Skip to main content

Cerebras

Official website: https://cerebras.ai/

Cerebras is a cutting-edge computing platform for large-scale artificial intelligence model training and inference. This guide provides a comprehensive overview of Cerebras, including detailed information on core concepts, workflows, features, and the Cerebras PyTorch API.

Key Features​

  • Cerebras Model Zoo with pre-configured Large Language Models (LLM).
  • Support for Computer Vision (CV) models.
  • Compatibility with multimodal models like LLaVA.
  • Ability to use custom PyTorch models and datasets.

The Cerebras Wafer-Scale Cluster is meticulously engineered to enable neural network training with remarkably efficient linear scaling across millions of cores without the complexities of traditional distributed computing.

How to Connect​

- On the VSCode CodeGPT Extension, change the model on the chat, click on `Manage Models` button

  • Pick the API Keys tab and select Cerebras as Provider

  • Paste API Key here, and click on Connect
  • Click outside the options and ask to chat.
Remove Key

To remove your API Key from CodeGPT:

  • Click again on the Manage models button.
  • Click on "Change connection settings"

API Errors​

If you are experiencing API errors, please refer to the following documentation: Cerebras Documentation.

Important Note

This provider is only available in version 3.5.73 of the CodeGPT extension for VSCode and later. Ensure that you have the extension updated.