Egyptian Arabic Translator Llama-3-8B Overview
This model, developed by ahmedsamirio, is a fine-tuned version of Meta's Llama-3-8B, specifically adapted for translation tasks. It focuses on bridging the language gap between English and Egyptian Arabic, making it a specialized tool for developers working with these languages.
Key Capabilities
- Specialized Translation: Primarily designed for translating text from English to Egyptian Arabic and vice-versa.
- Llama 3 Architecture: Built upon the robust Meta-Llama-3-8B base model, inheriting its general language understanding capabilities.
- Instruction-Tuning: Fine-tuned on the
ahmedsamirio/oasst2-9k-translation dataset, which includes instruction fine-tuning examples, enhancing its ability to follow translation instructions. - Parameter Efficient Fine-Tuning (PEFT): Utilizes LoRA (r=32, alpha=16) for efficient adaptation of the base model.
Intended Uses & Limitations
- Good for: Translating instruction fine-tuning and text generation datasets between English and Egyptian Arabic.
- Training Details: Trained for 2 epochs with a learning rate of 2e-5, showing a validation loss reduction from 1.3816 to 0.6798 over the training period.
This model offers a targeted solution for developers needing accurate and context-aware translation into Egyptian Arabic, particularly for AI model training and data preparation.