semantic-kernel

Semantic Kernel Feature Matrix by Language

Legend

✅ - Feature implemented
🔄 - Feature partially implemented (see associated Note column)
❌ - Feature not implemented

AI Services

  C# Python Java Notes
Text Generation Example: text-davinci-003
Text Embeddings Example: text-embeddings-ada-002
Chat Completion Example: GPT-4, GPT-3.5-turbo
Image Generation Example: Dall-E 2

AI Service Endpoints

  C# Python Java Notes
OpenAI  
Azure OpenAI  
Hugging Face Inference API 🔄 Coming soon to Python, not all scenarios are covered for .NET
Hugging Face Local  
Custom 🔄 Requires to define the service schema in the application

Tokenizers

  C# Python Java Notes
GPT2  
GPT3  
tiktoken Coming soon. Can be added manually to Python via pip install tiktoken

Core Skills

  C# Python Java Notes
TextMemory Skill 🔄  
ConversationSummary Skill  
FileIO Skill  
Http Skill  
Math Skill  
Text Skill 🔄  
Time Skill  
Wait Skill  

Planning

  C# Python Java Notes
Plan 🔄 Plan object model to be completed
BasicPlanner  
ActionPlanner 🔄  
SequentialPlanner 🔄  

Memory Connectors, Vector storage

  C# Python Java Notes
Azure Search 🔄 Azure Cognitive Search coming soon
Weaviate Currently supported on Python 3.9+, 3.8 coming soon
Chroma  
Qdrant  
Pinecone  
Milvus Coming soon
Sqlite Vector optimization requires sqlite-vss
Postgres Vector optimization requires pgvector
CosmosDB CosmosDB is not optimized for vector storage
Redis Vector optimization requires RediSearch

Connectors and Skill Libraries

  C# Python Java Notes
MsGraph Contains connectors for OneDrive, Outlook, ToDos, and Organization Hierarchies
Document and Data Loading Skills Pdf, csv, docx, pptx. Currently only supports Word documents
OpenAPI  
Web Search Skills (i.e. Bing, Google)  
Text Chunkers 🔄 🔄  

Design Choices

The overall architecture of the core kernel is consistent across all languages, however, the code follows common paradigms and style of each language.

During the initial development phase, many Python best practices have been ignored in the interest of velocity and feature parity. The project is now going through a refactoring exercise to increase code quality.

To make the SDK as lightweight as possible, the core packages have a minimal set of external dependencies.