|
|
--- |
|
|
languages: |
|
|
- ar |
|
|
- en |
|
|
license: mit |
|
|
tags: |
|
|
- pytorch |
|
|
- llama |
|
|
- text-generation |
|
|
- conversational |
|
|
- chatbot |
|
|
pipeline_tag: text-generation |
|
|
--- |
|
|
|
|
|
# Chat Model |
|
|
|
|
|
This is a custom chat model fine-tuned for conversational AI. The model is based on LLaMA architecture and is specifically designed for Arabic and English conversations. |
|
|
|
|
|
## Model Details |
|
|
- **Architecture**: LLaMA |
|
|
- **Task**: Text Generation |
|
|
- **Language**: Arabic/English |
|
|
- **License**: MIT |
|
|
- **Model Size**: Large |
|
|
- **Training Data**: Custom conversational data |
|
|
- **Optimization**: Quantized (int8) |
|
|
|
|
|
## Usage |
|
|
```python |
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
|
|
model = AutoModelForCausalLM.from_pretrained("AMRALMughira/chat-model") |
|
|
tokenizer = AutoTokenizer.from_pretrained("AMRALMughira/chat-model") |
|
|
|
|
|
# Example input |
|
|
input_text = "مرحبا كيف حالك؟" |
|
|
inputs = tokenizer(input_text, return_tensors="pt") |
|
|
outputs = model.generate( |
|
|
**inputs, |
|
|
max_length=256, |
|
|
temperature=0.7, |
|
|
do_sample=True, |
|
|
top_p=0.95 |
|
|
) |
|
|
response = tokenizer.decode(outputs[0], skip_special_tokens=True) |
|
|
``` |
|
|
|
|
|
## Inference API |
|
|
This model is compatible with Hugging Face's Inference API. You can use it with the following endpoint: |
|
|
|
|
|
``` |
|
|
POST https://api-inference.huggingface.co/models/AMRALMughira/chat-model |
|
|
``` |
|
|
|
|
|
## Model Performance |
|
|
- Optimized for conversational tasks |
|
|
- Supports both Arabic and English |
|
|
- Fast response times |
|
|
- High-quality responses |
|
|
|
|
|
## Requirements |
|
|
- PyTorch |
|
|
- Transformers |
|
|
- CUDA (optional for GPU acceleration) |
|
|
|
