What is Fine-Tuning?
Further training a general model on specialized data to adapt it to a task.
Formal Definition
Fine-tuning continues training a pre-trained base model on a narrower dataset so it adapts to a specific domain, style, or task. It changes the model's weights, unlike prompting. AI Stock Challenge deliberately compares off-the-shelf frontier models under one shared prompt rather than fine-tuned variants, so the test measures general reasoning, not bespoke training.
In Simple Terms
It is taking a smart, general-purpose AI and giving it extra lessons in one subject so it gets better at that specific job, like coaching a strong generalist to specialize in finance.
Example
A firm might fine-tune a base model on years of earnings-call transcripts, whereas the benchmark tests the standard models as they ship, with no custom training.