Checklist NOAI China 2024 Round 2 (China Stage) · Task 3
News Text Classification Task
English title: 新闻文本分类任务
Train a PyTorch text classifier that assigns news articles to categories within a 10-minute CPU budget.
The task
The dataset consists of news texts (column text) with their category (column category). The training set train_news.csv has 1,000 samples; the test set has 200 samples, provided without labels during the contest.
The contestant designs and trains a natural-language-processing model in PyTorch that outputs the category of each news text. Word embeddings with an LSTM are recommended.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
train_news.csv(1,000 rows) from the Bohrium datasets tab;test_news_nolabel.csv(200 rows) is not downloadable.- You submit
submission.ipynbcontaining the full training process and writing the test predictions tosubmission.csv, with labels named as intrain_news.csv.- Scoring
- Mean of the per-category F1 scores (macro F1) if training and testing finish within the time limit; 0 if the per-category F1 scores cannot be computed or the time limit is exceeded.
- Rules
- CPU training plus testing time must not exceed 10 minutes (connection and queueing time excluded).
- Format
- NOAI 2024 Round 2 (China Stage), a one-day practical round in Beijing on 10 June 2024 (date and place from a third-party overview page). Republished on Bohrium as the 'NOAI2025 teaching test (NOAI2024 real problems)' and, in English, as the 'APOAI2025 Mock Competition' (10 Nov 2024 - 31 May 2025). Leaderboard A uses 50% of the test set during the contest; leaderboard B (the remaining 50%) is final. Question 3.