amazon-reviews / README.md
tomaarsen's picture
tomaarsen HF staff
Update README.md
b620302 verified
---
language:
- en
multilinguality:
- monolingual
size_categories:
- 10M<n<100M
task_categories:
- feature-extraction
- sentence-similarity
pretty_name: Amazon Reviews 2018
tags:
- sentence-transformers
dataset_info:
config_name: pair
features:
- name: title
dtype: string
- name: review
dtype: string
splits:
- name: train
num_bytes: 43047414971
num_examples: 87877725
download_size: 28943084017
dataset_size: 43047414971
configs:
- config_name: pair
data_files:
- split: train
path: pair/train-*
---
# Dataset Card for Amazon Reviews 2018
This dataset is a collection of title-review pairs collected from Amazon, as collected in [Ni et al.](https://cseweb.ucsd.edu//~jmcauley/pdfs/emnlp19a.pdf). See [Amazon Reviews 2018](https://nijianmo.github.io/amazon/index.html) for additional information.
This dataset can be used directly with Sentence Transformers to train embedding models.
## Dataset Subsets
### `pair` subset
* Columns: "title", "review"
* Column types: `str`, `str`
* Examples:
```python
{
'title': "It doesn't fit my machine. I can't seem to ...",
'review': "It doesn't fit my machine. I can't seem to find the part for the Mr. Coffee Tea maker I have.",
}
```
* Collection strategy: Reading the Amazon Reviews 2018 dataset from [embedding-training-data](https://huggingface.co/datasets/sentence-transformers/embedding-training-data).
* Deduplified: No