Dampfinchen/Llama-3.1-8B-Ultra-Instruct

Dampfinchen/Llama-3.1-8B-Ultra-Instruct is an 8 billion parameter instruction-tuned language model based on the Llama 3.1 architecture, created by merging multiple fine-tuned models using the DARE TIES method. This model is optimized for general instruction following and demonstrates strong performance for its size, making it suitable for a wide range of conversational and text generation tasks. It leverages the combined strengths of its merged components to provide a robust 32K context window model.