Update README.md
Browse filesThis dataset is built from scrapped speeches on the Bank of International Settlements. The dataset is made of 12k speeches from 1997 to 2023.
Each pair is built with extracted sentences from speeches, if B is following A then the 'next_sentence_label' is 1 else it is 0.
Negative pairs are built by choosing a sentence from another speech randomly.
README.md
CHANGED
@@ -15,7 +15,19 @@ dataset_info:
|
|
15 |
num_bytes: 89252509.32784761
|
16 |
num_examples: 136482
|
17 |
download_size: 365034957
|
18 |
-
dataset_size: 595014767
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
---
|
20 |
# Dataset Card for "BIS_Speeches_97_23"
|
21 |
|
|
|
15 |
num_bytes: 89252509.32784761
|
16 |
num_examples: 136482
|
17 |
download_size: 365034957
|
18 |
+
dataset_size: 595014767
|
19 |
+
license: apache-2.0
|
20 |
+
task_categories:
|
21 |
+
- text-classification
|
22 |
+
- token-classification
|
23 |
+
language:
|
24 |
+
- en
|
25 |
+
tags:
|
26 |
+
- economics
|
27 |
+
- finance
|
28 |
+
- business
|
29 |
+
size_categories:
|
30 |
+
- 100K<n<1M
|
31 |
---
|
32 |
# Dataset Card for "BIS_Speeches_97_23"
|
33 |
|