pub fn speculative_decode<D: Drafter + ?Sized>(
decoder: &Decoder,
prompt_tokens: &[usize],
max_new_tokens: usize,
kv_caches: &mut [KvCache],
drafter: &D,
) -> SpeculativeDecodeResultExpand description
Greedy speculative decode over a fresh KV cache, with
prompt-lookup drafting. Thin wrapper over
speculative_decode_with, kept for callers that want the original
no-options shape.