ReDiX/Qwen2.5-0.5B-Instruct-ITA

ReDiX/Qwen2.5-0.5B-Instruct-ITA is a 0.5 billion parameter instruction-tuned causal language model, fine-tuned by ReDiX from Qwen/Qwen2.5-0.5B-Instruct. This model specializes in Italian language tasks, demonstrating improved performance on Italian evaluation benchmarks such as ARC-IT, HellaSwag-IT, and M-MMLU-IT. It is designed for applications requiring a compact, efficient model with enhanced Italian language understanding and generation capabilities.