msmarco-passage-cos-dpr-distil-hnsw-int8.template

# Anserini Regressions: MS MARCO Passage Ranking

**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries)

This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper:

> Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom.

In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding).

The exact configurations for these regressions are stored in [this YAML file](${yaml}).
Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation.

From one of our Waterloo servers (e.g., `orca`), the following command will perform the complete regression, end to end:

```bash
python src/main/python/run_regression.py --index --verify --search --regression ${test_name}
```

We make available a version of the MS MARCO Passage Corpus that has already been encoded with cosDPR-distil.

From any machine, the following command will download the corpus and perform the complete regression, end to end:

```bash
python src/main/python/run_regression.py --download --index --verify --search --regression ${test_name}
```

The `run_regression.py` script automates the following steps, but if you want to perform each step manually, simply copy/paste from the commands below and you'll obtain the same regression results.

## Corpus Download

Download the corpus and unpack into `collections/`:

```bash
wget ${download_url} -P collections/
tar xvf collections/${corpus}.tar -C collections/
```

To confirm, `${corpus}.tar` is 57 GB and has MD5 checksum `${download_checksum}`.
With the corpus downloaded, the following command will perform the remaining steps below:

```bash
python src/main/python/run_regression.py --index --verify --search --regression ${test_name} \
  --corpus-path collections/${corpus}
```

## Indexing

Sample indexing command, building HNSW indexes:

```bash
${index_cmds}
```

The path `/path/to/${corpus}/` should point to the corpus downloaded above.
Upon completion, we should have an index with 8,841,823 documents.

Note that here we are explicitly using Lucene's `NoMergePolicy` merge policy, which suppresses any merging of index segments.
This is because merging index segments is a costly operation and not worthwhile given our query set.
Furthermore, we are using Lucene's [Automatic Byte Quantization](https://www.elastic.co/search-labs/blog/articles/scalar-quantization-in-lucene) feature, which increase the on-disk footprint of the indexes since we're storing both the int8 quantized vectors and the float32 vectors, but only the int8 quantized vectors need to be loaded into memory.
See [issue #2292](https://github.com/castorini/anserini/issues/2292) for some experiments reporting the performance impact.

## Retrieval

Topics and qrels are stored [here](https://github.com/castorini/anserini-tools/tree/master/topics-and-qrels), which is linked to the Anserini repo as a submodule.
The regression experiments here evaluate on the 6980 dev set questions; see [this page](${root_path}/docs/experiments-msmarco-passage.md) for more details.

After indexing has completed, you should be able to perform retrieval as follows using HNSW indexes:

```bash
${ranking_cmds}
```

Evaluation can be performed using `trec_eval`:

```bash
${eval_cmds}
```

## Effectiveness

With the above commands, you should be able to reproduce the following results:

${effectiveness}

Note that due to the non-deterministic nature of HNSW indexing, results may differ slightly between each experimental run.
Nevertheless, scores are generally within 0.005 of the reference values recorded in [our YAML configuration file](${yaml}).

## Reproduction Log[*](${root_path}/docs/reproducibility.md)

To add to this reproduction log, modify [this template](${template}) and run `bin/build.sh` to rebuild the documentation.

+ Results reproduced by [@yilinjz](https://github.com/yilinjz) on 2023-09-01 (commit [`4ae518b`](https://github.com/castorini/anserini/commit/4ae518bb284ebcba0b273a473bc8774735cb7d19))