khaled314/qwen25-saudi-v4

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:May 19, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

The khaled314/qwen25-saudi-v4 is a 7.6 billion parameter causal language model, fine-tuned from Qwen/Qwen2.5-7B-Instruct. This model is specifically optimized for tasks involving the Saudi dialect, having been trained on the saudi_dialect_train dataset. It is designed to enhance performance in applications requiring understanding and generation of text in this specific Arabic dialect, leveraging a 32768 token context length.

Loading preview...

Overview

The khaled314/qwen25-saudi-v4 model is a specialized large language model, fine-tuned from the robust Qwen/Qwen2.5-7B-Instruct base. With 7.6 billion parameters and a substantial 32768 token context length, its primary distinction lies in its targeted training on the saudi_dialect_train dataset. This focused fine-tuning aims to significantly improve its proficiency and accuracy when processing or generating content in the Saudi dialect.

Key Capabilities

  • Saudi Dialect Proficiency: Enhanced understanding and generation of text specific to the Saudi dialect.
  • Instruction Following: Inherits instruction-following capabilities from its Qwen2.5-7B-Instruct base.
  • Large Context Window: Benefits from a 32768 token context length, suitable for handling longer inputs and maintaining conversational coherence.

Training Details

The model was trained with a learning rate of 2e-05 over 3 epochs, utilizing an AdamW optimizer and a cosine learning rate scheduler. It achieved a training loss of 4.6754 on the evaluation set.

Good For

  • Applications requiring high accuracy in Saudi dialect text processing.
  • Chatbots or virtual assistants designed for users in Saudi Arabia.
  • Content generation, translation, or analysis tasks specifically targeting the Saudi dialect.