Microsoft's small DP-finetuned transformer for keyboard prediction beats an older GRU in offline tests but shows no aggregate live gain, and its privacy guarantee for the modified sampling is unproven.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Privacy-Preserving Transformers: SwiftKey's Differential Privacy Implementation
Microsoft's small DP-finetuned transformer for keyboard prediction beats an older GRU in offline tests but shows no aggregate live gain, and its privacy guarantee for the modified sampling is unproven.