| 2025 | ICLR | MrT5: Dynamic Token Merging for Efficient Byte-level Language Models. | Julie Kallini, Shikhar Murty, Christopher D. Manning, Christopher Potts, Rbert Csords |
| 2025 | ICML | Measuring In-Context Computation Complexity via Hidden State Prediction. | Vincent Herrmann, Rbert Csords, Jrgen Schmidhuber |
| 2024 | ICANN | Self-organising Neural Discrete Representation Learning la Kohonen. | Kazuki Irie, Rbert Csords, Jrgen Schmidhuber |
| 2023 | ACL | Randomized Positional Encodings Boost Length Generalization of Transformers. | Anian Ruoss, Grgoire Deltang, Tim Genewein, Jordi Grau-Moya, Rbert Csords, Mehdi Bennani, Shane Legg, Joel Veness |
| 2023 | EMNLP | Approximating Two-Layer Feedforward Networks for Efficient Transformers. | Rbert Csords, Kazuki Irie, Jrgen Schmidhuber |
| 2023 | EMNLP | Practical Computational Power of Linear Transformers and Their Recurrent and Self-Referential Extensions. | Kazuki Irie, Rbert Csords, Jrgen Schmidhuber |
| 2022 | EMNLP | CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations. | Rbert Csords, Kazuki Irie, Jrgen Schmidhuber |
| 2022 | ICLR | The Neural Data Router: Adaptive Control Flow in Transformers Improves Systematic Generalization. | Rbert Csords, Kazuki Irie, Jrgen Schmidhuber |
| 2022 | ICML | The Dual Form of Neural Networks Revisited: Connecting Test Time Predictions to Training Patterns via Spotlights of Attention. | Kazuki Irie, Rbert Csords, Jrgen Schmidhuber |
| 2022 | ICML | A Modern Self-Referential Weight Matrix That Learns to Modify Itself. | Kazuki Irie, Imanol Schlag, Rbert Csords, Jrgen Schmidhuber |
| 2021 | EMNLP | The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers. | Rbert Csords, Kazuki Irie, Jrgen Schmidhuber |
| 2021 | ICLR | Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks. | Rbert Csords, Sjoerd van Steenkiste, Jrgen Schmidhuber |
| 2019 | ICLR | Improving Differentiable Neural Computers Through Memory Masking, De-allocation, and Link Distribution Sharpness Control. | Rbert Csords, Jrgen Schmidhuber |