LM-Kit.NET のリリース

Released: Mar 24, 2025

2025.3.6 での更新項目

機能

  • Added support for Mistral Small 3.1 model and added mistral-small-3.1 model card.
  • Added the GetText method to the Attachment class.
  • Added HTML format support to the Attachment class.
  • Strong performance improvements for the Vulkan backend.
  • Enhanced extraction accuracy in the structured text extraction engine.
  • Reduced VRAM/RAM usage.

Released: Mar 17, 2025

2025.3.5 での更新項目

機能

  • Added HandleUnknownCategory property to the Categorization class.
  • Added support for Gemma 3 vision.
  • Improved inference with Gemma 3 models family.
  • Improved internal context caching strategy.
  • Improved inference speed.

Released: Mar 12, 2025

2025.3.4 での更新項目

機能

  • Added support for Gemma 3 models family.
  • Added model cards for Qwen QwQ 32B, Gemma 3 1B, Gemma 3 4B, Gemma 3 12B and Gemma 3 27B.
  • Added the Guidance property to the Categorization class.

Released: Mar 11, 2025

2025.3.3 での更新項目

機能

  • Added SetElementsFromJsonScheme method to the TextExtraction class.
  • Added IsVisionRequired property to the Attachment class.
  • Added GetTopCategories and GetTopCategoriesAsync property to the Categorization class.
  • Added GetBestCategoryAsync and GetTopCategoriesAsync overloads to the Categorization class to enable classification from image attachments.

Released: Mar 3, 2025

2025.3.2 での更新項目

機能

  • Introduced the MemoryType enumeration.
  • Added SaveInformation overloads to the AgentMemory class to specify the type of memory to record.
  • Added SaveInformationAsync overloads to the AgentMemory class to specify the type of memory to record.
  • Added GetMemoryType static method to the AgentMemory class.
  • Added GetDataSource method to the RagEngine class.
  • Added TryGetDataSource method to the RagEngine class.
  • Added Get method to the MetadataCollection class.
  • Added TryGet method to the...

Released: Mar 1, 2025

2025.3.1 での更新項目

機能

  • Added support for Phi-4 Mini models.
  • Added model cards for Phi-4 Mini and Deepseek R1.
  • Added RemoveDataSource overloads to the RagEngine class to remove a DataSource from an identifier.
  • Added RemoveSection method to the DataSource class.
  • Improvements to generic classification accuracy and speed.

Released: Feb 26, 2025

2025.2.4 での更新項目

機能

  • Improved inference with various model families such as Granite, Mistral, Monarch, Orion, MiniCPM, and GigaChat.
  • Improved language detection accuracy.
  • Minor performance improvements of the inference system.
  • Added new constructor to the Attachment class to support stream input.
  • Added ExtractKeywords and ExtractKeywordsAsync overloads to the KeywordExtraction class to extract keywords from image attachments.
  • Added Summarize and SummarizeAsync overloads to the Summarizer class to produce summary...

Released: Feb 19, 2025

2025.2.3 での更新項目

機能

  • Introduced the AgentMemory class.
  • Introduced the MemoryRecallEventArgs class.
  • Introduced the TopNSigmaSampling class.
  • Added Memory property to the MultiTurnConversation class.
  • Added MaximumRecallTokens property to the MultiTurnConversation class.
  • Added MemoryRecall event to the MultiTurnConversation class.
  • Added DownloadAsync method to the ModelCard class.
  • Added LogCallback event to the Runtime class.
  • Improved RAG engine accuracy and speed.
  • Optimized dynamic sampling mechanisms.

Released: Feb 12, 2025

2025.2.2 での更新項目

機能

  • Added ModelID property to the ModelCard class.
  • Added GetPredefinedModelCardByModelID method to the ModelCard class.
  • Added LoadFromModelID static method to the LM class.
  • Added ModelUri property to the LM class.
  • Added ImportTextAsync method to the RagEngine class.
  • Added ImportTextFromFileAsync method to the RagEngine class.
  • Improved internal context caching strategy.

Released: Feb 4, 2025

2025.2.1 での更新項目

機能

  • Added model card for Mistral Small Instruct 2501.
  • Created a new open-source assembly bridging Microsoft SemanticKernel and LM-Kit.NET.
  • Improved inference of MiniCPM engine.