lance has no module 'dataset'

[Aryan-Deshpande]

Error Description:
Collecting lance
Downloading lance-1.2.1-py3-none-any.whl (22 kB)
Installing collected packages: lance
Successfully installed lance-1.2.1
AttributeError Traceback (most recent call last)
in <cell line: 3>()
1 get_ipython().system('pip install lance')
2 import lance
----> 3 dataset = lance.dataset('codeparrot_1M.lance/')
4 print(dataset.count_rows())
AttributeError: module 'lance' has no attribute 'dataset'

(colab)
Code:
!pip install lance
import lance
dataset = lance.dataset('codeparrot_1M.lance/')
print(dataset.count_rows())

[tanaymeh]

Hi @Aryan-Deshpande,

Can you try uninstalling lance and installing it like this?

$ pip install --upgrade pyarrow
$ pip install --upgrade pylance

The correct name for Lance for pip install is "pylance".

Please let us know if it works. Cheers!

[Aryan-Deshpande]

Hey @tanaymeh , The issue still persists

[changhiskhan]

you may have to uninstall the pypi package named lance first. Then reinstall pip install pylance afterwards

[tanaymeh]

@Aryan-Deshpande It works fine when I run it in a fresh Google Colab session.

Can you try deleting the colab runtime and starting afresh? Thanks!

[Aryan-Deshpande]

The previous issue was fixed, but I encountered another error

It says dataset was not found at the location content/codeparrot_1M.lance

[tanaymeh]

@Aryan-Deshpande It runs fine for me when I download the dataset following the instructions on the Kaggle Datasets page for the Codeparrot 1M dataset with changes for Google Colab since it downloads everything at /content.

You need to make the necessary changes in the path depending on what platform you are using (for example: /content/ for Google Colab or /kaggle/working/ for Kaggle, etc)

Here's the bash script that worked fine for me:

pip install -q kaggle pyarrow pylance
mkdir ~/.kaggle/
mv /content/kaggle.json ~/.kaggle/kaggle.json  # move kaggle.json API key to ~/.kaggle/ from /content
kaggle datasets download -d heyytanay/codeparrot-1m
mkdir /content/codeparrot_1M.lance/
unzip -qq /content/codeparrot-1m.zip -d /content/codeparrot_1M.lance/
rm /content/codeparrot-1m.zip

[Aryan-Deshpande]

Ah ok, so do we need to download the dataset first from kaggle then unpack and use it ?. Must have been my misunderstanding, I thought it would download it automatically.

[tanaymeh]

Converting this to a discussion since it's not an issue with Lance.