✨
AI Summary
- Channel-wise MLPs in recurrent convolutional networks improve generalization.
- DAMP architecture with gated MLPs outperforms standard DARC on Re-ARC benchmark.
- Explicit channel mixing enhances network performance.