Library/weak-to-strongForked

openai/weak-to-strong

weak-to-strong

This repository implements weak-to-strong generalization, a research framework where weaker AI models are used to supervise and train stronger models.

View on GitHub↗Upstream openai/weak-to-strong↗

Builder

OpenAI

openai • ai-lab

Stars

2,552

Using upstream star count

Forks

310

Using upstream fork count

Open Issues

Activity Score

0/100

0 commits in 30d

Created

Dec 13, 2023

Project creation date

README Summary

This repository implements weak-to-strong generalization, a research framework where weaker AI models are used to supervise and train stronger models. The work explores whether strong models can learn to perform better than their weak supervisors, which has implications for AI alignment and safety as models become more capable than humans.

AI Dev Skills

Unmapped

AI Safety ResearchModel AlignmentWeak-to-Strong GeneralizationSupervised LearningNeural Network TrainingAI GovernanceMachine Learning Research Methodology

Taxonomy

AI Trends

AI Safety AI Alignment Scalable Oversight Weak-to-Strong Generalization

Deployment Context

Self-hosted Research Environment

Modalities

Text

Skill Areas

AI Safety Research Model Alignment Weak-to-Strong Generalization Supervised Learning Neural Network Training AI Governance Machine Learning Research Methodology

Use Cases

AI Model Alignment Research Weak Supervision Experiments AI Safety Evaluation Model Generalization Studies

Recent Activity

Updated 1 years ago

7 Days

30 Days

90 Days

Quality

research

Quality: medium
Maturity: research

PM Skills

Product Discovery

Languages

Python100.0%

Timeline

Project created: Dec 13, 2023
Forked: Mar 14, 2026
Your last push: 1 years ago
Upstream last push: 1 years ago
Tracked since: May 19, 2024

Similar Repos

pgvector cosine similarity · $0

Loading…