Hugging Face's logo

Datasets:
seokwon99
/
Multi3IR

Tasks:
Text Retrieval
Modalities:
Image
Tabular
Text
Formats:
parquet
Languages:
English
Size:
1M - 10M
ArXiv:
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
seokwon99 commited on
Commit
082e2cf
·
verified ·
1 Parent(s): 1d0e232

Add files using upload-large-folder tool

Browse files
Files changed (49) hide show
  1. README.md +7 -6
  2. corpus/images/part-00000.parquet +3 -0
  3. corpus/images/part-00001.parquet +3 -0
  4. corpus/images/part-00002.parquet +3 -0
  5. corpus/images/part-00003.parquet +3 -0
  6. corpus/images/part-00004.parquet +3 -0
  7. corpus/images/part-00005.parquet +3 -0
  8. corpus/images/part-00006.parquet +3 -0
  9. corpus/images/part-00007.parquet +3 -0
  10. corpus/images/part-00008.parquet +3 -0
  11. corpus/images/part-00010.parquet +3 -0
  12. corpus/images/part-00011.parquet +3 -0
  13. corpus/images/part-00012.parquet +3 -0
  14. corpus/images/part-00013.parquet +3 -0
  15. corpus/images/part-00014.parquet +3 -0
  16. corpus/images/part-00015.parquet +3 -0
  17. corpus/images/part-00016.parquet +3 -0
  18. corpus/images/part-00017.parquet +3 -0
  19. corpus/images/part-00018.parquet +3 -0
  20. corpus/images/part-00019.parquet +3 -0
  21. corpus/images/part-00020.parquet +3 -0
  22. corpus/images/part-00021.parquet +3 -0
  23. corpus/images/part-00022.parquet +3 -0
  24. corpus/images/part-00023.parquet +3 -0
  25. corpus/images/part-00024.parquet +3 -0
  26. corpus/images/part-00025.parquet +3 -0
  27. corpus/images/part-00026.parquet +3 -0
  28. corpus/images/part-00027.parquet +3 -0
  29. corpus/images/part-00028.parquet +3 -0
  30. corpus/images/part-00029.parquet +3 -0
  31. corpus/images/part-00030.parquet +3 -0
  32. corpus/images/part-00031.parquet +3 -0
  33. corpus/images/part-00032.parquet +3 -0
  34. corpus/images/part-00033.parquet +3 -0
  35. corpus/images/part-00034.parquet +3 -0
  36. corpus/images/part-00035.parquet +3 -0
  37. corpus/images/part-00036.parquet +3 -0
  38. corpus/images/part-00037.parquet +3 -0
  39. corpus/images/part-00038.parquet +3 -0
  40. corpus/text/part-00001.parquet +3 -0
  41. corpus/text/part-00004.parquet +3 -0
  42. corpus/text/part-00006.parquet +3 -0
  43. corpus/text/part-00007.parquet +3 -0
  44. corpus/text/part-00011.parquet +3 -0
  45. corpus/text/part-00012.parquet +3 -0
  46. dropped_broken_images.json +14 -0
  47. gold/train.parquet +2 -2
  48. queries/eval.parquet +1 -1
  49. queries/train.parquet +1 -1
README.md CHANGED
@@ -21,9 +21,9 @@ configs:
21
  - config_name: corpus
22
  data_files:
23
  - split: text
24
- path: corpus/text.parquet
25
  - split: image
26
- path: corpus/image.parquet
27
  ---
28
 
29
  # Multi³IR-Bench
@@ -35,17 +35,18 @@ Multi-perspective, multi-domain, multi-modal information retrieval benchmark ([p
35
  | config | split | rows | columns |
36
  |---|---|---|---|
37
  | `queries` | train / eval | 103,877 / 1,039 | uid, query, perspectives |
38
- | `gold` | train / eval | 632,558 / 42,389 | uid, perspective_idx, summary, doc_id, modality, url, domain (train), rank + text (eval; text inline for text docs) |
39
- | `corpus` | text / image | 457,168 / 346,829 | doc_id, url — the benchmark corpus (every document that is gold for some question) |
40
 
41
- Documents are referenced by id and source url, not redistributed: C4 passages (`c4-<file>-<line>`) can be recovered from [allenai/c4](https://huggingface.co/datasets/allenai/c4), images from their url. In the code repository, `python -m multi3ir.data.hub pull` restores the benchmark layout and `python -m multi3ir.data.build_corpus` materialises the corpus.
42
 
43
  ```python
44
  from datasets import load_dataset
45
  gold = load_dataset("seokwon99/Multi3IR", "gold", split="eval")
 
46
  ```
47
 
48
- Sources: Stack Exchange (CC BY-SA 4.0), C4 (ODC-BY), images © their owners (research use). Perspectives were written by GPT-5-mini and filtered for faithfulness (Bespoke-MiniCheck-7B) and uniqueness (all-MiniLM-L6-v2, cosine > 0.8); document support was judged by Qwen3-VL-30B-A3B-Instruct (agreement 0.90 with human labels on `fully`).
49
 
50
  ```bibtex
51
  @article{song2026multi3ir,
 
21
  - config_name: corpus
22
  data_files:
23
  - split: text
24
+ path: corpus/text/*.parquet
25
  - split: image
26
+ path: corpus/images/*.parquet
27
  ---
28
 
29
  # Multi³IR-Bench
 
35
  | config | split | rows | columns |
36
  |---|---|---|---|
37
  | `queries` | train / eval | 103,877 / 1,039 | uid, query, perspectives |
38
+ | `gold` | train / eval | 1,111,041 / 42,389 | uid, perspective_idx, summary, doc_id, modality, url, domain (train), rank + text (eval; text inline for text docs) |
39
+ | `corpus` | text / image | 631,377 / 380,672 | doc_id, url, text (passage) / image (JPEG) — the paper's retrieval corpus: gold documents of every split plus distractors |
40
 
41
+ The corpus is the 1.01M-document pool used for all results in the paper. Text documents are C4 passages (`c4-<file>-<line>`) shipped with their full text; images are shipped as JPEG re-encodings (longest side ≤ 512 px, quality 90, transparency composited on white) together with their source url. 12 images whose source returned HTML or gzip instead of an image were removed from the corpus and from `gold/train` (none were test gold).
42
 
43
  ```python
44
  from datasets import load_dataset
45
  gold = load_dataset("seokwon99/Multi3IR", "gold", split="eval")
46
+ images = load_dataset("seokwon99/Multi3IR", "corpus", split="image") # ~17 GB with text
47
  ```
48
 
49
+ Sources: Stack Exchange (CC BY-SA 4.0), C4 (ODC-BY), images © their owners, redistributed downscaled for research use only. Perspectives were written by GPT-5-mini and filtered for faithfulness (Bespoke-MiniCheck-7B) and uniqueness (all-MiniLM-L6-v2, cosine > 0.8); document support was judged by Qwen3-VL-30B-A3B-Instruct (agreement 0.90 with human labels on `fully`).
50
 
51
  ```bibtex
52
  @article{song2026multi3ir,
corpus/images/part-00000.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:092d0e184acee923e836ccd2e5365afa8b1708cea846c66375c14d77bbfd5906
3
+ size 405629409
corpus/images/part-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26a150f953b656b023b76e100b0fc7f49069d6e4ab87c6e92614c4777a4f5bd9
3
+ size 404327006
corpus/images/part-00002.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c49ebd72265661f93085ef968f0fe8edede266474b8c76228c372a6b7157214f
3
+ size 407121071
corpus/images/part-00003.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df6d6cb734c79bedd9bc5d8ad4926056d10d228e182b514324fb24e2cca665fe
3
+ size 411077980
corpus/images/part-00004.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b7bbb82eebec22862bcf5a378d66e1e99d151635d3dc0f861842a521b447a07
3
+ size 403864424
corpus/images/part-00005.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:359a86104f4501a7d3c70a9317aaf59a640cf4206cd325ca89e142a97966ebfe
3
+ size 406780740
corpus/images/part-00006.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41914e495e1249c80ad84a7f43601a78f9899d4eb878dcd23998eb113523e8ea
3
+ size 405773249
corpus/images/part-00007.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75ad1de84a86d6f9f3b83dd2111a9ab1961b0780e81c97f97ba9fe6d4bef4744
3
+ size 407343703
corpus/images/part-00008.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dea27eb9160f072635bc0c8e9aecceee139a1aa1a9e6678686e21151bb02a644
3
+ size 406267793
corpus/images/part-00010.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:90e9de1052df06a1eccf5fc90ca622fb7a785c14315053d6fe0422b9abf03d51
3
+ size 407162350
corpus/images/part-00011.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:614b819bd05bdcfad514da55bc26df715ac352a00af37dcd224e9229cdd1be39
3
+ size 408809116
corpus/images/part-00012.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ec6e31583bf5533428a150e5ff68f0335ff73fae6fde8abaedd80819f20c70f
3
+ size 407256348
corpus/images/part-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3163826d2abcffc81532694d1ed135d75d78165fc6ba8ffd56cd202d27dc8ebd
3
+ size 405823593
corpus/images/part-00014.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9251eafdc04e22150a937cb36263bbff848b35d81ef9477370e7dad343745a90
3
+ size 408402041
corpus/images/part-00015.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f0c741a031bfd2915b5a35801ca822c896ae3f055c22391fb75b69a4d399bae
3
+ size 406847989
corpus/images/part-00016.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebd6786869566f6cbf4cdf0973ee7b38f4910879b75e496032d3c21a273ad92f
3
+ size 406091181
corpus/images/part-00017.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d7d8d02926ca92de717989c5939b1d1ea043db01e7fc7a10be691b596d8abf1
3
+ size 409793304
corpus/images/part-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c9dcfd109d0d441b91554bd533972084be65e29a829dde7251b4255e73d8e37
3
+ size 408516758
corpus/images/part-00019.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4726d66c69780b9563a1dcf5140727a63608ac97e38907e4e9af2437eecf441f
3
+ size 408352798
corpus/images/part-00020.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:452096e0952960ed17d6fca3c39e3d1800ccb30a569320013b7796e72888a016
3
+ size 405420708
corpus/images/part-00021.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08ffd5b9d2dd553fd5d790f53d1f73e59081e951e23879dde8da84eb01fd0b19
3
+ size 409453577
corpus/images/part-00022.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b3396593c1610433fb10b405c6b75bcd47057f3a2c63ef39c7fb522a488adc0
3
+ size 408667253
corpus/images/part-00023.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68746960d349fd2c2e9ba79fcefaed7899b5c2117ce1d734e2da15d1982b83e1
3
+ size 406300746
corpus/images/part-00024.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cc13c77f826f466f428997e6c2cf1917f8b131998751e2646652d12a03a249c
3
+ size 408356268
corpus/images/part-00025.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a633001fc5ffc35c8499305386876c951c042a4cae37a1040364010ab781191
3
+ size 409945304
corpus/images/part-00026.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:478d6dff18f32c1365bc27e80099831cd8139c1564f2635201bffdbdb9870f45
3
+ size 407258004
corpus/images/part-00027.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9aec3f5845207cff8e1f21238453eaa65d1bc63210c04f33ea4da145cde0be6b
3
+ size 406553254
corpus/images/part-00028.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6d8d016d008f2d100782b28636cad1a190350a018aa15a85b16738800e658d3
3
+ size 407883456
corpus/images/part-00029.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b7be30245441dcfa57e4c6369debde61ae0c68f005c6ccc4725205b7a725fea
3
+ size 405078758
corpus/images/part-00030.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a132fff92f2727b2b3645e6f4af8b9731d32fd5785f6184b73b968ff4e02214
3
+ size 406273324
corpus/images/part-00031.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7c4e95b560497f2879231436d757e9df41d733bf803bb431b26f399124e2a1c
3
+ size 404401676
corpus/images/part-00032.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddce29c9ad96ba58441c5de3ade4a85765b10667f4ab50f871f37d92cdca9f9e
3
+ size 408698861
corpus/images/part-00033.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46787dc3d84ec1d7b517d5f509d1710985c8059da65646c40cefd87ff421efdc
3
+ size 407980484
corpus/images/part-00034.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43d719d9d4c1f9ca1efee8423623bbcb4caad95374efec4ebcc845a130f77eed
3
+ size 414414388
corpus/images/part-00035.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d26d8aa0ee22de9516253e6b99cdbbcefc28a3108463bb78f6ab3c60cad09cb
3
+ size 430389705
corpus/images/part-00036.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:069b217d21d2741f03931f6a75e5745e0d7fe2d87459e51646afbd7942c9d744
3
+ size 430745881
corpus/images/part-00037.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fdb62b6e7eee35f589252af24b5923ecba23f1f4d057747134942dfd5301c6aa
3
+ size 428771003
corpus/images/part-00038.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a2e3e8a940f494c9525379771c7ea5ce50efafb2ce40851581dde5f3890972e
3
+ size 28438304
corpus/text/part-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a01193b60cdf8dd5485785bf58f48523635138dd2583144b0c50a6e1dd15d00c
3
+ size 139659485
corpus/text/part-00004.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cdb2c2e51b3e86ffc941fdb24f91ca2192351f4dfb3357a60b1d4d7348bc0ae
3
+ size 140668736
corpus/text/part-00006.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f68a041ddd822da2e0589214de0acbc8ba640912b4f9847668a1c4e408879d62
3
+ size 149796024
corpus/text/part-00007.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be017a29576ff63f2aaf094e78b836bec8258578667f049d3bf853fd7df73c5c
3
+ size 151446391
corpus/text/part-00011.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d2ec4a1c28581dfd2fc4e5b90a4c5d47636e7959e3778271691efd5e11d2b48
3
+ size 132990217
corpus/text/part-00012.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1438fe94ab074f622703dc942cb75990f35fa0326c8c66f3a961c0a6ce8f7b4
3
+ size 83929976
dropped_broken_images.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ "serper_0c86303dda54.jpg",
3
+ "serper_3965546dca1b.jpg",
4
+ "serper_406fabbcbac3.jpg",
5
+ "serper_41f52c58a3a1.jpg",
6
+ "serper_42530f569e85.jpg",
7
+ "serper_51154b652d92.jpg",
8
+ "serper_703f2a141ef9.jpg",
9
+ "serper_79e28bd77ff0.jpg",
10
+ "serper_8b510f0ce172.jpg",
11
+ "serper_d7d43ef4f3b4.jpg",
12
+ "serper_df8a90e89fc9.jpg",
13
+ "serper_e5d8b4c0fdbc.jpg"
14
+ ]
gold/train.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6818347871a2fd2096f0204dba175887ddd0c36b28b23a885bd4b31dcf414fbe
3
- size 72456642
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c75778e11524d499449d57d195214fb96abc2aca7c85f82716cea7b282d0d1b
3
+ size 122814260
queries/eval.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:186225d7b75c503dad81a0cc79a668c622293379d9246b65aa0d87db20b1068c
3
  size 620665
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33f39e99b844847b91f0ecec6e60b6c72509b96a020b56bdd102c7c16a15ae11
3
  size 620665
queries/train.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:17d3ba59e776b6bdfc72aa11a336dc2b0d7d5699df5a97053aaa6e7e98d56ce4
3
  size 57102792
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bec8617aabcf35e44b0e72751ba5ac42a826844d0ad08044094c178274b578e6
3
  size 57102792