Skip to main content

Cortex Model Repos

warning

🚧 Cortex.cpp is currently under development. Our documentation outlines the intended behavior of Cortex, which may not yet be fully implemented in the codebase.

cortex repo

Cortex Model Repos are Cortex.cpp's built-in models hosted on Huggingface, which uses a single Git repository to hold different versions of a model, which can be pulled using Docker or Ollama-like syntax.

info

We also plan to provide alternative hosting locations or servers to replicate the content, ensuring access in regions where HuggingFace is blocked or has slow download speeds.

Why HuggingFace?​

Cortex Model Repos are hosted on Huggingface for several reasons:

  • Tag-Based Formats: Inspired by Docker and Ollama, Cortex Model Repos uses a tag-based format (model:version) that simplifies managing multiple versions and formats of a model.
  • Git and Git-LFS Based (Inspired by Ollama): Using Git and Git-LFS makes mirroring and version control easy.
  • Mirror-Friendly Setup: If Huggingface is blocked or slow, repositories can be mirrored across different locations.

Usage​

Download a built-in model from the Cortex Model Repos using a model_id. You can obtain the model_id from the Cortex model repository or the model's specific branch.


# Stable
## Download the default mistral model from: https://huggingface.co/cortexso/mistral/tree/main
cortex pull mistral
## Download an ONNX version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/onnx
cortex pull mistral:onnx
## Download a TensorRT-LLM version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/tensorrt-llm-linux-ada
cortex pull mistral:tensorrt-llm-linux-ada
## Download a 7B version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/7b-gguf
cortex pull mistral:7b-gguf
# Beta
## Download the default mistral model from: https://huggingface.co/cortexso/mistral/tree/main
cortex-beta pull mistral
## Download an ONNX version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/onnx
cortex-beta pull mistral:onnx
## Download a TensorRT-LLM version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/tensorrt-llm-linux-ada
cortex-beta pull mistral:tensorrt-llm-linux-ada
## Download a 7B version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/7b-gguf
cortex-beta pull mistral:7b-gguf
# Nightly
## Download the default mistral model from: https://huggingface.co/cortexso/mistral/tree/main
cortex-nightly pull mistral
## Download an ONNX version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/onnx
cortex-nightly pull mistral:onnx
## Download a TensorRT-LLM version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/tensorrt-llm-linux-ada
cortex-nightly pull mistral:tensorrt-llm-linux-ada
## Download a 7B version of the mistral model from: https://huggingface.co/cortexso/mistral/tree/7b-gguf
cortex-nightly pull mistral:7b-gguf