Machine Learning · Cross-listed from cs.CL
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits
Shu Ishida, Aliasghar Khani, Tianyu Zhang et al. (+3 more)
Autoregressive decoding is less suited to non-causally ordered data, e.g., cross-references or future dependencies, since their outputs cannot be revised post hoc, unlike iterative human
reasoning.
Diffusion LLMs partially mitigate this limitation but still cannot flexibly insert or remove arbitrary numbers of
tokens from a pre-generated context. We design AutoIndexer as a generative model with a novel training and decoding mechanism that simulates a Chain of Edits, allowing the model to flexibly modify its prior outputs in hindsight via
token insertions, substitutions, and deletions. The model learns to predict marker
tokens for opening and closing edits, simultaneously moving a cursor to positions in the existing
token sequence, analogous to a text editor. Training sequences are obtained by perturbing ground-truth labels with inverse chained edit operations, and the model learns to edit by undoing those inverses. The perturbation training allows AutoIndexer to be robust to erroneous context and adaptable to editing tasks such as code correction. We demonstrate that our method exhibits post-generation context-editing capabilities that most autoregressive and
diffusion models lack without compromising performance on standard LLM
benchmarks. The code is available at https://github.com/AutodeskAILab/autoindexer.
Read the full abstract ▾
All authors Shu Ishida, Aliasghar Khani, Tianyu Zhang, The Cong Luong, James Seale Smith, Adam Gaier
Machine Learning
Transformer-Based Masked Diffusion Models Provably Learn Distributions with Latent Structure
Jingda Wu and Changxiao Cai
Masked
diffusion models (MDMs) generate discrete data by iteratively unmasking multiple
tokens in parallel using a learned mask predictor. Their output carries two errors: parallelization error due to parallel sampling, and estimation error from the mask predictor learned with finite samples. Existing theory treats these errors separately, and the available statistical bounds typically scale with the number $V^L$ of length-$L$ sequences over a vocabulary of size $V$, suffering from the curse of dimensionality. In this work, we establish end-to-end statistical guarantees showing that
transformer-based MDMs learn structured distributions with a sample complexity governed by their latent structure. Specifically, we model this structure through mixtures of hidden Markov trees, which capture latent dependencies within components and heterogeneity across components. When the data distribution is well approximated by such a mixture, a
transformer-based MDM with parallel unmasking schedules achieves $\varepsilon$-accurate sampling in KL divergence with $\widetilde{O}(LK_{\mathrm{par}}/\varepsilon)$ training samples, where $K_{\mathrm{par}}$ is
polynomial in $L$, $V$, the number of hidden states, and the number of mixture components. The
transformer network and sampling procedure require no knowledge of the tree
topologies or component parameters, thereby adapting to the unknown latent structure automatically. Overall, our results provide theoretical justification for the empirical effectiveness of MDMs on real-world data.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Equal-Budget Monte Carlo Tree Search versus Randomized-Restart List Scheduling for DAG Workflows: A Controlled Equal-Budget Computational Study
Yuyu Zhang
Monte Carlo Tree Search (MCTS) is often proposed for scheduling directed-acyclic-graph (DAG) workflows, yet most studies never run the comparison that could refute it - giving a much simpler method the same search effort. We run that comparison under a controlled protocol. All arms receive an identical budget - the same number of candidate schedules built and evaluated: an MCTS scheduling method, preregistered as the main arm; a randomized-restart list scheduler with no search tree as the control; and a textbook-UCT MCTS arm as a registered follow-up on the same domain. The analysis plan was locked before data collection; an independent checker confirmed all 768 confirmation-stage schedules valid with correct finish times. At the main budget of 2048 evaluated schedules, MCTS is 2.88% worse than the restarts (95% CI [-3.62%,-2.15%]; the 72 of 96 held-out instances with positive baseline-relative headroom). The deficit is not budget starvation: on the separate exploration pool restarts self-improve by 1.85% where MCTS gains 0.24%, and 700 of 720 of MCTS's first results are merely the warm start it was handed. The studied selection rule locks deterministically onto the first-expanded child, keeping its trials correlated while restarts draw gains from the tails; the causal link to makespan remains an inference. A structural map locates where the tree stops paying: restarts win on sparse irregular workflows, no difference is detectable on regular layered ones, and dense ones leave no room to improve. Swapping the rule for textbook UCT removes most of the deficit (-0.26%, 95% CI [-0.71%,+0.23%]) without reversing its sign. Results are confined to a controlled synthetic domain (N in {8,16,32} tasks, 2-8 processors, budgets up to 2048); we make no claim about larger budgets.
Read the full abstract ▾
Machine Learning
Enriched Holomorphic Neural Networks for 2D and 3D Laplace Problems
Kalp Shah, Andrei Vlad Nicula, Mihnea Andrei Bârsan et al. (+4 more)
Boundary-value problems for the Laplace equation develop singular behaviour near corners, edges and junctions that smooth approximators resolve poorly. We combine holomorphic neural networks with singular functions derived from the local geometry and boundary conditions. The representation is harmonic by construction, so no interior residual is needed: the network fits the smooth remainder from boundary data alone, and the enrichment terms capture the singular behaviour. Singular exponents are trainable, so the construction also applies when the local expansion is not known in advance. We extend the approach to three dimensions using independent networks evaluated along complex null directions. The enriched model is the most accurate tested method on every problem. Its margin is several orders of magnitude where the enrichment captures the singularity exactly, and a factor of about 1.7-2.1 over the next-best method on a CdZnTe radiation detector with many singular sites, in both two and three dimensions.
Read the full abstract ▾
All authors Kalp Shah, Andrei Vlad Nicula, Mihnea Andrei Bârsan, Robert-Stefan Sofroni, Agneya Bhardwaj, Jan Jelínek, Alejandro Valverde Mahou
Machine Learning
Efficient Learning of Survival Analysis Models Under Outcome-Dependent Truncation and Censoring
Itai Zilberstein, Ioannis Anagnostides, Arman Kilic et al. (+1 more)
Modern machine learning models are typically trained on data assumed to be representative of the predictive tasks the model will face in the future, but practical
datasets violate this assumption. We focus on outcome-dependent truncation in which an observation appears in the training data depending on its label. Our work connects two largely distinct literatures: efficient learning under truncation and survival analysis. Existing methods for truncated survival analysis do not characterize the optimization landscape and instead rely on iterative procedures such as expectation maximization. We address this gap and prove that Cox regression can be efficiently learned under truncated and right-censored data using first-order optimization when the baseline hazard is known (despite nonconvexity). In contrast, when the baseline hazard is learned jointly, we construct identifiable and correctly specified models whose negative log-likelihood contains strict suboptimal local minima. On experiments with four survival
datasets, we show that accounting for truncation reduces the prediction bias induced by naive Cox regression and extends to neural networks. We also show orders of magnitude speedups compared to standard expectation maximization.
Read the full abstract ▾
All authors Itai Zilberstein, Ioannis Anagnostides, Arman Kilic, Tuomas Sandholm
Machine Learning
Market Impact Estimation via Causal Foundation Models
Dennis Thumm and Ying Chen
When a trader executes a large order, the relevant question is not what will the price be? but what would the price have been had I not traded?
Market impact is thus a causal quantity, yet classical models (Almgren–Chriss, propagators) regress price on flow, which is biased whenever informed flow moves prices through information as well as trading. We present a proof-of-concept Causal
Market Simulator: a continuous-time causal prior-data fitted network (PFN) trained on a synthetic
market prior in which a hidden informational flow confounds the flow–price relation by construction, with confounding strength an exact per-episode knob. On this prior, the naive
estimator's bias follows a closed form we confirm empirically (up to 68% overestimation), while the PFN tracks the unbiased true-parameter oracle at high confounding. Its do-ablated twin, the same model with the intervention inputs removed, collapses onto the pre-window mean. Zero-shot, the model responds at the ground-truth scale to metaorders in an ABIDES limit-order book ($+0.38$ pre-onset standard deviations against seed-replay counterfactuals), and its intervention response replicates on about 2,300 detected flow bursts in BTC/ETH/SOL perpetuals and 3,800 on Nasdaq equities, evidence consistent with, though not proof of, causal transfer. Off-prior the response degrades gracefully, and split-conformal recalibration repairs interval calibration. Transfer hinges on encoding the metaorder as added flow rather than a clamped price level, not on more training. A recurrent state-space variant closes the remaining shape gap: its impact accrues over execution duration where the
transformer's stays flat, raising the correlation with measured metaorder impact from $+0.68$ to between $+0.92$ and $+0.96$, at the cost of weaker on-prior deconfounding that traces to the training recipe rather than to recurrence.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.SD
Structural and Representational Agency in LLM-Driven Timbre Generation and Interpolation
Joseph Cameron and Alan Blackwell
Generative creative interfaces routinely report the user's position within a space of possibilities. Such displays imply that the quantities shown were measured from the artefact; in a growing class of systems they are instead predicted by a
language model describing its own output, and the two are indistinguishable on screen. Most creative domains lack an agreed measurement against which such a self-report can be checked; audio admits one. In a measurement study of a working LLM-driven timbre interface, what a user can cause proves separable from what the interface reports. The first is exactly characterisable: the reachable set is a zonotope whose dimension increases by one per sound placed, so the space the model may act within is demonstrably the user's own. The second is inaccurate. The displayed position differs from measurement by approximately three times the width of that region, and the error has two independent sources: model-predicted coordinates, whose accuracy varies twofold between the two models evaluated, and the interpolation step, which arises from combining synthesiser parameters and is unaffected by model quality. End to end, 26 of 36 refinement instructions produced no perceptible change, and under the weaker model the facility performed no action while reporting convergence. Effective agency (what a user can cause) and epistemic agency (whether they can know it) are therefore separable and independently engineered: structurally enforced constraints survive model failure, whereas guarantees resting on the model's own report do not. The failure mode that remains is silence presented as success. We close with a short protocol for building and auditing such displays.
Read the full abstract ▾
Machine Learning
CacheScope: Exact-Compute Allocation for Training-Free Diffusion Acceleration
Qinhan Lyu, Jusheng Zhang, Zhixing Zhang et al. (+4 more)
Training-free feature reuse accelerates
diffusion transformer inference by replacing full denoising computation at selected steps with cached approximations. Existing methods mainly improve the reconstruction of skipped features, but we find that once a stable first-order reconstructor is used, increasing reconstruction complexity brings limited gains, whereas the placement of exact-computation steps has a much larger effect on fidelity. More importantly, absolute shallow-probe variation is jointly affected by input-trajectory displacement and network response strength, and therefore does not directly reflect cache risk. We address this confounding by normalizing probe-output variation with the corresponding input displacement, yielding a directional gain per unit trajectory movement. Based on this signal, we propose CacheScope, which identifies high-risk reuse steps online and selectively falls back to full denoising computation while leaving the underlying cache reconstructor unchanged. CacheScope requires no additional training, gradients, auxiliary networks, or
prompt-specific offline schedule optimization. Across FLUX.1-dev, Wan2.1-T2V-1.3B, and Open-Sora 1.2, CacheScope achieves up to 3.57x acceleration and reduces median LPIPS by up to 70.1% over the strongest probe-based baseline.
Read the full abstract ▾
All authors Qinhan Lyu, Jusheng Zhang, Zhixing Zhang, Hao Liu, Zijie Liang, Jin Jiang, Keze Wang
Machine Learning
GlucoPRISM: Protocol-Supervised Pretraining for Cross-Cohort Glucose Transfer
Adinath Madhavrao Dukre, Abdullah Mohammad, Yifan Lu et al. (+3 more)
Continuous glucose monitoring (CGM) foundation models encode each day as one vector that mixes who the person is, what happened that day and how the sensor measured it, so measurement differences travel with the representation into new cohorts. We ask whether pretraining can place the measurement component that harms transfer in known coordinates, so that deployment drops them with a fixed projection and no target data. GlucoPRISM supervises a GlucoFM backbone with response contracts, rules for how each part of the representation should change when only the day or only the sensor changes: a 64-dimensional Trait band should change with neither, a 48-dimensional State band only across days, and a 16-dimensional Sensor band should track the device. Synthetic sensor views calibrated on paired Dexcom and Libre recordings, different-day pairs and glucose-summary targets supply this supervision without clinical labels. Across twelve cross-cohort directions, dropping the Sensor band from the same frozen encoder raises mean transfer AUROC from 68.75 to 73.57, against 69.3 for GlucoFM. The gain has two parts. Into CGMacros, the one target cohort recorded with two sensor types, filtering adds 11 to 19 points, and this holds when the generator is recalibrated on half of the participants and scored on the other half. On the other eight directions filtering is neutral and the encoder itself exceeds GlucoFM by 3.5 points. The device remains linearly decodable after filtering, so the filter removes a device-related component that harms transfer rather than all measurement information. The retained bands improve window AUROC on thirteen of fourteen public task–cohort
cells (Holm-adjusted p=0.031) and correlation on fourteen of fifteen clinical endpoints in an external cohort of 9,303 participants.
Read the full abstract ▾
All authors Adinath Madhavrao Dukre, Abdullah Mohammad, Yifan Lu, Eran Segal, Ahmed Metwally, Imran Razzak
Machine Learning
Behind-the-Meter Photovoltaic Visibility from Net Smart Meters: Deployment Regimes and Operational Insights
Jack Zheng and Hao Wang
Distribution operators need visibility of behind-the-meter photovoltaic (PV) generation, yet most residential customers are observed only through net smart meters. Deep-learning-based PV disaggregation can restore this visibility, but its operational value depends on whether the model serves known customers, new customers, future periods, or new regions. We quantify this dependence using four deployment regimes that separate household transfer from temporal extrapolation, applied to Ausgrid smart-meter data with cross-
dataset adaptation to the Pecan Street data. Using a fixed training protocol, we
benchmark representative deep sequence architectures for PV disaggregation under regime-matched evaluation. At a fixed 48-hour context length, daylight mean absolute error (MAE) of the best model rises from 0.104~kW on seen households in the future test period to 0.185~kW on unseen households over the same period, and under a fixed training contract the deployment regime explains far more of the error variation than model choice. Under target-domain normalization, fine-tuning on five target-region households reduces source-model transfer error by 28%–36%. Long input windows help known-customer extrapolation, while their new-customer benefits vary by model. These findings suggest regime-matched error budgets, limited target supervision for model adaptation in new regions, and use-case-dependent context selection for energy data analytics.
Read the full abstract ▾
Machine Learning · Cross-listed from physics.chem-ph
MolP2SBench: Do LLMs Really Reason from Molecular Properties to Structure?
Yuhang Xie, Zhiyi Shi, Zeyi Zhang et al. (+4 more)
Large
language models (LLMs) are increasingly used in molecular discovery and structure elucidation, where decisions depend on physical observations. Existing evaluations primarily focus on end-to-end molecular reconstruction, where correct predictions do not necessarily indicate reliance on the supplied properties. We introduce MolP2SBench, a
benchmark for molecular property-to-structure (P2S)
reasoning in LLMs. Each task presents a masked molecular scaffold with given properties and asks the model to select the missing structure from plausible candidates. We shuffle or replace properties to test reliance on correct property-structure pairing and responsiveness to counterfactual property changes. A channel-wise leaderboard uses Property-Grounded Recovery (PGR), giving equal weight to source and counterfactual recovery. Evaluating seven models, we find that LLMs can use physical properties for structure prediction, but performance varies substantially across property types. Higher accuracy with a property does not necessarily imply greater responsiveness to it: a model may benefit more from $^1$H NMR than from $^{13}$C NMR, yet be less likely to change its prediction when the $^1$H data is replaced. LLM performance also need not reflect the relative difficulty of these properties for human chemists. Although UV-Vis spectra provide less direct structural information than NMR spectra, Astra achieves higher accuracy with UV-Vis data than with NMR peaks. MolP2SBench supports studying how LLMs extract and use physical-property information, revealing patterns that may differ from conventional chemical intuition and providing a foundation for more reliably property-grounded models for molecular and
quantum-chemical inference. Code and
dataset: https://github.com/DUAL-Xiao/MolP2SBench.
Read the full abstract ▾
All authors Yuhang Xie, Zhiyi Shi, Zeyi Zhang, Zhisheng Qi, Yushun Dong, Yu Wang, Xiongye Xiao
Machine Learning · Cross-listed from cs.RO
CSF: Contextual Safety Filtering for Motion Generators
Lizhi Yang, Yiling Hou, Yao Tang et al. (+4 more)
Text-conditioned motion generators produce trackable whole-body motion, but they have no notion of scene-dependent safety: the same action may target an object or a person. Existing safeguards either inspect the
prompt, require labeled motion data, or enforce
geometric constraints; therefore, they do not directly account for how scene context changes a motion's meaning. We introduce contextual safety filtering (CSF), a training-free filter that grounds natural-
language safety rules in safe and unsafe reference trajectories produced by the generator. For each active rule, safe and unsafe reference trajectories define an affine safety value that a safe reference tracking CBF-QP enforces. Across four pretrained generators with different architectures, CSF activates the intended rules in all explicit and scene-triggered unsafe cases and reduces the danger-event rate by up to 90%, while preserving 88-100% of benign motions. We demonstrate the complete system on a real-world Unitree G1, where it successfully prevents unsafe motions in a variety of scenarios, including interactions with humans and objects.
Read the full abstract ▾
All authors Lizhi Yang, Yiling Hou, Yao Tang, Junheng Li, Daniel Weng, Blake Werner, Aaron D. Ames
Machine Learning · Cross-listed from cs.AI
On the estimation and validity of AI time horizons—a statistical look at the METR plot
Drew T. Nguyen and William Fithian
METR's 50% time horizon measures the human completion time of software tasks that an AI solves with 50% probability, allowing AI capabilities to be expressed in interpretable units. On 228 tasks and 26 AIs, we recompute the time horizons using splines and item-response theory to relax the assumption that the AI difficulty of a task depends linearly on the log of human time. Our fitted spline can be interpreted as a function that converts human time to AI difficulty; it is nearly flat in a region from 2–30 min but close to linear elsewhere. Hence, a time-horizon jump from 3 min to 30 min is much easier than one from 30 min to 5 hours despite the same multiplier of $10 \times$. Overall, we contribute time-horizon point estimates that perform better under a cross-validated suite of proper scoring rules, as well as diagnostic plots for assessing time horizons' construct validity. We suggest that time horizons be interpreted together with the diagnostic plots, especially as new time-horizon-based
benchmarks are proposed or existing ones grow to include longer tasks.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.RO
A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control
Octi Zhang, Mateo Guaman Castro, Patrick Yin et al. (+4 more)
General-purpose
robots must perform a wide range of tasks from agile locomotion to dexterous manipulation. While sim-to-real reinforcement learning (RL) has proven to be a useful tool for this goal, current RL pipelines depend on engineering-heavy, per-task structural priors such as shaped rewards and demonstrations. Recent work has shown that diverse simulator resets, combined with massively parallel simulation, can alleviate much of this engineering burden on several manipulation problems. However, we find that naively scaling this paradigm to more precise or dynamic problems remains non-trivial. While simulator resets can help with exploration, uniformly sampling over this distribution wastes a growing fraction of learning experience on task configurations the policy has already mastered or cannot yet attempt. This makes it challenging to see the expected benefits of scaling parallel environments for RL, since much of the learning signal in a batch is wasted during learning. To mitigate this, we introduce Success Guided Sampling (SGS), a simple adaptive sampler that concentrates RL training on task configurations around the frontier of the policy's capabilities. Doing so allows large-scale simulated RL to make the most out of the experience in a batch, enabling much more effective scaling to large-scale parallel simulation. Across experiments using up to $2^{20}$ (over one million) parallel environments, SGS enables RL to solve challenging multi-terrain quadruped locomotion and contact-rich assembly tasks that prior methods fail to solve. Finally, we distill the learned manipulation policies into RGB-based policies and demonstrate zero-shot transfer to several challenging assembly tasks on real hardware. Project website: https://sgs-rl.github.io/.
Read the full abstract ▾
All authors Octi Zhang, Mateo Guaman Castro, Patrick Yin, Ignacio Dagnino, Abhishek Gupta, Rosario Scalise, Byron Boots
Machine Learning
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems
Anna Zimmel, Fleur Hendriks, Markus Holzleitner et al. (+4 more)
Bifurcations are ubiquitous in physical systems, from structural buckling to fluid and
climate dynamics, yet they remain largely unexplored in deep learning. At a
symmetry-breaking bifurcation, a single input admits multiple equally valid solutions, violating the one-to-one assumption underlying most learned physical surrogates. We introduce Bi-FORK, a generative framework for learning these one-to-many solution maps in high-dimensional systems. Bi-FORK generates complete trajectories through latent flow matching, preserving space and time coherence, and uses repulsion-guided sampling to recover distinct solution branches in a single amortized pass. We evaluate Bi-FORK on buckling beams, mechanical metamaterials, and Allen-Cahn phase separation, spanning continuous, discrete, and field-valued bifurcations with discretizations up to 260,000 points. Bi-FORK recovers the
multimodal solution structure while scaling several orders of magnitude beyond prior approaches, opening generative modeling to high-dimensional bifurcating physical systems.
Read the full abstract ▾
All authors Anna Zimmel, Fleur Hendriks, Markus Holzleitner, Florian Sestak, Martin Weichselbaumer, Vlado Menkovski, Johannes Brandstetter
Machine Learning · Cross-listed from cs.CV
One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts
Adrian Bulat, Yassine Ouali and Georgios Tzimiropoulos
In this work, we show that a single
Transformer block, applied recurrently, can match the accuracy of a full-depth
vision encoder at comparable inference FLOPs without intermediate feature
distillation. reViT restores depth-specific transformations by representing the FFN at each recurrent depth as a convex combination of a small shared expert bank. A continuous normalized-depth coordinate programs this mixture, defining a resampleable trajectory through FFN parameter space. We evaluate this design in two regimes: supervised ImageNet-1k training and
distillation from a DINOv2 teacher. Across both regimes, controlled adaptations identify weight-space merging as the strongest tested MoE family at a matching one-FFN budget, ahead of the
token-dispatch and output-mixture alternatives. Trained from scratch, reViT-B/16 attains DeiT III accuracy with about 70% fewer stored parameters. An 8-experts model distilled using only the teacher's output features retains nearly all of its DINOv2 teacher's linear-probe accuracy and transfers across classification, segmentation, and depth prediction. Elastic-depth training allows one checkpoint (trained model) to operate at multiple tested depths by resampling the same normalized coordinate interval. For fixed-depth deployment, the recurrent block can be materialized as a conventional dense graph, removing online routing and merging without changing the one-FFN-per-depth compute but expanding deployment storage.
Read the full abstract ▾
Machine Learning
Telling FIBS: Probes Effectively Detect Sabotage and Catch Unverbalized Deception
Oskar J. Hollinsworth, Alex F. Spies, Tigist Diriba et al. (+2 more)
Recent incidents have highlighted the challenge of monitoring LLM
agents and the danger of models deceiving people. We show that white-box deception detection via probes can be scaled up to frontier monitoring settings by collecting the largest deception
dataset to date for training probes and introducing a novel probe architecture which can aggregate information across many layers and
tokens. Our probes surpass black-box monitors on agentic coding
benchmarks, achieving 98.3% AUROC in SHADE-Arena and 98.0% in BashArena (vs. 97.7%, 97.6% resp. for action-only text monitors), and show improved efficacy as the underlying model is scaled up. We further test probes on several cases where deception cannot be determined from the context alone; in these cases of introspective deception, the ground truth can only be determined through careful elicitation or knowledge of a model's training data. In one such evaluation, we show that probes can distinguish transcripts containing a model's true hidden goal from other goals with an AUROC of up to 99.7%. Our probes also detect deception on prominent open-weight models which lie about politically sensitive topics when put under pressure. We release our training
dataset, dubbed FIBS, to help drive frontier deployment of effective probes, and encourage the community to expand upon it with further examples of deception and sabotage.
Read the full abstract ▾
All authors Oskar J. Hollinsworth, Alex F. Spies, Tigist Diriba, Adam Gleave, Chris Cundy
Machine Learning
Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization
Hanyang Li, Shao Tang, Daniel Thomas Braithwaite et al. (+7 more)
Quantizing AdamW's optimizer states reduces persistent storage, but quantization errors propagate through the moment recurrences and perturb subsequent adaptive updates. We redesign 4-bit optimizer-state quantization for AdamW from the perspective of rounding space: the coordinate in which a quantizer chooses between adjacent reconstruction levels. For the second moment, a local analysis of the quantization
cell adjacent to zero shows that small mean state error need not imply small mean preconditioner error at the next step. A one-dimensional quadratic construction further shows qualitatively different optimization dynamics under state-space and preconditioner-space rounding. These results motivate Zero-Inclusive Preconditioner-space
Stochastic Rounding (ZIP-SR), which retains zero in the second-moment codebook and computes
stochastic-rounding probabilities in preconditioner space. As a complementary route, Zero-Excluding EDEN calibration (ZE-EDEN) uses a zero-excluding second-moment codebook and rescales the quantized second-moment block to mitigate the preconditioner distortion caused by the positive quantization floor. Both configurations use 4-bit NormalFloat (NF4) for the first moment, with targeted
stochastic rounding of the LM-head first moment during the final 10% of training. Across GPT- and Llama-style pretraining experiments ranging from 130M to 2.7B parameters, both methods reduce TorchAO 4-bit AdamW's mean validation-loss gap to 32-bit AdamW at every evaluated model size, with the largest reported gap reduction reaching 70%. In full-parameter supervised fine-tuning, both recipes achieve lower validation loss than TorchAO while remaining close to 32-bit AdamW on downstream tasks.
Read the full abstract ▾
All authors Hanyang Li, Shao Tang, Daniel Thomas Braithwaite, Gregory Dexter, Leonardo Neves, Aman Gupta, Hiroto Udagawa, Abhishek Shivanna, Daniel Silva, Rohan Ramanath
Machine Learning · Cross-listed from stat.ML
Density Ratio Estimation with Stein Displacement Fields
Song Liu
Density ratios quantify distribution shift from a probability-mass point of view, whereas displacement fields describe, from a dynamical point of view, how one distribution is transported onto another. Although both offer complementary insights, they are usually estimated separately, and converting one into the other requires post-processing. In this paper, we estimate the density ratio between a target and a base distribution by parametrizing it through a displacement field acting on the base: the log-ratio is modeled as minus the Stein operator of the base applied to the field, up to a normalizing constant. This gives both statistical and dynamical descriptions of the distribution shift through a single convex optimization problem. Iterating this estimate-and-move step gives two inference
algorithms: push-forward moves the model and corrects a pretrained sampler without retraining it, whereas pull-back moves the data closer to the base and fits a transformation model one layer at a time. Applications to distribution shift in simulation-based inference and to nonlinear independent component analysis illustrate the benefits and limitations of the approach.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.RO
VioLA: Learning Generalist Humanoid Control Policies from Human Data
Mert Albaba, Jens Beißwenger, Anna Manasyan et al. (+6 more)
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 $\Psi_0$ 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.
Read the full abstract ▾
All authors Mert Albaba, Jens Beißwenger, Anna Manasyan, Daniel Marta, Michael J. Black, Wieland Brendel, Andreas Krause, Georg Martius, Martin Riedmiller
Machine Learning · Cross-listed from cs.RO
FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems
Songyuan Zhang, Baljeet Singh, Sarthak Ranjeet Kaingade et al. (+2 more)
Safe reinforcement learning commonly places safety and task performance in the same policy objective, where they can introduce competing updates. Safety filters separate them at action execution, but classical designs require an analytic safety function and dynamics model, and standard minimal-intervention filters are myopic to long-horizon task return because they minimize only instantaneous action deviation. Hard projections are also undefined when no safe action exists. We present FAITH, a feasibility-aware, model-free framework that approximates the optimal state-action safety value and amortizes minimal-intervention filtering with a feedforward network. The task policy optimizes the task return through the filtered dynamics, which recovers the feasible constrained problem without a competing safety term in the task-policy update. When no action satisfies the learned safety condition, the same filter approaches the action with minimum predicted peak harm. On a double integrator example and a Safety Gym environment, FAITH achieves the highest return among methods with no feasible-start violations and matches the lowest harm from infeasible starts. On a 29-DoF humanoid, it reaches a 99.95% safety rate while retaining 97% of the unfiltered return in Walking-Avoid, and obtains the highest measured safety rate in Push-Avoid by learning to sacrifice balancing and fall away from the protected region. The same policies are also demonstrated on a real-world Unitree G1 humanoid.
Read the full abstract ▾
All authors Songyuan Zhang, Baljeet Singh, Sarthak Ranjeet Kaingade, Chuchu Fan, Bryan Trinh
Machine Learning · Cross-listed from quant-ph
Toward Joint Optimization of Circuit Depth and Training Data Size in Adaptively Grown Quantum Classifiers
Saeefa Rubaiyet Nowmi, Md Mahmuduzzaman Kamol and Mohammad Saidur Rahman
Building a
quantum model involves a tradeoff: how complex the
circuit should be, and how much training data it needs. Caro et al. show that models with fewer trainable gates need less training data to generalize well. Q-FLAIR shows that a
quantum feature-map
circuit can be grown gate-by-gate, stopping once further growth stops improving the training loss. We ask whether these two results combine into a predictable scaling law. Does Q-FLAIR's own stopping rule pick larger or smaller
circuits as training data grows? Does the resulting generalization behavior track Caro et al.'s bound? We reimplement Q-FLAIR's growth mechanism faithfully, including its analytic reconstruction and exact stopping rule. We run it on full-resolution (784-pixel) MNIST 3-vs-5 classification, at five training-set sizes from N = 2000 to 10000. We then fine-tune each resulting
circuit, so we can measure Caro et al.'s notion of active gates, K. We find no predictable relationship between training-set size and the
circuit size Q-FLAIR converges to.
Circuit size and test accuracy both vary non-monotonically with N, and seed-to-seed variance is nearly as large as any trend across N. The empirical generalization gap never exceeds Caro et al.'s bound in 14 of 15 runs, so the bound holds as a valid guarantee in those runs. But the gap correlates only weakly with the bound's value (r = 0.12). This shows that K does not explain most of the variation we observe. Why a valid guarantee can coexist with such weak predictive power remains an open question, and answering it may be necessary before
circuit depth and training data size can be jointly optimized in practice.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CV
Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching
Luping Liu, Bingyi Kang, Yifan Wang et al. (+1 more)
Dense correspondence matching has historically been bounded by simplifying spatio-temporal priors, such as smooth motion and rigid geometry. While effective for classical tasks, these assumptions break down in image editing and reference-guided generation (IEG), where transformations can preserve visual identity while breaking physical continuity. To establish identity-preserving correspondence across such transformations, we introduce FreeMatching, a generalizable framework combining generative and semantic foundation representations with heterogeneous supervision from classical
datasets, tracked
videos, and synthetic scenes. Teacher-guided iterative refinement further improves correspondence in IEG without dense correspondence annotations. Experimentally, a single FreeMatching model substantially improves correspondence quality on challenging IEG image pairs while retaining competitive performance on classical
benchmarks. Furthermore, we demonstrate its utility as a quantitative metric for evaluating identity preservation, with scores that correlate with human judgment. The code is available at https://github.com/luping-liu/FreeMatching.
Read the full abstract ▾
All authors Luping Liu, Bingyi Kang, Yifan Wang, Dong Xu
Machine Learning
A Unified Bellman Operator for Safety-Critical Reinforcement Learning
Nishanth Arun Rao, Royina Karegoudra Jayanth, Benjamin Eysenbach et al. (+1 more)
Reinforcement learning in safety-critical domains requires maximizing task performance while strictly adhering to safety constraints. Existing safe reinforcement learning paradigms typically force a trade-off: they either require a priori knowledge to provide strict safety guarantees (e.g., safety filters), or they enable joint learning but only satisfy safety constraints on average. In this work, we propose a novel Bellman operator that unifies performance and safety objectives into a joint value function. We show that temporal difference learning with the joint Bellman operator converges under a two-timescale
stochastic approximation framework. On the fast timescale, the safety value of the learning joint policy is estimated, while the joint value is estimated on the slow timescale. Convergence is ensured by formulating the limiting dynamics as an occupation-averaged differential inclusion, and showing that it asymptotically converges to a set of limiting optimal safety-constrained task value functions. Theoretically, once converged, the resulting optimal policy maximizes task return while maintaining safety at all times. Empirical evaluations on continuous control tasks with neural approximations demonstrate stable convergence with near-zero safety violations at test time.
Read the full abstract ▾
All authors Nishanth Arun Rao, Royina Karegoudra Jayanth, Benjamin Eysenbach, Jaime Fernández Fisac
Machine Learning · Cross-listed from cs.CV
WOVEN: Weaving Visual World Modeling into Multimodal LLMs
Zheyu Fan, Yue Zhang, Mingkai Deng et al. (+9 more)
Multimodal large
language models (MLLMs) struggle with spatial, embodied, physical, and temporal
reasoning. We hypothesize that these failures reflect a shared deficit in visual transition
reasoning, and test whether this capability can serve as a shared training primitive, one that different models can learn from different supervision sources and reuse across different tasks, with a systematic training recipe. Existing
benchmarks document these deficits separately but do not support controlled comparisons across scenes, actions, and
reasoning operations. We therefore introduce WOVEN, a training source and
benchmark for visual transition
reasoning that organizes transition supervision by scene, action, and
reasoning type, using diverse, realistic rollouts from
video-pretrained generative models: 36,076 examples across 20 scene types, 5 action types, and 8
reasoning types. We first evaluate 38 frontier MLLMs (e.g., GPT-5.4 and Qwen3-VL-235B-A22B) and find a substantial and systematic deficit: even the strongest models fall far below humans, and the failures recur across model families and persist with scale. We then train MLLMs at multiple scales on WOVEN and find that they learn a shared capability that transfers broadly: training subsets of only about 2,000 items each collectively improve 22 of 26 external
benchmarks by up to 27.3 percentage points, and WOVEN data can replace 30-50% of a task's own training data with comparable accuracy. Controlled comparisons further yield a training recipe for visual world modeling, validated prospectively on held-out
benchmarks: select supervision by the
reasoning operation it teaches rather than by the actions, scenes, or domains it shows, and prefer larger changes to the visual state for robustness. Our work establishes visual transition
reasoning as a reusable foundation for systematic visual world-model training in MLLMs.
Read the full abstract ▾
All authors Zheyu Fan, Yue Zhang, Mingkai Deng, Kangrui Wang, Qineng Wang, Canyu Chen, Jie Hao, Xing Fan, Chenlei Guo, Eric P. Xing, Mohit Bansal, Manling Li
Machine Learning · Cross-listed from cs.CL
Predicting Alignment Generalization with Value Representations
Andy Liu, Mehar Bhatia, Karolina Stanczak et al. (+3 more)
LLM developers post-train their models to exhibit prosocial values and behavioral traits, which are enumerated in an
alignment target. However, while recent post-training developments have yielded models that score highly on
alignment evaluations, training models on sets of narrow behaviors still influences their behavior across unseen contexts and environments in unexpected ways. In this paper, we establish the task of
alignment generalization prediction, i.e., predicting how fine-tuning a model to follow a given value changes its behavior across a wide range of held-out values. We conduct a large-scale analysis of
alignment generalization effects across 66 values found in modern
alignment targets, and
benchmark representational techniques on the
alignment generalization prediction task. We find that representations based on model activations when applying values in context significantly outperform methods based on textual descriptions of the values. Specifically, the best activations-based methods achieve correlations of 0.45 with our generalization
matrix, compared with 0.05 from description-based baselines. We then show the applicability of representations that predict
alignment generalization toward downstream tasks by using them to measure how similar the values in a multi-value
alignment target are, which we find is significantly correlated with model robustness. Finally, we show initial evidence towards a shared, model-independent value space, which we use to develop the first taxonomy of LLM values grounded in empirical generalization dynamics. Our work demonstrates the importance of studying value generalization in LLMs and its application toward the more empirical design and training of model behavior.
Read the full abstract ▾
All authors Andy Liu, Mehar Bhatia, Karolina Stanczak, Mona Diab, Vered Shwartz, Daniel Fried
Machine Learning
Learning Kilometer-Scale Weather Prediction with Global-Regional Alignment
Guowen Li, Yang Liu, Yujie Wang et al. (+5 more)
Kilometer-scale regional weather forecasting is essential for local weather warnings and weather-sensitive decisions. Existing data-driven approaches often rely on numerical forecasts for large-scale guidance or require additional training of global forecasting components. Pretrained global weather models offer an efficient source of large-scale forecasts, motivating their reuse to guide high-resolution regional prediction. However, this coupling requires aligning global and regional representations across different grids and integrating global guidance with local interactions to advance regional states. We propose ScaleCast, a regional forecasting framework that addresses these challenges through Global-Regional
Alignment. Its Global-Regional Conversion module aligns joint global and regional representations with regional locations, while the Global-Regional
Alignment and Dynamics block combines aligned guidance with regional neighborhood interactions. Experiments using ERA5 global analyses on a 0.25-degree grid and CERRA regional reanalysis at 5.5 km spacing demonstrate improved regional forecasts across surface and upper-air variables, with a single trained model supporting multiple global forecast drivers (i.e., Pangu-Weather, GraphCast, and HRES) without specific retraining. Fine-tuning on HRRR at 3 km spacing further demonstrates the framework's adaptability to a different regional domain and spatial resolution. Windstorm case studies show improved cyclone positioning and core-pressure estimates, while comparisons with HadISD station observations show closer agreement with local temperature and humidity changes.
Read the full abstract ▾
All authors Guowen Li, Yang Liu, Yujie Wang, Qiuyan Sun, Haoyuan Liang, Juepeng Zheng, Hong Cheng, Haohuan Fu
Machine Learning
Prospective Prediction of OOD Degradation from Source-Side Training Dynamics
Sasha Monin
We study whether persistent out-of-distribution (OOD) degradation can be predicted before it is directly observed using only source-side training dynamics. In a controlled shortcut-learning setting, a simple logistic regression predictor develops a clear prospective signal, while training time alone does not. Temporal summaries of the source-side quantities are substantially more informative than their current values. When transferred without additional training from a CNN to an MLP, confidence and
entropy dynamics retain substantial predictive information. These results provide a proof of principle that source-side training dynamics can contain an early warning signal for future OOD failure.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling
Qun Dai, Liangjian Wen, Jiang Duan et al. (+7 more)
Self-supervised
multimodal representation learning has achieved remarkable success across diverse domains, yet capturing synergistic information remains challenging due to the complexity of cross-modal interactions. Unlike the shared information across individual modalities, synergy arises when task-relevant signals emerge only from the joint configuration of multiple modalities and cannot be recovered from any modality in isolation. This work focuses on how to preserve the information capacity for such synergistic signals in
multimodal representations. The key observation is that synergistic information is reflected in higher-order statistical dependence among modalities, which provides a principled target for explicitly modeling joint interactions. Motivated by this insight, we propose Higher-order Representation and Information Learning (HRIL), which constructs an empirical cross-moment tensor over modality
embeddings to represent multi-way interactions. HRIL employs Tucker decomposition to obtain a core tensor, complemented by a synergy-aware regularizer that prevents energy concentration and preserves higher-order coupling capacity for synergistic information capture. Experiments on the controlled synergy task and real-world
benchmarks demonstrate consistent improvements over existing
multimodal contrastive methods, with notable gains on tasks dominated by synergistic interactions. Code is released at https://github.com/brightest66/HRIL.
Read the full abstract ▾
All authors Qun Dai, Liangjian Wen, Jiang Duan, Yong Dai, Dongkai Wang, Maolin Wang, Mingjie Wang, Jianzhuang Liu, He Yan, Zhao Kang
Machine Learning
Long Text to Predictive Features: LLM-Guided Blockwise Feature Engineering via Executable Program Search
Ziming Dai, Dabiao Ma, Ziheng Guo et al. (+2 more)
Industrial risk-control systems typically rely on structured-data models for efficient prediction, yet substantial valuable information remains embedded in unstructured long text. Extracting this information through manual feature engineering is labor-intensive, while requiring a large
language model (LLM) to process every real-time input may not meet practical deployment requirements. To address this challenge, we propose LLM-BlockFE, an LLM-guided offline feature construction framework that converts long text into executable feature programs, thereby avoiding LLM calls during online inference. LLM-BlockFE constructs feature programs by incrementally appending immutable code blocks and evaluates candidate features using a downstream model. To address the tendency of conventional greedy search to become trapped in suboptimal solutions, our method introduces a block-level rollback mechanism based on depth-calibrated credit allocation and advances multiple independent search trajectories in an interleaved manner, reducing redundant exploration by sharing fixed descriptions of each trajectory's exploration direction. After the search, the resulting programs are frozen and deployed to extract structured features for downstream prediction models. Across two public and two private
datasets, LLM-BlockFE achieves absolute AUC improvements of 0.0069 to 0.0358 over the strongest baseline on each
dataset in the full-
dataset comparison. Post-launch monitoring across five deployed financial risk-control applications shows absolute KS improvements of 0.02 to 1.56 percentage points over the existing manually designed strategy.
Read the full abstract ▾
All authors Ziming Dai, Dabiao Ma, Ziheng Guo, Jack Dong, Zimu Zhou