Features

AGI framework offers 2 user interfaces:

agi-core an API interface to interface directly in your python program on a dask cluster

agilab a WEB interface that generate the agi-core API call in a snippet and offer openai code assitant.

There is share components agi-env a share component to set the environment and agi-gui the Graphic User Interface engine

agi-core

Automated Virtual Environment Setup:
- Automatically installs virtual environments for cluster nodes which are computers with multi-cores CPU, GPU and NPU.
Flexible Application Run Modes:
- Process Management:
  - Single Process
  - Multiple Processes
- Language Support:
  - Pure Python (From python 3.12)
  - Cython (Ahead of execution compilation)
- Deployment Modes:
  - Single Node with MacOS, Windows (from W11) or Linux (Ubuntu from ubuntu 24.04)
  - Cluster with heterogeneous os Modes
Dynamic Node Capacity Calibration:
- Automatically calibrates the capacity of each node to optimize performance.
Static Load Balancing:
- Distributes workloads evenly across nodes to ensure efficient resource utilization.
Integrated Map-Reduce Mechanism:
- Facilitates parallel data processing and aggregation.
Optimized Run-Mode Selection:
- Chooses the best run-mode from 8 available combinations, when rapids is enables it adds 8 more.

agilab

agi-core API Generation:
- Automatically generates APIs to streamline development processes.
ChatGPT Coding Assistant:
- Integrates with ChatGPT to offer real-time code suggestions and support.
Embedded Dataframe Export:
- Easily export dataframes cross project.
5 Ways to Reuse Code:
- Framework Instantiation:
  - Inherit from agi-core Agi[Agent | Dag | Data]Worker classes.
- Project Templates:
  - Clone existing code or create new project from templates.
- Q&A Snippets History:
  - Utilize historical code snippets for quick integration.
- Collaborative Coding:
  - Export / Import project to work together efficiently cross organisation.
- Views Creation:
  - Share views seamlessly across multiple projects.