Writing detail

My First Attempt at Fine-Tuning an SLM using Unsloth

A hands-on breakdown of turning a small language model into a focused information extraction system for NER-style tasks.

Back to writing Read the original article

Key takeaways

Small-model fine-tuning becomes far more practical when the task is narrow and evaluation is explicit.
Dataset preparation and output consistency matter as much as the training run itself.
The article frames fine-tuning as an engineering workflow, not just a model experiment.

This portfolio page keeps a concise internal summary while the full article remains published externally on Medium.

This piece documents an early fine-tuning experiment centered on information extraction instead of open-ended chat. The goal was to see whether a small language model could be shaped into a focused assistant for structured NER-style tasks without taking on the cost profile of a larger model.

The article is valuable because it stays practical. It focuses on the parts of the workflow that usually determine whether a fine-tuning attempt becomes useful in production: data preparation, output reliability, and evaluating against the exact format the system is expected to return.

Next What I Learned Building a Transformer Model from Scratch