Trustworthy-ML-Lab / ThinkEditLinks
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.
☆12Updated 2 weeks ago
Alternatives and similar repositories for ThinkEdit
Users that are interested in ThinkEdit are comparing it to the libraries listed below
Sorting:
- Official code for SEAL: Steerable Reasoning Calibration of Large Language Models for Free☆28Updated 2 months ago
- A Sober Look at Language Model Reasoning☆74Updated last week
- [ICML'25] Our study systematically investigates massive values in LLMs' attention mechanisms. First, we observe massive values are concen…☆73Updated this week
- [ICLR 2025 Workshop] "Landscape of Thoughts: Visualizing the Reasoning Process of Large Language Models"☆25Updated last week
- ☆10Updated last year
- Model merging is a highly efficient approach for long-to-short reasoning.☆65Updated 3 weeks ago
- Official repository of "Localizing Task Information for Improved Model Merging and Compression" [ICML 2024]☆45Updated 8 months ago
- Localize-and-Stitch: Efficient Model Merging via Sparse Task Arithmetic☆26Updated 5 months ago
- ☆44Updated 3 months ago
- This is the official code for the paper "Vaccine: Perturbation-aware Alignment for Large Language Models" (NeurIPS2024)☆44Updated 7 months ago
- PoC for "SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning" [arXiv '25]☆39Updated last month
- ☆22Updated 3 months ago
- [NeurIPS 2024] "Can Language Models Perform Robust Reasoning in Chain-of-thought Prompting with Noisy Rationales?"☆34Updated 5 months ago
- Codebase for decoding compressed trust.☆24Updated last year
- ☆65Updated 2 months ago
- [ICLR 2025] Code and Data Repo for Paper "Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluation"☆64Updated 6 months ago
- ☆49Updated last year
- [NeurIPS 2024 Spotlight] EMR-Merging: Tuning-Free High-Performance Model Merging☆59Updated 3 months ago
- [ACL 25] SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities☆17Updated 2 months ago
- Code for paper "Parameter Efficient Multi-task Model Fusion with Partial Linearization"☆21Updated 9 months ago
- Code for paper: Aligning Large Language Models with Representation Editing: A Control Perspective☆32Updated 4 months ago
- Official implementation for "ALI-Agent: Assessing LLMs'Alignment with Human Values via Agent-based Evaluation"☆18Updated last month
- Chain of Thoughts (CoT) is so hot! so long! We need short reasoning process!☆54Updated 2 months ago
- This is the official code for the paper "Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable".☆18Updated 3 months ago
- ☆13Updated last year
- Representation Surgery for Multi-Task Model Merging. ICML, 2024.☆45Updated 8 months ago
- 📜 Paper list on decoding methods for LLMs and LVLMs☆49Updated last month
- ☆10Updated 2 months ago
- Official implementation for "Mixture of In-Context Experts Enhance LLMs’ Awareness of Long Contexts" (Accepted by Neurips2024)☆12Updated 5 months ago
- Code for Fine-grained Uncertainty Quantification for LLMs from Semantic Similarities (NeurIPS'24)☆24Updated 6 months ago