auto_lr_find gives error (NameError: name 'trainer' is not defined)

[lyndonlauder]

Hi i am trying to use lightning auto_lr_find, i use the command trainer.tune(model) and it gives me this error

NameError: name 'trainer' is not defined

Here is the model code and link

https://github.com/kuielab/mdx-net/blob/main/src/models/mdxnet.py
https://github.com/kuielab/mdx-net/blob/8e0476f29f330e2fe31cbb2833737ad0b2df7ad0/src/utils/utils.py

Please do you have any advice for this?

[akihironitta]

@lyndonlauder Could you please provide your env detail and full error message? You can use this template here: https://github.com/Lightning-AI/lightning/blob/master/.github/ISSUE_TEMPLATE/bug_report.md

[akihironitta]

@lyndonlauder I've converted your issue to a GitHub Discussion because this is the right place to ask questions and GitHub issues are basically used for bug reports or feature requests!

---

As the error states, there seems no trainer defined before you run trainer.fit(...) :)

[lyndonlauder]

@akihironitta do you know how i can define trainer because i am not experienced with code. Thank you for your time.

[akihironitta]

NameError Traceback (most recent call last)
in ()
----> 1 trainer.fit(model)

NameError: name 'trainer' is not defined

Where is the trainer.fit call exactly? I couldn't find any from the links you've provided above.

[lyndonlauder]

@akihironitta here https://github.com/kuielab/mdx-net/blob/002b33f2b4c0dbc8bb2b93ac880fafde2d42ff6f/src/train.py#L88

[akihironitta]

It should be defined at: https://github.com/kuielab/mdx-net/blob/002b33f2b4c0dbc8bb2b93ac880fafde2d42ff6f/src/train.py#L71

Unfortunately, I'm not familiar with hydra. I'd suggest you creating an issue in their repository. You have already :) kuielab/mdx-net#37

[lyndonlauder]

Thank you @akihironitta here it what you asked for

To Reproduce: trainer.tune(model)
Expected behavior: auto_lr_find works
Environment: conda google colab
PyTorch Lightning Version (e.g., 1.5.0): 1.6.4
PyTorch Version (e.g., 1.10): 1.8.1
Python version (e.g., 3.9): 3.8
OS (e.g., Linux): Google colab
CUDA/cuDNN version: cudatoolkit-11.7.0
GPU models and configuration: google colab p100
How you installed PyTorch (conda, pip, source): conda

Full error:

NameError Traceback (most recent call last)
in ()
----> 1 trainer.fit(model)

NameError: name 'trainer' is not defined