XMC - XRTransformer - Recommended way to train on large datasets(200k labels)
codemonk2023 opened this issue · 1 comments
codemonk2023 commented
Hello,
Is there any recommended way to run XRTransformer on large datasets in terms of AWS machine and environment setup including parameters? I just ran and my python notebook stopped without any logs? Also is there a way to log the errors and warnings?
jiong-zhang commented
You can refer to examples here for parameter setups for datasets from 4k to 3m labels (requires p3.16xlarge or better instance).