为了防止您希望将数据写入csv并将其检索回来，我有一个示例：

# Sample dataframe written to csv file and read back
d = {'id': [1, 2], 'pos_tag': [(1, 2), ("a", "b", "c")]}
df1 = pd.DataFrame(d)
df1.to_csv("/content/df1.csv", sep="|", index=False)
df2 = pd.read_csv("/content/df1.csv", sep="|", )
df2.head()

我用管道分隔符将其保存为csv，只是为了确保它与其他文字（如逗号、分号等）不同。

# Retrieve tuple values
df2['pos_tag'] = [eval(s) for s in df2['pos_tag']] 
df2.head()