Fine-tuning a small model with reinforcement learning boosts structured output accuracy

Original: Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps

Source: huggingface.co ↗