DaiFu rewrites a DL training function into exception-wrapped cells so a crash can be patched and resumed in place, restoring runs in seconds with under 0.4% overhead.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DaiFu: In-Situ Crash Recovery for Deep Learning Systems
DaiFu rewrites a DL training function into exception-wrapped cells so a crash can be patched and resumed in place, restoring runs in seconds with under 0.4% overhead.