A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity.
☆90Mar 7, 2025Updated last year
Alternatives and similar repositories for dpo_toxic
Users that are interested in dpo_toxic are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- ☆13Mar 29, 2024Updated 2 years ago
- ☆25Jun 13, 2024Updated 2 years ago
- ☆18Mar 30, 2025Updated last year
- Algebraic value editing in pretrained language models☆71Nov 1, 2023Updated 2 years ago
- [ICML 2024] Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications☆91Mar 30, 2025Updated last year
- Proton VPN Special Offer - Get 70% off • AdSpecial partner offer. Trusted by over 100 million users worldwide. Tested, Approved and Recommended by Experts.
- LLM experiments done during SERI MATS - focusing on activation steering / interpreting activation spaces☆104Sep 21, 2023Updated 2 years ago
- Evaluate interpretability methods on localizing and disentangling concepts in LLMs.☆58Oct 30, 2025Updated 8 months ago
- ☆30Aug 2, 2024Updated last year
- Co-Supervised Learning: Improving Weak-to-Strong Generalization with Hierarchical Mixture of Experts☆15Feb 26, 2024Updated 2 years ago
- Inspecting and Editing Knowledge Representations in Language Models☆120Jul 24, 2023Updated 2 years ago
- Improving Alignment and Robustness with Circuit Breakers☆266Sep 24, 2024Updated last year
- Steering Llama 2 with Contrastive Activation Addition☆240May 23, 2024Updated 2 years ago
- ☆26Dec 20, 2023Updated 2 years ago
- LLM Unlearning☆185Oct 20, 2023Updated 2 years ago
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- [EMNLP 25] An effective and interpretable weight-editing method for mitigating overly short reasoning in LLMs, and a mechanistic study un…☆19Dec 17, 2025Updated 7 months ago
- Official Code for What Makes and Breaks Safety Fine-tuning? A Mechanistic Study (NeurIPS 2024)☆12Oct 31, 2024Updated last year
- Representation Engineering: A Top-Down Approach to AI Transparency☆1,012Aug 14, 2024Updated last year
- ☆57Oct 23, 2023Updated 2 years ago
- Redwood Research's transformer interpretability tools☆15Apr 15, 2022Updated 4 years ago
- Code for paper "Universal Jailbreak Backdoors from Poisoned Human Feedback"☆67Apr 24, 2024Updated 2 years ago
- ☆30Apr 6, 2026Updated 3 months ago
- This is the oficial repository for "Safer-Instruct: Aligning Language Models with Automated Preference Data"☆17Feb 22, 2024Updated 2 years ago
- Code and data for editing model beliefs with SDF and other methods, and for evaluating the depth of the implanted beliefs.☆15Oct 23, 2025Updated 8 months ago
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- ☆48Sep 29, 2024Updated last year
- This is a repository for "PMET: Precise Model Editing in a Transformer"☆58Sep 28, 2023Updated 2 years ago
- Steering vectors for transformer language models in Pytorch / Huggingface☆157Feb 21, 2025Updated last year
- A library for mechanistic anomaly detection☆22Jan 9, 2025Updated last year
- [ICML 2024] COLD-Attack: Jailbreaking LLMs with Stealthiness and Controllability☆176Dec 18, 2024Updated last year
- Sparse Autoencoder Training Library☆58May 1, 2025Updated last year
- Inference-Time Intervention: Eliciting Truthful Answers from a Language Model☆581Jan 28, 2025Updated last year
- Implementation of the Decrypto benchmark for multi-agent reasoning and theory of mind.☆22Jan 19, 2026Updated 6 months ago
- ☆47Oct 1, 2024Updated last year
- Virtual machines for every use case on DigitalOcean • AdGet dependable uptime with 99.99% SLA, simple security tools, and predictable monthly pricing with DigitalOcean's virtual machines, called Droplets.
- Data Valuation on In-Context Examples (ACL23)☆24Jan 12, 2025Updated last year
- ☆100Jun 27, 2024Updated 2 years ago
- ☆256Feb 22, 2024Updated 2 years ago
- Erasing concepts from neural representations with provable guarantees☆258Jan 27, 2025Updated last year
- The official implementation of "Mind the Gap: Offline Policy Optimization for Imperfect Rewards" (ICLR2023)☆16Mar 3, 2023Updated 3 years ago
- ☆13Jul 8, 2023Updated 3 years ago
- Code for "Preference Tuning For Toxicity Mitigation Generalizes Across Languages." Paper accepted at Findings of EMNLP 2024☆18Mar 25, 2025Updated last year