Developed RAG Mini-Suite and Addressed API Limitations

2 min read

📅 2025-05-11 — Session: Developed RAG Mini-Suite and Addressed API Limitations

🕒 14:00–15:10
🏷️ Labels: RAG, Openai, API, Development, Gradio, LSFA 2025
📂 Project: Dev
⭐ Priority: MEDIUM

Session Goal

The session aimed to advance the development of a Retrieval-Augmented Generation (RAG) mini-suite and address limitations in using OpenAI’s Assistants API for document management.

Key Activities

  • LSFA 2025 Contribution Planning: Explored potential contributions to the LSFA 2025 conference, focusing on Matías’ expertise in semantic pipelines and formal logic.
  • RAG Mini-Suite Development: Outlined a roadmap for the RAG mini-suite, including MVP completion and productization steps.
  • Gradio UI Module Implementation: Developed a rag_ui.py module for Gradio, integrating context retrieval and LLM-based question answering.
  • Python UI Fixes: Resolved tab configuration issues and implemented the missing get_ui() function in the RAG UI.
  • OpenAI API Integration: Updated the run_llm() function for OpenAI API version 1.0.0 and optimized retrieval strategies using the Assistants API.
  • Assistants API Enhancements: Implemented a starter script for the RAG pipeline and updated tool types, replacing deprecated ‘retrieval’ with ‘file_search’.
  • Scaling and File Management: Addressed file attachment limits in the Assistants v2 API and proposed a hybrid RAG approach using a vector database.

Achievements

  • Successfully implemented the Gradio UI module and resolved configuration issues.
  • Updated OpenAI API integration to comply with new standards.
  • Developed strategic options for document retrieval using RAG and OpenAI Assistants.

Pending Tasks

  • Finalize the productization phase of the RAG mini-suite.
  • Implement the proposed hybrid RAG approach using a vector database for document management.
  • Explore further contributions to LSFA 2025, focusing on networking and presentation opportunities.

Graph View

Backlinks