Download
Aviary is an LLM serving solution that makes it easy to deploy and manage a variety of open source LLMs. Providing an extensive suite of pre-configured open source LLMs, with defaults that work out of the box. Supporting Transformer models hosted on Hugging Face Hub or present on local disk. Aviary has native support for autoscaling and multi-node deployments thanks to Ray and Ray Serve. Aviary can scale to zero and create new model replicas (each composed of multiple GPU workers) in response to demand. Ray ensures that the orchestration and resource management is handled automatically. Aviary is able to support hundreds of replicas and clusters of hundreds of nodes, deployed either in the cloud or on-prem.
Features
- Fully supporting multi-GPU & multi-node model deployments
- Providing a REST API that is similar to OpenAI's to make it easy to migrate and cross test them
- Offering high performance features like continuous batching, quantization and streaming
- Simplifying the addition of new LLMs to within minutes in most cases
- Simplifying the deployment of multiple LLMs within a single unified framework
- Providing an extensive suite of pre-configured open source LLMs, with defaults that work out of the box
Project Samples
