File size: 1,036 Bytes
dedb4bc | 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 39 40 41 42 | ---
configs:
- config_name: default
data_files:
- split: train
path: data/train.parquet
- split: validation
path: data/validation.parquet
- split: test
path: data/test.parquet
- split: test_unseen
path: data/test_unseen.parquet
tags:
- image-forensics
- multimodal-vision
pretty_name: TJ Multimodal Research Dataset
size_categories:
- 100K<n<1M
---
# TJ Research Dataset — Canonical Manifests
Canonical consolidated multimodal research dataset.
Conforms to `data/COMBINED_DATASET_SCHEMA.md`.
## Task Target
Binary classification:
- `authentic` -> 0 (`ai_positive = 0`)
- `tampered` (human memes) -> 1 (`ai_positive = 0`)
- `tampered` (AI edits) -> 1 (`ai_positive = 1`)
- `fully_aigc` -> 2 (`ai_positive = 1`)
## Split Breakdown
- Total rows: 127902
- Train: 77106
- Validation: 21001
- Test: 16773
- Test Unseen: 13022
## Exclusions
Strictly excludes COCO val2017 and WildFake DALL-E Advanced per competition contract.
|