will need to start Minecraft.
- Start a single player game with Game Mode set to
Creativeand Difficulty set toPeaceful. - After the world is created, press the
Esckey and pressOpen to LAN. - Select
Allow cheats: ONand pressStart LAN World.
Once Minecraft is ready, you can run Voyager by executing the following on the Windows Command Prompt:
voyager
Note: The first time you run, answer "no" when prompted. On subsequent runs answer "yes".
>voyager
Starting Voyager...
_ _
| | | |
| | | | ___ _ _ __ _ __ _ ___ _ __
| | | |/ _ \| | | |/ _` |/ _` |/ _ \ '__|
\ \_/ / (_) | |_| | (_| | (_| | __/ |
\___/ \___/ \__, |\__,_|\__, |\___|_|
__/ | __/ |
|___/ |___/
An Open-Ended Embodied Agent with Large Language Models
original authors:
Guanzhi Wang and Yuqi Xie and Yunfan Jiang and
Ajay Mandlekar and Chaowei Xiao and Yuke Zhu
and Linxi Fan and Anima Anandkumar
Do you want to continue from your previous session? (yes/no):
Enjoy!
If you stop the learning process and want to resume from a checkpoint later, you can update __main__.py (as shown below) and restart Voyager:
voyager = Voyager(
mc_port=mc_port,
openai_api_key=openai_api_key,
ckpt_dir="YOUR_CKPT_DIR",
resume=True
)If you want to run Voyager for a specific task with a learned skill library, you should first pass the skill library directory to Voyager by making the following changes to __main__.py and restarting:
# First instantiate Voyager with skill_library_dir.
voyager = Voyager(
mc_port=mc_port,
openai_api_key=openai_api_key,
skill_library_dir="./skill_library/trial1", # Load a learned skill library.
ckpt_dir="YOUR_CKPT_DIR", # Feel free to use a new dir. Do not use the same dir as skill library because new events will still be recorded to ckpt_dir.
resume=False, # Do not resume from a skill library because this is not learning.
)Then, you can run task decomposition. Notice: Occasionally, the task decomposition may not be logical. If you notice the printed sub-goals are flawed, you can rerun the decomposition.
## Run task decomposition
task = "YOUR TASK" # e.g. "Craft a diamond pickaxe"
sub_goals = voyager.decompose_task(task=task)Finally, you can run the sub-goals with the learned skill library:
voyager.inference(sub_goals=sub_goals)For all valid skill libraries, see Learned Skill Libraries.
If you have any questions, please check our FAQ first before opening an issue.
If you find our work useful, please consider citing us!
@article{wang2023voyager,
title = {Voyager: An Open-Ended Embodied Agent with Large Language Models},
author = {Guanzhi Wang and Yuqi Xie and Yunfan Jiang and Ajay Mandlekar and Chaowei Xiao and Yuke Zhu and Linxi Fan and Anima Anandkumar},
year = {2023},
journal = {arXiv preprint arXiv: Arxiv-2305.16291}
}Disclaimer: This project is strictly for research purposes, and not an official product from NVIDIA.