What is CGC
Comtegra GPU Cloud (CGC) is a GPU cloud platform by Comtegra S.A. for AI and machine learning workloads, managed with the CGC CLI and the CGC SDK.
CGC CLI is a command line interface for the Comtegra GPU Cloud. CGC CLI enables management of your Comtegra GPU Cloud resources. The current version of the app provides support for compute, storage and network resources to be created, listed and deleted. Every compute resource is given to you as an URL, which is accessible from the open Internet.
To enable better access to your storage resources, every account has the ability to spawn a free of charge filebrowser which is a local implementation of dropbox. Remember to mount newly created volumes to it.
For now, we provide the ability to spawn compute resources like:
- Jupyter notebook with tensorflow or pytorch installed as default
- Triton inferencing server for large scale inferencing
- Label studio for easy management of your data annotation tasks with variety of modes
- Rapids suite of accelerated libraries for data processing
- LLM API for large language models
- DeepStream for video processing
- Vision API for computer vision tasks
- llama.cpp for serving GGUF models through an OpenAI compatible API
- vLLM for serving Hugging Face models through an OpenAI compatible API
- OpenWebUI for a self-hosted chat interface
- n8n for low-code workflow automation
- AimStack for tracking your experiments
- Label Studio ML backend for automatic pre-annotation with YOLO
- Unsloth for fine-tuning of large language models
- nginx for serving static content
- Custom images for running any application from your own container image
Notebooks are equipped with all CUDA libraries and GPU drivers which enables the usage of GPU for accelerated computations. Apart from compute resources, we provide these database engines accessible from within your namespace:
- PostgreSQL
- Weaviate
- PG-Vector
- Apache AGE
- MinIO
- MongoDB (your own image)
- Redis (your own image)
More is coming! Please follow instructions to get started.