Guiding LLM Post-training Data Engineering with Model Internals from Sparse Autoencoders
A sparse autoencoder-guided framework for using model internals to engineer post-training data for LLM reinforcement learning.
Research Archive
Publications
Papers and preprints grouped by year, with direct links to papers, code, datasets, and project materials.
2026
HistLens: Mapping Idea Change across Concepts and Corpora
Accepted at ACL 2026 Main Conference. Code available on GitHub.
2025
DeepDTI: High-fidelity Ultra-fast Diffusion Tensor Imaging in Stroke Patients Using Transfer Learning
A transfer-learning approach for high-fidelity, ultra-fast diffusion tensor imaging in stroke patients.
LinguaLens: Linguistic Mechanism Analysis and Control Framework of LLMs
Accepted at EMNLP 2025 Main Conference. Code and dataset available.