GitHub / Trustworthy-ML-Lab / ThinkEdit
An effective 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.
JSON API: http://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Trustworthy-ML-Lab%2FThinkEdit
PURL: pkg:github/Trustworthy-ML-Lab/ThinkEdit
Stars: 10
Forks: 1
Open issues: 0
License: None
Language: Python
Size: 545 KB
Dependencies parsed at: Pending
Created at: 4 months ago
Updated at: 3 months ago
Pushed at: 3 months ago
Last synced at: 3 months ago
Topics: deep-learning, generative-ai, interpretable-machine-learning, large-language-models, mechanistic-interpretability, reasoning-language-models