Skip to main content
Qwen Cloud home page
Docs

Create a dataset

Upload training data to Qwen Cloud for use in fine-tuning jobs.

This guide walks you through creating a dataset in the Qwen Cloud console.

Prerequisites

  • A Qwen Cloud account with access to the console.
  • Data files prepared in JSONL format. See Datasets overview for format requirements.

Create a dataset

1
Open the Datasets page
2
Navigate to the Datasets page and click Create Dataset.
3
Enter a dataset name
4
Provide a descriptive name for your dataset.
5
Select the task type
6
Choose Text Generation as the task type.
7
Select a fine-tuning algorithm
8
SFT (Supervised Fine-Tuning) is the fine-tuning algorithm.
9
Upload files
10
Drag and drop files into the upload area or click to browse. Supported format: JSONL (max 200 MB per file, up to 10 files).
11
Submit
12
Click Save as Draft to create the dataset.
Your dataset is now in Draft status. You must publish it before it can be used in a fine-tuning job. See Publish a dataset for instructions.
The right panel on the creation page displays file format requirements and a sample file download link for the selected algorithm.

SFT example format

{
  "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Explain gravity in simple terms."},
    {"role": "assistant", "content": "Gravity is the force that pulls objects toward each other."}
  ]
}

Next steps