SafeLM unifies privacy-preserving federated LLM training with Paillier encryption, attack defenses, contrastive grounding, and binarized aggregation to achieve 98% harmful content detection, 96.9% less communication, and much lower gradient inversion risk.
On faithfulness and factuality in abstractive summarization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SafeLM: Unified Privacy-Aware Optimization for Trustworthy Federated Large Language Models
SafeLM unifies privacy-preserving federated LLM training with Paillier encryption, attack defenses, contrastive grounding, and binarized aggregation to achieve 98% harmful content detection, 96.9% less communication, and much lower gradient inversion risk.