EvoLlama: Enhancing LLMs' Understanding of Proteins via Multimodal Structure and Sequence Representations