Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
## About
|
2 |
+
This is a curated subset of 3 representative samples per class (77 classes in total) for the Banking77 dataset, as collected by a domain expert.
|
3 |
+
It was used in the paper "Making LLMs Worth Every Penny: Resource-Limited Text Classification in Banking", published in ACM ICAIF 2023 (https://arxiv.org/abs/2311.06102).
|
4 |
+
Our findings show that Few-Shot Text Classification on representative samples are better than randomly selected samples.
|
5 |
+
|
6 |
+
## Citation
|
7 |
+
|
8 |
+
```
|
9 |
+
@inproceedings{10.1145/3604237.3626891,
|
10 |
+
author = {Loukas, Lefteris and Stogiannidis, Ilias and Diamantopoulos, Odysseas and Malakasiotis, Prodromos and Vassos, Stavros},
|
11 |
+
title = {Making LLMs Worth Every Penny: Resource-Limited Text Classification in Banking},
|
12 |
+
year = {2023},
|
13 |
+
isbn = {9798400702402},
|
14 |
+
publisher = {Association for Computing Machinery},
|
15 |
+
address = {New York, NY, USA},
|
16 |
+
url = {https://doi.org/10.1145/3604237.3626891},
|
17 |
+
doi = {10.1145/3604237.3626891},
|
18 |
+
pages = {392–400},
|
19 |
+
numpages = {9},
|
20 |
+
keywords = {Anthropic, Cohere, OpenAI, LLMs, NLP, Claude, GPT, Few-shot},
|
21 |
+
location = {Brooklyn, NY, USA},
|
22 |
+
series = {ICAIF '23}
|
23 |
+
}
|
24 |
+
```
|
25 |
+
|
26 |
+
---
|
27 |
+
language:
|
28 |
+
- en
|
29 |
+
Tags:
|
30 |
+
- banking77
|
31 |
+
- classification
|
32 |
+
- conversational
|
33 |
+
---
|