PyRun Documentation

Appearance

PyRun for AI: Streamlining Your Machine Learning Workflows 🧠 ​

PyRun is actively expanding its capabilities to provide robust support for a wide range of Artificial Intelligence (AI) and Machine Learning (ML) workloads. Our platform simplifies the execution of complex AI pipelines, allowing you to leverage the power of the cloud for tasks like audio recognition, distributed machine learning, image classification, and interacting with Large Language Models (LLMs).

This initial offering provides a solid foundation for your AI projects. PyRun's integrated environment, automatic runtime management, and support for frameworks like TensorFlow, Dask-ML, and tools like Ollama (via notebooks) make it easier to:

  • Develop and Iterate: Quickly set up environments and test your AI models.
  • Scale Computations: Utilize Dask and Lithops for distributed training and processing.
  • Manage Dependencies: Define complex AI/ML environments with ease.
  • Explore Pre-built Pipelines: Get started quickly with our initial set of AI-focused pipelines.

Current AI Pipelines on PyRun ​

Explore our ready-to-run AI pipelines to see PyRun in action:

  • Audio Recognition: An AI pipeline for audio keyword recognition using TensorFlow. Demonstrates how to preprocess audio data, train a model, and perform inference.
  • DaskML for Distributed Learning: Shows how to use Dask-ML for distributed machine learning tasks, scaling your model training across multiple nodes.
  • Image Classification: An AI pipeline for image classification tasks using TensorFlow. Covers data loading, model building, and evaluation.
  • LLM Execution with Ollama: An interactive notebook environment to run and experiment with Large Language Models (LLMs) locally within your PyRun workspace using Ollama.

LLM Execution pipeline

The Future of AI on PyRun 🚀 ​

This is just the beginning! We are committed to making PyRun an even more powerful platform for AI development. Future enhancements will focus on:

  • Expanding Pipeline Library: We will continuously add more pre-built pipelines for diverse AI use cases.
  • Deeper AI Workflow Integration: We plan to introduce more specialized features and tighter integrations for key stages of the ML lifecycle, including:
    • Data Preprocessing: Enhanced tools for large-scale data ingestion and transformation.
    • Model Training: More streamlined experiences for distributed training and experiment tracking.
    • Model Evaluation: Advanced options for evaluating model performance and comparison.
  • Broader Framework and Tool Support: Expanding native support for more AI/ML frameworks and MLOps tools.

As PyRun grows, we will continue to expand our AI/ML capabilities, providing more tools and integrations to support your end-to-end machine learning lifecycle. Stay tuned for updates, and check our Roadmap for more details.