Practice question from Semantic Search Principles

What happens if you try to embed a document that exceeds the model's context window? A. The model automatically summarizes it B. It's compressed without data loss C. The text is truncated or causes an error D. The embedding becomes more accurate E. It splits automatically into optimal chunks

Answer

C

Explanation

Most embedding models have hard context limits (e.g., 512 tokens for BERT, 8K for modern models). Exceeding this limit typically results in silent truncation (losing information from the end of the document) or an explicit error, depending on the implementation. You must manually chunk long documents into appropriate sizes.

More questions from 2026 Modern AI Search & RAG Roadmap