FaseehGPT-streamlit

A modern Streamlit web app for generating high-quality Arabic text using the FaseehGPT model by Alphatechlogics.

---

🚀 Features

• Arabic text generation powered by FaseehGPT (decoder-only GPT-style transformer)
• Adjustable generation parameters: max tokens, temperature, top-k, top-p
• Clean, user-friendly UI with documentation and sample outputs
• Model details, usage, limitations, and citation info included

---

🧠 About FaseehGPT

• Type: Decoder-only Transformer (GPT-style)
• Language: Arabic
• Parameters: ~70.7M | Layers: 12 | Embedding Dim: 512
• Tokenizer: asafaya/bert-base-arabic
• Trained on: arbml/Arabic_News, arbml/Arabic_Literature
• License: Apache 2.0

For more, see the official model card.

---

💡 Usage

1. Clone this repo and install requirements:

   git clone https://github.com/alphatechlogics/FaseehGPT-streamlit.git
   cd FaseehGPT-streamlit
   pip install -r requirements.txt

2. Run the app:

   streamlit run app.py

3. Open your browser to the provided local URL.

---

📝 Example

• Prompt: اللغة العربية
• Output:

  "اللغة العربية اقرب ويح الي كما ذلك هذه البيان شعره قاله الاستاذر من وتج معهم..."

---

⚙️ Parameters

• max_new_tokens: Max tokens to generate (e.g., 100)
• temperature: Controls randomness (default: 1.0)
• top_k: Limits sampling to top-k tokens (default: 50)
• top_p: Nucleus sampling threshold (default: 0.9)

---

⚠️ Limitations & Ethics

• Generated text may lack full coherence or contain errors due to limited training.
• Only supports Arabic; other languages untested.
• May reflect biases from source data.
• For research/non-commercial use; always validate outputs.

---

📖 Citation

@misc{faseehgpt2025,
  title     = {FaseehGPT: An Arabic Language Model},
  author    = {Rohma, Ahsan Umar},
  year      = {2025},
  url       = {https://huggingface.co/alphatechlogics/FaseehGPT}
}

---

🔗 Useful Links

• FaseehGPT on Hugging Face
• GitHub Source Code
• Alphatechlogics LinkedIn