This library aims to provide a unified approach for representing generative AI components, and enable seamless integration and interoperability with AI services from various providers such as Azure OpenAI, Amazon, Google and Aliyun/Alicloud. Compatible to .NET standard 2.0 and .NET 8.0+.
Besides the interfaces from Microsoft.Extensions.AI library (IEmbeddingGenerator, IChatClient and IImageGenerator), this library also provides and implements interfaces for video and speech generation (IVideoGenerator, ISpeechGenerator).
| Package Name | Description | Related Platform Services |
|---|---|---|
| Starsbane.AI | Contains platform-agnostic implementations for PDF text extraction and sentence chunking | N/A |
| Starsbane.AI.Abstractions | Contains platform-agnostic abstractions for the whole project | N/A |
| Starsbane.AI.Alicloud | Provides implementations for Aliyun/Alibaba Cloud | DashScope |
| Starsbane.AI.Amazon | Provides implementations for AWS | AWS Bedrock, Polly, Rekognition, Textract |
| Starsbane.AI.Azure | Provides implementations for Azure | Azure OpenAI, Sora, AI Vision, AI Document Intelligence |
| Starsbane.AI.Google | Provides implementations for GCP | Google Vertex AI, Vision AI, Document AI |
The sentence chunking support is powered by source code from [github.com/GregorBiswanger/SemanticChunker.NET]SemanticChunker.NET, modified for .NET Standard 2.0 support and adapt abstraction in this project.
- Implement tools support for Starsbane.AI.Google
- Comprehensive documentation, getting started, installation procedure
- Better platform and model specific config validation
- and many more...
Please refer to the projects in samples folder for the usage.
To be updated
To be updated
This project is licensed under the Apache-2.0 License - see the LICENSE file for details.