Contiguous Decoding

ConceptMentioned in 1 video

A technique supported by SGLang that uses finite state machines (derived from schemas like JSON via tools like XGrammar) to control output, enabling faster decoding by skipping unnecessary tokens.

Videos Mentioning Contiguous Decoding

DeepSeek V3, SGLang, and the state of Open Model Inference in 2025 (Quantization, MoEs, Pricing)

DeepSeek V3, SGLang, and the state of Open Model Inference in 2025 (Quantization, MoEs, Pricing)

Latent Space

A technique supported by SGLang that uses finite state machines (derived from schemas like JSON via tools like XGrammar) to control output, enabling faster decoding by skipping unnecessary tokens.