Discussion on Finetuning Reasoning LLMs with SFT or RL

Core summary

One sentence to understand this update

A discussion on Reddit seeks advice on the best approach (Supervised Fine-Tuning or Reinforcement Learning) for finetuning small LLMs using conversational data with reasoning traces.

Impact & opportunity

What this could mean

This thread provides a valuable resource for developers seeking guidance and community insights on advanced LLM finetuning techniques.