Smart Vision
Series
Appearance

2025.1.23

Smart Vision Ver3.2 have been released, with the following key upgrades:

  • Application

1.Workflow: Supports passing in conversation_id, and adds a new "memory" function to LLM node, so that conversation memory can be brought in by turning on the memory switch

image_17376262429610

2.Workflow: Added organize blocks function to make the workflow canvas more beautiful

image_1737626360569

  • Data and Capability

    1.Knowledge Base: The knowledge governance function is launched online to support knowledge governance of uploaded files, including innovative functions such as layout recognition and sorting, flexible manual adjustment of chunks, summarization and tags generation, etc., which effectively enhances knowledge quality and improves retrieval accuracy.

image_17376355212195

image_17376359985785

image_17376374769104

image_17376358809379

2.Knowledge Base: Added knowledge base access API, enhancing Smart Vision's ability to be integrated

image_17376364324089

3.Datasets: When generating data through "Tool Calling", supports selecting workflow configured as the source of training data

image_17376366352157

4.Tool Store: 3 tools are available, including Google Scholar, Google Search, and Global News Search

image_17376367214059

5.Tools: Internal tool API supports Post calling method and HTTPS protocol

image_17376369347921

  • Model

    1.Model Store: 2 large models are available, including Alibaba|Marco-o1 and DeepSeek-LLM-7B-Chat, providing more choices for enterprise application scenarios

image_17376375586900