KTO: Model Alignment as Prospect Theoretic Optimization
Paper • 2402.01306 • Published • 22
alignment_24_best
-
-
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Paper • 2305.18290 • Published • 64 -
SimPO: Simple Preference Optimization with a Reference-Free Reward
Paper • 2405.14734 • Published • 12 -
Anchored Preference Optimization and Contrastive Revisions: Addressing Underspecification in Alignment
Paper • 2408.06266 • Published • 10 -
Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs
Paper • 2402.14740 • Published • 18 -
Binary Classifier Optimization for Large Language Model Alignment
Paper • 2404.04656 • Published • 2 -
Noise Contrastive Alignment of Language Models with Explicit Rewards
Paper • 2402.05369 • Published • 2 -
Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation
Paper • 2401.08417 • Published • 37 -
Direct Language Model Alignment from Online AI Feedback
Paper • 2402.04792 • Published • 35 -
Nash Learning from Human Feedback
Paper • 2312.00886 • Published • 18 -
ORPO: Monolithic Preference Optimization without Reference Model
Paper • 2403.07691 • Published • 72 -
Exploratory Preference Optimization: Harnessing Implicit Q*-Approximation for Sample-Efficient RLHF
Paper • 2405.21046 • Published • 4 -
From r to Q^*: Your Language Model is Secretly a Q-Function
Paper • 2404.12358 • Published • 2 -
Offline Regularised Reinforcement Learning for Large Language Models Alignment
Paper • 2405.19107 • Published • 15 -
Towards Scalable Automated Alignment of LLMs: A Survey
Paper • 2406.01252 • Published • 3 -
Towards a Unified View of Preference Learning for Large Language Models: A Survey
Paper • 2409.02795 • Published • 72 -
Preference Fine-Tuning of LLMs Should Leverage Suboptimal, On-Policy Data
Paper • 2404.14367 • Published • 1 -
Self-Supervised Alignment with Mutual Information: Learning to Follow Principles without Preference Labels
Paper • 2404.14313 • Published -
Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback
Paper • 2406.09279 • Published • 3 -
Is DPO Superior to PPO for LLM Alignment? A Comprehensive Study
Paper • 2404.10719 • Published • 6 -
Advancing LLM Reasoning Generalists with Preference Trees
Paper • 2404.02078 • Published • 46 -
Building Math Agents with Multi-Turn Iterative Preference Learning
Paper • 2409.02392 • Published • 16 -
Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
Paper • 2406.17312 • Published -
Show, Don't Tell: Aligning Language Models with Demonstrated Feedback
Paper • 2406.00888 • Published • 33 -
Contrastive Prefence Learning: Learning from Human Feedback without RL
Paper • 2310.13639 • Published • 25 -
trl-lib/kto-mix-14k
Viewer • Updated • 15k • 129 • 9 -
Towards Efficient and Exact Optimization of Language Model Alignment
Paper • 2402.00856 • Published • 2 -
HelpSteer2-Preference: Complementing Ratings with Preferences
Paper • 2410.01257 • Published • 24 -
General Preference Modeling with Preference Representations for Aligning Language Models
Paper • 2410.02197 • Published • 9 -
Modulated Intervention Preference Optimization (MIPO): Keep the Easy, Refine the Difficult
Paper • 2409.17545 • Published • 20 -
RLHF Workflow: From Reward Modeling to Online RLHF
Paper • 2405.07863 • Published • 71 -
Regressing the Relative Future: Efficient Policy Optimization for Multi-turn RLHF
Paper • 2410.04612 • Published -
Understanding Likelihood Over-optimisation in Direct Alignment Algorithms
Paper • 2410.11677 • Published • 1