Use on OpenRouter · Use on Vercel AI Gateway · Release blog post Laguna S 2.1 DFlash DFlash speculator (draft model) for Laguna S 2.1, bf16. Architecture: DFlashLagunaForCausalLM (6 sliding attention layers, block size 16). Shares token embedding + lm head with the target; draft vocab size == vocab size (no d2t/t2d). Loads under vLLM (native laguna dflash ) and TRT LLM (pytorch DFlash backend) as the draft model in a speculative config. Usage vLLM SGLang llama.cpp (GGUF) A llama.cpp conversion of this draft model ( laguna s 2.1 DFlash BF16.gguf ) is published in poolside/Laguna S 2.1 GGUF alongside the target GGUFs. It embeds the target tokenizer and the DFlash metadata ( dflash.decoder arch = laguna , capture layers, block size), so it works directly as the md draft model: spec draft n max is clamped to the trained block size (15 draft tokens + 1). [!NOTE] Requires Poolside's llama.cpp fork, branch laguna . Upstream llama.cpp ships the generic DFlash framework but not the Laguna decoder contract this draft model needs, and upstream PR ggml org/llama.cpp 25165 covers the target architecture only.
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy