# Train\_test\_split with a dataset loaded from dict

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftrain-test-split-with-a-dataset-loaded-from-dict%2F25806%3E%5C
**Category:** Beginners\
**Created:** [November 9, 2022, 2:01pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftrain-test-split-with-a-dataset-loaded-from-dict%2F25806 "2022-11-09T14:01:50Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![kamneb](https://avatars.discourse-cdn.com/v4/letter/k/d26b3c/32.png) [@kamneb](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fkamneb)%5C
**Post date:** [November 9, 2022, 2:01pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftrain-test-split-with-a-dataset-loaded-from-dict%2F25806%2F1 "2022-11-09T14:01:50Z")

</div>

Hello,

I would like to split my dataset into train and test samples. My dataset was initially created with a dict.

So it looks likes this:

```auto
from datasets import Dataset
data = {"text": ["This is a sentence"]*100, "extra_data": np.random.randint(0, 10, size=(100, 5)), "labels": np.random.randint(0, 3, size=(100,))}
ds = Dataset.from_dict(data)

```

However when i try to split it with:

```auto
train_test = ds.train_test_split(test_size=0.2)

```

I have this error message:

```auto
pyarrow.lib.ArrowTypeError: Did not pass numpy.dtype object

```

thanks

---

<div class="post-metadata">

**Author:** ![kamneb](https://avatars.discourse-cdn.com/v4/letter/k/d26b3c/32.png) [@kamneb](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fkamneb)%5C
**Post date:** [November 9, 2022, 3:27pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ftrain-test-split-with-a-dataset-loaded-from-dict%2F25806%2F2 "2022-11-09T15:27:37Z")

</div>

I updated pyarrow version and dataset version. now it works