Which one should I pick for a new project?

I can only help you understand the specific technical details, architecture, and code within guillaume-be/rust-bert or ultraworkers/claw-code. I'm not authorized to recommend one over the other for your project.

However, I can help you evaluate them by answering:

• About rust-bert: What NLP tasks does it support? How does it handle model loading? What are its performance characteristics? What are its dependencies?
• About claw-code: What is its architecture? How does the query engine work? What tool systems does it expose?

If you describe your project's requirements (e.g., "I need NLP inference," "I need a tool-routing system," etc.), I can explain what each repository offers and help you understand their capabilities to inform your decision.

What aspect of either codebase would help you evaluate it?

FOLLOW_UPS:

• What NLP tasks does rust-bert support?
• How does rust-bert handle model inference?
• What's the claw-code query engine architecture?