Framework to make youre prediction performance-efficient and scalable.
- Increases the throughput of your machine learning-based service
- Uses shared memory for instantaneous transfer of large amounts of data between processes
- All optimizations in one library
- Supports multiple frameworks
- Videos about aqueduct
- Getting started
- Batching
- Logging
- Metrics
- F.A.Q.
- Additional features
Install using pip
:
pip install aqueduct
Moreover, aqueduct has "optional extras"
numpy
- support types from numpy in shared memoryaiohttp
- extension for aiohttp support(see more in examples)
pip install aqueduct[numpy,aiohttp]
Feel free to ask questions in Telegram: t.me/avito-ml