Delete config.json

Add config.json with model_type: qwen2

Upload MATH-500 Japanese LoRA model (GRPO fine-tuned)

initial commit