In this video, we'll walk you through the process of fine-tuning a custom reasoning model from start to finish. Specifically, a distillation of Deepseek R1 into Llama. The whole process takes under 30 minutes and does not require coding. The same process shown here can be used for other reasoning models, like Claude Sonnet 3.7 Thinking.
- Check out the documentation for distilling reasoning models.
- Download Kiln to try it yourself.
- See our Fine Tuning Guide if you want to fine-tune a model, but aren't looking to distill a reasoning model.