---
language:
- en
license_name: cc-by-4.0
license_link: https://creativecommons.org/licenses/by/4.0/legalcode
tags:
- coco
- mscoco
- simple-english
annotations_creators:
- machine-generated
pretty_name: MS COCO Karpathy in Simple English
size_categories:
- n<650k
source_datasets:
- mscoco
task_categories:
- text-generation
- image-to-text
- text-to-image
---

# Dataset Card for MS COCO Karpathy in Simple English

This dataset contains captions that were rephrased into simple english so that a young child would understand it.

## Dataset Details

### Dataset Description

- **Curated by:** {{ curators | default("[More Information Needed]", true)}}
- **Language(s) (NLP):** {{ language | default("[More Information Needed]", true)}}
- **License:** {{ license | default("[More Information Needed]", true)}}

### Dataset Sources

The processed [MS COCO datasets](https://cocodataset.org/#download) (Karpathy Split) in this repo are based on the following sources:

* train: http://images.cocodataset.org/zips/train2014.zip
* val: http://images.cocodataset.org/zips/val2014.zip 
* test: http://images.cocodataset.org/zips/test2014.zip

Files uses:

* aa31ac474cf6250ebb81d18348a07ed8  coco_karpathy_train.json
* 3ff34b0ef2db02d01c37399f6a2a6cd1  coco_karpathy_test.json
* b273847456ef5580e33713b1f7de52a0  coco_karpathy_val.json

MS COCO:

- **Download:** https://cocodataset.org/#download
- **Paper:** http://arxiv.org/abs/1405.0312


## Dataset Creation

This dataset was generated by processing the annotations via [Mistal7B](https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-AWQ).


Prompt used:

```
Rewrite the sentence " + caption + " for a 3 to 4 year old child. Give only one simple sentence. Don't use the word see. Give only a single answer.
```

A filter was applied to only store captions which matched the common output format. A best effort filter was applied to reduce the chance of including multiple example sentences in the output.

### Curation Rationale

This dataset is useful for experiments with small LLMs which have only a reduced corpus. The dataset is suitable to be used for LAVIS experiments (QFormer Training) with a finetuned TinyStories 33M LLM.