# DL Assignment | Toxic Comment Classification — Data

来源：https://www.kaggle.com/competitions/dl-assignment-toxic-comment-classification/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

The dataset consists of thousands of toxic comments categorized into three classes. To simulate real-world filter bypassing, a portion of this data contains modern internet slang, emojis, and leetspeak (e.g., replacing vowels with numbers or symbols).

**Files**

**train.csv** - The training set. This file contains the id, the raw Toxic Comment, and the true Classification label. Use this to train your model.

**test.csv** - The test set. This file contains only the id and the Toxic Comment. You must predict the classification for these rows.

**sample_submission.csv** - A sample submission file in the correct format to help you prepare your final upload.