llm-parallelism-explorer-poc. It is a cutting-edge research tool designed to optimize parallelism strategies for large language models, with a particular focus on Mixture of Experts (MoE) architectures.