Tag: logit bias

Logit Bias and Token Banning: Steering LLM Outputs Without Retraining

Learn how to use logit bias and token banning to precisely control LLM outputs without retraining. Discover practical strategies for safety, branding, and format control.

Logit Bias and Token Banning: Steer LLM Outputs Without Retraining

Learn how logit bias and token banning allow precise control over LLM outputs without retraining. Discover technical implementation, pitfalls, and alternatives.