The Propensity for Density in Feed-forward Models von Nandi Schoots, Jackson, Alex, Kholmovaia, Ali, McBurney, Peter, Shanahan, Murray

Volltext

Dissecting Language Models: Machine Unlearning via Selective Pruning von Pochinkov, Nicholas, Schoots, Nandi

Volltext bestellen

Low-Entropy Latent Variables Hurt Out-of-Distribution Performance von Schoots, Nandi, Cope, Dylan

Volltext bestellen

Any Deep ReLU Network is Shallow von Villani, Mattia Jacopo, Schoots, Nandi

Volltext bestellen

Training Neural Networks for Modularity aids Interpretability von Golechha, Satvik, Cope, Dylan, Schoots, Nandi

Volltext bestellen

Learning to Communicate with Strangers via Channel Randomisation Methods von Cope, Dylan, Schoots, Nandi

Volltext bestellen

Dissecting Language Models: Machine Unlearning via Selective Pruning von Pochinkov, Nicholas, Nandi Schoots

Volltext

Extending Activation Steering to Broad Skills and Multiple Behaviours von van der Weij, Teun, Poesio, Massimo, Schoots, Nandi

Volltext bestellen

Low-Entropy Latent Variables Hurt Out-of-Distribution Performance von Nandi Schoots, Cope, Dylan

Volltext

Improving Activation Steering in Language Models with Mean-Centring von Jorgensen, Ole, Cope, Dylan, Schoots, Nandi, Shanahan, Murray

Volltext bestellen

Comparing Optimization Targets for Contrast-Consistent Search von Fry, Hugo, Fallows, Seamus, Fan, Ian, Wright, Jamie, Schoots, Nandi

Volltext bestellen

Hidden in Plain Text: Emergence & Mitigation of Steganographic Collusion in LLMs von Mathew, Yohan, Matthews, Ollie, McCarthy, Robert, Velja, Joan, de Witt, Christian Schroeder, Cope, Dylan, Schoots, Nandi

Volltext bestellen

Any Deep ReLU Network is Shallow von Villani, Mattia Jacopo, Nandi Schoots

Volltext

Training Neural Networks for Modularity aids Interpretability von Golechha, Satvik, Cope, Dylan, Nandi Schoots

Volltext

Learning to Communicate with Strangers via Channel Randomisation Methods von Cope, Dylan, Nandi Schoots

Volltext

Extending Activation Steering to Broad Skills and Multiple Behaviours von Teun van der Weij, Poesio, Massimo, Nandi Schoots

Volltext

A theory of representation learning gives a deep generalisation of kernel methods von Yang, Adam X, Robeyns, Maxime, Milsom, Edward, Anson, Ben, Schoots, Nandi, Aitchison, Laurence

Volltext bestellen

Improving Activation Steering in Language Models with Mean-Centring von Jorgensen, Ole, Cope, Dylan, Nandi Schoots, Shanahan, Murray

Volltext

Comparing Optimization Targets for Contrast-Consistent Search von Fry, Hugo, Fallows, Seamus, Fan, Ian, Wright, Jamie, Nandi Schoots

Volltext

A theory of representation learning gives a deep generalisation of kernel methods von Yang, Adam X, Robeyns, Maxime, Milsom, Edward, Anson, Ben, Nandi Schoots, Aitchison, Laurence

Volltext