Max Tokens
Domain: LLM Concepts
🎯 Motivation
A parameter that limits how many tokens the model can generate. Prevents runaway generation and controls cost.
📋 Concrete Example
Setting max_tokens=100 means the model generates at most 100 tokens (about 75 words). If the answer needs 200 tokens, it gets cut off. Choose wisely based on the task.
🔗 Analogy
A word limit on an essay - ‘maximum 500 words’. The student must fit their answer within that limit. Max_tokens does the same for the model, capping output length.