UFGCEMIGONA/Gemma-4-search

VISIONConcurrency Cost:1Model Size:5.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 29, 2026Architecture:Transformer Cold

UFGCEMIGONA/Gemma-4-search is a 5.1 billion parameter Gemma-based model developed by UFGCEMIGONA, specifically designed for advanced search and document querying tasks. It features a 32768-token context length and is optimized for retrieving relevant information from databases using structured tool calls. This model excels at processing natural language queries to interact with document repositories, providing summaries and categorized results.

Loading preview...

Overview

UFGCEMIGONA/Gemma-4-search is a 5.1 billion parameter model built upon the Gemma architecture, developed by UFGCEMIGONA. It is specifically engineered for sophisticated search and document interaction, leveraging a substantial 32768-token context window to process complex queries. The model integrates a robust tool-calling mechanism, enabling it to execute document_query and search_document functions for precise information retrieval.

Key Capabilities

  • Advanced Document Querying: Utilizes document_query to search within specific documents, filtering results based on a given query.
  • Relevance-Based Document Search: Employs search_document to identify and return the most relevant documents for a broad query.
  • Structured Tool Interaction: Processes and parses tool calls from assistant messages, executing them to retrieve data.
  • Contextual Information Extraction: Extracts and presents document metadata such as summaries, categories, and subcategories from search results.

Good For

  • Information Retrieval Systems: Ideal for building applications that require intelligent searching and querying of large document bases.
  • Knowledge Management: Can be used to enhance internal knowledge bases by providing efficient access to specific information.
  • Automated Data Extraction: Suitable for scenarios where structured data needs to be extracted from unstructured text through targeted queries.