dnotitia/Llama-DNA-1.0-8B-Instruct

dnotitia/Llama-DNA-1.0-8B-Instruct is an 8 billion parameter bilingual language model developed by Dnotitia Inc., based on the Llama architecture with a 131,072 token context length. Optimized for Korean language understanding and generation, it also maintains strong English capabilities. The model was created through SLERP merging with Llama 3.1 8B Instruct, knowledge distillation using Llama 3.1 405B, and extensive continual pre-training on a high-quality Korean dataset. It excels in Korean-specific benchmarks like KMMLU, KoBEST, and Belebele, often outperforming similar-sized models.