3 ms·
I thought most ML engineers use their laptops as dumb terminals and just remote into a Linux GPU server.
by mysteria 2y ago
I thought most ML engineers use their laptops as dumb terminals and just remote into a Linux GPU server.
- daemonologist 2y agoYeah, the workday there looks pretty similar though, except that installing pytorch and pillow is usually no problem. Today it was flash-attn I spent the afternoon on.
- ungamedplayer 2y agoIsn't this what containers are for. Someone somewhere gets it configured right and then you download and run pre setup container and add your job data ? Or am I looking at the problem wrong?
- rolisz 2y agoBut then how do you test out the latest model that came out from who knows where and has the weirdest dependencies and a super obscure command to install?
- fshbbdssbbgdd 2y agoJust email all your data to the author and ask them to run it for you.
- davedx 2y agoSpoiler: my main role isn’t ML engineer :) and that doesn’t sound like a bad idea at all