0:00
/

Paid episode

The full episode is only available to paid subscribers of The Neural Maze

Finetuning Sessions – Week 2 Office Hours

Covering Supervised Finetuning, Reasoning Models, Agentic SFT, and Dataset Generation

Here's the recording from yesterday's live session, covering the Week 2 content of the Finetuning Sessions.

During office hours, we focused on:

  • The foundations of Supervised Finetuning (SFT)

  • Chat templates, with a deep dive into the Qwen3 template

  • Reasoning models and why they require an SFT step before applying RL

  • Synthetic dataset generation

  • Agentic SFT, in…

User's avatar

Continue reading this post for free, courtesy of Miguel Otero Pedrido.