> See the full index of documentation pages at https://docs.zettabyte.com/llms.txt.

# Tasks overview

> Training, fine-tuning, and inference all run as tasks. Start here, then follow the walkthrough for what you want to do.

Source: https://docs.zettabyte.com/docs/run-on-gpus/training-and-tasks

Last updated: 2026-07-21

The **Tasks** page runs `Training`, `Fine-Tuning`, and `Inference` jobs. It has three tabs, **Current Tasks**, **Templates**, and **Archived Tasks**, and a **Show Only My** toggle. Tasks run on a launched resource (see GPU instances) and draw on **Library** assets such as images, models, and datasets.

## The three task types

- `Training`: a job with a start and an end, such as running a script to train a model. See Run a training job.
- `Fine-Tuning`: a guided form of a training job for adapting an existing model. See Fine-tune a model.
- `Inference`: a long-running service that exposes a model over an API endpoint. See Deploy a model and Call your endpoint.

## Reuse a configuration

Turn a working task configuration into a reusable `Template` so you do not re-enter everything next time. See Task templates.

## Track your tasks

The **Current Tasks** tab lists running and recent tasks for the selected type. Each row shows `Name/ID`, `Task Type`, `Status`, `Scaling`, `Runtime`, and `Creation Time`. `Inference` rows also show `Model Name`, `Inference Engine`, `GPU Count`, and `Resource`. Search by task name or ID.