upload hf.co/datasets/nyu-visionx/VSI-Bench-Debiased version

README.md

@@ -9,60 +9,175 @@ tags:
 - Text
 size_categories:
 - 1K<n<10K
+configs:
+  - config_name: full
+    data_files:
+      - split: test
+        path: "test*.parquet"
+    default: true
+  - config_name: debiased
+    data_files:
+      - split: test
+        path: "test_debiased.parquet"
+  - config_name: pruned
+    data_files:
+      - split: test
+        path: "test_pruned.parquet"
 ---
 
-<a href="https://arxiv.org/abs/2412.14171" target="_blank">
-    <img alt="arXiv" src="https://img.shields.io/badge/arXiv-thinking--in--space-red?logo=arxiv" height="20" />
-</a>
-<a href="https://vision-x-nyu.github.io/thinking-in-space.github.io/" target="_blank">
-    <img alt="Website" src="https://img.shields.io/badge/🌎_Website-thinking--in--space-blue.svg" height="20" />
-</a>
-<a href="https://github.com/vision-x-nyu/thinking-in-space" target="_blank" style="display: inline-block; margin-right: 10px;">
-    <img alt="GitHub Code" src="https://img.shields.io/badge/Code-thinking--in--space-white?&logo=github&logoColor=white" />
-</a>
+<!-- <div align="center"> -->
 
+| Dataset | arXiv | Website | Code |
+| :------ | :---- | :------ | :--- |
+| **VSI-Bench** | <a href="https://arxiv.org/abs/2412.14171" target="_blank"><img alt="arXiv" src="https://img.shields.io/badge/arXiv-thinking--in--space-red?logo=arxiv" height="20" /></a> | <a href="https://vision-x-nyu.github.io/thinking-in-space.github.io/" target="_blank"><img alt="Website" src="https://img.shields.io/badge/🌎_Website-thinking--in--space-blue.svg" height="20" /></a> | <a href="https://github.com/vision-x-nyu/thinking-in-space" target="_blank"><img alt="GitHub Code" src="https://img.shields.io/badge/Code-thinking--in--space-white?&logo=github&logoColor=white" /></a> |
+| **VSI-Bench-Debiased** | <a href="https://arxiv.org/abs/2511.04655" target="_blank"><img alt="arXiv" src="https://img.shields.io/badge/arXiv-test--set--stress--test-red?logo=arxiv" height="20" /></a> | <a href="https://vision-x-nyu.github.io/test-set-training/" target="_blank"><img alt="Website" src="https://img.shields.io/badge/🌎_Website-test--set--stress--test-blue.svg" height="20" /></a> | <a href="https://github.com/vision-x-nyu/test-set-training" target="_blank"><img alt="GitHub Code" src="https://img.shields.io/badge/Code-test--set--stress--test-white?&logo=github&logoColor=white" /></a> |
 
-# Visual Spatial Intelligence Benchmark (VSI-Bench)
-This repository contains the visual spatial intelligence benchmark (VSI-Bench), introduced in [Thinking in Space: How Multimodal Large Language Models See, Remember and Recall Spaces](https://arxiv.org/pdf/). 
+<!-- </div> -->
 
+<br>
 
-## Files
-The `test-00000-of-00001.parquet` file contains the complete dataset annotations and pre-loaded images, ready for processing with HF Datasets. It can be loaded using the following code:
+> [!IMPORTANT]
+> ***[Nov. 7, 2025] UPDATE:** This Dataset has been updated to include a "Debiased" subset following the [TsT Pruning Methodology](https://vision-x-nyu.github.io/test-set-training/)*
 
-```python
-from datasets import load_dataset
-vsi_bench = load_dataset("nyu-visionx/VSI-Bench")
-```
-Additionally, we provide the videos in `*.zip`.
+<br>
+
+# Visual-Spatial Intelligence Benchmark (VSI-Bench & VSI-Bench-Debiased)
+
+This repository contains the visual spatial intelligence benchmark (VSI-Bench), introduced in [Thinking in Space: How Multimodal Large Language Models See, Remember and Recall Spaces](https://arxiv.org/abs/2412.14171), and its debiased counterpart **VSI-Bench-Debiased**, introduced in our follow-up work on systematic benchmark robustification [Benchmark Designers Should "Train on the Test Set" to Expose Exploitable Non-Visual Shortcuts](https://arxiv.org/abs/2511.04655).
+
+## Overview
 
-## Dataset Description
+**VSI-Bench** evaluates visual-spatial intelligence of multimodal models through egocentric video understanding, comprising over 5,000 question-answer pairs from real-world indoor scenes.
+
+**VSI-Bench-Debiased** is a robustified version that reduces non-visual shortcuts using our Test-set Stress-Test (TsT) and Iterative Bias Pruning (IBP) methodology. This version better isolates visual reasoning capabilities by systematically removing samples that can be solved without visual input.
+
+### Description
 VSI-Bench quantitatively evaluates the visual-spatial intelligence of MLLMs from egocentric video. VSI-Bench comprises over 5,000 question-answer pairs derived from 288 real videos. These videos are sourced from the validation sets of the public indoor 3D scene reconstruction datasets `ScanNet`, `ScanNet++`, and `ARKitScenes`, and represent diverse environments -- including residential spaces, professional settings (e.g., offices, labs), and industrial spaces (e.g., factories) and multiple geographic regions. By repurposing these existing 3D reconstruction and understanding datasets, VSI-Bench benefits from accurate object-level annotations, which are used in question generation and could support future studies exploring the connection between MLLMs and 3D reconstruction.
 
+
+#### Fields
+
 The dataset contains the following fields:
 
 | Field Name | Description |
 | :--------- | :---------- |
-| `idx` | Global index of the entry in the dataset |
+| `id` | Global index of the entry in the dataset |
 | `dataset` | Video source: `scannet`, `arkitscenes` or `scannetpp` |
 | `scene_name` | Scene (video) name for each question-answer pair |
 | `question_type` | The type of task for question |
 | `question` | Question asked about the video |
 | `options` | Choices for the question (only for multiple choice questions) |
 | `ground_truth` | Ground truth answer for the question |
+| `pruned` | Boolean indicating if example was removed by Iterative Bias Pruning (IBP) |
+
+### Why VSI-Bench-Debiased?
+
+While the original VSI-Bench was designed to require visual understanding, our follow-up analysis revealed that a portion of questions could be answered using non-visual shortcuts—such as statistical biases in answer distributions or world knowledge priors—without actually processing the visual input.
+
+**VSI-Bench-Debiased** addresses this through systematic robustification:
+
+1. **Test-set Stress-Test (TsT)**: We applied k-fold cross-validation directly on the test set to identify samples with high non-visual solvability, assigning each sample a bias score.
+2. **Iterative Bias Pruning (IBP)**: We iteratively removed samples with the highest bias scores, creating a subset that better compels genuine visual reasoning.
+
+**Key improvements in VSI-Bench-Debiased:**
+- **Reduced non-visual solvability**: Blind models (text-only, no vision) perform closer to chance
+- **Wider vision-blind gap**: Greater performance difference between vision-enabled and vision-disabled models
+- **Better isolation of visual reasoning**: Fine-tuning on in-distribution data improves vision-enabled performance much more than blind performance, confirming reduced shortcut reliance
+
+For researchers interested in robust evaluation of visual-spatial intelligence, **we recommend reporting results on both the full and debiased subsets** to provide comprehensive assessment.
+
+## Usage
+
+### Dataset Configurations
+
+This dataset provides three configurations for flexible evaluation:
+
+| Config | Description | Usage |
+|--------|-------------|-------|
+| `full` (default) | All 5,131 examples with `pruned` column | Load all data, filter as needed |
+| `debiased` | 2,363 examples (non-pruned subset) | Evaluate on robustified benchmark |
+| `pruned` | 2,768 examples (pruned by IBP) | Analyze removed samples |
+
+#### Loading the Dataset Annotations
+
+##### Load specific configuration
+
+If you want to load just a specific subset, you can use the config name with the `load_dataset` function as follows:
+
+```python
+from datasets import load_dataset
+
+# Load full dataset (default)
+vsi_bench_full = load_dataset("nyu-visionx/VSI-Bench")
+# or use the config name "full"
+vsi_bench_full = load_dataset("nyu-visionx/VSI-Bench", "full")
+
+# Load debiased version only
+vsi_bench_debiased = load_dataset("nyu-visionx/VSI-Bench", "debiased")
+
+# Load pruned examples only
+vsi_bench_pruned = load_dataset("nyu-visionx/VSI-Bench", "pruned")
+```
+
+##### Load full dataset and filter using `pruned` column (recommended)
+
+> [!TIP]
+> **For LMMS-Eval users:** We have updated the `vsi-bench` task to automatically report scores on both full and debiased subsets. (TODO: LINK).
+
+We recommend loading the "full" set, evaluating on all samples, and then using the `pruned` column to compute scores on both the full and debiased subsets.
+
+```python
+from datasets import load_dataset
+
+# Load full dataset with pruned annotations
+vsi_bench_full = load_dataset("nyu-visionx/VSI-Bench")
 
-## Evaluation
+# Evaluate on full set
+model_predictions = evaluate_model(vsi_bench_full)
+
+# Score on both the full and debiased subsets
+full_acc = compute_accuracy(model_predictions)
+debiased_acc = compute_accuracy(model_predictions.filter(lambda x: not x["pruned"]))
+```
+
+### Evaluation
+
+> [!TIP]
+> ***TODO: link to the LMMS Eval Code***
 
 VSI-Bench evaluates performance using two metrics: for multiple-choice questions, we use `Accuracy`, calculated based on exact matches. For numerical-answer questions, we introduce a new metric, `MRA (Mean Relative Accuracy)`, to assess how closely model predictions align with ground truth values.
 
 We provide an out-of-the-box evaluation of VSI-Bench in our [GitHub repository](https://github.com/vision-x-nyu/thinking-in-space), including the [metrics](https://github.com/vision-x-nyu/thinking-in-space/blob/main/lmms_eval/tasks/vsibench/utils.py#L109C1-L155C36) implementation used in our framework. For further detailes, users can refer to our paper and GitHub repository.
 
+## Files
+
+- `test-*.parquet`: Parquet files containing dataset annotations (questions, answers, metadata).
+  * `test_debiased.parquet`: Annotations for the debiased subset (2,363 examples)
+  * `test_pruned.parquet`: Annotations for the pruned subset (2,768 examples)
+- `*.zip`: Compressed video files for the dataset
+  * `arkitscenes.zip`: Videos for the ARKitScenes dataset
+  * `scannet.zip`: Videos for the ScanNet dataset
+  * `scannetpp.zip`: Videos for the ScanNet++ dataset
+- `pruned_ids.txt`: List of example IDs removed by Iterative Bias Pruning
+- `create_pq.py`: Convenience script to regenerate parquet files from `test.jsonl` and `pruned_ids.txt`. Can be run with `uv run create_pq.py`.
+
+
 ## Citation
 
+If you use these datasets in your research, please cite the original VSI-Bench paper and our debiasing paper that produced VSI-Bench-Debiased:
+
 ```bibtex
-@article{yang2024think,
+@inproceedings{yang2025thinking,
     title={{Thinking in Space: How Multimodal Large Language Models See, Remember and Recall Spaces}},
     author={Yang, Jihan and Yang, Shusheng and Gupta, Anjali and Han, Rilyn and Fei-Fei, Li and Xie, Saining},
-    year={2024},
-    journal={arXiv preprint arXiv:2412.14171},
+    booktitle={CVPR},
+    year={2025},
 }
-```
+
+@article{brown2025benchmark,
+    title={{Benchmark Designers Should "Train on the Test Set" to Expose Exploitable Non-Visual Shortcuts}},
+    author={Brown, Ellis and Yang, Jihan and Yang, Shusheng and Fergus, Rob and Xie, Saining},
+    year={2025},
+    journal={arXiv preprint arXiv:2511.04655},
+}
+```

create_pq.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python3
+# /// script
+# requires-python = ">=3.10"
+# dependencies = [
+#     "fastparquet",
+#     "pandas",
+#     "pathlib",
+#     "pyarrow",
+# ]
+# ///
+"""
+Create parquet files for config subsets of the VSI-Bench dataset.
+* debiased: all examples not pruned by Iterative Bias Pruning (aka VSI-Bench-Debiased)
+* pruned: all examples pruned by Iterative Bias Pruning
+
+> [!NOTE]
+> If you do not pass `index=False`, the parquet files will have a `__index_level_0__` column
+"""
+
+import pandas as pd
+from pathlib import Path
+
+script_dir = Path(__file__).parent
+pruned_ids_path = script_dir / "pruned_ids.txt"
+test_jsonl_path = script_dir / "test.jsonl"
+pq_debiased_path = script_dir / "test_debiased.parquet"
+pq_pruned_path = script_dir / "test_pruned.parquet"
+
+print("Creating parquet files...")
+
+print(f"Loading pruned ids from '{pruned_ids_path}'...")
+with open(pruned_ids_path, "r") as f:
+    pruned_ids = f.read().splitlines()
+print(f"    -> Loaded {len(pruned_ids)} pruned ids.")
+
+print(f"Loading test data from '{test_jsonl_path}'...")
+df = pd.read_json(str(test_jsonl_path), lines=True)
+print(f"    -> Loaded {len(df)} examples.")
+df["pruned"] = df["id"].astype(str).isin(pruned_ids)
+print(f"    -> Added pruned column.")
+
+# save the debiased and pruned subsets separately to parquet files
+df_debiased = df[~df["pruned"]]
+df_pruned = df[df["pruned"]]
+
+print(f"Saving debiased examples to '{pq_debiased_path}'...")
+df_debiased.to_parquet(pq_debiased_path, index=False)
+print(f"    -> Saved {len(df_debiased)} debiased examples.")
+
+print(f"Saving pruned examples to '{pq_pruned_path}'...")
+df_pruned.to_parquet(pq_pruned_path, index=False)
+print(f"    -> Saved {len(df_pruned)} pruned examples.")
+
+print("Done.") 

pruned_ids.txt

@@ -0,0 +1,2768 @@
+1
+2
+3
+4
+5
+7
+8
+9
+10
+13
+14
+15
+16
+20
+21
+22
+23
+24
+26
+27
+28
+29
+30
+31
+32
+33
+34
+35
+39
+41
+43
+44
+45
+46
+47
+48
+50
+52
+53
+54
+55
+56
+57
+58
+59
+61
+62
+63
+64
+65
+66
+67
+69
+70
+72
+73
+74
+75
+77
+78
+79
+81
+82
+83
+84
+86
+89
+91
+92
+93
+94
+95
+96
+97
+98
+100
+102
+103
+105
+106
+107
+108
+109
+111
+115
+116
+119
+120
+122
+123
+124
+125
+126
+127
+129
+130
+131
+132
+133
+135
+136
+139
+141
+142
+143
+144
+145
+146
+147
+148
+150
+151
+152
+153
+154
+155
+156
+157
+159
+160
+161
+162
+164
+165
+167
+168
+169
+170
+171
+172
+173
+174
+175
+176
+177
+178
+179
+181
+184
+185
+187
+188
+190
+191
+192
+193
+194
+197
+198
+199
+200
+201
+202
+203
+204
+205
+206
+207
+208
+209
+210
+211
+213
+214
+215
+216
+218
+219
+222
+224
+226
+228
+229
+231
+232
+233
+234
+236
+237
+238
+241
+242
+243
+244
+245
+246
+247
+248
+249
+250
+251
+253
+254
+255
+256
+259
+261
+262
+263
+264
+266
+267
+270
+271
+272
+273
+275
+276
+277
+278
+279
+280
+281
+282
+284
+286
+287
+288
+289
+291
+292
+293
+294
+295
+297
+299
+300
+301
+302
+303
+304
+305
+306
+307
+308
+309
+310
+311
+312
+313
+316
+317
+318
+319
+320
+321
+322
+324
+325
+326
+327
+328
+329
+330
+332
+334
+335
+336
+338
+339
+341
+342
+343
+344
+345
+346
+347
+348
+349
+350
+351
+352
+353
+355
+356
+357
+358
+359
+360
+362
+363
+364
+365
+368
+369
+371
+373
+375
+376
+377
+378
+379
+380
+382
+383
+384
+386
+388
+393
+394
+395
+397
+399
+400
+402
+403
+404
+406
+407
+408
+409
+410
+411
+412
+414
+417
+418
+420
+421
+424
+426
+427
+428
+430
+432
+433
+434
+435
+436
+437
+438
+439
+440
+442
+443
+444
+445
+446
+447
+448
+449
+450
+451
+452
+455
+456
+459
+461
+462
+463
+464
+465
+466
+468
+469
+470
+471
+472
+473
+474
+475
+476
+477
+478
+479
+481
+482
+484
+485
+486
+487
+488
+489
+490
+493
+494
+495
+496
+500
+501
+502
+503
+504
+505
+506
+508
+509
+511
+512
+513
+514
+515
+516
+517
+521
+522
+523
+524
+525
+526
+527
+528
+529
+531
+535
+537
+539
+540
+544
+547
+549
+551
+555
+557
+563
+564
+565
+568
+569
+576
+580
+585
+586
+590
+594
+608
+610
+612
+617
+618
+621
+622
+623
+627
+629
+632
+637
+640
+643
+646
+655
+658
+659
+663
+668
+671
+672
+675
+677
+678
+679
+680
+681
+682
+683
+684
+685
+686
+688
+690
+691
+693
+695
+697
+698
+704
+705
+706
+708
+711
+712
+713
+715
+716
+717
+718
+719
+722
+724
+726
+728
+731
+732
+733
+734
+738
+740
+741
+742
+744
+745
+746
+750
+755
+760
+761
+762
+764
+765
+766
+767
+772
+776
+777
+778
+783
+784
+785
+786
+787
+799
+801
+803
+804
+805
+806
+808
+809
+811
+813
+814
+817
+818
+821
+824
+826
+828
+830
+831
+832
+833
+835
+837
+838
+840
+841
+842
+845
+846
+847
+848
+851
+852
+853
+854
+855
+856
+857
+861
+862
+866
+871
+875
+877
+878
+881
+882
+883
+888
+892
+893
+894
+895
+900
+903
+904
+907
+910
+912
+913
+916
+917
+919
+920
+922
+926
+928
+933
+936
+938
+939
+940
+941
+942
+944
+945
+946
+948
+949
+952
+953
+954
+955
+956
+957
+958
+959
+960
+961
+962
+963
+964
+965
+967
+968
+970
+971
+973
+974
+975
+977
+978
+979
+980
+981
+982
+983
+984
+985
+986
+993
+995
+997
+998
+1000
+1001
+1004
+1005
+1006
+1008
+1010
+1012
+1013
+1014
+1015
+1017
+1018
+1020
+1021
+1022
+1023
+1024
+1026
+1029
+1032
+1034
+1035
+1038
+1043
+1044
+1045
+1046
+1047
+1049
+1050
+1051
+1052
+1054
+1055
+1057
+1061
+1063
+1064
+1066
+1068
+1069
+1073
+1074
+1075
+1078
+1079
+1080
+1081
+1082
+1083
+1084
+1085
+1086
+1087
+1088
+1089
+1090
+1092
+1093
+1094
+1096
+1099
+1100
+1101
+1102
+1103
+1104
+1105
+1106
+1107
+1108
+1109
+1110
+1111
+1112
+1114
+1115
+1116
+1117
+1118
+1119
+1120
+1121
+1122
+1123
+1124
+1125
+1126
+1129
+1131
+1133
+1134
+1135
+1136
+1138
+1139
+1140
+1141
+1143
+1144
+1146
+1147
+1148
+1149
+1150
+1151
+1152
+1154
+1155
+1156
+1157
+1158
+1159
+1160
+1161
+1162
+1163
+1164
+1165
+1166
+1168
+1169
+1170
+1171
+1172
+1173
+1174
+1175
+1176
+1177
+1178
+1181
+1182
+1183
+1184
+1185
+1186
+1187
+1188
+1190
+1191
+1192
+1194
+1195
+1196
+1197
+1198
+1199
+1200
+1201
+1202
+1203
+1204
+1205
+1207
+1209
+1210
+1211
+1212
+1213
+1214
+1215
+1216
+1217
+1218
+1219
+1220
+1221
+1222
+1223
+1225
+1226
+1227
+1228
+1229
+1230
+1231
+1232
+1233
+1234
+1236
+1283
+1336
+1339
+1341
+1342
+1346
+1349
+1354
+1356
+1357
+1358
+1359
+1361
+1367
+1372
+1374
+1377
+1378
+1384
+1386
+1389
+1390
+1391
+1392
+1393
+1400
+1408
+1410
+1411
+1412
+1417
+1419
+1422
+1423
+1427
+1429
+1431
+1432
+1435
+1441
+1442
+1443
+1444
+1445
+1446
+1447
+1448
+1450
+1451
+1452
+1453
+1454
+1456
+1457
+1463
+1464
+1465
+1468
+1469
+1476
+1477
+1481
+1483
+1486
+1487
+1496
+1498
+1501
+1502
+1503
+1504
+1505
+1507
+1508
+1509
+1510
+1511
+1512
+1513
+1514
+1516
+1517
+1518
+1519
+1520
+1521
+1522
+1523
+1524
+1525
+1526
+1527
+1528
+1530
+1531
+1532
+1533
+1534
+1535
+1536
+1537
+1538
+1539
+1540
+1541
+1542
+1543
+1544
+1545
+1546
+1547
+1548
+1549
+1550
+1551
+1552
+1553
+1554
+1556
+1557
+1558
+1559
+1561
+1562
+1563
+1564
+1565
+1566
+1567
+1568
+1570
+1571
+1572
+1573
+1574
+1576
+1577
+1578
+1579
+1580
+1581
+1583
+1584
+1587
+1589
+1590
+1591
+1592
+1593
+1594
+1596
+1597
+1599
+1605
+1611
+1612
+1613
+1614
+1616
+1619
+1620
+1621
+1624
+1625
+1629
+1630
+1631
+1634
+1638
+1639
+1643
+1646
+1647
+1648
+1652
+1653
+1654
+1664
+1667
+1668
+1669
+1672
+1675
+1677
+1688
+1690
+1691
+1692
+1698
+1699
+1700
+1702
+1706
+1710
+1712
+1713
+1714
+1718
+1719
+1720
+1721
+1722
+1723
+1724
+1726
+1734
+1735
+1736
+1737
+1740
+1743
+1744
+1745
+1747
+1748
+1750
+1752
+1757
+1759
+1763
+1765
+1767
+1770
+1773
+1774
+1775
+1776
+1777
+1778
+1780
+1781
+1782
+1783
+1786
+1788
+1790
+1792
+1797
+1799
+1800
+1804
+1805
+1806
+1807
+1809
+1811
+1814
+1815
+1816
+1817
+1819
+1821
+1822
+1823
+1824
+1825
+1826
+1827
+1828
+1831
+1832
+1833
+1835
+1836
+1838
+1840
+1841
+1842
+1843
+1845
+1846
+1847
+1850
+1851
+1852
+1856
+1858
+1861
+1862
+1863
+1864
+1865
+1866
+1869
+1870
+1871
+1872
+1873
+1874
+1875
+1876
+1878
+1879
+1881
+1882
+1883
+1884
+1885
+1886
+1887
+1888
+1889
+1890
+1891
+1892
+1893
+1894
+1896
+1897
+1899
+1900
+1903
+1904
+1905
+1906
+1907
+1910
+1912
+1913
+1914
+1915
+1917
+1920
+1922
+1924
+1925
+1926
+1928
+1929
+1930
+1931
+1933
+1936
+1939
+1941
+1942
+1944
+1947
+1949
+1952
+1953
+1954
+1955
+1956
+1957
+1958
+1959
+1960
+1962
+1963
+1966
+1970
+1972
+1973
+1975
+1977
+1978
+1980
+1981
+1982
+1983
+1992
+1995
+1999
+2002
+2003
+2009
+2011
+2013
+2015
+2016
+2017
+2018
+2019
+2022
+2023
+2024
+2026
+2028
+2030
+2031
+2032
+2033
+2034
+2035
+2036
+2037
+2039
+2041
+2042
+2043
+2044
+2045
+2046
+2047
+2048
+2049
+2050
+2051
+2052
+2053
+2054
+2056
+2059
+2060
+2061
+2062
+2063
+2064
+2065
+2067
+2068
+2069
+2071
+2072
+2074
+2076
+2077
+2078
+2079
+2080
+2081
+2082
+2083
+2084
+2085
+2086
+2087
+2088
+2089
+2090
+2092
+2096
+2097
+2101
+2102
+2103
+2105
+2107
+2108
+2110
+2111
+2112
+2113
+2115
+2116
+2117
+2118
+2119
+2122
+2123
+2124
+2126
+2127
+2128
+2132
+2133
+2134
+2135
+2137
+2139
+2142
+2144
+2146
+2147
+2148
+2150
+2151
+2152
+2155
+2163
+2164
+2166
+2168
+2169
+2170
+2171
+2172
+2173
+2174
+2175
+2176
+2177
+2178
+2179
+2180
+2182
+2183
+2184
+2185
+2186
+2187
+2188
+2189
+2195
+2196
+2199
+2201
+2208
+2213
+2214
+2220
+2222
+2226
+2229
+2230
+2231
+2235
+2236
+2238
+2240
+2242
+2243
+2244
+2248
+2250
+2253
+2255
+2256
+2259
+2261
+2263
+2264
+2265
+2266
+2267
+2268
+2269
+2270
+2274
+2279
+2281
+2284
+2285
+2286
+2291
+2292
+2293
+2294
+2295
+2297
+2299
+2302
+2304
+2305
+2306
+2307
+2308
+2310
+2312
+2313
+2318
+2319
+2322
+2324
+2325
+2328
+2330
+2333
+2336
+2337
+2341
+2342
+2343
+2344
+2345
+2347
+2348
+2349
+2351
+2352
+2353
+2354
+2355
+2356
+2364
+2367
+2369
+2372
+2374
+2376
+2378
+2379
+2381
+2384
+2387
+2389
+2390
+2391
+2392
+2393
+2395
+2396
+2397
+2399
+2401
+2402
+2403
+2405
+2407
+2409
+2414
+2415
+2418
+2463
+2477
+2484
+2486
+2488
+2489
+2490
+2491
+2495
+2496
+2499
+2500
+2506
+2508
+2509
+2513
+2516
+2518
+2520
+2527
+2533
+2534
+2536
+2537
+2539
+2540
+2541
+2543
+2547
+2548
+2550
+2551
+2554
+2561
+2574
+2576
+2578
+2581
+2582
+2583
+2588
+2590
+2591
+2593
+2598
+2599
+2602
+2607
+2616
+2618
+2627
+2628
+2629
+2634
+2638
+2641
+2644
+2652
+2656
+2658
+2659
+2662
+2663
+2666
+2668
+2671
+2683
+2684
+2686
+2688
+2689
+2693
+2697
+2698
+2706
+2715
+2718
+2721
+2722
+2724
+2725
+2726
+2727
+2730
+2732
+2734
+2737
+2738
+2739
+2740
+2742
+2744
+2746
+2747
+2748
+2750
+2752
+2753
+2754
+2756
+2757
+2758
+2759
+2760
+2761
+2763
+2765
+2767
+2782
+2783
+2784
+2789
+2798
+2802
+2807
+2812
+2815
+2816
+2817
+2819
+2820
+2822
+2823
+2825
+2826
+2827
+2830
+2831
+2832
+2833
+2834
+2835
+2837
+2838
+2840
+2859
+2864
+2868
+2869
+2876
+2877
+2879
+2880
+2883
+2885
+2889
+2892
+2893
+2895
+2896
+2897
+2898
+2899
+2902
+2905
+2906
+2907
+2908
+2913
+2929
+2930
+2933
+2934
+2935
+2944
+2951
+2952
+2953
+2954
+2955
+2956
+2958
+2959
+2960
+2966
+2967
+2968
+2969
+2972
+2975
+2978
+2980
+2981
+2984
+2985
+2986
+2987
+2989
+2990
+2992
+2993
+2994
+2995
+2996
+2999
+3000
+3002
+3003
+3004
+3005
+3006
+3009
+3010
+3011
+3020
+3022
+3024
+3026
+3029
+3031
+3034
+3036
+3037
+3038
+3040
+3041
+3042
+3047
+3049
+3050
+3051
+3053
+3054
+3056
+3058
+3059
+3060
+3065
+3066
+3067
+3069
+3070
+3072
+3073
+3074
+3075
+3076
+3077
+3079
+3080
+3083
+3084
+3085
+3086
+3087
+3088
+3089
+3090
+3091
+3095
+3103
+3104
+3106
+3108
+3111
+3118
+3120
+3121
+3123
+3124
+3125
+3127
+3128
+3129
+3130
+3131
+3134
+3136
+3137
+3139
+3140
+3144
+3145
+3146
+3147
+3148
+3149
+3150
+3152
+3153
+3154
+3155
+3156
+3157
+3158
+3160
+3161
+3162
+3165
+3167
+3169
+3171
+3173
+3174
+3175
+3176
+3177
+3179
+3180
+3181
+3184
+3185
+3186
+3187
+3188
+3189
+3190
+3191
+3192
+3193
+3194
+3196
+3197
+3198
+3199
+3200
+3201
+3203
+3206
+3207
+3208
+3211
+3212
+3213
+3215
+3216
+3218
+3219
+3220
+3222
+3224
+3228
+3229
+3230
+3231
+3232
+3235
+3236
+3239
+3241
+3242
+3243
+3244
+3245
+3247
+3250
+3254
+3255
+3256
+3257
+3258
+3259
+3261
+3262
+3263
+3264
+3265
+3266
+3269
+3270
+3280
+3283
+3284
+3285
+3287
+3288
+3289
+3290
+3293
+3294
+3295
+3299
+3300
+3301
+3302
+3303
+3305
+3306
+3309
+3310
+3312
+3314
+3315
+3317
+3320
+3321
+3323
+3324
+3325
+3328
+3331
+3332
+3334
+3337
+3338
+3340
+3342
+3344
+3346
+3347
+3349
+3352
+3355
+3357
+3358
+3359
+3360
+3362
+3363
+3364
+3365
+3367
+3368
+3369
+3371
+3373
+3374
+3376
+3378
+3379
+3381
+3383
+3385
+3386
+3387
+3389
+3390
+3393
+3394
+3395
+3396
+3397
+3399
+3400
+3401
+3402
+3403
+3404
+3405
+3406
+3409
+3410
+3415
+3416
+3418
+3421
+3422
+3424
+3425
+3426
+3427
+3429
+3430
+3433
+3435
+3436
+3437
+3438
+3439
+3440
+3441
+3443
+3444
+3445
+3446
+3447
+3451
+3452
+3453
+3455
+3456
+3458
+3461
+3466
+3468
+3469
+3473
+3474
+3475
+3476
+3477
+3481
+3482
+3483
+3487
+3488
+3490
+3491
+3493
+3494
+3497
+3498
+3502
+3503
+3504
+3506
+3509
+3510
+3511
+3514
+3515
+3517
+3520
+3521
+3523
+3524
+3525
+3526
+3529
+3530
+3531
+3533
+3535
+3540
+3541
+3543
+3545
+3547
+3548
+3551
+3553
+3556
+3557
+3559
+3560
+3561
+3562
+3563
+3564
+3567
+3570
+3572
+3573
+3574
+3576
+3577
+3578
+3579
+3580
+3581
+3583
+3584
+3585
+3586
+3588
+3590
+3591
+3592
+3594
+3596
+3597
+3598
+3599
+3600
+3602
+3603
+3604
+3605
+3606
+3608
+3609
+3610
+3611
+3612
+3613
+3614
+3615
+3616
+3617
+3618
+3619
+3620
+3621
+3622
+3623
+3624
+3625
+3626
+3627
+3628
+3629
+3630
+3631
+3632
+3633
+3634
+3638
+3639
+3640
+3642
+3643
+3646
+3649
+3650
+3651
+3652
+3653
+3655
+3656
+3657
+3658
+3659
+3666
+3667
+3668
+3670
+3671
+3673
+3676
+3679
+3682
+3683
+3685
+3686
+3688
+3689
+3690
+3692
+3693
+3694
+3695
+3696
+3697
+3711
+3763
+3767
+3768
+3769
+3773
+3778
+3779
+3785
+3789
+3792
+3794
+3796
+3799
+3801
+3808
+3809
+3814
+3819
+3820
+3822
+3826
+3830
+3832
+3833
+3836
+3838
+3844
+3850
+3852
+3853
+3854
+3855
+3859
+3860
+3861
+3862
+3865
+3869
+3871
+3872
+3873
+3874
+3877
+3878
+3879
+3880
+3881
+3882
+3883
+3884
+3885
+3886
+3887
+3888
+3889
+3891
+3892
+3894
+3895
+3896
+3900
+3903
+3908
+3909
+3910
+3912
+3914
+3915
+3916
+3917
+3918
+3919
+3922
+3923
+3924
+3925
+3927
+3928
+3929
+3930
+3931
+3932
+3933
+3934
+3935
+3936
+3937
+3947
+3951
+3954
+3957
+3960
+3965
+3966
+3967
+3973
+3974
+3975
+3976
+3977
+3978
+3979
+3980
+3981
+3982
+3983
+3984
+3985
+3986
+3987
+3988
+3990
+3991
+3992
+3993
+3994
+3995
+3996
+3997
+3998
+3999
+4000
+4001
+4002
+4015
+4017
+4018
+4021
+4025
+4026
+4028
+4029
+4035
+4036
+4040
+4047
+4050
+4052
+4054
+4055
+4056
+4058
+4060
+4062
+4063
+4064
+4067
+4068
+4072
+4073
+4076
+4078
+4080
+4081
+4082
+4083
+4084
+4085
+4087
+4088
+4089
+4092
+4093
+4095
+4096
+4098
+4099
+4100
+4101
+4102
+4103
+4105
+4106
+4107
+4109
+4122
+4124
+4125
+4127
+4129
+4130
+4131
+4132
+4133
+4134
+4135
+4136
+4137
+4139
+4141
+4143
+4145
+4148
+4149
+4151
+4152
+4154
+4156
+4157
+4158
+4159
+4160
+4162
+4163
+4164
+4165
+4166
+4167
+4168
+4169
+4170
+4174
+4175
+4176
+4180
+4188
+4189
+4191
+4194
+4195
+4201
+4203
+4204
+4205
+4207
+4209
+4212
+4214
+4216
+4217
+4218
+4222
+4225
+4229
+4231
+4234
+4235
+4241
+4242
+4243
+4245
+4246
+4247
+4248
+4249
+4250
+4251
+4253
+4254
+4255
+4256
+4257
+4258
+4260
+4261
+4262
+4265
+4266
+4267
+4268
+4269
+4270
+4271
+4272
+4273
+4274
+4275
+4276
+4277
+4278
+4279
+4280
+4282
+4283
+4284
+4285
+4286
+4287
+4291
+4292
+4293
+4294
+4295
+4296
+4297
+4298
+4299
+4300
+4301
+4302
+4303
+4304
+4305
+4307
+4308
+4309
+4310
+4313
+4314
+4316
+4317
+4318
+4319
+4320
+4321
+4322
+4323
+4324
+4325
+4326
+4327
+4328
+4329
+4330
+4331
+4333
+4334
+4335
+4336
+4337
+4338
+4339
+4340
+4341
+4342
+4343
+4344
+4345
+4346
+4347
+4350
+4352
+4353
+4354
+4355
+4356
+4357
+4358
+4359
+4360
+4362
+4363
+4365
+4366
+4367
+4368
+4369
+4370
+4371
+4372
+4373
+4374
+4375
+4376
+4377
+4378
+4379
+4384
+4385
+4386
+4390
+4391
+4394
+4395
+4396
+4398
+4399
+4401
+4402
+4403
+4404
+4406
+4407
+4410
+4414
+4415
+4416
+4419
+4420
+4422
+4423
+4425
+4426
+4427
+4432
+4434
+4435
+4436
+4437
+4438
+4439
+4442
+4443
+4445
+4447
+4449
+4450
+4451
+4453
+4455
+4458
+4459
+4463
+4464
+4465
+4466
+4467
+4468
+4470
+4472
+4473
+4474
+4475
+4478
+4479
+4480
+4481
+4483
+4484
+4485
+4486
+4488
+4489
+4493
+4494
+4495
+4496
+4497
+4498
+4499
+4501
+4507
+4510
+4511
+4512
+4513
+4518
+4519
+4521
+4523
+4525
+4526
+4527
+4528
+4529
+4530
+4533
+4534
+4535
+4536
+4537
+4538
+4539
+4540
+4542
+4543
+4544
+4547
+4552
+4553
+4554
+4555
+4558
+4559
+4564
+4565
+4568
+4572
+4575
+4576
+4577
+4579
+4580
+4581
+4583
+4584
+4585
+4586
+4587
+4588
+4591
+4592
+4593
+4594
+4595
+4598
+4600
+4604
+4605
+4609
+4610
+4612
+4614
+4615
+4616
+4619
+4620
+4621
+4624
+4625
+4626
+4627
+4628
+4630
+4631
+4632
+4635
+4640
+4642
+4643
+4644
+4648
+4650
+4651
+4662
+4663
+4670
+4671
+4672
+4674
+4675
+4676
+4678
+4679
+4682
+4685
+4686
+4691
+4692
+4694
+4695
+4696
+4699
+4701
+4704
+4706
+4708
+4709
+4711
+4717
+4722
+4724
+4725
+4727
+4729
+4730
+4732
+4735
+4736
+4737
+4740
+4742
+4743
+4747
+4748
+4749
+4750
+4751
+4754
+4755
+4756
+4757
+4760
+4762
+4763
+4765
+4769
+4775
+4777
+4780
+4781
+4787
+4791
+4793
+4794
+4796
+4797
+4798
+4799
+4802
+4803
+4805
+4806
+4810
+4811
+4812
+4814
+4821
+4822
+4823
+4825
+4826
+4827
+4829
+4832
+4833
+4834
+4836
+4838
+4839
+4841
+4843
+4845
+4847
+4848
+4851
+4852
+4854
+4855
+4857
+4859
+4870
+4871
+4873
+4874
+4876
+4877
+4882
+4884
+4885
+4887
+4888
+4889
+4890
+4891
+4893
+4894
+4895
+4897
+4898
+4899
+4900
+4901
+4902
+4908
+4910
+4913
+4915
+4916
+4918
+4919
+4922
+4926
+4928
+4930
+4931
+4933
+4941
+4942
+4949
+4952
+4954
+4955
+4958
+4959
+4965
+4967
+4969
+4970
+4971
+4972
+4973
+4974
+4975
+4976
+4977
+4979
+4983
+4984
+4985
+4986
+4987
+4988
+4990
+4991
+4992
+4993
+4996
+4997
+4998
+4999
+5000
+5002
+5003
+5004
+5005
+5007
+5009
+5010
+5014
+5015
+5016
+5019
+5023
+5032
+5034
+5036
+5037
+5038
+5042
+5043
+5046
+5048
+5049
+5051
+5056
+5064
+5066
+5069
+5070
+5071
+5077
+5078
+5081
+5085
+5087
+5088
+5091
+5093
+5096
+5109
+5111
+5116
+5123
+5127
+5130
+5132
+5133
+5136
+5137
+5138
+5144
+5147
+5154

test-00000-of-00001.parquet → test_debiased.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:64eb8a4ff3c705038d2c489fb97345c19e33f0a297f440a168e6940e76d329ca
-size 160845
+oid sha256:dd4fcb087f724652f9f178af4ff6ebf2be1eb11bf7ffba51ce2b8396329ee9e1
+size 85343

test_pruned.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b029b219535db754c49b967a1739eb878078796eae3bc8b1252ea9d55cd7f5cc
+size 89920