Decoding and inference behavior · MLD-1155
Speculative decoding
Accelerating generation by proposing multiple tokens with a draft mechanism and verifying them with a target model.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Use a small draft model to reduce target-model serial work.
Original caveat
Exact distribution preservation requires the appropriate acceptance/correction algorithm.
Context and aliases
Level: Intermediate. Stages: Inference · Evaluation · Deployment.
Tags: SLM · LLM · NLP.
Aliases: None recorded.