map() not recognizing "text"

[EngineerKhan]

Describe the bug

The map() documentation reads: ds = ds.map(lambda x: tokenizer(x['text'], truncation=True, padding=True), batched=True)

I have been trying to reproduce it in my code as:

tokenizedDataset = dataset.map(lambda x: tokenizer(x['text']), batched=True)

But it doesn't work as it throws the error:

KeyError: 'text'

Can you please guide me on how to fix it?

Steps to reproduce the bug

1. `from datasets import load_dataset

dataset = load_dataset("amazon_reviews_multi")`

2. Then this code: `from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")`

3. The line I quoted above (which I have been trying)

Expected behavior

As mentioned in the documentation, it should run without any error and map the tokenization on the whole dataset.

Environment info

Python 3.10.2

[mariosasko]

There is no "text" column in the amazon_reviews_multi, hence the KeyError. You can get the column names by running dataset.column_names.