Released: May 5, 2025

Mises à jour de 2025.5.1

Fonctionnalités

• Added support for Qwen 2.5 VL and Pixtral model families.
• Added model cards for pixtral, qwen2.5-vl:3b and qwen2.5-vl:7b.
• Improved inference of vision models.
• Improved inference of LLama 4 models.

Released: Apr 30, 2025

Mises à jour de 2025.4.13

Fonctionnalités

• Added model cards for Alibaba Qwen3 (qwen3:0.6b, qwen3:1.7b, qwen3:4b, qwen3:8b and qwen3:14b).
• Added the DisableReasoning property to the SingleTurnConversation class.
• Added the DisableReasoning property to the MultiTurnConversation class.
• Added the HasReasoningSupport member to the TemplateFormatFlags enumeration.
• Added the HasReasoning property to the LM class.
• Improved translation engine precision with small models.

Released: Apr 27, 2025

Mises à jour de 2025.4.12

Fonctionnalités

• Added Upsert and UpsertAsync method overloads to the DataSource class taking raw vector as parameter.

Correctifs

• Fixed a multi-byte decoding issue in produced model tokens.

Released: Apr 23, 2025

Mises à jour de 2025.4.11

Fonctionnalités

• Added the DataSource.VectorEntry class.
• Added the Upsert and UpsertAsync method to the DataSource class.
• Introduced the VectorSearch static class.
• Breaking change - The TextPartitionSimilarity class has been renamed to PartitionSimilarity.

Correctifs

• Fixed ONNX runtime loading issue on Linux.

Released: Apr 22, 2025

Mises à jour de 2025.4.10

Fonctionnalités

• Added support for running inference on ONNX models.
• Added the HasImageEmbeddings property to the LM class.
• Introduced the Summarizer.SummarizationIntent enumeration.
• Added the Summarizer.Intent property to the Summarizer class.
• Added model cards for nomic-embed-vision, granite3.3:2b and granite3.3:8b.
• Optimized dynamic sampling mechanisms.
• Added support for image embeddings.
• Added the GetEmbeddings and GetEmbeddingsAsync overloads to the Embedder class to enable embeddings...

Released: Apr 15, 2025

Mises à jour de 2025.4.9

Fonctionnalités

• Improved translation engine.
• Improved structured text extraction accuracy.
• Improved compatibility with JSON schemas for structured text extraction.
• Improved internal scheduler engine.
• Global speed improvement, up to 20%.
• Added GetTopCategories, GetTopCategoriesAsync, GetBestCategory and GetBestCategoryAsync overloads to the Categorization class to specify category descriptions.

Released: Apr 9, 2025

Mises à jour de 2025.4.7

Fonctionnalités

• Added the HasSection method to the DataSource class.

Released: Apr 9, 2025

Mises à jour de 2025.4.6

Fonctionnalités

• Introduced a new built-in vector database engine that permits handling small- to medium-sized vector collections on the file system without any dependency.
• Introduced the StorageMode enumeration.
• Added the DataSource.StorageMode property.
• Added the LoadFromFile and LoadFromFileAsync methods to the DataSource class.
• Added the OptimizeDataSource and OptimizeDataSourceAsync methods to the DataSource class.
• Added the CreateFileDataSource, CreateInMemoryDataSource...

Released: Apr 2, 2025

Mises à jour de 2025.4.1

Fonctionnalités

• Introduced different DataSource storage modes, enabling storage of DataSource content directly in vector databases.
• Introduced a new collection of open-source assemblies to bridge external vector databases with LM-Kit.NET. Currently, Qdrant is supported.
• Introduced IVectorStore interface.
• Introduced PointEntry class.
• Added CollectionChanged event to the MetadataCollection class.
• Added PropertyChanged event to the MetadataCollection class.
• Added AddOrReplace method to the...

Released: Mar 24, 2025

Mises à jour de 2025.3.6

Fonctionnalités

• 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.