Last released Mar 28, 2026
Compress large embedding models into small, fast students via layer pruning, vocab pruning, hidden dim reduction, and knowledge distillation.
Supported by
