GitHub · Python
Trustworthy-ML-Lab/ThinkEdit
[EMNLP 25] An effective and interpretable weight-editing method for mitigating overly short reasoning in LLMs, and a mechanistic study uncovering how reasoning length is encoded in the model’s representation space.
- Stars
- 21
- Forks
- 2
- Stars gained in 24 hours
- +1
- Forks gained in 24 hours
- +1
Topics: deep-learning, generative-ai, interpretable-machine-learning, large-language-models, mechanistic-interpretability, reasoning-language-models
Latest data point: Nov 21, 2026 · 21 stars · 2 forks