2nd and 3rd commits are about @eleapttn YoloV5's notebook. A user has reported that this notebook was very laggy and then blocked during the COCO dataset unzipping.
I have added a -q to the unzip command, to make outputs silent. It now works perfectly 😉 (not massive output, so not laggy anymore).
I have also removed the /data/workspace from the rm command, since this command does not delete folders (only files). It was therefore useless for the /data/workspace path, especially as the unzipped images were stored there. This usage seems to me to be a mistake, and this is why i have removed it.
I have also removed the --cache parameter from the python train.py cmd, since the caching images process was always blocked at 90% in my case.
Then 3 commits have been done but reverted, since i have finally decided to commit these changes in another PR.
Here is a PR to correct smth in our notebooks.
1 st commit as about LLaMA2 notebook.
2nd and 3rd commits are about @eleapttn YoloV5's notebook. A user has reported that this notebook was very laggy and then blocked during the COCO dataset unzipping.
-qto the unzip command, to make outputs silent. It now works perfectly 😉 (not massive output, so not laggy anymore)./data/workspacefrom thermcommand, since this command does not delete folders (only files). It was therefore useless for the/data/workspacepath, especially as the unzipped images were stored there. This usage seems to me to be a mistake, and this is why i have removed it.python train.pycmd, since the caching images process was always blocked at 90% in my case.Then 3 commits have been done but reverted, since i have finally decided to commit these changes in another PR.