PC
Pablo Samuel Castro
cs.LGcs.AIstat.MLcs.SDcs.CLcs.HCcs.ROeess.ASanimal behavior and cognitioncs.CV
On Valency
published · living versionsW_8ujrvbgj·v1 · currentpublished
Freeze, Don't Clone: Fine-Tuning with Expressive Policies Without Retaining Offline Data
with Sami Nur Islam, Johan Obando-Ceron, Doina Precup, Aaron Courville
1 version
Preprints & journals
72 papers in the corpus · 2012–2026GlyphBench: A Playground for Language-Model Reinforcement Learning2609.34214v1 · Roger Creus Castanyer, Marc-Alexandre Cot'e, Matthew James Sargent et al.2026 · 0 citationsarXiv
V-Simba: Unleashing the Architectural Potential of RL in Visual Continuous Control2608.07870v1 · Donghu Kim, Youngdo Lee, Hojoon Lee et al.2026 · 0 citationsarXiv
Representation Learning Enables Scalable Multitask Deep Reinforcement Learning2606.05555v1 · Johan Obando-Ceron, Lu Li, Scott Fujimoto et al.2026 · 0 citationsarXiv
Local Guidance, Global Impact: Gaussian-Reshaped Trust Region Unlocks Behavior Transitions2606.03382v2 · Bingxu Liu, Jiashun Liu, Johan Obando-Ceron et al.2026 · 0 citationsarXiv
AI-Discovered Cognitive Models Reveal Novel Insights into Human and Animal Learning.42239314 · Kasenberg, Daniel, Castro, Pablo Samuel, Eckstein, Maria K et al.2026 · 1 citationbioRxiv : the preprint server for biology. 2026
Agentick: A Unified Benchmark for General Sequential Decision-Making Agents2605.06869v3 · Roger Creus Castanyer, Pablo Samuel Castro, Glen Berseth2026 · 0 citationsarXiv
A Mechanistic Analysis of Looped Reasoning Language Models2604.11791v1 · Hugh Blayney, 'Alvaro Arroyo, Johan Obando-Ceron et al.2026 · 0 citationsarXiv
Align and Filter: Improving Performance in Asynchronous On-Policy RL2603.01365v1 · Homayoun Honari, Roger Creus Castanyer, Michael Przystupa et al.2026 · 0 citationsarXiv
Stable Deep Reinforcement Learning via Isotropic Gaussian Representations2602.19373v3 · Ali Saheb Pasand, Johan Obando-Ceron, Aaron Courville et al.2026 · 0 citationsarXiv
Discovering Differences in Strategic Behavior Between Humans and LLMs2602.10324v2 · Caroline Wang, Daniel Kasenberg, Kim Stachenfeld et al.2026 · 0 citationsarXiv
A Comedy of Estimators: On KL Regularization in RL Training of LLMs2512.21852v4 · Vedant Shah, Johan Obando-Ceron, Vineet Jain et al.2025 · 0 citationsarXiv
The Formalism-Implementation Gap in Reinforcement Learning Research2510.16175v2 · Pablo Samuel Castro2025 · 0 citationsarXiv
ARM-FM: Automated Reward Machines via Foundation Models for Compositional Reinforcement Learning2510.14176v3 · Roger Creus Castanyer, Faisal Mohamed, Pablo Samuel Castro et al.2025 · 0 citationsarXiv
Simplicial Embeddings Improve Sample Efficiency in Actor-Critic Agents2510.13704v2 · Johan Obando-Ceron, Walter Mayor, Samuel Lavoie et al.2025 · 0 citationsarXiv
Asymmetric Proximal Policy Optimization: mini-critics boost LLM reasoning2510.01656v3 · Jiashun Liu, Johan Obando-Ceron, Han Lu et al.2025 · 0 citationsarXiv
A Survey of State Representation Learning for Deep Reinforcement Learning2506.17518v1 · Ayoub Echchahed, Pablo Samuel Castro2025 · 0 citationsarXiv
Stable Gradients for Stable Learning at Scale in Deep Reinforcement Learning2506.15544v2 · Roger Creus Castanyer, Johan Obando-Ceron, Lu Li et al.2025 · 0 citationsarXiv
Adaptive Accompaniment with ReaLchords2506.14723v1 · Yusong Wu, Tim Cooijmans, Kyle Kastner et al.2025 · 0 citationsarXiv
The Courage to Stop: Overcoming Sunk Cost Fallacy in Deep Reinforcement Learning2506.13672v1 · Jiashun Liu, Johan Obando-Ceron, Pablo Samuel Castro et al.2025 · 0 citationsarXiv
The Impact of On-Policy Parallelized Data Collection on Deep Reinforcement Learning Networks2506.03404v1 · Walter Mayor, Johan Obando-Ceron, Aaron Courville et al.2025 · 0 citationsarXiv
Optimistic critics can empower small actors2506.01016v3 · Olya Mastikhina, Dhruv Sreenivas, Pablo Samuel Castro2025 · 0 citationsarXiv
Mitigating Plasticity Loss in Continual Reinforcement Learning by Reducing Churn2506.00592v1 · Hongyao Tang, Johan Obando-Ceron, Pablo Samuel Castro et al.2025 · 0 citationsarXiv
Continual Learning in Vision-Language Models via Aligned Model Merging2506.03189v1 · Ghada Sokar, Gintare Karolina Dziugaite, Anurag Arnab et al.2025 · 0 citationsarXiv
Measure gradients, not activations! Enhancing neuronal activity in deep reinforcement learning2505.24061v2 · Jiashun Liu, Zihao Wu, Johan Obando-Ceron et al.2025 · 1 citationarXiv
Mind the GAP! The Challenges of Scale in Pixel-based Deep Reinforcement Learning2505.17749v2 · Ghada Sokar, Pablo Samuel Castro2025 · 0 citationsarXiv
Meta-World+: An Improved, Standardized, RL Benchmark2505.11289v2 · Reginald McLean, Evangelos Chatzaroulas, Luc McCutcheon et al.2025 · 6 citationsarXiv
Studying the Interplay Between the Actor and Critic Representations in Reinforcement Learning2503.06343v2 · Samuel Garcin, Trevor McInroe, Pablo Samuel Castro et al.2025 · 0 citationsarXiv
Multi-Task Reinforcement Learning Enables Parameter Scaling2503.05126v3 · Reginald McLean, Evangelos Chatzaroulas, Jordan Terry et al.2025 · 0 citationsarXiv
Discovering Symbolic Cognitive Models from Human and Animal Behavior10.1101/2025.02.05.636732v1 · Castro, P. S., Tomasev, N., Anand, A. et al.2025 · 14 citationsbioRxiv
CALE: Continuous Arcade Learning Environment2410.23810v1 · Jesse Farebrother, Pablo Samuel Castro2024 · 0 citationsarXiv
Don't flatten, tokenize! Unlocking the key to SoftMoE's efficacy in deep RL2410.01930v2 · Ghada Sokar, Johan Obando-Ceron, Aaron Courville et al.2024 · 0 citationsarXiv
NAVIX: Scaling MiniGrid Environments with JAX2407.19396v1 · Eduardo Pignatelli, Jarek Liesen, Robert Tjarko Lange et al.2024 · 3 citationsarXiv
Mixture of Experts in a Mixture of RL settings2406.18420v1 · Timon Willi, Johan Obando-Ceron, Jakob Foerster et al.2024 · 1 citationarXiv
On the consistency of hyper-parameter selection in value-based deep reinforcement learning2406.17523v3 · Johan Obando-Ceron, Joao G.M. Ara'ujo, Aaron Courville et al.2024 · 0 citationsarXiv
Stop Regressing: Training Value Functions via Classification for Scalable Deep RL2403.03950v1 · Jesse Farebrother, Jordi Orbay, Quan Vuong et al.2024 · 7 citationsarXiv
In value-based deep reinforcement learning, a pruned network is a good network2402.12479v3 · Johan Obando-Ceron, Aaron Courville, Pablo Samuel Castro2024 · 0 citationsarXiv
Mixtures of Experts Unlock Parameter Scaling for Deep RL2402.08609v3 · Johan Obando-Ceron, Ghada Sokar, Timon Willi et al.2024 · 4 citationsarXiv
A density estimation perspective on learning from pairwise human preferences2311.14115v3 · Vincent Dumoulin, Daniel D. Johnson, Pablo Samuel Castro et al.2023 · 1 citationarXiv
Learning and Controlling Silicon Dopant Transitions in Graphene using Scanning Transmission Electron Microscopy2311.17894v1 · Max Schwarzer, Jesse Farebrother, Joshua Greaves et al.2023 · 0 citationsarXiv
Small batch deep reinforcement learning2310.03882v1 · Johan Obando-Ceron, Marc G. Bellemare, Pablo Samuel Castro2023 · 5 citationsarXiv
A Kernel Perspective on Behavioural Metrics for Markov Decision Processes2310.19804v1 · Pablo Samuel Castro, Tyler Kastner, Prakash Panangaden et al.2023 · 2 citationsarXiv
Offline Reinforcement Learning with On-Policy Q-Function Regularization2307.13824v1 · Laixi Shi, Robert Dadashi, Yuejie Chi et al.2023 · 5 citationsarXiv
Minigrid & Miniworld: Modular & Customizable Reinforcement Learning Environments for Goal-Oriented Tasks2306.13831v1 · Maxime Chevalier-Boisvert, Bolun Dai, Mark Towers et al.2023 · 23 citationsarXiv
Bigger, Better, Faster: Human-level Atari with human-level efficiency2305.19452v3 · Max Schwarzer, Johan Obando-Ceron, Aaron Courville et al.2023 · 9 citationsarXiv
JaxPruner: A concise library for sparsity research2304.14082v3 · Joo Hyung Lee, Wonpyo Park, Nicole Mitchell et al.2023 · 1 citationarXiv
Proto-Value Networks: Scaling Representation Learning with Auxiliary Tasks2304.12567v1 · Jesse Farebrother, Joshua Greaves, Rishabh Agarwal et al.2023 · 0 citationsarXiv
The Dormant Neuron Phenomenon in Deep Reinforcement Learning2302.12902v2 · Ghada Sokar, Rishabh Agarwal, Pablo Samuel Castro et al.2023 · 8 citationsarXiv
The State of Sparse Training in Deep Reinforcement Learning2206.10369v1 · Laura Graesser, Utku Evci, Erich Elsen et al.2022 · 5 citationsarXiv
Reincarnating Reinforcement Learning: Reusing Prior Computation to Accelerate Progress2206.01626v2 · Rishabh Agarwal, Max Schwarzer, Pablo Samuel Castro et al.2022 · 16 citationsarXiv
Malakai: Music That Adapts to the Shape of Emotions2112.02070v1 · Zack Harris, Liam Atticus Clarke, Pietro Gagliano et al.2021 · 0 citationsarXiv
Career total: 103 works. 72 are in this corpus.Showing the 50 most recent.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.