Divide and Rule: Effective Pre-Training for Context-Aware Multi-Encoder Translation Models

Open in new window