Run LLMs on weak devices or make powerful devices even more powerful by distributing the workload and dividing the RAM usage. - View it on GitHub