docs: add Azure AI Search + Azure OpenAI RAG recipe notebook #675
Conversation
Merge ProtectionsYour pull request matches the following merge protections and will not be merged until they are valid. 🟢 Enforce conventional commitWonderful, this rule succeeded.Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/
|
|
@vagenas @pablocastro can you review :) |
| " print(\"MPS GPU is enabled.\")\n", | ||
| "else:\n", | ||
| " raise EnvironmentError(\n", | ||
| " \"No GPU or MPS device found. Please check your environment and ensure GPU or MPS support is configured.\"\n", |
There was a problem hiding this comment.
Not all the other examples seem to require a GPU/MPS backend. Is this a hard requirement?
| "AZURE_OPENAI_ENDPOINT = os.getenv(\"AZURE_OPENAI_ENDPOINT\")\n", | ||
| "AZURE_OPENAI_API_KEY = os.getenv(\"AZURE_OPENAI_API_KEY\")\n", | ||
| "AZURE_OPENAI_CHAT_MODEL = os.getenv(\"AZURE_OPENAI_CHAT_MODEL\")\n", | ||
| "AZURE_OPENAI_API_VERSION = os.getenv(\"AZURE_OPENAI_API_VERSION\")\n", |
There was a problem hiding this comment.
You could include a default version to save folks from having to go to the docs to chase a good version number.
| "source": [ | ||
| "# Part 0: Prerequisites\n", | ||
| " - Azure AI Search resource\n", | ||
| " - Azure OpenAI resource with deployed embeddings & chat models\n", |
There was a problem hiding this comment.
Further down you make the assumption that the embedding model is text-embedding-3-small, maybe we could call this out here so folks know what to deploy in their Azure OpenAI instance?
| " embedding_vector = embed_text(chunk_text)\n", | ||
| " upload_docs.append(\n", | ||
| " {\n", | ||
| " \"chunk_id\": str(uuid.uuid4()),\n", |
There was a problem hiding this comment.
In a previous cell you created a list of tuples (chunk id, chunk text), but here you're not using the id and generating a GUID instead. The previously generated chunk id doesn't seem to be used, seems you only need one thing, either the chunk ids you generated before (then you don't need the new GUID here, use the first element of the tuple instead), or don't generate chunk ids in the previous cell (and in that case it looks like you don't need to create all_chunks at all, since you could just enumerate the original doc_chunks instead.
| " subset = upload_docs[i : i + BATCH_SIZE]\n", | ||
| " resp = search_client.upload_documents(documents=subset)\n", | ||
| " console.print(\n", | ||
| " f\"Uploaded batch {i} -> {i+len(subset)}; success: {resp[0].succeeded}, status code: {resp[0].status_code}\"\n", |
There was a problem hiding this comment.
Did you mean to check if all succeeded/any failed? Aren't you just checking the first doc above?
| "source": [ | ||
| "from azure.search.documents.models import VectorizableTextQuery\n", | ||
| "\n", | ||
| "def generate_chat_response(prompt: str, system_message: str = None):\n", |
There was a problem hiding this comment.
Since this method creates a new message list each time, there's no chat history preserved between turns (e.g. can't make follow up questions). I assume this was on purpose, but calling it out just in case.
Azure AI Search RAG Example Using Docling
This notebook demonstrates how to build a Retrieval Augmented Generation (RAG) system using Docling and Azure AI Search. The example showcases document parsing, chunking, vector search integration, and RAG query implementation.
Checklist: