anakin87/Phi-3.5-mini-ITA

anakin87/Phi-3.5-mini-ITA is a 3.82 billion parameter language model, fine-tuned from Microsoft's Phi-3.5-mini-instruct, specifically optimized for enhanced performance in Italian. It supports a 128k context length and leverages the Spectrum technique for efficient training. This model excels in Italian language tasks, outperforming larger models like Meta-Llama-3.1-8B-Instruct on the Open ITA LLM and Pinocchio ITA Leaderboards.

Warm
Public
4B
BF16
4096
License: mit
Hugging Face