Diamond

Visit Tool

DIAMOND is a reinforcement learning agent trained in a diffusion world model. It enables agents to play in autoregressive imagination of Atari and CSGO world models, with NeurIPS 2024 Spotlight recognition.

Claim this tool

1View

At a glance

Pricing

Open Source

Free tier

Yes

API

Skill level

Technical

About

What is diamond?

DIAMOND (DIffusion As a Model Of eNvironment Dreams) is a cutting-edge reinforcement learning agent that operates entirely within a diffusion world model. This innovative approach allows the agent to learn and play in an autoregressive imagination of environments like Atari and Counter-Strike: Global Offensive. Recognized as a NeurIPS 2024 Spotlight, DIAMOND offers researchers and developers a powerful tool for exploring diffusion-based world models in reinforcement learning. It provides quick installation for pretrained models, options for launching training runs, and extensive configuration management via Hydra. Users can visualize policy play, human interaction, and dataset replays, making it a versatile platform for advanced AI research.

Best used for

Ideal for developers and data scientists who need to train reinforcement learning agents within diffusion world models, simulate complex environments like Atari and CSGO, and visualize agent behavior. Especially valuable for academic research and advanced AI development in the field of world modeling.

Common actions

train AI agents

research diffusion models

simulate environments

visualize agent behavior

workflowsautomated workflowopen-sourcedeepfakelow-code/no-code"AI Agents"collaborationgithub copilotface swapping

Capabilities

Key features

Diffusion world model
Reinforcement learning agent
Atari environment support
CSGO environment support
Policy/human play visualization
Configurable training runs

Target Audience

developerdata scientist

Integrations

Not yet documented

Pricing & Plans

Open Source

Free

FAQs

What environments does DIAMOND support for training and play?

DIAMOND primarily supports Atari environments, with specific examples like BreakoutNoFrameskip-v4. It also has a dedicated branch for Counter-Strike: Global Offensive (CSGO), allowing users to train and interact with world models in both classic and modern gaming contexts.

How can I try out the pretrained world models with DIAMOND?

You can quickly install DIAMOND via miniconda, then run `python src/play.py --pretrained` for Atari or `git checkout csgo && python src/play.py` for CSGO. This will download and cache pretrained models from Hugging Face Hub, allowing you to immediately interact with the agent.

Can I adjust the sampling parameters of the diffusion world model?

Yes, you can adjust sampling parameters such as the number of denoising steps, stochasticity, and order. These settings are configured in the `world_model_env.diffusion_sampler` section of the `config/trainer.yaml` file, allowing for fine-tuning of sampling speed and quality.

Trending

Subcategories trending in AI Agents & Automation

AI Frameworks & Infra Chatbots & Conversational AI Workflow Agents Personal Assistants RAG & Document AI Voice Agents

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce