LLM video Series Multimodal Rag, Building Multimodal AI RAG with LlamaIndex, NVIDIA NIM, and Milvus | LLM App Development

AI & Data Science NVIDIA NeMo
, , ,
1

Hello Everyone,

I am trying to implement multimodal rag from this link GenerativeAIExamples/community/llm_video_series at main · NVIDIA/GenerativeAIExamples · GitHub
and the video https://www.youtube.com/watch?v=NaT5Eo97_I0&t=460s
unfortunately many of the functions are outdated, first I tried python env, it did search for frontend.py then after installing conda environment it resolved but there are many issues with milvus, I tried to vibe code with GPT I feel I am in the loop of destrying a service and make something else correct currently I am getting ImportError: cannot import name ‘AnnSearchRequest’ from ‘pymilvus’ (C:\Users\sasan\anaconda3\envs\multimodal-rag-fresh\lib\site-packages\pymilvus_init_.py)
but after so many days I think the issue is the imncomaptivbities with current version and the version when this git has been posted

2

This looks less like an implementation issue and more like a version mismatch—the NVIDIA repo was built on older dependencies, and libraries like pymilvus have evolved, which is why errors like the AnnSearchRequest import are appearing. It would be helpful to either align your environment with the repo’s original versions or update the code to match the latest APIs, as that should resolve the loop you’re facing. You might also find TechQware’s blog “Large Language Models in Mobile App Development: A Complete Guide for Businesses” useful for gaining better clarity on managing such integrations.