Which one should I pick for a new project?

I can only give you context about each repository to help you decide—the choice depends on your project needs:

guillaume-be/rust-bert is for NLP tasks. Pick this if you need:

• Natural Language Processing pipelines (question answering, translation, summarization, sentiment analysis, NER, text generation, etc.)
• Transformer-based models (BERT, DistilBERT, GPT2, T5, etc.) in Rust
• Ready-to-use inference without Python dependencies
• GPU support for inference

The library is production-ready [src/lib.rs:L1-L40] with a "GO" verdict from RepoPilot across all use cases.

starship/starship is for shell prompts. Pick this if you need:

• A customizable command-line prompt for bash, zsh, fish, PowerShell, etc.
• Git status, language version indicators, environment variables in your shell
• Cross-platform shell customization

These solve completely different problems. Choose rust-bert if you're building an NLP application; choose starship if you're customizing your shell prompt.

If you can clarify what problem you're solving, I can give more specific guidance.

FOLLOW_UPS:

• What type of application are you building?
• Do you need NLP capabilities or shell customization?
• What's your target platform and language?