Datasets:

Modalities:
Text
Formats:
parquet
Size:
< 1K
Libraries:
Datasets
Dask
File size: 724 Bytes
e9b9f30
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1938ebe
 
 
 
e9b9f30
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
---
dataset_info:
  features:
  - name: input_ids
    sequence: int32
  - name: attention_mask
    sequence: int8
  - name: token_type_ids
    sequence: int8
  - name: labels
    sequence: int64
  - name: pixel_values
    sequence:
      sequence:
        sequence: float32
  - name: masked_indices
    sequence: int64
  - name: metadata
    struct:
    - name: include_only_masked_tokens_images
      dtype: bool
    - name: logic_name
      dtype: string
    - name: mask_images_equally
      dtype: bool
  splits:
  - name: train
    num_bytes: 1095181943
    num_examples: 617
  download_size: 50697224
  dataset_size: 1095181943
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---