File size: 1,885 Bytes
5d1e8e9
 
 
 
 
228a4fb
5d1e8e9
228a4fb
 
c00634f
5d1e8e9
 
228a4fb
c00634f
 
 
5d1e8e9
c00634f
0bed00e
5d1e8e9
 
 
 
 
 
c6fbb6c
5d1e8e9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e98213f
5d1e8e9
 
 
e98213f
 
 
 
5d1e8e9
 
 
 
a5a44bd
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
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
---
license: mit
---


# 📰 NYT Datasets

The 📰**NYT datasets** serve as a benchmark designed to evaluate methods to identify memorized training data or to infer membership, specifically from OpenAI models that are released before 2023. 

🔧 Memorized Training Data

This dataset contains examples provided in [Exhibit-J](https://nytco-assets.nytimes.com/2023/12/Lawsuit-Document-dkt-1-68-Ex-J.pdf) of the New York Times Lawsuit (label=1). 

- Snippet is the article.
- Prefix is the prefix provided to the model according to the evidence in Exhibit-J.
- Completion is the original article content that follows the prefix.

🔧 Non-Member Data
  
This dataset also contains excerpts of CNN articles scraped in 2023 (label=0).



### 📌 Applicability

The datasets can be applied to OpenAI models released before **2023**. 


## Loading the datasets

To load the dataset:

```python
from datasets import load_dataset

dataset = load_dataset("lasha-nlp/NYT_Memorization")
```

* *Label 0*: Refers to the CNN data that cannot have been memorized. *Label 1*: Refers to text data from NYT articles.

## 🛠️ Codebase

For evaluating information-guided probes, visit our [GitHub repository](https://github.com/AbhilashaRavichander/information-probing).

## ⭐ Citing our Work

If you find our code or data useful, kindly cite our work:

```bibtex
@misc{ravichander2025informationguidedidentificationtrainingdata,
      title={Information-Guided Identification of Training Data Imprint in (Proprietary) Large Language Models}, 
      author={Abhilasha Ravichander and Jillian Fisher and Taylor Sorensen and Ximing Lu and Yuchen Lin and Maria Antoniak and Niloofar Mireshghallah and Chandra Bhagavatula and Yejin Choi},
      year={2025},
      eprint={2503.12072},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2503.12072}, 
}
```