clinicalml / TabLLM

MIT License
265 stars 42 forks source link

Error with Creating Serialized Datasets #20

Open Ping-Song opened 7 months ago

Ping-Song commented 7 months ago

When I was creating serialized datasets with the following code, it always show the error 'no matches found' Code: create_external_datasets.py --dataset (car|income|diabetes|heart|bank|blood|calhousing|creditg|jungle) (--list) (--list (--tabletotext|--t0serialization|--values|--permuted|--shuffled)) Error: zsh: no matches found: (car|income|diabetes|heart|bank|blood|calhousing|creditg|jungle)

I did run the previous code in Preparing the Environments section.

Ping-Song commented 7 months ago

I also found out that cudatoolkit cannot be installed, do you know why?

stefanhgm commented 7 months ago

Hello @Ping-Song ,

thanks for using our code and for reaching out with this issue.

Just to be sure, what is the exact command that you are running?

To get a serialization for the car dataset you should run:

create_external_datasets.py --dataset car

Does this work or what output does it create?

Thank you!

Ping-Song commented 6 months ago

Hi, thank you for your reply. I tried it again, it says "zsh: command not found: create_external_datasets.py"

Ping-Song commented 6 months ago

When I was running "conda install pytorch==1.10.1 torchvision==0.11.2 torchaudio==0.10.1 cudatoolkit=11.3 -c pytorch -c conda-forge". There are PackageNotFoundError:

PackagesNotFoundError: The following packages are not available from current channels:

Could you please tell me how to install this 2 packages or what alternative package I can use?

Ping-Song commented 6 months ago

Hello @Ping-Song ,

thanks for using our code and for reaching out with this issue.

Just to be sure, what is the exact command that you are running?

To get a serialization for the car dataset you should run:

create_external_datasets.py --dataset car

Does this work or what output does it create?

Thank you!

Hi, should I run "create_external_datasets.py --dataset car" or "create_external_datasets.py --dataset car (--list) (--list (--tabletotext|--t0serialization|--values|--permuted|--shuffled))" as shown in this github instruction?

Ping-Song commented 6 months ago

Hi, I managed to run the serialization, but there is one additional error: AttributeError: module 'huggingface_hub.constants' has no attribute 'HF_HUB_CACHE'.

Can you help me with this please? Thanks!

stefanhgm commented 6 months ago

Hello @Ping-Song ,

Hi, thank you for your reply. I tried it again, it says "zsh: command not found: create_external_datasets.py"

Please make sure that you are in the correct directory.

Hi, should I run "create_external_datasets.py --dataset car" or "create_external_datasets.py --dataset car (--list) (--list (--tabletotext|--t0serialization|--values|--permuted|--shuffled))" as shown in this github instruction?

The second command does not work. It just shows all possible options you can run the script with. You have to choose the options depending on what you need. The first command should work.

When I was running "conda install pytorch==1.10.1 torchvision==0.11.2 torchaudio==0.10.1 cudatoolkit=11.3 -c pytorch -c conda-forge". There are PackageNotFoundError:

PackagesNotFoundError: The following packages are not available from current channels:

cudatoolkit=11.3 torchaudio==0.10.1 Could you please tell me how to install this 2 packages or what alternative package I can use?

Sorry for that. The package can often cause problems as version are not compatible anymore. Often you have to try around a bit and check online for alternatives. What worked for me the last time I setup the environment was the following:

conda create -n tabllm python==3.8
conda activate tabllm
pip install fsspec==2023.5.0
pip install click==7.1.2

The using the requirements file of t-few (https://github.com/r-three/t-few/blob/master/requirements.txt).

pip install --use-deprecated=legacy-resolver  -r requirements.txt -f https://download.pytorch.org/whl/cu113/torch_stable.html
pip install urllib3==1.26.6
pip install importlib-metadata==4.13.0
pip install scikit-learn
pip install setuptools==59.5.0
pip install protobuf==3.20.3

I used the resulting environment for both t-few and TabLLM code.

Hope that works!

stefanhgm commented 6 months ago

Hello @Ping-Song,

Sorry, I do not know this error. Maybe setting the environment variable can help?

export HF_HUB_CACHE=~/.cache/huggingface

Ping-Song commented 6 months ago

Hello @Ping-Song,

Sorry, I do not know this error. Maybe setting the environment variable can help?

export HF_HUB_CACHE=~/.cache/huggingface

Hi Thanks for your prompt reply! I did set the environment variable.