Takip et
Adrià Garriga-Alonso
Adrià Garriga-Alonso
Research Scientist, FAR AI
far.ai üzerinde doğrulanmış e-posta adresine sahip - Ana Sayfa
Başlık
Alıntı yapanlar
Alıntı yapanlar
Yıl
Beyond the imitation game: Quantifying and extrapolating the capabilities of language models
A Srivastava, A Rastogi, A Rao, AAM Shoeb, A Abid, A Fisch, AR Brown, ...
arXiv preprint arXiv:2206.04615, 2022
14492022
Deep Convolutional Networks as shallow Gaussian Processes
A Garriga-Alonso, L Aitchison, CE Rasmussen
International Conference on Learning Representations, 2019
3132019
Towards automated circuit discovery for mechanistic interpretability
A Conmy, A Mavor-Parker, A Lynch, S Heimersheim, A Garriga-Alonso
Advances in Neural Information Processing Systems 36, 16318-16352, 2023
2292023
Bayesian neural network priors revisited
V Fortuin, A Garriga-Alonso, SW Ober, F Wenzel, G Rätsch, RE Turner, ...
arXiv preprint arXiv:2102.06571, 2021
1712021
Causal scrubbing: A method for rigorously testing interpretability hypotheses
L Chan, A Garriga-Alonso, N Goldowsky-Dill, R Greenblatt, ...
AI Alignment Forum 2, 2022
662022
Understanding variational inference in function-space
DR Burt, SW Ober, A Garriga-Alonso, M van der Wilk
arXiv preprint arXiv:2011.09421, 2020
532020
Exact Langevin dynamics with stochastic gradients
A Garriga-Alonso, V Fortuin
arXiv preprint arXiv:2102.01691, 2021
412021
Data augmentation in Bayesian neural networks and the cold posterior effect
S Nabarro, S Ganev, A Garriga-Alonso, V Fortuin, M van der Wilk, ...
Uncertainty in Artificial Intelligence, 1434-1444, 2022
342022
BNNpriors: A library for Bayesian neural network inference with different prior distributions
V Fortuin, A Garriga-Alonso, M van der Wilk, L Aitchison
Software Impacts 9, 100079, 2021
272021
Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models (arXiv: 2206.04615). arXiv
A Srivastava, A Rastogi, A Rao, AAM Shoeb, A Abid, A Fisch, AR Brown, ...
232023
Analyzing the Generalization and Reliability of Steering Vectors
D Tan, D Chanin, A Lynch, D Kanoulas, B Paige, A Garriga-Alonso, R Kirk
arXiv preprint arXiv:2407.12404, 2024
72024
Correlated weights in infinite limits of deep convolutional neural networks
A Garriga-Alonso, M van der Wilk
Uncertainty in Artificial Intelligence, 1998-2007, 2021
72021
Hypothesis testing the circuit hypothesis in LLMs
C Shi, N Beltran-Velez, A Nazaret, C Zheng, A Garriga-Alonso, A Jesson, ...
arXiv preprint arXiv:2410.13032, 2024
52024
Planning in a recurrent neural network that plays Sokoban
M Taufeeque, P Quirke, M Li, C Cundy, AD Tucker, A Gleave, ...
arXiv preprint arXiv:2407.15421, 2024
22024
Interpbench: Semi-synthetic transformers for evaluating mechanistic interpretability techniques
R Gupta, I Arcuschin, T Kwa, A Garriga-Alonso
arXiv preprint arXiv:2407.14494, 2024
22024
Open Problems in Mechanistic Interpretability
L Sharkey, B Chughtai, J Batson, J Lindsey, J Wu, L Bushnaq, ...
arXiv preprint arXiv:2501.16496, 2025
12025
Planning behavior in a recurrent neural network that plays Sokoban
A Garriga-Alonso, M Taufeeque, A Gleave
ICML 2024 Workshop on Mechanistic Interpretability, 2024
12024
Probability Density Imputation of Missing Data with Gaussian Mixture Models
A Garriga-Alonso
University of Oxford, 2017
12017
Adversarial Circuit Evaluation
A Garriga-Alonso
arXiv preprint arXiv:2407.15166, 2024
2024
Catastrophic Goodhart: regularizing RLHF with KL divergence does not mitigate heavy-tailed reward misspecification
T Kwa, D Thomas, A Garriga-Alonso
arXiv preprint arXiv:2407.14503, 2024
2024
Sistem, işlemi şu anda gerçekleştiremiyor. Daha sonra yeniden deneyin.
Makaleler 1–20