FasterDecoding/Medusa

Inference Engines

Medusa: Simple Framework for Accelerating LLM Generation with Multiple Decoding Heads

GitHub →Website →

5.4

GitHub Metrics

Stars

2.7k

Forks

197

Open Issues

Watchers

Contributors

Weekly Commits

Language

Jupyter Notebook

License

Apache-2.0

Last Commit

Jun 25, 2024

Created

Sep 10, 2023

Latest Release

v0.1

Release Date

Sep 11, 2023

Synced: Apr 16, 2026

Quality Scores

Documentation Qualityw: 20%

6.6

Has docs site (https://sites.google.com/view/medusa-llm). Description: 85 chars. Stars signal: 2,723. Contributors: 11. Score: 6.6/10

Community Healthw: 20%

5.7

Stars: 2,723. Contributors: 11. Watchers: 29. Forks: 197. Issue ratio: 2.1%. Score: 5.7/10

Maintenance Velocityw: 15%

0.7

Last commit: 660d ago. Weekly commits: 0. Latest release: v0.1. Maturity bonus: 2.6y old. Score: 0.7/10

API Design & DXw: 20%

7.2

Stars/issues ratio: 48. Has documentation site. Permissive license: Apache-2.0. Popularity signal: 2,723 stars. Score: 7.2/10

Production Readinessw: 15%

5.0

Battle-tested: 2,723 stars. Peer review: 11 contributors. Versioned: v0.1. Licensed: Apache-2.0. Age: 2.6 years. Maintenance: last commit 660d ago. Score: 5/10

Ecosystem Integrationw: 10%

6.9

Fork interest: 197. Ecosystem: Jupyter Notebook. Integration-friendly: Apache-2.0. Adoption: 2,723 stars. Has web presence. Score: 6.9/10