Cargoship

Cargoship lets developers incorporate AI models into their software without having to learn about machine learning. The project offers a library of pre-trained, open-source models that developers can use through APIs. The models tackle a variety of AI tasks, including image and text recognition, text generation, audio transcription and general-purpose data processing.

Among Cargoship's features are:

• Growing library of models: Models that are curated and fine-tuned from HuggingFace and GitHub.
• Up to date: The project keeps pace with the latest developments in AI.
• Easy to use: Models are packaged in Docker containers with well-documented APIs.
• Open-source and self-hostable: Users can run models themselves or get a personal endpoint and API key with a minimal amount of setup.

Cargoship models span several categories:

• Text Processing: Analyze text, identify language and other tasks (4 models)
• Text Generation: Generate new text based on input (2 models)
• Image Recognition: Identify objects and text in images, classify them or find subtitles (3 models)
• Image Generation: Generate new images based on input (1 model)
• Audio Transcription: Convert spoken words in an audio recording into text (1 model)

Cargoship is geared for developers who want to add AI abilities to their software without having to build a lot of machine learning expertise. The project is in early access and welcomes feedback through its Discord channel. Developers can get started with pre-trained models, using them directly in their products to add new abilities.

Published on June 13, 2024