How I Reproduced BM25, Dense Retrieval, and SPLADE on a 16GB MacBook

A researcher successfully reproduced BM25, dense retrieval, and SPLADE document retrieval methods on a 16GB MacBook using Castorini's Anserini and Pyserini toolkits, matching expected benchmark scores. The process revealed practical challenges like memory crashes and gated model access issues not documented officially. This work provides a reliable baseline for evaluating retrieval quality in systems like RAG, emphasizing the importance of verifying document ranking before other enhancements.