Barcenas Nemotron Qwen3 4B Instruct
Based on: Qwen/Qwen3-4B-Instruct-2507 Trained with dataset: nvidia/Nemotron-Instruction-Following-Chat-v1
The primary objective was to enhance the base model's conversational capabilities by training it on 100,000 carefully selected examples from the Nemotron dataset, which is renowned for its excellence in instruction following and structured responses.
The result is a model that delivers more coherent and better-formatted responses, with higher accuracy when following complex instructions.
Note: For me, this represents the fine-tuning process with the heaviest and most complex dataset I have undertaken to date.
Made with ❤️ in Guadalupe, Nuevo Leon, Mexico 🇲🇽