Released: Feb 4, 2025

Actualizaciones en 2025.2.1

Características

• 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.

Released: Jan 30, 2025

Actualizaciones en 2025.1.10

Características

• Added WebP and TIFF image support to Attachment.
• Minor performance improvements of the inference system.

Released: Jan 27, 2025

Actualizaciones en 2025.1.9

Características

• Enhanced compatibility with .NET Framework 4.6.2 apps.
• Enhanced backend management mechanism.

Released: Jan 26, 2025

Actualizaciones en 2025.1.8

Características

• Enhanced compatibility with .NET Framework 4.6.2 apps.
• Minor performance improvements for the Vulkan backend.

Released: Jan 24, 2025

Actualizaciones en 2025.1.7

Características

• Improved speed and accuracy of vision models inference.

Correcciones

• Fixed a memory leak during image eval using MiniCPM models.

Released: Jan 22, 2025

Actualizaciones en 2025.1.6

Características

• Added support for the MiniCPM-O model.
• Added ReplacementModel property to the ModelCard class.
• Improved speed and accuracy of vision models inference for content retrieval.

Correcciones

• Resolved a crash when regenerating responses using MiniCPM models.
• Fixed a memory leak with VLM disposal.

Released: Jan 19, 2025

Actualizaciones en 2025.1.4

Correcciones

• Resolved an inference issue in AOT mode.
• Resolved a rare race condition in the Vulkan backend during inference.
• Resolved a segmentation fault in the TextExtraction class.

Released: Jan 18, 2025

Actualizaciones en 2025.1.3

Características

• Added SetContent method overload that accepts an Attachment to the TextExtraction class, allowing extraction of unstructured content from images.
• Various performance improvements.
• Improved Runtime initialization.

Correcciones

• Resolved a crash during Qwen2-VL inference on macOS.

Released: Jan 11, 2025

Actualizaciones en 2025.1.2

Características

• Added initial support for multimodal capabilities with vision handling (VLM).
• Added support for the LMK model format.
• Introduced the Prompt class.
• Introduced the Attachment class.
• Added overloads to the Submit and SubmitAsync methods in the MultiTurnConversation class to support prompts with image attachments.
• Added overloads to the Submit and SubmitAsync methods in the SingleTurnConversation class to support prompts with image attachments.
• Enhanced GPU device memory management...

Released: Jan 1, 2025

Actualizaciones en 2025.1.1

Características

• Optimized dynamic sampling mechanisms.
• Improved error handling.
• Added HasVision property to the LM class.
• Added EnableModelChecksumValidation property to the Configuration class.