Model architecture
Instruction-tuned model
What is instruction-tuned model?
An instruction-tuned model has already been fine-tuned to follow instructions and hold a conversation. Starting from one means adapting existing conversational behaviour rather than creating it, which usually needs far less data.
For most task-specific work this is the better starting point: the model already knows how to be asked a question, and the fine-tune only has to teach the specifics of your task.
Starting from a raw base makes sense when the target behaviour is unusual enough that existing instruction tuning would be something to overcome rather than build on.
Related terms
Start with the free tier
A magic link creates your account, your tenant and your first API key. No card until you ask for compute.