Open dicktangdev opened 1 month ago
Is this service an offering that Azure provides outside of their built in Document Chat? If so then we could leverage it possibly. Otherwise, we do not OCR/vision PDF images or chats, which is why that data would not be referenced since not every LLM is multimodal - making full transcription/captioning not tenable for those who are not using those types of models
Hi @timothycarambat,
Thanks for your reply. Yes, the Azure document intelligence(DI) is a standalone service and can be used to provide as a document loader and chunking. Please find a RAG sample that uses DI below - https://github.com/microsoft/Form-Recognizer-Toolkit/blob/main/SampleCode/Python/sample_rag_langchain.ipynb
What would you like to see?
Our another Azure OpenAI solution with Azure Document Intelligence works great at indexing PDFs containing charts and tables, enabling accurate data extraction from these elements. However, when utilizing Anything LLM, it frequently indicates that the tables'/charts' information is not provided. Could you advise on a method to leverage Azure Document Intelligence for enhanced indexing and parsing capabilities?