Model Compression via Knowledge Distillation

How knowledge distillation compresses teacher models into compact students by transferring behavior and using tailored training objectives for efficient models.