Sam Witteveen breaks down NVIDIA's Nemotron 3.5 Lightning MoE model and its pitch for fine-tuning cheap models to handle agentic 'grunt work' like tool calls and validation.
Continue to AI University →