Zum Inhalt

Add Base Models

Adding a base model is required before you can initiate training or perform fine-tuning in DeepExtension. We classify this as a developer task since it requires access to the backend file system — typically only admin users have the necessary permissions.

This guide walks you through the process of registering a model such as Qwen2.5-7B-Instruct and making it available for training within DeepExtension.

Step 1: Download the Model Files

To take model files from Hugging Face as an example, use a search engine and look for:

huggingface qwen2 5 7b instruct

You will likely find the official page for Qwen/Qwen2.5-7B-Instruct. Under the "Files and versions" tab (ensure version = main), download all files containing:

  • config.json
  • One or more *.safetensors model weight files
  • tokenizer.json, tokenizer_config.json, etc.

Create a local directory named:

Qwen2.5-7B-Instruct

and save all the files inside.

Step 2: Move to the DeepExtension Models Directory

Assume {deepextension_base_dir} is the root directory of your DeepExtension installation. Inside it, there should be a folder named models.

Move your model directory into that location:

{deepextension_base_dir}/models/Qwen2.5-7B-Instruct

This directory path will later be referenced as the Model Path.

Step 3 (Optional): Get Deployment Template (For Ollama)

If you plan to deploy the saved model via Ollama, you also need to provide a Model Deployment Template.

  1. Search online: 
    ollama modelfile qwen2 5 7b instruct
  2. Locate the official Ollama model page (e.g., for qwen2.5:7b-instruct)
  3. Click on the template link
  4. Copy the template content and save it for the next step

If you don’t plan to deploy via Ollama, this step is not required.

Step 4: Register the Base Model in DeepExtension

Go to the Base Models page in the DeepExtension Web UI and click "Add a New".

Fill in the following fields:

Field Example Value Notes
Technical Name qw2.5-7-i Must be short and unique
Model Name Qwen2.5-7B-Instruct Descriptive, full name
Model Path models/Qwen2.5-7B-Instruct Path must match the directory in Step 2
Deployment Template (Paste from Step 3) Required only for Ollama deployment

You can leave the Deployment Template field empty and fill it later using "Update Deployment Template File".

Step 5: Confirm Registration

Return to the Base Models main page and refresh. Your model should now be listed.

You should also see:

  • Disk usage (approx. 14GB for Qwen2.5-7B-Instruct)
  • Model metadata and a generated Model Card

You’re done! The base model is now available for fine-tuning, evaluation, and deployment.

For further help with deployment templates and Ollama setup, see: