Upload lucene-inverted/tf/lucene-inverted.bright.20250705.44ae8e.README.md with huggingface_hub

lucene-inverted/tf/lucene-inverted.bright.20250705.44ae8e.README.md

@@ -0,0 +1,52 @@
+# BRIGHT
+
+The Lucene inverted indexes for BRIGHT were generated on 2025/07/05 at Anserini commit [`44ae8e`](https://github.com/castorini/anserini/commit/44ae8e487760a2cd21bdcdb11e18735e46606a86) on `orca` with the following commands:
+
+```bash
+#!/bin/bash
+
+for c in biology earth-science economics psychology robotics stackoverflow sustainable-living pony leetcode aops theoremqa-theorems theoremqa-questions
+do
+
+bin/run.sh  io.anserini.index.IndexCollection \
+  -threads 16 \
+  -collection JsonCollection \
+  -input collections/bright/${c}/  \
+  -generator DefaultLuceneDocumentGenerator \
+  -index indexes/bright/lucene-inverted.bright-$c.20250705.44ae8e \
+  -storePositions -storeDocvectors -storeRaw -optimize
+
+done
+```
+
+Documents were generated with the following Python script:
+
+```python
+from datasets import load_dataset
+import json
+from tqdm import tqdm
+import os
+
+bright_corpus = load_dataset("xlangai/BRIGHT", 'documents')
+splits = list(bright_corpus.keys())
+print(splits)
+
+for split in splits:
+    exist_docids = set()
+    os.makedirs(f'collections/bright/{split}', exist_ok=True)
+    corpus = bright_corpus[split]
+    print(split)
+    with open(f'collections/bright/{split}/{split}.jsonl', 'w') as f:
+        for doc in tqdm(corpus):
+            id_ = doc['id']
+            # replace spaces with underscores
+            id_ = id_.replace(' ', '_')
+            contents = doc['content']
+            if id_ in exist_docids:
+                print(f'Duplicate document id found: {id_}')
+                continue
+            exist_docids.add(id_)
+            f.write(json.dumps({'id': id_, 'contents': contents}) + '\n')
+```
+
+The processed documents can be found on Hugging Face [here](https://huggingface.co/datasets/castorini/collections-bright).