mrdbourke/learn_hf_food_not_food_image_captions

---
dataset_info:
features:
- name: text
dtype: string
- name: label
dtype: string
splits:
- name: train
num_bytes: 20253
num_examples: 250
download_size: 11945
dataset_size: 20253
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: apache-2.0
---

# Food/Not Food Image Caption Dataset

Small dataset of synthetic food and not food image captions.

Text generated using Mistral Chat/Mixtral.

Can be used to train a text classifier on food/not_food image captions as a demo before scaling up to a larger dataset.

See [Colab notebook](https://colab.research.google.com/drive/14xr3KN_HINY5LjV0s2E-4i7v0o_XI3U8?usp=sharing) on how dataset was created.

## Example usage

```python
import random
from datasets import load_dataset

# Load dataset
loaded_dataset = load_dataset("mrdbourke/learn_hf_food_not_food_image_captions")

# Get random index
rand_idx = random.randint(0, len(loaded_dataset["train"]))

# All samples are in the 'train' split by default (unless otherwise stated)
random_sample = loaded_dataset["train"][rand_idx]

print(f"Showing sample: {rand_idx}\n{random_sample}")
```

```
>>> Showing sample: 71
{'text': 'A kabob of grilled vegetables, including zucchini, squash, and onion, perfect for a summer barbecue.', 'label': 'food'}
```

原始信息汇总

Food/Not Food Image Caption Dataset

数据集概述

• 数据集名称: Food/Not Food Image Caption Dataset
• 数据集描述: 这是一个合成食物和非食物图像标题的小型数据集。
• 用途: 可用于训练文本分类器，识别食物/非食物图像标题，作为在扩展到更大数据集之前的演示。

数据集结构

• 特征:
  • text: 字符串类型，表示图像标题。
  • label: 字符串类型，表示标签（食物或非食物）。
• 分割:
  • train: 训练集，包含250个样本，总大小为20253字节。

数据集大小

• 下载大小: 11945字节
• 数据集大小: 20253字节

许可证

• 许可证: Apache 2.0