To install and use the Sentence Transformers library, follow these steps:
Installation
Start by installing the library via pip. Open a terminal and run pip install sentence-transformers. This command automatically installs the core library along with dependencies like PyTorch and Hugging Face Transformers. If you’re using a virtual environment, activate it first. For GPU support, ensure PyTorch is configured for CUDA (e.g., install it separately via pip install torch torchvision --extra-index-url https://download.pytorch.org/whl/cu117 before installing Sentence Transformers). If installation fails, upgrade pip (pip install --upgrade pip) and verify Python version compatibility (requires Python 3.6+).
Importing and Basic Usage
After installation, import the library in your Python script using from sentence_transformers import SentenceTransformer. To generate embeddings, load a pre-trained model like model = SentenceTransformer('all-MiniLM-L6-v2') and encode text with embeddings = model.encode(["Your text here"]). The encode() method accepts single strings or lists, returning NumPy arrays (or PyTorch/TensorFlow tensors if configured). For example, embeddings = model.encode(["Hello world", "How are you?"]) produces a 2D array where each row corresponds to a sentence’s embedding.
Troubleshooting and Considerations
If imports fail, confirm the installation succeeded with pip show sentence-transformers. Common issues include environment conflicts (resolve with virtual environments) or missing dependencies like transformers or scikit-learn. Models download to ~/.cache/torch/sentence_transformers on first use, so ensure internet access. For offline use, pre-download models or specify a local path with model = SentenceTransformer('/path/to/model'). In Jupyter notebooks, restart the kernel after installation. If you encounter ModuleNotFoundError, reinstall the library and check for typos in import statements.