Hi there,
Thank you for authoring a wonderful introduction to language processing. I am having difficulty with chapter 2.
On trying to run the first learning section of chapter 2, the learn.fine_tune(4, 1e-2) command (#7 in the notebook) runs for a short period before generating the error below.
RuntimeError Traceback (most recent call last)
Input In [7], in <cell line: 1>()
----> 1 learn.fine_tune(4, 1e-2)
File ~/miniforge3/envs/nlpbook/lib/python3.9/site-packages/fastai/callback/schedule.py:164, in fine_tune(self, epochs, base_lr, freeze_epochs, lr_mult, pct_start, div, **kwargs)
162 base_lr /= 2
163 self.unfreeze()
--> 164 self.fit_one_cycle(epochs, slice(base_lr/lr_mult, base_lr), pct_start=pct_start, div=div, **kwargs)
...
...
...
RuntimeError: CUDA out of memory. Tried to allocate 118.00 MiB (GPU 0; 7.80 GiB total capacity; 6.47 GiB already allocated; 94.81 MiB free; 6.67 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
I have installed miniforge3, run the environment.yml file, followed by the two aws commands indicated in your github page. I am running this on an alienware laptop with rtx3070 and 8gb.
Is there anything that I can do to get this chapter working through a jupyter notebook? Please let me know if you would like a copy of the full output as I have just cut the first error, followed by the last error.
Thanks,
Michael.
Note from the Author or Editor:
Did you try modifying the max_split_size_mb?