---
dataset_info:
  features:
  - name: reviews
    sequence: string
  - name: labels
    sequence: int64
  - name: languages
    sequence: string
  - name: review_tokens
    sequence: string
  - name: token_labels
    sequence: int64
  - name: token_languages
    sequence: string
  - name: unique_language_count
    dtype: int64
  splits:
  - name: train
    num_bytes: 223198016
    num_examples: 60000
  - name: test
    num_bytes: 18490176
    num_examples: 5000
  - name: validation
    num_bytes: 18490176
    num_examples: 5000
  download_size: 74540072
  dataset_size: 260178368
language:
- de
- en
- es
- fr
---
# marc-code-mixed-small

This dataset is based on [The Multilingual Amazon Reviews Corpus](https://huggingface.co/datasets/amazon_reviews_multi).

It contains German (DE), English (EN), Spanish (ES), and French (FR) languages. 

The labels are 0 (DE), 1 (EN), 2 (ES), and 3 (FR). 

Each review contains all four languages.

Total number of tokens:
* In training set: 10195342
* In test set: 842760
* In validation set: 842760