Segueix
Ekdeep Singh Lubana
Ekdeep Singh Lubana
Harvard / NTT
Correu electrònic verificat a fas.harvard.edu - Pàgina d'inici
Títol
Citada per
Citada per
Any
Foundational Challenges in Assuring Alignment and Safety of Large Language Models
U Anwar, A Saparov*, J Rando*, ES Lubana*, D Paleka*, M Turpin*, ...
Transactions of Machine Learning Research (TMLR), 2024
151*2024
Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering
ES Lubana, CI Tang, F Kawsar, RP Dick, A Mathur
International Conference on Machine Learning (ICML) (spotlight), 2022
702022
A Gradient Flow Framework For Analyzing Network Pruning
ES Lubana, RP Dick
International Conference on Learning Representations (ICLR) (spotlight), 2021
642021
Mechanistic Mode Connectivity
ES Lubana, EJ Bigelow, RP Dick, D Krueger, H Tanaka
International Conference on Machine Learning (ICML), 2023
57*2023
Augmentations in graph contrastive learning: Current methodological flaws & towards better practices
P Trivedi, ES Lubana, Y Yan, Y Yang, D Koutra
Proceedings of the ACM Web Conference 2022, 1538-1549, 2022
572022
Mechanistically analyzing the effects of fine-tuning on procedurally defined tasks
S Jain*, R Kirk*, ES Lubana*, RP Dick, H Tanaka, E Grefenstette, ...
International Conference on Learning Representations (ICLR), 2024
552024
Beyond BatchNorm: towards a unified understanding of normalization in deep learning
ES Lubana, R Dick, H Tanaka
Advances in Neural Information Processing Systems (NeurIPS), 2021
542021
Compositional Abilities Emerge Multiplicatively: Exploring Diffusion Models on a Synthetic Task
M Okawa*, ES Lubana*, RP Dick, H Tanaka*
Advances in Neural Information Processing Systems (NeurIPS), 2023
452023
Minimalistic image signal processing for deep learning applications
ES Lubana, RP Dick, V Aggarwal, PM Pradhan
2019 IEEE International Conference on Image Processing (ICIP), 4165-4169, 2019
39*2019
What shapes the loss landscape of self-supervised learning?
L Ziyin, ES Lubana, M Ueda, H Tanaka
International Conference on Learning Representations (ICLR), 2023
252023
How capable can a transformer become? a study on synthetic, interpretable tasks
R Ramesh, M Khona, RP Dick, H Tanaka, ES Lubana
International Conference on Machine Learning (ICML), 2023
24*2023
Analyzing data-centric properties for graph contrastive learning
P Trivedi, ES Lubana, M Heimann, D Koutra, J Thiagarajan
Advances in Neural Information Processing Systems (NeurIPS), 2022
18*2022
How do quadratic regularizers prevent catastrophic forgetting: The role of interpolation
ES Lubana, P Trivedi, D Koutra, R Dick
Conference on Lifelong Learning Agents (CoLLAs), 2021
15*2021
What Makes and Breaks Safety Fine-tuning? A Mechanistic Study
S Jain, ES Lubana, K Oksuz, T Joy, P Torr, A Sanyal, PK Dokania
Advances in Neural Information Processing Systems (NeurIPS), 2024
92024
Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space
CF Park*, M Okawa*, A Lee, H Tanaka, ES Lubana
Advances in Neural Information Processing Systems (NeurIPS) (spotlight), 2024
62024
A Percolation Model of Emergence: Analyzing Transformers Trained on a Formal Language
ES Lubana*, K Kawaguchi*, RP Dick, H Tanaka
International Conference on Learning Representations (ICLR), 2024
52024
What shapes the loss landscape of self supervised learning?
Z Liu, ES Lubana, M Ueda, H Tanaka
ICLR, 2023
42023
ICLR: In-Context Learning of Representations
CF Park*, A Lee*, ES Lubana*, Y Yang*, M Okawa, K Nishi, M Wattenberg, ...
International Conference on Learning Representations (ICLR), 2024
32024
Competition Dynamics Shape Algorithmic Phases of In-Context Learning
CF Park*, ES Lubana*, I Pres, H Tanaka
International Conference on Learning Representations (ICLR) (spotlight), 2024
32024
Abrupt Learning in Transformers: A Case Study on Matrix Completion
P Gopalani, ES Lubana, W Hu
Advances in Neural Information Processing Systems (NeurIPS), 2024
3*2024
En aquests moments el sistema no pot dur a terme l'operació. Torneu-ho a provar més tard.
Articles 1–20