Question

A customer support team wants to deploy a large language model chatbot that can reference an entire product manual (roughly 40,000 words) in a single prompt so the model answers questions using that document. During testing, the team finds the model rejects prompts that include the full manual. Which characteristic of the LLM is most directly responsible for this limitation?

Accepted Answer

The model's context window, which caps the number of tokens that can be processed in a single input/output exchange

Answer

The GPU's memory bandwidth, which limits how fast token embeddings are streamed

Answer

The training dataset size, which determines how much factual knowledge the model retains

Answer

The model's parameter count, which sets the maximum vocabulary the model can recognize