GraySoft
Projects Models Compare Cloud benchmarks FAQ Download guIDE →
Model Intelligence Sheet

MachadoDeCastro/krull-7b.Q8_0.gguf overview

KRULL 7B Specialist in RAG and Manuals GGUF KRULL is a lightweight language model 7 billion parameters optimized for inference on Edge Devices standard CPUs, l…

ggufollamaopen-webuiragreasoningptbase_model:deepseek-ai/DeepSeek-R1-Distill-Qwen-7Bbase_model:quantized:deepseek-ai/DeepSeek-R1-Distill-Qwen-7Blicense:apache-2.0endpoints_compatibleregion:usconversational

Runs locally from ~7.54 GB disk (8 GB VRAM class GPUs with llama.cpp / guIDE).

Downloads
101
Likes
0
Pipeline

Repository Files & Downloads

1 GGUF files detected
Direct downloads for local inference
FileTypeQuantizationSizeLink
krull-7b.Q8_0.ggufGGUFGGUF7.54 GBDownload

Model Details

Model IDMachadoDeCastro/krull-7b.Q8_0.gguf
AuthorMachadoDeCastro
Pipeline
Licenseapache-2.0
Base modeldeepseek-ai/DeepSeek-R1-Distill-Qwen-7B
Last modified2026-08-16T14:11:24.000Z

Model README

---

language:

  • pt

license: apache-2.0

base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-7B

tags:

  • gguf
  • ollama
  • open-webui
  • rag
  • reasoning

---

KRULL 7B - Specialist in RAG and Manuals (GGUF)

KRULL is a lightweight language model (7 billion parameters) optimized for inference on Edge Devices (standard CPUs, laptops, and local servers with low memory).

It underwent fine-tuning (via QLoRA) from DeepSeek-R1-Distill-Qwen-7B with an absolute focus on RAG (Retrieval-Augmented Generation). The model is trained to read PDF documents, Human Resources (HR) manuals, and Public Servant rules, answering questions strictly and without hallucinating information outside the given context.

Model Details

  • Base Model: DeepSeek-R1-Distill-Qwen-7B
  • Tamanho: 7 Billion Parameters
  • Formato: GGUF (Optimized for CPU and RAM)
  • Quantização Recomendada: Q8_0 (8-bit) to preserve the quality of the Portuguese vocabulary.
  • Idioma Principal: Portuguese (pt-BR)
  • Especialidade: Context Analysis (RAG), PDF Reading, and Logical Reasoning (Chain of Thought).

How to Use (Ollama and Open WebUI)

You do not need to download the file manually. Hugging Face and Ollama have native integration.

Running via Ollama CLI (Terminal)

If you have Ollama installed, simply run the command below in your terminal to download and run the model instantly:

ollama run hf.co/MachadoDeCastro/krull-7b.Q8_0.gguf

Run MachadoDeCastro/krull-7b.Q8_0.gguf with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models

Source: Hugging Face · Compare models