Stream

Scaling up Mechanistic Interpretability for Long Context LLMs via Sparse Tracing.

Research Paper

If you use this library in academic research, consider citing it:

@article{2025stream,
  title     = {Stream: Scaling up Mechanistic Interpretability for Long Context LLMs via Sparse Tracing},
  author    = {},
  journal   = {arXiv preprint},
  year      = {2025},
}