Modelpublic

ManniX-ITA/gemma-4-31B-it-assistant-GGUF

sourceHugging Facegemmaupdated 11d agoView on Hugging Face

1likes626downloads

Model Card

Gemma-4 31B IT Assistant — MTP draft head (llama.cpp `gemma4-assistant` format)

MTP / speculative-decoding draft head for gemma-4-31B-it, in the upstream llama.cpp gemma4-assistant GGUF data-plane format (arch gemma4-assistant, with nextn.* (+ masked_embd_* where applicable) tensors and embedding_length_out).

Use it as the drafter for self-speculative decoding against the matching gemma-4-31B-it target:

--mtp-head gemma-4-31B-it-assistant.Q8_0.gguf --spec-type draft-assistant -ngld 99

Quants: F16, Q80, Q5KM, Q4KM, Q4K_S.

Adapted from AtomicChat/gemma-4-31B-it-assistant-GGUF. Weights byte-identical to the source GGUFs — only the GGUF metadata/tensor names were rewritten to the upstream gemma4-assistant data-plane format. The Gemma license applies.

Gemma-4 31B IT Assistant — MTP draft head (llama.cpp gemma4-assistant format)

Gemma-4 31B IT Assistant — MTP draft head (llama.cpp `gemma4-assistant` format)