---

license: apache-2.0

language:

• multilingual
• pl
• en
• sq
• bel
• bs
• bg
• hr
• cs
• da
• et
• fi
• fr
• el
• es
• is
• lt
• nl
• de
• no
• pt
• ru
• ro
• sr
• hbs
• sv
• sk
• sl
• tr
• uk
• hu
• it
• lv

library_name: transformers

pipeline_tag: text-generation

tags:

• dflash
• speculative-decoding
• block-diffusion
• draft-model
• efficiency
• bielik
• diffusion-language-model

base_model:

• speakleash/Bielik-Minitron-7B-v3.0-DFlash

---

Bielik-Minitron-7B-v3.0-DFlash-GGUF

This repository provides GGUF files with F16 and Q8_0 quantization of Bielik-Minitron-7B-v3.0-DFlash model by SpeakLeash for use in llama.cpp.

Bielik-Minitron-7B-v3.0-DFlash is a DFlash draft model designed for use with Bielik-Minitron-7B-v3.0-Instruct.

Its development and training were supported by computational grant PLG/2026/019559 and carried out on the Helios supercomputer, providing access to the advanced infrastructure and computational resources required for large-scale machine learning workloads.

This is not a standalone language model. It is designed to be paired with the target model in a speculative decoding setup.

How to use

Install llama.cpp and one of the GGUF files in the root directory, then run Bielik Minitron like this:

llama-cli -m Bielik-Minitron-7B-v3.0-Instruct-Q8_0.gguf --spec-type draft-dflash --spec-draft-model Bielik-Minitron-7B-v3.0-DFlash-Q8_0.gguf --spec-draft-ngl all --spec-draft-n-min 16 -fa on