Downloads · 30 days
28
8% of all-time downloads
Mohamed132411/Qwen3-4B-FitGPT-AR-EN-Instruct-GGUF
Qwen3-4B-FitGPT-AR-EN-Instruct-GGUF is a text generation model from Mohamed132411. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
<p align="center" <img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="180"/ </p
Downloads · 30 days
28
8% of all-time downloads
All-time downloads
357
Public
Repo size
2.5 GB
Likes
1
Public
Click a slice to open those files.
.gguf2.5 GB · 100%
From the Hugging Face model README
This repository contains the GGUF quantized version of Qwen3-4B-FitGPT-AR-EN-Instruct — a bilingual fitness AI model that delivers science-based workout and nutrition guidance in both Arabic and English, with built-in strict instruction-following for agent/JSON workflows.
This version uses Q4_K_M quantization — the recommended sweet spot between model size, inference speed, and output quality. Ideal for local deployment, offline use, and resource-constrained environments.
| Property | Value |
|---|---|
| Developed by | Mohamed Ramadan |
| Model Type | GGUF (Quantized for local inference) |
| Quantization | Q4_K_M |
| Source Model | Mohamed132411/Qwen3-4B-FitGPT-AR-EN-Instruct |
| Base Architecture | Qwen3-4B (Custom DARE-TIES Merge) |
| Languages | Arabic 🇸🇦 & English 🇺🇸 |
| License | Apache 2.0 |
ollama run hf.co/Mohamed132411/Qwen3-4B-FitGPT-AR-EN-Instruct-GGUF
That's it. Ollama handles the download and runtime automatically.
# Download the model file first, then:
./llama-cli \
--model Qwen3-4B-FitGPT-AR-EN-Instruct-Q4_K_M.gguf \
--system-prompt "You are an elite fitness coach. Give science-based, practical advice." \
-p "Create a beginner workout plan for someone who wants to build muscle."
Mohamed132411/Qwen3-4B-FitGPT-AR-EN-Instruct-GGUFQ4_K_M fileFitness Coach (English):
You are Qwen3-4B-FitGPT-AR-EN-Instruct, an elite fitness coach and sports nutritionist. Give science-based, detailed, personalised advice on training, nutrition, exercise technique, and recovery. Be specific and practical.
Fitness Coach (Arabic):
أنت Qwen3-4B-FitGPT-AR-EN-Instruct، مدرب لياقة بدنية نخبة وأخصائي تغذية رياضية. تقدّم نصائح علمية دقيقة ومخصصة في التدريب والتغذية وأداء التمارين والتعافي. كن تفصيلياً وعملياً.
Agent / JSON Mode:
You are a precise AI assistant. Follow every instruction exactly. If asked for JSON — ONLY valid JSON, no markdown, no explanation. Never add unsolicited commentary.
Arabic:
سؤال: أنا مبتدئ وزني 85 كغ، أريد خطة أسبوعية لبناء العضلات.
يرد الموديل بخطة تفصيلية بالعربي مع التمارين والتكرارات والراحة والتغذية.
English:
Prompt:
Create a high-protein meal plan for an 80 kg male in a lean bulk.Model responds with a detailed, macro-tracked meal plan.
Strict JSON:
Prompt:
Return ONLY JSON {exercise, sets, reps} for squats.Output:
{"exercise": "Barbell Squat", "sets": 4, "reps": 8}
This model is based on a custom-merged Qwen3-4B foundation, fine-tuned on ~7,000 curated samples covering:
See the full model card for complete training details.
| Format | Size (approx.) | Use Case |
|---|---|---|
| Q4_K_M | ~2.5 GB | ✅ Recommended — best balance |
| Q5_K_M | ~2.9 GB | Higher quality, slightly slower |
| Q8_0 | ~4.2 GB | Near-original quality |
This repo currently provides Q4_K_M only.
| Repo | Description |
|---|---|
| 🔗 Qwen3-4B-FitGPT-AR-EN-Instruct | Full 16-bit merged model |
| ⚡ Qwen3-4B-FitGPT-AR-EN-Instruct-GGUF | This repo — quantized for local inference |
Built with ❤️ by Mohamed Ramadan using Unsloth + Hugging Face