Skip to main content

FAQ

Why does my code work locally but fail with Grid Run?​

This is likely an environment mismatch. For Python users we recommend testing your code in a virtual environment with Python or Conda and then running a pip freeze to get all the necessary packages. For example:

# Create conda environment
conda create --name test-environment
pip install -r requirments.txt

# ensure code works

# copy dependencies to requirements.txt
pip freeze > requirements.txt

How do I download my run/experiment artifacts?​

From the CLI​

It's as easy as running grid artifacts my-run-name! This will download all artifacts fromthe run into a new directory called grid_artifacts.

From the UI​

How long are artifacts stored?​

Artifacts are stored until the run or experiment that generated the artifacts is deleted.

Why isn't Grid locating my requirements.txt file?​

Grid by default will look for a requirements.txt file in the root directory of your project. To customize this behavior try using the --dependency_file flag. An example is below: grid run --dependency_file ./path/to/requirements.txt model.py

How can I resolve Windows pathing issues when using Grid?​

There are two options users can try:

  1. default to the Windows Subsystem for Linux
  2. Try running your commands with by replacing \ characters with \\ in your path.

The cost changed during a run?​

Grid provides estimates of ongoing costs. Once a run terminates we compute the final cost.

How do I find out what packages are pre-installed in the cloud machine?​

Cloud machines are configured in a simple way with only what is needed to run the scripts in the framework of choice.

What machine learning frameworks does Grid support?​

Grid is optimized for PyTorch Lightning. It also supports Tensorflow, Keras, or any framework built on top of PyTorch.

Try this repository for running Keras example: https://github.com/gridai/hello_mnists/blob/443d9522/keras.py

Grid can also run non-deep learning focused workloads such as plain numpy, sklearn, etc..

I'm using an in-house ML library. Can I use it with Grid?​

Grid can run arbitrary python scripts. You're free to run whatever you want inside a script. However, Grid is optimized for Pytorch, Pytorch Lightning, Tensorflow, Keras, numpy and sklearn.

I am getting lot of errors using CLI​

Grid supports Linux based operating systems. We recommend using virtual environment when using CLI. Please see the guidance here

Experiments are queued for a long time​

If experiments are queued for a long time it could be a sign that instance type requested is not available; reach out on slack or open a github issue if you see this: https://github.com/gridai/gridai/issues. Choosing another instance type may help in some cases.

If the following FAQ didn't help resolve your issue please file a support ticket at support@grid.ai or reach out to the community at community Slack.