You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
TFRA dynamic_embedding has used cuckoohash_map to support fast, parallel read/write/export on host on CPU. But there is a fatal problem that when too many threads writes concurrently, leading to core dump when the rehashing threshold has been triggered, which is hard to track down.
Need to seek a safer kv on CPU, to make the training more robust.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
TFRA
dynamic_embedding
has usedcuckoohash_map
to support fast, parallel read/write/export on host on CPU. But there is a fatal problem that when too many threads writes concurrently, leading to core dump when the rehashing threshold has been triggered, which is hard to track down.Need to seek a safer kv on CPU, to make the training more robust.
Beta Was this translation helpful? Give feedback.
All reactions