Róbert Csordás

46 papers A* 11Journal 28Unranked 7

Year	Rank	Type	Title / Venue / Authors
2025	J	`jnl`	Decoupling the "What" and "Where" With Polar Coordinate Positional Embeddings. CoRR Anand Gopalakrishnan, Róbert Csordás, Jürgen Schmidhuber, Michael C. Mozer
2025	J	`jnl`	Do Language Models Use Their Depth Efficiently? CoRR Róbert Csordás, Christopher D. Manning, Christopher Potts
2025	J	`jnl`	GIM: Improved Interpretability for Large Language Models. CoRR Joakim Edin, Róbert Csordás, Tuukka Ruotsalo, Zhengxuan Wu, Maria Maistro, Jing Huang, Lars Maaløe
2025	A*	`conf`	Measuring In-Context Computation Complexity via Hidden State Prediction. ICML Vincent Herrmann, Róbert Csordás, Jürgen Schmidhuber
2025	J	`jnl`	Measuring In-Context Computation Complexity via Hidden State Prediction. CoRR Vincent Herrmann, Róbert Csordás, Jürgen Schmidhuber
2025	J	`jnl`	Metalearning Continual Learning Algorithms. Trans. Mach. Learn. Res. Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2025	J	`jnl`	Mindstorms in Natural Language-Based Societies of Mind. Comput. Vis. Media Mingchen Zhuge, Haozhe Liu, Francesco Faccio, Dylan R. Ashley, Róbert Csordás, Anand Gopalakrishnan, Abdullah Hamdi, Hasan Abed Al Kader Hammoud, Vincent Herrmann, Kazuki Irie, Louis Kirsch, Bing Li, Guohao Li, Shuming Liu, Jinjie Mai, Piotr Piekos, Aditya A. Ramesh, Imanol Schlag, Weimin Shi, Aleksandar Stanic, Wenyi Wang, Yuhui Wang, Mengmeng Xu, Deng-Ping Fan, Bernard Ghanem, Jürgen Schmidhuber
2025	J	`jnl`	Mixture of Sparse Attention: Content-Based Learnable Sparse Attention via Expert-Choice Routing. CoRR Piotr Piekos, Róbert Csordás, Jürgen Schmidhuber
2025	A*	`conf`	MrT5: Dynamic Token Merging for Efficient Byte-level Language Models. ICLR Julie Kallini, Shikhar Murty, Christopher D. Manning, Christopher Potts, Róbert Csordás
2025	J	`jnl`	Thoughtbubbles: an Unsupervised Method for Parallel Thinking in Latent Space. CoRR Houjun Liu, Shikhar Murty, Christopher D. Manning, Róbert Csordás
2024	A*	`conf`	MoEUT: Mixture-of-Experts Universal Transformers. NeurIPS Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber, Christopher Potts, Christopher D. Manning
2024	J	`jnl`	MoEUT: Mixture-of-Experts Universal Transformers. CoRR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber, Christopher Potts, Christopher D. Manning
2024	J	`jnl`	MrT5: Dynamic Token Merging for Efficient Byte-level Language Models. CoRR Julie Kallini, Shikhar Murty, Christopher D. Manning, Christopher Potts, Róbert Csordás
2024	J	`jnl`	Recurrent Neural Networks Learn to Store and Generate Sequences using Non-Linear Representations. CoRR Róbert Csordás, Christopher Potts, Christopher D. Manning, Atticus Geiger
2024	—	`conf`	Self-organising Neural Discrete Representation Learning à la Kohonen. ICANN (1) Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2024	A*	`conf`	SwitchHead: Accelerating Transformers with Mixture-of-Experts Attention. NeurIPS Róbert Csordás, Piotr Piekos, Kazuki Irie, Jürgen Schmidhuber
2023	—	`conf`	Approximating Two-Layer Feedforward Networks for Efficient Transformers. EMNLP (Findings) Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2023	J	`jnl`	Approximating Two-Layer Feedforward Networks for Efficient Transformers. CoRR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2023	J	`jnl`	Automating Continual Learning. CoRR Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2023	J	`jnl`	Mindstorms in Natural Language-Based Societies of Mind. CoRR Mingchen Zhuge, Haozhe Liu, Francesco Faccio, Dylan R. Ashley, Róbert Csordás, Anand Gopalakrishnan, Abdullah Hamdi, Hasan Abed Al Kader Hammoud, Vincent Herrmann, Kazuki Irie, Louis Kirsch, Bing Li, Guohao Li, Shuming Liu, Jinjie Mai, Piotr Piekos, Aditya A. Ramesh, Imanol Schlag, Weimin Shi, Aleksandar Stanic, Wenyi Wang, Yuhui Wang, Mengmeng Xu, Deng-Ping Fan, Bernard Ghanem, Jürgen Schmidhuber
2023	A*	`conf`	Practical Computational Power of Linear Transformers and Their Recurrent and Self-Referential Extensions. EMNLP Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2023	J	`jnl`	Practical Computational Power of Linear Transformers and Their Recurrent and Self-Referential Extensions. CoRR Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2023	—	`conf`	Randomized Positional Encodings Boost Length Generalization of Transformers. ACL (2) Anian Ruoss, Grégoire Delétang, Tim Genewein, Jordi Grau-Moya, Róbert Csordás, Mehdi Bennani, Shane Legg, Joel Veness
2023	J	`jnl`	Randomized Positional Encodings Boost Length Generalization of Transformers. CoRR Anian Ruoss, Grégoire Delétang, Tim Genewein, Jordi Grau-Moya, Róbert Csordás, Mehdi Bennani, Shane Legg, Joel Veness
2023	J	`jnl`	SwitchHead: Accelerating Transformers with Mixture-of-Experts Attention. CoRR Róbert Csordás, Piotr Piekos, Kazuki Irie, Jürgen Schmidhuber
2023	J	`jnl`	Topological Neural Discrete Representation Learning à la Kohonen. CoRR Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2022	—	`conf`	A Generalist Neural Algorithmic Learner. LoG Borja Ibarz, Vitaly Kurin, George Papamakarios, Kyriacos Nikiforou, Mehdi Bennani, Róbert Csordás, Andrew Joseph Dudzik, Matko Bosnjak, Alex Vitvitskyi, Yulia Rubanova, Andreea Deac, Beatrice Bevilacqua, Yaroslav Ganin, Charles Blundell, Petar Velickovic
2022	J	`jnl`	A Generalist Neural Algorithmic Learner. CoRR Borja Ibarz, Vitaly Kurin, George Papamakarios, Kyriacos Nikiforou, Mehdi Bennani, Róbert Csordás, Andrew Dudzik, Matko Bosnjak, Alex Vitvitskyi, Yulia Rubanova, Andreea Deac, Beatrice Bevilacqua, Yaroslav Ganin, Charles Blundell, Petar Velickovic
2022	A*	`conf`	A Modern Self-Referential Weight Matrix That Learns to Modify Itself. ICML Kazuki Irie, Imanol Schlag, Róbert Csordás, Jürgen Schmidhuber
2022	J	`jnl`	A Modern Self-Referential Weight Matrix That Learns to Modify Itself. CoRR Kazuki Irie, Imanol Schlag, Róbert Csordás, Jürgen Schmidhuber
2022	A*	`conf`	CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations. EMNLP Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2022	J	`jnl`	CTL++: Evaluating Generalization on Never-Seen Compositional Patterns of Known Functions, and Compatibility of Neural Representations. CoRR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2022	A*	`conf`	The Dual Form of Neural Networks Revisited: Connecting Test Time Predictions to Training Patterns via Spotlights of Attention. ICML Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2022	J	`jnl`	The Dual Form of Neural Networks Revisited: Connecting Test Time Predictions to Training Patterns via Spotlights of Attention. CoRR Kazuki Irie, Róbert Csordás, Jürgen Schmidhuber
2022	A*	`conf`	The Neural Data Router: Adaptive Control Flow in Transformers Improves Systematic Generalization. ICLR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2021	A*	`conf`	Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks. ICLR Róbert Csordás, Sjoerd van Steenkiste, Jürgen Schmidhuber
2021	A*	`conf`	Going Beyond Linear Transformers with Recurrent Fast Weight Programmers. NeurIPS Kazuki Irie, Imanol Schlag, Róbert Csordás, Jürgen Schmidhuber
2021	J	`jnl`	Going Beyond Linear Transformers with Recurrent Fast Weight Programmers. CoRR Kazuki Irie, Imanol Schlag, Róbert Csordás, Jürgen Schmidhuber
2021	J	`jnl`	Improving Baselines in the Wild. CoRR Kazuki Irie, Imanol Schlag, Róbert Csordás, Jürgen Schmidhuber
2021	—	`conf`	The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers. EMNLP (1) Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2021	J	`jnl`	The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers. CoRR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2021	J	`jnl`	The Neural Data Router: Adaptive Control Flow in Transformers Improves Systematic Generalization. CoRR Róbert Csordás, Kazuki Irie, Jürgen Schmidhuber
2020	J	`jnl`	Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks. CoRR Róbert Csordás, Sjoerd van Steenkiste, Jürgen Schmidhuber
2019	—	`conf`	Improving Differentiable Neural Computers Through Memory Masking, De-allocation, and Link Distribution Sharpness Control. ICLR (Poster) Róbert Csordás, Jürgen Schmidhuber
2019	J	`jnl`	Improving Differentiable Neural Computers Through Memory Masking, De-allocation, and Link Distribution Sharpness Control. CoRR Róbert Csordás, Jürgen Schmidhuber
2015	—	`conf`	Detecting Objects Thrown over Fence in Outdoor Scenes. VISAPP (3) Róbert Csordás, László Havasi, Tamás Szirányi