Grass: Compute Efficient Low-Memory LLM Training with Structured Sparse Gradients

Documenti analoghi

• Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
  di: Muhamed, Aashiq, et al.
  Pubblicazione: (2024)
• CoRAG: Collaborative Retrieval-Augmented Generation
  di: Muhamed, Aashiq, et al.
  Pubblicazione: (2025)
• SAEs $\textit{Can}$ Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs
  di: Muhamed, Aashiq, et al.
  Pubblicazione: (2025)
• DSPA: Dynamic SAE Steering for Data-Efficient Preference Alignment
  di: Wedgwood, James, et al.
  Pubblicazione: (2026)
• Pando: Do Interpretability Methods Work When Models Won't Explain Themselves?
  di: Zhong, Ziqian, et al.
  Pubblicazione: (2026)