preference-modeling

[ICML 2025] Beyond Bradley-Terry Models: A General Preference Model for Language Model Alignment (https://arxiv.org/abs/2410.02197)

Lightweight pairwise evaluator for relational signals in Ouro-2.6B-Thinking loop-state trajectories.

Kaggle Silver Medal LLM preference prediction project with large-scale inference and performance optimization.

This NTU project presents a comprehensive approach to predicting human preferences for large language model (LLM) outputs in a multi-model ranking scenario. We combine knowledge distillation techniques with novel NLP-based preference modeling that integrates logical consistency and human-aligned desirability metrics.

O chatbot-arena-preference-prediction-llms.ipynb é um projeto de aprendizado de máquina que utiliza grandes modelos de linguagem (LLMs) para prever preferências dos usuários em um ambiente de chatbot. O objetivo é analisar interações para melhorar as respostas e personalizar a experiência do usuário com base em suas preferências.

Bayesian MCMC model for estimating browsing preferences, diversity preference, and switching costs.

Fine-tuned Qwen2.5-1.5B that predicts which chatbot answer humans prefer (Kaggle LLM Classification Finetuning), plus GPU, CUDA, vLLM and ONNX inference work and a free-CPU live demo.

Optimize flight search costs with 8 AI skills for travelers and teams, cutting airfare spend by 30–66% for personal trips and 10–15% for business trips