Setup:

• Username: @prav0761
• Date: 2025-10-13
• Commit: 4a2f9a0
• OS: macOS Ventura
• Python 3.11
• Result: Everything worked successfully

1. Begin your journey here. worked perfectly with datasets downloading and data prep.
2. BM25 Baselines for MS MARCO Passage Ranking in Anserini chapter worked perfectly, just few changes
   2.1 I had a few issues. The index wouldn't build because of a memory issue, so I had to reduce the threads from 9 to 4, and it worked perfectly

bin/run.sh io.anserini.index.IndexCollection \
  -collection JsonCollection \
  -input collections/msmarco-passage/collection_jsonl \
  -index indexes/msmarco-passage/lucene-index-msmarco \
  -generator DefaultLuceneDocumentGenerator \
  -threads 9 -storePositions -storeDocvectors -storeRaw

2.2 The retrieval part didnt work because of memory issue as well, so had to change the run.sh command to

java -cp `ls target/*-fatjar.jar` -Xms512M -Xmx8G --add-modules jdk.incubator.vector $@

Where I have changed to Xmx8G from Xmx192G
2.3 After these changes, retrieval worked perfectly and all the scores matched the expected as well

3. Dense Retrieval for MS MARCO Passage Ranking in Anserini worked successfully, and scores matched the expected score.