Commit
•
8812ade
1
Parent(s):
27b9dce
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,136 +1,136 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
tags:
|
| 4 |
-
- nifty
|
| 5 |
-
- stock-movement
|
| 6 |
-
- news-and-events
|
| 7 |
-
- RLMF
|
| 8 |
-
task_categories:
|
| 9 |
-
- multiple-choice
|
| 10 |
-
- time-series-forecasting
|
| 11 |
-
- document-question-answering
|
| 12 |
-
task_ids:
|
| 13 |
-
- topic-classification
|
| 14 |
-
- semantic-similarity-classification
|
| 15 |
-
- multiple-choice-qa
|
| 16 |
-
- univariate-time-series-forecasting
|
| 17 |
-
- document-question-answering
|
| 18 |
-
language:
|
| 19 |
-
- en
|
| 20 |
-
pretty_name: nifty-rl
|
| 21 |
-
size_categories:
|
| 22 |
-
- 1K<n<100k
|
| 23 |
-
configs:
|
| 24 |
-
- config_name: nifty-rl
|
| 25 |
-
data_files:
|
| 26 |
-
- split: train
|
| 27 |
-
path: "train.jsonl"
|
| 28 |
-
- split: test
|
| 29 |
-
path: "test.jsonl"
|
| 30 |
-
- split: valid
|
| 31 |
-
path: "valid.jsonl"
|
| 32 |
-
default: true
|
| 33 |
-
|
| 34 |
-
---
|
| 35 |
-
|
| 36 |
-
<h1>
|
| 37 |
-
<img alt="RH" src="./nifty-icon.png" style="display:inline-block; vertical-align:middle; width:120px; height:120px; object-fit:contain" />
|
| 38 |
-
The News-Informed Financial Trend Yield (NIFTY) Dataset.
|
| 39 |
-
</h1>
|
| 40 |
-
|
| 41 |
-
The News-Informed Financial Trend Yield (NIFTY) Dataset.
|
| 42 |
-
|
| 43 |
-
## 📋 Table of Contents
|
| 44 |
-
|
| 45 |
-
- [🧩 NIFTY Dataset](#nifty-dataset)
|
| 46 |
-
- [📋 Table of Contents](#table-of-contents)
|
| 47 |
-
- [📖 Usage](#usage)
|
| 48 |
-
- [Downloading the dataset](#downloading-the-dataset)
|
| 49 |
-
- [Dataset structure](#dataset-structure)
|
| 50 |
-
- [Large Language Models](#large-language-models)
|
| 51 |
-
- [✍️ Contributing](#contributing)
|
| 52 |
-
- [📝 Citing](#citing)
|
| 53 |
-
- [🙏 Acknowledgements](#acknowledgements)
|
| 54 |
-
|
| 55 |
-
## 📖 [Usage](#usage)
|
| 56 |
-
|
| 57 |
-
Downloading and using this dataset should be straight-forward following the Huggingface datasets framework.
|
| 58 |
-
|
| 59 |
-
### [Downloading the dataset](#downloading-the-dataset)
|
| 60 |
-
|
| 61 |
-
The NIFTY dataset is available on huggingface [here](https://huggingface.co/datasets/raeidsaqur/NIFTY) and can be downloaded with the following python snipped:
|
| 62 |
-
|
| 63 |
-
```python
|
| 64 |
-
|
| 65 |
-
from datasets import load_dataset
|
| 66 |
-
|
| 67 |
-
# If the dataset is gated/private, make sure you have run huggingface-cli login
|
| 68 |
-
dataset = load_dataset("raeidsaqur/nifty-rl")
|
| 69 |
-
|
| 70 |
-
```
|
| 71 |
-
|
| 72 |
-
### [Dataset structure](#dataset-structure)
|
| 73 |
-
|
| 74 |
-
The dataset is split into 3 partition, train, valid and test and each partition is a jsonl file where a single row has the following keys.
|
| 75 |
-
|
| 76 |
-
```python
|
| 77 |
-
['prompt', 'chosen', 'rejected', 'chosen_label', 'chosen_value']
|
| 78 |
-
```
|
| 79 |
-
|
| 80 |
-
Currently, the dataset has 2111 examples in total, the dates randing from 2010-01-06 to 2020-09-21.
|
| 81 |
-
<!-- The number of examples for each split is given below.
|
| 82 |
-
| Split | Num Examples | Date range |
|
| 83 |
-
|-------|--------------|------------|
|
| 84 |
-
|Train |1477 |2010-01-06 - 2017-06-27 |
|
| 85 |
-
|Valid|317 | 2017-06-28- 2019-02-12|
|
| 86 |
-
|Test |317|2019-02-13 - 2020-09-21|
|
| 87 |
-
-->
|
| 88 |
-
<!--
|
| 89 |
-
<img alt="St" src="./imgs/visualize_nifty_1794_2019-02-13.png"
|
| 90 |
-
style="display:inline-block; vertical-align:middle; width:640px;
|
| 91 |
-
height:640px; object-fit:contain" />
|
| 92 |
-
|
| 93 |
-
-->
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
## ✍️ [Contributing](#contributing)
|
| 98 |
-
|
| 99 |
-
We welcome contributions to this repository (noticed a typo? a bug?). To propose a change:
|
| 100 |
-
|
| 101 |
-
```
|
| 102 |
-
git clone https://huggingface.co/datasets/raeidsaqur/nifty-rl
|
| 103 |
-
cd nifty-rl
|
| 104 |
-
git checkout -b my-branch
|
| 105 |
-
pip install -r requirements.txt
|
| 106 |
-
pip install -e .
|
| 107 |
-
```
|
| 108 |
-
|
| 109 |
-
Once your changes are made, make sure to lint and format the code (addressing any warnings or errors):
|
| 110 |
-
|
| 111 |
-
```
|
| 112 |
-
isort .
|
| 113 |
-
black .
|
| 114 |
-
flake8 .
|
| 115 |
-
```
|
| 116 |
-
|
| 117 |
-
Then, submit your change as a pull request.
|
| 118 |
-
|
| 119 |
-
## 📝 [Citing](#citing)
|
| 120 |
-
|
| 121 |
-
If you use the NIFTY Financial dataset in your work, please consider citing our paper:
|
| 122 |
-
|
| 123 |
-
```
|
| 124 |
-
@article{
|
| 125 |
-
title = {NIFTY Financial News Headlines Dataset},
|
| 126 |
-
author = {Raeid Saqur},
|
| 127 |
-
year = 2024,
|
| 128 |
-
journal = {ArXiv},
|
| 129 |
-
url = {https://arxiv.org/abs/2024.5599314}
|
| 130 |
-
}
|
| 131 |
-
```
|
| 132 |
-
|
| 133 |
-
## 🙏 [Acknowledgements](#acknowledgements)
|
| 134 |
-
|
| 135 |
-
The authors acknowledge and thank the generous computing provided by the Vector Institute, Toronto.
|
| 136 |
-
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
- nifty
|
| 5 |
+
- stock-movement
|
| 6 |
+
- news-and-events
|
| 7 |
+
- RLMF
|
| 8 |
+
task_categories:
|
| 9 |
+
- multiple-choice
|
| 10 |
+
- time-series-forecasting
|
| 11 |
+
- document-question-answering
|
| 12 |
+
task_ids:
|
| 13 |
+
- topic-classification
|
| 14 |
+
- semantic-similarity-classification
|
| 15 |
+
- multiple-choice-qa
|
| 16 |
+
- univariate-time-series-forecasting
|
| 17 |
+
- document-question-answering
|
| 18 |
+
language:
|
| 19 |
+
- en
|
| 20 |
+
pretty_name: nifty-rl
|
| 21 |
+
size_categories:
|
| 22 |
+
- 1K<n<100k
|
| 23 |
+
configs:
|
| 24 |
+
- config_name: nifty-rl
|
| 25 |
+
data_files:
|
| 26 |
+
- split: train
|
| 27 |
+
path: "train.jsonl"
|
| 28 |
+
- split: test
|
| 29 |
+
path: "test.jsonl"
|
| 30 |
+
- split: valid
|
| 31 |
+
path: "valid.jsonl"
|
| 32 |
+
default: true
|
| 33 |
+
|
| 34 |
+
---
|
| 35 |
+
|
| 36 |
+
<h1>
|
| 37 |
+
<img alt="RH" src="./nifty-icon.png" style="display:inline-block; vertical-align:middle; width:120px; height:120px; object-fit:contain" />
|
| 38 |
+
The News-Informed Financial Trend Yield (NIFTY) Dataset.
|
| 39 |
+
</h1>
|
| 40 |
+
|
| 41 |
+
The News-Informed Financial Trend Yield (NIFTY) Dataset.
|
| 42 |
+
|
| 43 |
+
## 📋 Table of Contents
|
| 44 |
+
|
| 45 |
+
- [🧩 NIFTY Dataset](#nifty-dataset)
|
| 46 |
+
- [📋 Table of Contents](#table-of-contents)
|
| 47 |
+
- [📖 Usage](#usage)
|
| 48 |
+
- [Downloading the dataset](#downloading-the-dataset)
|
| 49 |
+
- [Dataset structure](#dataset-structure)
|
| 50 |
+
- [Large Language Models](#large-language-models)
|
| 51 |
+
- [✍️ Contributing](#contributing)
|
| 52 |
+
- [📝 Citing](#citing)
|
| 53 |
+
- [🙏 Acknowledgements](#acknowledgements)
|
| 54 |
+
|
| 55 |
+
## 📖 [Usage](#usage)
|
| 56 |
+
|
| 57 |
+
Downloading and using this dataset should be straight-forward following the Huggingface datasets framework.
|
| 58 |
+
|
| 59 |
+
### [Downloading the dataset](#downloading-the-dataset)
|
| 60 |
+
|
| 61 |
+
The NIFTY dataset is available on huggingface [here](https://huggingface.co/datasets/raeidsaqur/NIFTY) and can be downloaded with the following python snipped:
|
| 62 |
+
|
| 63 |
+
```python
|
| 64 |
+
|
| 65 |
+
from datasets import load_dataset
|
| 66 |
+
|
| 67 |
+
# If the dataset is gated/private, make sure you have run huggingface-cli login
|
| 68 |
+
dataset = load_dataset("raeidsaqur/nifty-rl")
|
| 69 |
+
|
| 70 |
+
```
|
| 71 |
+
|
| 72 |
+
### [Dataset structure](#dataset-structure)
|
| 73 |
+
|
| 74 |
+
The dataset is split into 3 partition, train, valid and test and each partition is a jsonl file where a single row has the following keys.
|
| 75 |
+
|
| 76 |
+
```python
|
| 77 |
+
['prompt', 'chosen', 'rejected', 'chosen_label', 'chosen_value']
|
| 78 |
+
```
|
| 79 |
+
|
| 80 |
+
Currently, the dataset has 2111 examples in total, the dates randing from 2010-01-06 to 2020-09-21.
|
| 81 |
+
<!-- The number of examples for each split is given below.
|
| 82 |
+
| Split | Num Examples | Date range |
|
| 83 |
+
|-------|--------------|------------|
|
| 84 |
+
|Train |1477 |2010-01-06 - 2017-06-27 |
|
| 85 |
+
|Valid|317 | 2017-06-28- 2019-02-12|
|
| 86 |
+
|Test |317|2019-02-13 - 2020-09-21|
|
| 87 |
+
-->
|
| 88 |
+
<!--
|
| 89 |
+
<img alt="St" src="./imgs/visualize_nifty_1794_2019-02-13.png"
|
| 90 |
+
style="display:inline-block; vertical-align:middle; width:640px;
|
| 91 |
+
height:640px; object-fit:contain" />
|
| 92 |
+
|
| 93 |
+
-->
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
## ✍️ [Contributing](#contributing)
|
| 98 |
+
|
| 99 |
+
We welcome contributions to this repository (noticed a typo? a bug?). To propose a change:
|
| 100 |
+
|
| 101 |
+
```
|
| 102 |
+
git clone https://huggingface.co/datasets/raeidsaqur/nifty-rl
|
| 103 |
+
cd nifty-rl
|
| 104 |
+
git checkout -b my-branch
|
| 105 |
+
pip install -r requirements.txt
|
| 106 |
+
pip install -e .
|
| 107 |
+
```
|
| 108 |
+
|
| 109 |
+
Once your changes are made, make sure to lint and format the code (addressing any warnings or errors):
|
| 110 |
+
|
| 111 |
+
```
|
| 112 |
+
isort .
|
| 113 |
+
black .
|
| 114 |
+
flake8 .
|
| 115 |
+
```
|
| 116 |
+
|
| 117 |
+
Then, submit your change as a pull request.
|
| 118 |
+
|
| 119 |
+
## 📝 [Citing](#citing)
|
| 120 |
+
|
| 121 |
+
If you use the NIFTY Financial dataset in your work, please consider citing our paper:
|
| 122 |
+
|
| 123 |
+
```
|
| 124 |
+
@article{raeidsaqur2024NiftyRL,
|
| 125 |
+
title = {NIFTY-RL: Financial News Headlines Dataset for LLM Alignment using Reinforcement Learning.},
|
| 126 |
+
author = {Raeid Saqur},
|
| 127 |
+
year = 2024,
|
| 128 |
+
journal = {ArXiv},
|
| 129 |
+
url = {https://arxiv.org/abs/2024.5599314}
|
| 130 |
+
}
|
| 131 |
+
```
|
| 132 |
+
|
| 133 |
+
## 🙏 [Acknowledgements](#acknowledgements)
|
| 134 |
+
|
| 135 |
+
The authors acknowledge and thank the generous computing provided by the Vector Institute, Toronto.
|
| 136 |
+
|