You are preparing long PDFs for a RAG application. What step comes before generating embeddings?
Why B is the answer
Chunking the documents and extracting metadata
Documents must be split into retrievable units, with metadata, before embeddings can be generated.
In detail
- Text is chunked and metadata extracted before embedding.
- Metadata lets retrieval filter results.
Use case
An illustrative scenario, not a real incident.
A support team splits long manuals into chunks with section titles as metadata, then embeds each chunk.
See it in different ways
flowchart LR A["Documents"] --> B["Chunk"] B --> C["Embed"] C --> D["Vector store"] E["Question"] --> D D --> F["Model with context"]
mindmap
root((Data preparation))
Missing values
Impute
Categories
One-hot encoding
Ranges
Scaling
Imbalance
Resampling
Class weights
Leakage
Remove future dataflowchart LR S0["PDF"] --> S1["Chunks and metadata"] --> S2["Embeddings"] --> S3["Vector store"]
Key point: Chunk and add metadata before embedding.
Why the other options are wrong
- A. Model distillation on the embedding model
- Distillation changes a model.
- C. Configuring auto scaling for the vector store
- Auto scaling is a later operations step.
- D. Hyperparameter tuning of the embedding model
- Tuning is not a data step.