Fix gpu loading for multi-GPU in LoRA, loading to cpu in week 10 + minor typing fixes and typos

[Abusagit]

Fixed the issue In the PEFT homework, namely, custom LoRA layers which are added to the network in the case of multiple GPU training, new low-rank tensors are placed on a device different to the full weight tensors. It caused the exception.

Also fixed some typos which could find and also added notations in functions where it can concern users.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[poedator]

Thank you for the proposed changes. To make this PR suitable for testing and review, please, re-submit it in clean and focused manner:

• one PR per lesson
• no unneeded merge commits
• no metadata changes