Reading: Methodology Sections — English for Researchers

4. Experimental Setup

We compare our proposed method against three baselines: a random-selection heuristic, the FIFO cache-eviction policy, and LRU-K, the current state-of-the-art for this workload class. All experiments are run on identical hardware (8 vCPUs, 32GB RAM) to control for confounding variables. We hold out 20% of each dataset as a test set that is never used during training or hyperparameter tuning, following standard practice to prevent data leakage. Hyperparameters for each baseline are tuned via grid search on a separate validation set, ensuring a fair comparison rather than reporting each baseline's default configuration. Each experiment is repeated across five random seeds, and we report the mean and standard deviation to account for variance in the results. All code, configuration files, and the exact commit hash used for these experiments are released to support reproducibility.