metadata

dataset_info:
  features:
    - name: sentence
      dtype: string
    - name: cluster
      dtype: int64
    - name: embedding_384
      sequence: float32
  splits:
    - name: train
      num_bytes: 1584641313
      num_examples: 990526
  download_size: 2163024841
  dataset_size: 1584641313
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
task_categories:
  - sentence-similarity
language:
  - en
size_categories:
  - 100K<n<1M

Dataset Card for "wikianswers_embeddings_384"

Dataset Summary

nikhilchigali/wikianswers_embeddings_384 is a subset of the embedding-data/WikiAnswers (Link) As opposed to the original dataset with 3,386,256 rows, this dataset contains only .13% of the total rows(sets). The sets of sentences have been unraveled into individual items with corresponding cluster IDs to identify sentences from the same set. Each Sentence has its associated cluster ID and embeddings of dimension 384.

Languages

English.

Dataset Structure

Each example in the dataset contains a sentence and its cluster id of other equivalent sentences. The sentences in the same cluster are paraphrases of each other. The embeddings for the dataset are created using the all-MiniLM-L6-v2 model.

{"sentence": [sentence], "cluster": [cluster_id], "embedding_384": [embeddings]}
{"sentence": [sentence], "cluster": [cluster_id], "embedding_384": [embeddings]}
{"sentence": [sentence], "cluster": [cluster_id], "embedding_384": [embeddings]}
...
{"sentence": [sentence], "cluster": [cluster_id], "embedding_384": [embeddings]}

Usage Example

Install the 🤗 Datasets library with pip install datasets and load the dataset from the Hub with:

from datasets import load_dataset
dataset = load_dataset("nikhilchigali/wikianswers_embeddings_384")

The dataset is loaded as a DatasetDict and has the format for N examples:

DatasetDict({
    train: Dataset({
        features: ['sentence', "cluster", "embedding_384"],
        num_rows: N
    })
})

Review an example i with:

dataset["train"][i]

Source Data

embedding-data/WikiAnswers on HuggingFace (Link)

Datasets:

nikhilchigali
/

wikianswers_embeddings_384