Views
No views yet
| Parameter | Value |
|---|---|
| Layers | 6 |
| Hidden Size | 512 |
| Attention Heads | 8 |
| FFN Dimension | 2048 |
| Max Sequence Length | 512 |
| Vocabulary Size | 128,004 |
| Training Data | 1,000 curated Arabic-English conversation pairs |
1from transformers import pipeline
2
3# Initialize chat pipeline
4chatbot = pipeline("text-generation", model="BINOMDA/OMDA")
5
6# Arabic input example
7ar_responser = chatbot("ما هو رأيك في التكنولوجيا الحديثة؟")
8
9# English input example
10en_response = chatbot("Explain artificial intelligence simply")
11
12## Intended Use
13- Chatbots, assistants, translation, and educational tools for Arabic/English.
14
15## Training
16- Trained for 5 epochs on 1000 samples.
17- Loss curve and checkpoints included.
18
19## Limitations
20- This is a small-scale demonstration model and may not generalize well to all real-world chat scenarios.
21- Not suitable for production use without further scaling, extensive evaluation, and safety checks.
22- Limited training data and model size may result in hallucinations or inaccurate translations.
23- No advanced filtering for inappropriate or biased outputs.
24- For research and educational purposes only.
25
26## Export & Deployment
27- See below for HuggingFace, llama.cpp, and ollama export