#sft
- 4 Feb 2026 · Study
LLMs · Instruction tuning
Supervised fine-tuning on instruction–response pairs turns a base model into an assistant; comparing SmolLM2-135M base and instruct on the same question, and masking the loss to the response.
Supervised fine-tuning on instruction–response pairs turns a base model into an assistant; comparing SmolLM2-135M base and instruct on the same question, and masking the loss to the response.