Magistral: Advancing Reasoning with Efficient GRPO Training

Qwen3 Technical Report: Reasoning in Pre-Training and Post-Training

Qwen2.5-VL: High-Resolution Vision Encoding with Efficient Windowed Attention

TÜLU 3: The Post-Training Recipe

TÜLU 3's High-Quality Synthetic Datasets for Post-Training LLMs

Go Zero-Shot for Cheaper LLM Evaluations

Evaluating AdEMAMix: A New Optimizer for Faster, More Efficient LLM Training

Qwen2-VL: How Does It Work?

Q-GaLore: Pre-Train 7B Parameter LLMs from Scratch on a 16 GB GPU

Add Code to Your Training Data for Better LLMs

How Generative LLMs Achieve Top MMLU Scores without Generating Anything

CriticGPT: How OpenAI Is Improving GPT-4 with GPT-4