Datasets:

Modalities:
Tabular
Text
Formats:
csv
Languages:
English
ArXiv:
Libraries:
Datasets
pandas
License:
rosehaeun commited on
Commit
9353f66
·
verified ·
1 Parent(s): 1f6488a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -21,7 +21,7 @@ configs:
21
  This is a repository for the paper [DYNAMICQA: Tracing Internal Knowledge Conflicts in Language Models](https://arxiv.org/abs/2407.17023) accepted at Findings of EMNLP 2024.
22
 
23
  <p align="center">
24
- <img src="main_figure.png" width="650" alt="main_figure">
25
  </p>
26
 
27
  Our paper investigates the Language Model's behaviour when the conflicting knowledge exist within the LM's parameters. We present a novel dataset containing inherently conflicting data, DYNAMICQA. Our dataset consists of three partitions, **Static**, **Disputable** 🤷‍♀️, and **Temporal** 🕰️.
 
21
  This is a repository for the paper [DYNAMICQA: Tracing Internal Knowledge Conflicts in Language Models](https://arxiv.org/abs/2407.17023) accepted at Findings of EMNLP 2024.
22
 
23
  <p align="center">
24
+ <img src="main_figure.png" width="800" alt="main_figure">
25
  </p>
26
 
27
  Our paper investigates the Language Model's behaviour when the conflicting knowledge exist within the LM's parameters. We present a novel dataset containing inherently conflicting data, DYNAMICQA. Our dataset consists of three partitions, **Static**, **Disputable** 🤷‍♀️, and **Temporal** 🕰️.