nafyi.
Topics / fine-tuning efficacy

Attributed viewpoint

Task data and fine-tuning drove the performance gains

Piero Molino says the vast majority of value in his experiments came from data for the task and from fine-tuning after training. He describes the base model as more or less an independent variable.

Piero Molino·Published

Behind the viewpoint

Translations are for reading; original excerpts remain the evidence.

Piero Molino on AI in Games

Original excerpt

The vast majority of the value was coming from the data for the task. The base model was more or less an independent variable. The actual effect was coming from the fine-tuning and the increased performance that you get after post-training.
Open original · Paragraph 160 (opens in a new tab)View source page