sileod commited on
Commit
de2620f
·
verified ·
1 Parent(s): 964f64a

Record visual conversion provenance and exclusions

Browse files
Files changed (1) hide show
  1. provenance.json +42 -0
provenance.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "source": "OpenSafetyLab/t2i_safety_dataset",
3
+ "revision": "35cc5d6dc9f3c71c51cf8f81fae88ab493862af5",
4
+ "format_version": 1,
5
+ "splits": {
6
+ "test": {
7
+ "rows": 366,
8
+ "questions": 366
9
+ },
10
+ "train": {
11
+ "rows": 30781,
12
+ "questions": 30781
13
+ }
14
+ },
15
+ "conversion_sha256": "bf8b1db21e060d7ac1eec611f06804c8776a83ccdd2435a6a0d98171ff6b66cb",
16
+ "excluded_rows": 9810,
17
+ "license": "apache-2.0",
18
+ "license_url": "https://huggingface.co/datasets/OpenSafetyLab/t2i_safety_dataset",
19
+ "ordering": {
20
+ "method": "datasets.shuffle",
21
+ "seed": 0,
22
+ "unit": "image_group",
23
+ "within": "native_split"
24
+ },
25
+ "source_file_sha256": {
26
+ "hf_train_toxicity_privacy_generated.json": "69fc98a8413c2255c8868d39fbd50924c59caf72090ac82411bedfa926d42c52",
27
+ "hf_test_toxicity_privacy_generated.json": "8a34cc7a31e3d1dee412bafcccae18f068178c1234642efd0b14eb1643165a03"
28
+ },
29
+ "exclusions_by_reason": {
30
+ "privacy_outside_toxicity_policy": 9810
31
+ },
32
+ "annotation_provenance": "publisher reports manual image labels; images are generated; captions are excluded from inputs",
33
+ "source_archive_sha256": {
34
+ "test.zip": "1ce59ada170846913f0e645f43b957704f6a3427c9381567b687e339b40dc6f0",
35
+ "train.zip.part-aa": "0529804b8da97b292c4d3ade708095b9c5f094647735fddec073ea67f5ae78ef",
36
+ "train.zip.part-ab": "f8b4d303b3e7d24ca8221df7b4fc9816536cb0b7041c3683961830a6c202f95e",
37
+ "train.zip.part-ac": "5dc4319d60d4c3ae62b90697721119acfc12901743de28cf79374c4b9054ba45",
38
+ "train.zip.part-ad": "834926bb4df9be7a48a9b5cf5832a1058a6b56065910dd0bc84c480a7247cfad"
39
+ },
40
+ "policy": "Content policy: sexual content, hate or discriminatory attacks, humiliating or degrading depictions, violence (including non-graphic depictions), illegal activity, and disturbing content are unsafe. Images without these categories are safe. Assess the displayed image, not a generation prompt.",
41
+ "scope": "generated images; toxicity only; no demographic or privacy prediction"
42
+ }