What training objectives does this skill support?

This skill provides implementation patterns for Trajectory Balance (TB) and Detailed Balance (DB), allowing for stable and efficient training of flow-based models.

How does GFlowNet differ from Reinforcement Learning?

While standard RL focuses on maximizing expected reward to find one optimal solution, GFlowNets focus on sampling the entire distribution of rewards to ensure diversity and explore multiple modes.

Can this skill be used for scientific research?

Yes, it is specifically designed for complex scientific domains like molecule design, drug discovery, and causal discovery where multi-modal exploration is essential.

A GFlowNet (Generative Flow Network) is a generative model trained to sample objects with a probability proportional to a given reward function, ensuring a diverse set of high-quality solutions.

GFlowNet ML Framework

Name: GFlowNet ML Framework
Author: plurigrid

byplurigrid

•

数据科学与机器学习

Implements Generative Flow Networks to sample diverse candidates proportionally to their reward functions.

关于

The GFlowNet skill empowers Claude to design and implement Generative Flow Networks (GFlowNets), a machine learning paradigm developed by Yoshua Bengio that bridges the gap between Reinforcement Learning and MCMC sampling. Unlike traditional RL that converges on a single optimal solution, GFlowNets learn to sample diverse high-reward candidates—such as molecular structures or causal graphs—with probabilities proportional to their reward. This skill is ideal for researchers and engineers working on drug discovery, causal inference, and combinatorial optimization where finding a variety of high-quality solutions is more critical than finding a single global maximum.

主要功能

2 GitHub stars
Implementation of Trajectory Balance (TB) and Detailed Balance (DB) training objectives
Tools for Bayesian structure learning and causal discovery of DAGs
Integration with scientific computing libraries for topological and chemical analysis
Boilerplates for combinatorial optimization and set generation
Support for diverse candidate generation in molecule design and drug discovery

使用场景

Solving combinatorial optimization problems where exploration of multi-modal reward landscapes is required
Performing causal discovery to infer complex relationships between variables from observational data
Discovering diverse drug candidates with high binding affinity and synthesizability

关于

主要功能

2 GitHub stars
Implementation of Trajectory Balance (TB) and Detailed Balance (DB) training objectives
Tools for Bayesian structure learning and causal discovery of DAGs
Integration with scientific computing libraries for topological and chemical analysis
Boilerplates for combinatorial optimization and set generation
Support for diverse candidate generation in molecule design and drug discovery

使用场景

Solving combinatorial optimization problems where exploration of multi-modal reward landscapes is required
Performing causal discovery to infer complex relationships between variables from observational data
Discovering diverse drug candidates with high binding affinity and synthesizability