preference-alignment

[NeurIPS 2024] SimPO: Simple Preference Optimization with a Reference-Free Reward

[Paper][ACL 2024 Findings] Knowledgeable Preference Alignment for LLMs in Domain-specific Question Answering

Video Generation Benchmark

Code for "ReSpace: Text-Driven Autoregressive 3D Indoor Scene Synthesis and Editing"

DPO-Shift: Shifting the Distribution of Direct Preference Optimization

[NeurIPS 2024] Official code of $\beta$-DPO: Direct Preference Optimization with Dynamic $\beta$

Source code for "A Dense Reward View on Aligning Text-to-Image Diffusion with Preference" (ICML'24).

[ICCV 2025] Official repository of "Mitigating Object Hallucinations via Sentence-Level Early Intervention".

[ICML 25] "Preference Optimization for Combinatorial Optimization Problems"

[ICLR 2025] Official code of "Towards Robust Alignment of Language Models: Distributionally Robustifying Direct Preference Optimization"

[ICLR 2026] Official repository of "Uni-DPO: A Unified Paradigm for Dynamic Preference Optimization of LLMs".

[ICLR 2025] Bridging and Modeling Correlations in Pairwise Data for Direct Preference Optimization

[ICML 2025] TGDPO: Harnessing Token-Level Reward Guidance for Enhancing Direct Preference Optimization

CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models

[NeurIPS 2025] Ranking-based Preference Optimization for Diffusion Models from Implicit User Feedback

A curated collection of papers & repos on Auto-Skill (self-evolving agents) and Auto-Rubric (rubric learning from preferences) for LLM alignment & customization.

[ECCV2026] RefReward-SR: LR-Conditioned Reward Modeling for Preference-Aligned Super-Resolution

LLM-Driven Preference Data Synthesis for Proactive Prediction of the User’s Next Utterance in Human–Machine Dialogue

Production-grade SFT + ORPO fine-tuning scripts for small reasoning & instruction models. Built with Unsloth + TRL, stabilized for single-GPU (RTX 4090) training with long thinking tokens. More models coming.

A living, evidence-based tracker of personalized LLM & agent research — benchmarks, methods, and surveys across preference alignment, long-term memory, multimodal, tool use, safety and privacy.