Tag: LLM tokenization
Logit Bias and Token Banning in LLMs: How to Control Outputs Without Retraining
Logit bias and token banning let you steer LLM outputs without retraining. Learn how to block unwanted words, avoid model workarounds, and apply this technique safely in real-world AI systems.
