Mistral-Large-4
Mistral AIpropose a block of 3 tokens
coreinference servingin Mistral Large 4Mistral AI
1 more from this model
The speculative depth is fixed at 3 by the checkpoint.
defaultunclearin Mistral Large 4Mistral AI
implementation detail · filed under inference & serving
Sets the speculative draft or verification depth to a block of three tokens.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See decoding strategy for the methods filed beside it.
Also called Three-token block speculative decoding, propose a block of 3 tokens, speculative depth is fixed at 3.
2 quotes from 2 sources, one card per model, each opening with its strongest.
propose a block of 3 tokens
The speculative depth is fixed at 3 by the checkpoint.