Hugging Face BlogSep 3, 2026, 08:00 UTC+8

A 350M model fine-tuned with GRPO in 100 steps lifts structured-output compliance from 22.6% to 29.7%

Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps

A hands-on guide from Hugging Face and Liquid AI that fine-tunes LFM2.5-350M with GRPO via the TRL library. Using only 500 samples and 100 training steps on a free Colab GPU, structured-output compliance on the IFStruct benchmark jumps from 22.6% to 29.7%. The post includes the full notebook, reward-function design, and a local evaluation setup with llama.cpp on a MacBook.

Why it matters: A hands-on guide with concrete numbers and a reproducible recipe — hits H and K. But the audience is narrow and R is absent; tutorial content at the featured threshold gets 72.

Read the original ↗Export Markdown

Take

This one's worth opening because it drops the barrier for structured-output fine-tuning to near zero. Hugging Face and Liquid AI took LFM2.5-350M, ran GRPO via TRL—a method that compares groups of responses and rewards the better ones—on just 500 samples and 100 steps, all on a free Colab GPU. Compliance on IFStruct went from 22.6% to 29.7%.

The absolute number isn't production-grade yet, but the direction is clear: the model follows format instructions more reliably. I'd discount the 29.7% a bit since IFStruct is one specific benchmark, not a universal structured-output test. The real value here is the full recipe—training data, reward function design, LoRA config, and a local eval setup with llama.cpp on a MacBook are all public. If you're wrangling a small model that needs to spit out clean JSON or fixed formats, this is a solid starting point.

Source: Hugging Face Blog · huggingface.co
Published: Sep 3, 2026, 08:00 UTC+8 · 27 days ago