arXiv subjects

Andreas Krause

Publications and source records attributed to Andreas Krause.

At least 19 recordsLinked to original sources

arXiv2026

Test-Time Self-Play: Scaling without a Verifier

Test-time scaling methods spend additional compute at deployment to improve a large language model's performance on a hard target problem. Most methods rely on a verifier that scores candidate solutions for correctness, using it to select among or refine the model's outputs. In many deployment regimes, however, a strong verifier on the target problem is unavailable. We introduce Test-Time Self-Play (TTSP), an approach to test-time scaling using asymmetric self-play without reliance on a verifier. On a difficulty-calibrated subset of LiveCodeBench v6, TTSP is the best-performing method at every verifier strength we evaluate, and the gap to the strongest baseline widens as the verifier weakens; in the no-verifier limit, TTSP yields a 35\% absolute improvement in mean@8. The gains hold across a range of verifier strengths, except when the base policy has near-zero coverage of the correct answer.

cs.LG↗

arXiv2026

Learning Soft Robotic Dynamics with Active Exploration

Soft robots offer adaptability and safety in unstructured environments, yet their compliant, high-dimensional, and nonlinear dynamics make modeling for control difficult. Existing data-driven approaches often fail to generalize, constrained by narrowly focused task demonstrations or inefficient random exploration. We introduce an uncertainty-aware active exploration framework that autonomously learns task-agnostic and generalizable dynamics models of soft robotic systems. We evaluate this framework on three simulated platforms---a continuum arm, an articulated fish in fluid, and a musculoskeletal leg---and on a real-world pneumatically actuated continuum arm. Compared with random exploration and task-specific baseline, our method achieves higher modeling accuracy by actively collecting informative data. This improved model quality yields higher returns in zero-shot control on unseen tasks and robust performance under sensing noise, actuation delays, and material nonlinearities. These results demonstrate that uncertainty-driven active exploration enables data-efficient dynamics learning across soft robotic morphologies, supporting autonomous and adaptable control for compliant robots.

cs.RO↗

arXiv2026

Reinforcement Learning Enables Scaling Generalist Robot Policy Improvement

The introduction of generalist robot policies has enabled a single, large model to learn to perform a variety of tasks from a vast pre-training corpus. While these models often exhibit powerful zero-shot generalization capabilities, they typically require an adaptation phase to achieve sufficient performance in deployment. Iterated self-improvement via reinforcement learning (RL) has demonstrated promise in enabling this adaptation, yet such approaches have been largely explored in single-task settings - where the goal is to simply improve on a narrow deployment task - which lies in stark contrast to the multi-task capabilities of the base policy. In this work we ask whether generalist robot policies are capable of multi-task, generalist improvement in deployment - targeting many diverse tasks simultaneously - and what algorithmic recipe enables this. To answer this question, we first develop a set of structured evaluations to test generalist improvement capabilities, and systematically evaluate algorithmic and practical design choices regarding value estimation, policy extraction and resource allocation. We show that not only can generalist policies be improved online across diverse tasks, but that the right recipe can enable generalization and increase performance on tasks not considered during online training. We characterize when this generalization occurs and show that both the training task distribution and the use of value-based RL are critical. To the best of our knowledge, this is among the first evidence for online generalist improvement for robot foundation models.

cs.LG↗

arXiv2026

Multi-Objective Generative Optimization via Expected Hypervolume Fine-Tuning

Scientific discovery requires plausible designs spanning useful trade-offs among competing properties: a molecule with strong binding affinity, for example, may be difficult to synthesize. Pretrained generative models, e.g., flow and diffusion generators, provide a foundation for proposing such designs, but steering them toward high-quality, complementary trade-offs remains challenging. Inference-time search incurs high per-sample costs, while fine-tuning typically relies on fixed per-candidate rewards that do not capture how designs complement one another. We introduce Hypervolume Fine-Tuning (HVFT), which adapts a single pretrained generator to maximize the expected hypervolume of its samples while controlling departure from the pretrained distribution. Expected hypervolume jointly rewards candidate quality and collective trade-off coverage. Nonetheless, we show that it cannot be expressed as the expectation of a fixed per-candidate reward and therefore cannot be optimized directly with standard techniques. A key observation of this work is that its first variation corresponds to the expected marginal hypervolume contribution of a candidate, which yields a tractable adaptive reward via an unbiased estimator based on model samples. HVFT uses this reward to decompose the objective into a sequence of KL-proximal scalar-reward subproblems. We establish concavity, relative smoothness, and convergence guarantees for the resulting distributional procedure. Across therapeutic-peptide and molecular design tasks spanning discrete diffusion and continuous flow-matching models, HVFT achieves state-of-the-art expected hypervolume and trade-off coverage.

cs.LG↗

arXiv2026

Controlling Sharpening in LLM Post-Training

LLM post-training via reinforcement learning has shown consistent improvements in accuracy at the cost of reducing behavioral coverage - a phenomenon known as sharpening that can stall progress on difficult tasks. Repeated sampling objectives such as $\mathrm{pass@k}$ and $\mathrm{max@k}$ are known to mitigate sharpening, but the mechanism behind it remains poorly understood. Moreover, existing approaches for $\mathrm{@k}$ optimization use dedicated algorithms, and no general framework exists for adapting established policy optimization methods from $\mathrm{@}1$ to $\mathrm{@k}$. We derive the optimal policies for KL-regularized $\mathrm{pass@k}$ and $\mathrm{max@k}$ objectives in closed form. Our explicit solutions reveal that $\mathrm{@k}$ optimization naturally redistributes learning pressure based on prompt difficulty, directing it toward harder problems. Crucially, we show that standard $\mathrm{@}1$ objectives achieve these exact same optimal policies through prompt-adaptive regularization or reward transformation. Ultimately, this yields an optimizer-independent explanation for why $\mathrm{@k}$ objectives reduce sharpening on easy tasks while prioritizing success on harder ones. Building on these insights, we introduce $\mathrm{@k}$ Adaptive Pressure for Policy Optimization (KAPPO). Following a one-time reference calibration, KAPPO seamlessly adapts existing $\mathrm{@}1$ methods to $\mathrm{@k}$ via prompt-adaptive regularization or reward transformation, without otherwise altering their training procedures. Experiments with several RL methods on mathematical reasoning and code generation demonstrate that KAPPO provides direct control over sharpening. Crucially, we demonstrate that $\mathrm{@k}$ pressure can be fully realized with single completion training, matching the performance of multi-completion $\mathrm{@k}$ optimizers at a fraction of the cost.

cs.LG↗

arXiv2026

Forgetting to Improve: Principled Data Removal in Active Learning

The uncertainty of a statistical model is most commonly factorized into an aleatoric and an epistemic part. This factorization changes how predictions are interpreted in downstream decision tasks. Importantly, except for the idealistic scenario with no model mismatch, the quantification is a characteristic of the model and not the data generating process. In this paper, we propose Forgetting to Improve a method that reduces this discrepancy by incorporating the task into the modeling framework. Our key insight is to acknowledge that in scenarios of model mismatch, data can have a detrimental effect on the modeling for a specific task. Based on this insight, we propose an influence function for Gaussian process models that allows for principled removal of detrimental data samples. We showcase the flexibility of this approach by demonstrating significant improvements across a range of tasks, including Bayesian optimization, model-based reinforcement learning, and transductive learning.

cs.AI↗

arXiv2026

POETS: Uncertainty-Aware LLM Optimization via Compute-Efficient Policy Ensembles

Balancing exploration with exploitation is a core challenge in black-box optimization and reinforcement learning. We introduce POETS ($\textbf{Po}$licy $\textbf{E}$nsembles for $\textbf{T}$hompson $\textbf{S}$ampling), a novel framework that bridges uncertainty quantification and policy optimization. Our approach is grounded in the insight that policies trained with Kullback-Leibler (KL) regularization implicitly encode an underlying reward function. Building on this, POETS bypasses the complex, nested process of training an uncertainty-aware reward model and separately fitting a policy to this model. Instead, we directly train a policy ensemble to capture epistemic uncertainty by matching implicitly encoded reward functions to online, bootstrapped data. To overcome the prohibitive compute and memory constraints of ensembling Large Language Models (LLMs), POETS utilizes an efficient architecture: the ensemble shares a pre-trained backbone while maintaining diversity through independent Low-Rank Adaptation (LoRA) branches. Theoretically, we show that POETS implicitly approximates KL-regularized Thompson sampling, a principled optimization strategy with optimal cumulative regret bounds of $O(\sqrt{T γ_T})$. Empirically, we demonstrate that POETS achieves state-of-the-art sample efficiency across diverse scientific discovery domains, including protein search, therapeutic molecular design, and quantum circuit design. POETS also improves the optimization trajectory and out-of-distribution transfer of reinforcement learning across mathematical reasoning and code generation tasks, proving particularly robust in off-policy settings with experience replay or in small dataset regimes.

cs.LG↗

arXiv2026

A Theoretical Analysis of Test-Driven Code Generation

Code assistants are increasingly utilized in test-driven software development, yet the theoretical mechanisms behind their environment-interaction strategies remain underexplored. We provide a probabilistic framework for two dominant paradigms: code selection after generation using the execution environment, and code generation conditioned on environment feedback. First, we formalize several well-established selection heuristics as environment-aware estimators of code correctness. We theoretically prove that estimators based on fuzzy functional similarity add an inductive bias and strictly dominate estimators based on functional equivalence in terms of signal-to-noise ratio. Second, we frame backprompting as an in-context approximation of Thompson sampling. We derive a novel regret bound for reward functions with unobservable components, theoretically explaining why the effectiveness of backprompting is limited by the ambiguity of the informal task description (an irreducible regret). Using seven state-of-the-art open weight models, we corroborate these findings across BigCodeBenchHard, LeetCodeDataset, QiskitHumanEvalSim, and SWE-bench Verified. Our formalization also suggests how to improve task descriptions effectively, leading to an improved benchmark for LLM-assisted quantum programming, QiskitHumanEvalSimX.

cs.SE↗

arXiv2026

Tactile Curiosity Drives Robot Interaction

Mastering robot manipulation skills via reinforcement learning (RL) remains largely sample-inefficient. The most common RL algorithms rely on random action sampling to discover new strategies, resulting in agents that allocate most of their training budget to motions in free space, away from the contacts from which manipulation skills emerge. Existing intrinsic motivation methods based on model disagreement or epistemic uncertainty improve on isotropic noise, but they can also reward uncertainty in functionally irrelevant transitions, such as erratic motions in free space. In this work, we argue that tactile feedback provides a natural signal for exploration, and introduce TacEx, a framework that incorporates touch into epistemic uncertainty-driven exploration by decomposing model uncertainty across sensory modalities and directing curiosity toward the tactile channel. By anchoring curiosity to the sense of touch, TacEx drives the robot to discover complex contact dynamics, learning to manipulate and grasp objects without task rewards or expert demonstrations during exploration. The interaction-dense dataset collected through this tactile-driven curiosity supports offline learning of downstream pick-and-place policies without additional environment interaction. We further use tactile-driven exploration to post-train vision-language-action (VLA) models. Although the VLAs are initially pre-trained without tactile feedback, post-training with TacEx substantially improves downstream performance while remaining highly sample-efficient.

cs.RO↗

arXiv2026

LLM-to-Flow Distillation: Teaching Natural Language Goals to Scientific Generators

Generative discovery in science requires coupling high-level scientific reasoning with the ability to generate samples respecting the structure of a design space. Yet these capabilities remain largely separated: foundation LLMs encode broad chemical and biological knowledge, while domain-specific flow- and diffusion-based generators capture the priors needed to produce valid sequences and geometries. We study how to align such scientific generators to high-level natural-language discovery goals without requiring an explicit proxy reward, which is often unavailable or poorly captured by learned predictors in experimental sciences. We introduce LLM-to-Flow Distillation (LFD), where a frozen LLM acts as an adaptive teacher for a domain-specific generator. Given a goal such as 'generate peptides with promising anticancer activity', the teacher constructs a curriculum of intermediate sub-goals aware of the flow's current capabilities, and progressively directs them toward the original discovery goal. At each round, the LLM judges sampled candidates under the current objective. These judgments induce preferences that are iteratively distilled into the generator through DPO. This procedure provides dense learning signals even when the initial model places negligible probability on designs satisfying the final goal, enabling adaptation toward poorly covered, out-of-distribution regions. Experiments on therapeutic peptides and small-molecule structure design demonstrate that LFD enables a new form of generative control: aligning arbitrary domain-specific scientific generators to scientific discovery tasks expressed via high-level natural-language goals.

cs.LG↗

arXiv2026

Sampling-Based Safe Reinforcement Learning

Safe exploration remains a fundamental challenge in reinforcement learning (RL), limiting the deployment of RL agents in the real world. We propose Sampling-Based Safe Reinforcement Learning (SBSRL), a model-based RL algorithm that maintains safety throughout the learning process by enforcing constraints jointly across a finite set of dynamics samples. This formulation approximates an intractable worst-case optimization over uncertain dynamics and enables practical safety guarantees in continuous domains. We further introduce an exploration strategy based on constraining epistemic uncertainty, eliminating the need for explicit exploration bonuses. Under regularity conditions, we derive high-probability guarantees of safety throughout learning and a finite-time sample complexity bound for recovering a near-optimal policy. Empirically, SBSRL achieves safe and efficient exploration both in simulation and in real robotic hardware, and readily extends to practical deep-ensemble implementations that scale to high-dimensional continuous control problems.

cs.LG↗

arXiv2026

VioLA: Learning Generalist Humanoid Control Policies from Human Data

Teaching a humanoid to follow instructions with its whole body runs into two obstacles. Its action space is large and tightly coupled: legs, arms, and fingers must move together while the robot keeps its balance, which makes joint-level actions hard to learn. And humanoid demonstrations are scarce, so current humanoid generalist policies do not follow new instructions out of the box and are fine-tuned on teleoperated demonstrations of each task before deployment. Human demonstrations exist in far larger numbers, but a person's motion is not a robot command. We remove both obstacles by changing what the generalist policy predicts. We introduce VioLA, a generalist humanoid policy that predicts body and hand motion latents instead of joint commands. A pretrained body- and hand-controller execute these latents on the robot. Their corresponding motion encoders map human and robot motion into the same latent spaces. A human recording is therefore labeled in the policy's action space, and the training demonstration pool contains 140.6 million frames, 93.2% of them human. As a result, VioLA follows locomotion instructions on the real robot zero-shot, without task-specific fine-tuning, reaching 100% success where GR00T N1.7 and $Ψ_0$ reach 16.7% and 0%, respectively. It also reaches 88.6% manipulation success without task-specific fine-tuning. The same approach works across two VLA and one world-action model backbones. A generalist policy trained on human demonstrations alone performs locomotion tasks on the real robot zero-shot. Code and checkpoints will be released.

cs.RO↗

arXiv2026

Local Sparsity Enables Unsupervised LLM Safety Detection

Deployment-time safety methods for large language models (LLMs) are predominantly supervised and assume access to unsafe training data. Nevertheless, new attacks and harm categories regularly arise, not captured by models trained in such a supervised fashion. An alternative approach is to view this problem through the lens of anomaly detection, namely, to rely solely on modeling safe data and flagging out-of-distribution inputs. However, LLM activations lie in a high-dimensional space, raising concerns about whether anomaly detection is statistically feasible. We show that, under the linear representation hypothesis (LRH), there may indeed be hope. In the LRH concept space, which is typically recovered via a sparse autoencoder (SAE), nearby points share a small common active support. Using this local sparsity insight, we propose a framework for locally masked SAE-based anomaly detection, supported by theoretical justifications. We validate it on various architectures and datasets, including both capability-testing datasets and safety-specific datasets. Finally, when we allow algorithms to use 1% out-of-distribution data for calibration, locally sparse methods achieve near-optimal performance, demonstrating their ability to capture meaningful safety information while using only 1-2% of SAE neurons for computation.

cs.LG↗

arXiv2026

Task Vector Descent: Learning from Non-IID Batches

A central challenge in continual learning is to acquire new knowledge without forgetting what the model has already learned. This challenge appears in language model training when training data comes from various domain-, user-, or task-specific distributions that are encountered unevenly over time. In such settings, successive minibatches are temporally clustered by distribution instead of being sampled i.i.d. from the overall data mixture. Training on temporally clustered data induces a stability-plasticity tradeoff. Adapting the model to the active distribution can improve the model on the active distribution but may lead to a performance degradation on data it previously trained on. We find that this tradeoff intensifies with longer exposure to the same distribution. We therefore ask if the parameter displacement produced by such a sequence (the task vector) should be fully retained or applied partially. We compare applying the full displacement ($λ=1$) with partial integration, which scales the task vector by $λ$ before applying it to the continuing model and scales the optimizer state by the same coefficient. Across continual pretraining, pretraining from random initialization, supervised post-training, and reinforcement post-training, we find that intermediate values of $λ$ often improve average continuing-model performance relative to full integration, particularly after longer same-distribution sequences. In continual-pretraining experiments with both controlled streams and naturally defined adaptation sequences, task-vector scaling outperforms full integration at the matched learning rate, showing that its benefits are not reproduced by learning-rate scaling alone.

cs.LG↗

arXiv2026

Uncertainty Quantification for Flow-Based Generalist Robot Policies

Generalist robot policies, such as vision-language-action models (VLAs) and world-action models (WAMs), combine powerful pretrained backbones with expressive generative action heads trained via flow matching on large-scale robotic datasets. Despite their strong empirical performance in robotic manipulation, these policies lack mechanisms to quantify confidence in their predictions and to detect when their actions may be unreliable. This presents a critical limitation for real-world deployment in non-stationary environments, where models inevitably encounter scenarios outside their pretraining distribution and may fail without warning. To address this, we derive an efficient method to quantify epistemic uncertainty in flow-matching models by leveraging velocity-field disagreement (VFD) across a small ensemble. We successfully use this uncertainty estimate for detecting failures during deployment and active fine-tuning of flow-based generalist policies. For the latter, we propose SAVE, a simple yet effective method for uncertainty-guided active multitask fine-tuning that reduces the number of costly expert demonstrations required to adapt generalist policies to new tasks. We conduct experiments in simulation and the real world, across VLAs and a WAM. VFD yields better-calibrated uncertainty estimates predictive of downstream performance and detects failures with 8 pp higher overall accuracy than existing methods. Across three real-world tasks, SAVE improves final average success from 39 % to 47 % with a fixed demonstration budget. Our results show that measuring epistemic uncertainty with VFD enhances both failure awareness and adaptation of generalist robot policies. Project website: tum-lsy.github.io/uq_generalist_policies.

cs.RO↗

arXiv2026

Deep Epistemic Value Functions for Optimistic Exploration

Principled exploration in reinforcement learning requires an agent to quantify its epistemic uncertainty and act to resolve it. Uncertainty over the value function provides a natural signal for exploration, yet existing deep approximations remain brittle and perform inconsistently. The central challenge is therefore to scale these ideas robustly. We conduct a systematic empirical study of how epistemic uncertainty is represented, propagated, and optimized in deep epistemic value functions, and uncover distinct failure modes along each of these axes. These findings motivate DEVOTE, a model-free reinforcement learning algorithm that controls how uncertainty generalizes beyond observed data, stabilizes its temporal propagation, and preserves adaptation to the resulting non-stationary exploration objective. Across reward-free exploration and challenging continuous-control tasks, DEVOTE reaches novel states more effectively and achieves higher task return than strong model-free and model-based exploration baselines. These results provide evidence that deep epistemic value functions are a promising path toward scalable, principled exploration.

cs.LG↗

arXiv2026

On Repulsive and Attractive Teachers: Separating Correctness from Behavior in Self-Distillation

On-policy self-distillation provides dense, token-level supervision by conditioning a model on privileged information and distilling the resulting teacher distribution back into the model. However, privileged information can change not only what the teacher knows, but also how it behaves, entangling correctness-relevant learning signals with unintended behavioral shifts. We study this effect in reasoning tasks by contrasting attractive self-distillation, which moves the model toward a privileged teacher, with repulsive self-distillation, which moves it away from a privileged teacher. We find that both objectives can induce strong and opposing behavioral shifts: attraction suppresses exploratory reasoning and promotes shorter, more confident responses, whereas repulsion increases response length, can trigger unintended switches into a model's latent thinking mode, and ultimately becomes unstable. Motivated by these observations, we study contrastive self-distillation, which combines attraction toward a correct-solution-conditioned teacher with repulsion from an incorrect-solution-conditioned teacher. In contrast to prior work that combines such distillation signals with a GRPO objective, we isolate the self-distillation objective and study its behavior on its own. We find that the shared behavioral shifts of the two teachers largely cancel, leaving a token-level signal that more directly reflects correctness. Across non-thinking, instruct-only, and already-thinking models, this contrastive objective improves reasoning performance while maintaining stable response lengths.

cs.LG↗

arXiv2026

TARC: Time-Adaptive Robotic Control

Most robotic systems rely on fixed-frequency discrete-time controllers, creating a trade-off between the efficiency of low-frequency control and the responsiveness of high-frequency feedback. As a result, systems typically default to high control rates for robustness, at the cost of wasted inference and unnecessary actuation. Addressing this, we introduce Time-Adaptive Robotic Control (TARC), a reinforcement learning framework in which the policy jointly predicts a control action and its duration of application. TARC learns temporally extended actions by optimizing task performance under soft or hard constraints on the number of control switches, enabling adaptive modulation of control rates. We evaluate TARC on two robotic hardware platforms: a high-speed RC car and the Unitree Go1 quadruped, and on a vision-language action model in simulation, where each query incurs a costly transformer forward pass. Across all settings, TARC matches the performance of high-frequency discrete-time controllers while operating at less than half their control frequency. Unlike fixed-rate controllers, TARC adapts its control frequency online, allocating high-frequency feedback only when required.

cs.RO↗