The thumbs project in the course doesn’t seem to work with 2gb memory. I have everything working with a logitech USB camera, Jetpack 4.6, 4gb and swap enabled. At the last steo (the bit where you take 30 or so snapshots of thumb up or down) regular memory warning start appearing and things slow to crawl between camera freezes and cursor slow down. I manged to get 30 in for both thumbs u and down, but clicking the ‘train’ button did nothing for hours and a hard reset was required in the end. I’ve seen other posts on here saying similar (and trying to switch voltage mode, up priority of swap file) but the only solution was something about running the code outside of docker/jupyter or using the Hello AI World project as it does the same. The course mentions the 2GB version, so really it should all work surely? Great of Nvidia to launch a cheaper mass market 2gb version of the Nano, but not great not to optimize their own certified course content for it? Nvidia, can you comment. Or anyone else, easy solution to get thumbs project completed on a 2gb Nano? Thanks.
Hi @mikesteveworley, can you try these suggestions to see if it improves the performance of the notebooks?