Terminology

Toloka LLM uses the following terminology in its interface and documentation.

Dataset

A set of text to classify with optional labels.

Deployment

Making a variant available for use either through API or text-upload UI.

Project

A set of parameters, prompts, and datasets which result in different variants you can deploy and use for labeling. This is also a separate business goal you want to achieve with Toloka LLM.

Prompt

An instruction that LLM uses to provide a correct answer for your dataset.

Variant

A combination of a model, its settings and a prompt. Our tool facilitates the process of finding the best variant for your task in a measured way. Once you have a variant that satisfies you, it can be deployed.

Last updated: September 5, 2023

Introduction
Toloka LLMRegistering and signing inAdding OpenAI API keyTerminology
Working with Toloka LLM
Workflow overview
Working with datasets
Problem definition
Iterating on quality
Creating endpoint