3 ms·
wouldn't that mean trying to fit it on one machine?
by mlstudies 3y ago
wouldn't that mean trying to fit it on one machine?
- brucethemoose2 3y agoIndeed :P Honestly I'm not sure how context "sharding" works on multiple GPUs atm. Decent, really long context OSS models like Yi 200K and YARN finetunes are very new.