Datasets:
Upload dataset (part 00003-of-00004)
Browse files- README.md +36 -0
- symile-m3-5-xs/test-00001-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00002-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00003-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00004-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00005-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00006-of-00008.parquet +3 -0
- symile-m3-5-xs/test-00007-of-00008.parquet +3 -0
README.md
CHANGED
@@ -38,6 +38,14 @@ configs:
|
|
38 |
path: symile-m3-5-s/val-*
|
39 |
- split: test
|
40 |
path: symile-m3-5-s/test-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
41 |
dataset_info:
|
42 |
- config_name: symile-m3-5-m
|
43 |
features:
|
@@ -95,6 +103,34 @@ dataset_info:
|
|
95 |
num_examples: 50000
|
96 |
download_size: 159628727029
|
97 |
dataset_size: 156989775197.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
98 |
---
|
99 |
# Dataset Card for Symile-M3
|
100 |
Symile-M3 is a multilingual dataset of (audio, image, text) samples. The dataset is specifically designed to test a model's ability to capture higher-order information between three distinct high-dimensional data types: by incorporating multiple languages, we construct a task where text and audio are both needed to predict the image, and where, importantly, neither text nor audio alone would suffice.
|
|
|
38 |
path: symile-m3-5-s/val-*
|
39 |
- split: test
|
40 |
path: symile-m3-5-s/test-*
|
41 |
+
- config_name: symile-m3-5-xs
|
42 |
+
data_files:
|
43 |
+
- split: train
|
44 |
+
path: symile-m3-5-xs/train-*
|
45 |
+
- split: val
|
46 |
+
path: symile-m3-5-xs/val-*
|
47 |
+
- split: test
|
48 |
+
path: symile-m3-5-xs/test-*
|
49 |
dataset_info:
|
50 |
- config_name: symile-m3-5-m
|
51 |
features:
|
|
|
103 |
num_examples: 50000
|
104 |
download_size: 159628727029
|
105 |
dataset_size: 156989775197.0
|
106 |
+
- config_name: symile-m3-5-xs
|
107 |
+
features:
|
108 |
+
- name: lang
|
109 |
+
dtype: string
|
110 |
+
- name: audio
|
111 |
+
dtype: audio
|
112 |
+
- name: image
|
113 |
+
dtype: image
|
114 |
+
- name: text
|
115 |
+
dtype: string
|
116 |
+
- name: cls
|
117 |
+
dtype: string
|
118 |
+
- name: cls_id
|
119 |
+
dtype: int64
|
120 |
+
- name: target_text
|
121 |
+
dtype: string
|
122 |
+
splits:
|
123 |
+
- name: train
|
124 |
+
num_bytes: 70410563197.0
|
125 |
+
num_examples: 500000
|
126 |
+
- name: val
|
127 |
+
num_bytes: 3607295872.0
|
128 |
+
num_examples: 25000
|
129 |
+
- name: test
|
130 |
+
num_bytes: 3624041386.0
|
131 |
+
num_examples: 25000
|
132 |
+
download_size: 80003029310
|
133 |
+
dataset_size: 77641900455.0
|
134 |
---
|
135 |
# Dataset Card for Symile-M3
|
136 |
Symile-M3 is a multilingual dataset of (audio, image, text) samples. The dataset is specifically designed to test a model's ability to capture higher-order information between three distinct high-dimensional data types: by incorporating multiple languages, we construct a task where text and audio are both needed to predict the image, and where, importantly, neither text nor audio alone would suffice.
|
symile-m3-5-xs/test-00001-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70f5472710c4036e33ef4157725760381a3e43ded72382c6fe2b1376d2f6d263
|
3 |
+
size 444336449
|
symile-m3-5-xs/test-00002-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8ab16ac9147ceb99e75bde858eae31d9e64ce7202f8219a1a2f85349597a69c3
|
3 |
+
size 446590727
|
symile-m3-5-xs/test-00003-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1c0d912ded54a1a21487e5dc2cadfaa729e89e22656d7eaae69355bf1f0413e1
|
3 |
+
size 452114483
|
symile-m3-5-xs/test-00004-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7c29f2f66fa0ee7cfd8bd70bf4b06ba5dbf99150ea69728a531e9bb074c07bcb
|
3 |
+
size 474134401
|
symile-m3-5-xs/test-00005-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1a19baf0caae5a5dbd7fa602a041239734573a60f7022103b37f5b2a267d0635
|
3 |
+
size 461031005
|
symile-m3-5-xs/test-00006-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6b139516f448cc01047bfcd693982613a9971f5987a34fd5b19af9c30224a074
|
3 |
+
size 455049613
|
symile-m3-5-xs/test-00007-of-00008.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0675cffc2c5be74f21fe06c3642c3b6aaef648cf2e6a1c974a1c69f19f8eaeb
|
3 |
+
size 460316618
|