Plexe is an AI-powered system that allows users to build and train machine learning models by simply describing their intent and data schemas in natural language. It targets developers and researchers who want to rapidly prototype and deploy ML models without extensive coding, leveraging an agentic approach for automated model selection, generation, and evaluation.

How It Works

Plexe employs a multi-agent architecture where specialized AI agents collaborate to understand requirements, plan solutions, generate code, and evaluate performance. This approach automates the typically complex ML workflow, enabling users to define models via plain English descriptions and schemas. It supports distributed training via Ray for parallel processing and offers flexibility with multiple LLM provider integrations.

Quick Start & Requirements

• Install: pip install plexe
• Prerequisites: API keys for LLM providers (e.g., OPENAI_API_KEY).
• Documentation: docs.plexe.ai
• Community: Discord

Highlighted Details

• Natural language model definition via intent and schema.
• Automated model building with model.build().
• Distributed training support using Ray.
• Data generation and schema inference capabilities.
• Multi-provider support for LLMs (OpenAI, Anthropic, Google).

Maintenance & Community

The project has an active Discord community and a public roadmap indicating ongoing development, including support for fine-tuning, self-hosted platforms, and non-tabular data types.

Licensing & Compatibility

Licensed under Apache-2.0, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

Support for non-tabular data types is currently listed as a future feature on the roadmap. The effectiveness of model generation is dependent on the chosen LLM provider and the clarity of the natural language prompt.