Hugging Face's logo

Datasets:
intfloat
/
mmE5-MMEB-hardneg

Modalities:
Text
Formats:
parquet
Languages:
English
Size:
1M - 10M
ArXiv:
Tags:
embedding
multimodal
Libraries:
Datasets
pandas
License:
Dataset card Data Studio Files
xet
Community
3

Link to paper and add task category

#2
by nielsr HF Staff - opened
base: refs/heads/main
←
from: refs/pr/2
Discussion Files changed
+56
-56
Files changed (1) hide show
  1. README.md +56 -56
README.md CHANGED
@@ -1,117 +1,117 @@
1
  ---
2
- license: mit
3
  language:
4
  - en
 
 
 
 
 
 
5
  tags:
6
  - embedding
7
  - multimodal
8
- pretty_name: mmE5 labeled data
9
- size_categories:
10
- - 1M<n<10M
11
  configs:
12
  - config_name: TAT-DQA
13
  data_files:
14
- - split: train
15
- path: "TAT-DQA/TAT-DQA.parquet"
16
  - config_name: ArxivQA
17
  data_files:
18
- - split: train
19
- path: "ArxivQA/ArxivQA.parquet"
20
  - config_name: InfoSeek_it2t
21
  data_files:
22
- - split: train
23
- path: "InfoSeek_it2t/InfoSeek_it2t.parquet"
24
  - config_name: InfoSeek_it2it
25
  data_files:
26
- - split: train
27
- path: "InfoSeek_it2it/InfoSeek_it2it.parquet"
28
  - config_name: ImageNet_1K
29
  data_files:
30
- - split: train
31
- path: "ImageNet_1K/ImageNet_1K.parquet"
32
  - config_name: N24News
33
  data_files:
34
- - split: train
35
- path: "N24News/N24News.parquet"
36
  - config_name: HatefulMemes
37
  data_files:
38
- - split: train
39
- path: "HatefulMemes/HatefulMemes.parquet"
40
  - config_name: SUN397
41
  data_files:
42
- - split: train
43
- path: "SUN397/SUN397.parquet"
44
  - config_name: VOC2007
45
  data_files:
46
- - split: train
47
- path: "VOC2007/VOC2007.parquet"
48
  - config_name: InfographicsVQA
49
  data_files:
50
- - split: train
51
- path: "InfographicsVQA/InfographicsVQA.parquet"
52
  - config_name: ChartQA
53
  data_files:
54
- - split: train
55
- path: "ChartQA/ChartQA.parquet"
56
  - config_name: A-OKVQA
57
  data_files:
58
- - split: train
59
- path: "A-OKVQA/A-OKVQA.parquet"
60
  - config_name: DocVQA
61
  data_files:
62
- - split: train
63
- path: "DocVQA/DocVQA.parquet"
64
  - config_name: OK-VQA
65
  data_files:
66
- - split: train
67
- path: "OK-VQA/OK-VQA.parquet"
68
  - config_name: Visual7W
69
  data_files:
70
- - split: train
71
- path: "Visual7W/Visual7W.parquet"
72
  - config_name: VisDial
73
  data_files:
74
- - split: train
75
- path: "VisDial/VisDial.parquet"
76
  - config_name: CIRR
77
  data_files:
78
- - split: train
79
- path: "CIRR/CIRR.parquet"
80
  - config_name: NIGHTS
81
  data_files:
82
- - split: train
83
- path: "NIGHTS/NIGHTS.parquet"
84
  - config_name: WebQA
85
  data_files:
86
- - split: train
87
- path: "WebQA/WebQA.parquet"
88
  - config_name: VisualNews_i2t
89
  data_files:
90
- - split: train
91
- path: "VisualNews_i2t/VisualNews_i2t.parquet"
92
  - config_name: VisualNews_t2i
93
  data_files:
94
- - split: train
95
- path: "VisualNews_t2i/VisualNews_t2i.parquet"
96
  - config_name: MSCOCO_i2t
97
  data_files:
98
- - split: train
99
- path: "MSCOCO_i2t/MSCOCO_i2t.parquet"
100
  - config_name: MSCOCO_t2i
101
  data_files:
102
- - split: train
103
- path: "MSCOCO_t2i/MSCOCO_t2i.parquet"
104
  - config_name: MSCOCO
105
  data_files:
106
- - split: train
107
- path: "MSCOCO/MSCOCO.parquet"
108
  ---
109
- # mmE5 Labeled Data
110
-
111
-
112
 
 
113
 
114
- This dataset contains datasets used for the supervised finetuning of mmE5 ([mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data](https://arxiv.org/abs/2502.08468)):
115
  - **MMEB** (with hard negative)
116
  - **InfoSeek** (from M-BEIR)
117
  - **TAT-DQA**
 
1
  ---
 
2
  language:
3
  - en
4
+ license: mit
5
+ task_categories:
6
+ - image-text-to-text
7
+ size_categories:
8
+ - 1M<n<10M
9
+ pretty_name: mmE5 labeled data
10
  tags:
11
  - embedding
12
  - multimodal
 
 
 
13
  configs:
14
  - config_name: TAT-DQA
15
  data_files:
16
+ - split: train
17
+ path: TAT-DQA/TAT-DQA.parquet
18
  - config_name: ArxivQA
19
  data_files:
20
+ - split: train
21
+ path: ArxivQA/ArxivQA.parquet
22
  - config_name: InfoSeek_it2t
23
  data_files:
24
+ - split: train
25
+ path: InfoSeek_it2t/InfoSeek_it2t.parquet
26
  - config_name: InfoSeek_it2it
27
  data_files:
28
+ - split: train
29
+ path: InfoSeek_it2it/InfoSeek_it2it.parquet
30
  - config_name: ImageNet_1K
31
  data_files:
32
+ - split: train
33
+ path: ImageNet_1K/ImageNet_1K.parquet
34
  - config_name: N24News
35
  data_files:
36
+ - split: train
37
+ path: N24News/N24News.parquet
38
  - config_name: HatefulMemes
39
  data_files:
40
+ - split: train
41
+ path: HatefulMemes/HatefulMemes.parquet
42
  - config_name: SUN397
43
  data_files:
44
+ - split: train
45
+ path: SUN397/SUN397.parquet
46
  - config_name: VOC2007
47
  data_files:
48
+ - split: train
49
+ path: VOC2007/VOC2007.parquet
50
  - config_name: InfographicsVQA
51
  data_files:
52
+ - split: train
53
+ path: InfographicsVQA/InfographicsVQA.parquet
54
  - config_name: ChartQA
55
  data_files:
56
+ - split: train
57
+ path: ChartQA/ChartQA.parquet
58
  - config_name: A-OKVQA
59
  data_files:
60
+ - split: train
61
+ path: A-OKVQA/A-OKVQA.parquet
62
  - config_name: DocVQA
63
  data_files:
64
+ - split: train
65
+ path: DocVQA/DocVQA.parquet
66
  - config_name: OK-VQA
67
  data_files:
68
+ - split: train
69
+ path: OK-VQA/OK-VQA.parquet
70
  - config_name: Visual7W
71
  data_files:
72
+ - split: train
73
+ path: Visual7W/Visual7W.parquet
74
  - config_name: VisDial
75
  data_files:
76
+ - split: train
77
+ path: VisDial/VisDial.parquet
78
  - config_name: CIRR
79
  data_files:
80
+ - split: train
81
+ path: CIRR/CIRR.parquet
82
  - config_name: NIGHTS
83
  data_files:
84
+ - split: train
85
+ path: NIGHTS/NIGHTS.parquet
86
  - config_name: WebQA
87
  data_files:
88
+ - split: train
89
+ path: WebQA/WebQA.parquet
90
  - config_name: VisualNews_i2t
91
  data_files:
92
+ - split: train
93
+ path: VisualNews_i2t/VisualNews_i2t.parquet
94
  - config_name: VisualNews_t2i
95
  data_files:
96
+ - split: train
97
+ path: VisualNews_t2i/VisualNews_t2i.parquet
98
  - config_name: MSCOCO_i2t
99
  data_files:
100
+ - split: train
101
+ path: MSCOCO_i2t/MSCOCO_i2t.parquet
102
  - config_name: MSCOCO_t2i
103
  data_files:
104
+ - split: train
105
+ path: MSCOCO_t2i/MSCOCO_t2i.parquet
106
  - config_name: MSCOCO
107
  data_files:
108
+ - split: train
109
+ path: MSCOCO/MSCOCO.parquet
110
  ---
 
 
 
111
 
112
+ # mmE5 Labeled Data
113
 
114
+ This dataset contains datasets used for the supervised finetuning of mmE5 ([mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data](https://huggingface.co/papers/2502.08468)):
115
  - **MMEB** (with hard negative)
116
  - **InfoSeek** (from M-BEIR)
117
  - **TAT-DQA**