pyvene-interventions

pyvene is Stanford NLP's library for performing causal interventions on PyTorch models. It provides a declarative, dict-based framework for activation patching, causal tracing, and interchange intervention training - making intervention experiments reproducible and shareable. GitHub : stanfordnlp/pyvene (840+ stars) Paper : pyvene: A Library for Understanding and Improving PyTorch Models via Interventions (NAACL 2024) Use pyvene when you need to: Consider alternatives when: Standard import: The main class that wraps any PyTorch model with intervention capabilities: Locate where factual associations are stored by corrupting inputs and restoring activations. Test which components are necessary for a specific behavior. Train interventions to discover causal structure. Steer model behavior during generation. pyvene works with any PyTorch model. Tested on: For detailed API documentation, tutorials, and advanced usage, see the references/ folder: - Perform causal tracing (ROME-style localization) - Run activation patching experiments - Conduct interchange intervention training (IIT) - Test causal hypotheses about model components - Share/reproduce intervention experiments via HuggingFace - Work with any PyTorch architecture (not just transformers) - You need exploratory activation analysis → Use TransformerLens - You want to train/analyze SAEs → Use SAELens - You need remote...