Accelerating AI and HPC with Wafer-Scale Compute from Cerebras Systems