get_stats.py

import pandas as pd
from datasets import load_dataset

table = []
task_description = {
    'tweet_intimacy': "regression on a single text",
    'tweet_ner7': "sequence labeling",
    'tweet_qa': "generation",
    'tweet_similarity': "regression on two texts",
    'tweet_topic': "multi-label classification",
    "tempo_wic": "binary classification on two texts",
    "tweet_sentiment": "ABSA on a five-point scale",
    "tweet_hate": "multi-class classification",
    "tweet_emoji": "multi-class classification",
    "tweet_nerd": "binary classification"
}
for task in task_description.keys():
    data = load_dataset("cardiffnlp/super_tweet_eval", task)
    tmp_table = {"task": task, "description": task_description[task]}
    tmp_table['number of instances'] = " / ".join([str(len(data[s])) for s in ['train', 'validation', 'test']])
    table.append(tmp_table)

df = pd.DataFrame(table)
print(df.to_markdown(index=False))