Predicting Algorithm Runtime Distributions: An In-Context Learning Approach with TabPFN

A Structure-Aware RAG system for predicting HPC job runtime using Hierarchical Retrieval (H-CORA) and LLM-based reasoning to handle Cold Start problems

Exploring HPC runtime prediction. Demonstrates how to leverage job submission parameters and historical data to build runtime forecasts. Training a Model, and serving it as a MicroService using gRPC - Demo implementation from data engineering to deployment.