Insights
Scorers, Graders, and Evaluation Tasks: A Practitioner’s Map
There is a terminology problem in AI evaluation that slows down almost every team building with large language models. “Scorer” and…
Read more →Creating Gradient Palettes: Fine-Tuning granite-7b for Analogous Gradient Palettes
An analogous color harmony palette is the most stable, consisting of a main color and two or more nearby colors. I have found that…
Read more →Split to Succeed: Crafting Train-Test Datasets for Optimal fine-tuning of an LLMs
Why “train” & “test” datasets
Read more →InstructLab with Custom Datasets
Note: This is an experiment. It is not the intended way of using InstructLab.
Read more →