Searcharxiv⌕ Search

EXPLORE CONNECTIONS

Privacy-preserving inference on hosted models: a learned input-embedding transform for frozen Qwen3.8-27B

Follow the relationships that help you find your next source.

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.