Searcharxiv⌕ Search

arXiv subjects

Sepp Hochreiter

Publications and source records attributed to Sepp Hochreiter.

At least 19 recordsLinked to original sources
arXiv2026

ParetoTransport: Generative Optimization by Mass Transport Toward The Pareto Front

Offline multi-objective optimization requires not only moving the objective vectors of candidate designs toward the Pareto front, but also distributing them effectively along it. Generative methods have recently emerged as a natural approach because they learn a distribution over feasible designs while allowing generation to be steered toward promising designs. Existing methods, however, largely retain classical sample-wise guidance strategies, leaving the distribution-level modeling capability of generative methods underused. We propose ParetoTransport, a training-free guidance method for pre-trained flow-matching models that explicitly specifies and refines a population-level distribution in objective space. ParetoTransport guides a flow-matching sampler to iteratively transport the empirical offline distribution toward the Pareto front, with Wasserstein matching to intermediate proxy distributions. This directly controls distributional displacement and mass allocation along the front. We establish a convergence result and demonstrate state-of-the-art performance on standard offline MOO benchmarks, extending recent evaluations beyond hypervolume to generational distance, inverted generational distance, and Wasserstein distance.

cs.LG↗
arXiv2026

KVpop -- Key-Value Cache Compression with Predictive Online Pruning

Key-value (KV) cache growth is a major bottleneck in autoregressive decoding, as memory and bandwidth scale linearly with context length. Existing KV eviction methods often rely on static heuristics or proxy scores, which poorly track future token utility and cause brittle eviction as relevance shifts. To address this, we introduce KVpop, which learns a fixed-budget KV eviction policy by directly supervising the keep-or-drop decision. The scorer is trained against a novel future-attention target, computed efficiently without materializing dense attention maps. We further introduce a delayed memory-based scorer that, uniquely among learned eviction methods, defers scoring for a fixed number of steps to exploit near-future context. On AIME and HMMT mathematical reasoning, KVpop retains 98% of full-attention performance on Qwen3-4B at 75% KV cache compression and 97% at 88% compression, consistently outperforming established eviction baselines. Qwen3-8B shows even stronger results, reaching near-full teacher performance. These results show that supervising eviction with future-attention signals cuts memory costs while maintaining quality.

cs.LG↗
arXiv2026

TiRex-2: Generalizing TiRex to Multivariate Data and Streaming

We introduce TiRex-2, a recurrent xLSTM-based time series foundation model that generalizes the univariate TiRex to multivariate forecasting with both past and future covariates. Real-world forecasting is inherently sequential: observations arrive continuously, variables evolve jointly, and a subset of covariates is known ahead of time. Existing Transformer-based time series foundation models capture cross-variate dependencies but incur quadratic complexity in context length and require full-history recomputation as new observations arrive. TiRex-2 addresses these limitations through a memory-centric recurrent design that operates at constant per-patch cost under streaming. The model combines a bidirectional time mixer with an asymmetric grouped-attention variate mixer, enabling the integration of future-known covariates while preserving strict causality over target variables. To our knowledge, this is the first time series foundation model that achieves this combination of properties. To support scalable multivariate pretraining, we propose a synthetic coupling pipeline that composes diverse multivariate samples on the fly from large univariate corpora. Empirically, TiRex-2 achieves state-of-the-art zero-shot performance on GIFT-Eval and fev-bench, remains stable when streamed to arbitrary context lengths, and maintains constant inference cost per patch. The model uses 38.4M active parameters in univariate mode, with an additional 44.1M parameters activated for multivariate forecasting.

cs.LG↗
arXiv2026

Accelerating LLM Inference via Vector Index Based Output Embeddings

Large output embedding matrices create a significant memory bandwidth bottleneck during autoregressive decoding, especially for compact LLMs with large multilingual vocabularies. We reformulate the output projection followed by top-k token selection as a maximum inner product search over token embeddings and replace the dense vocabulary projection with an HNSW-based vector index. The resulting output head retrieves only a small candidate set of high-scoring tokens and can be integrated into existing decoding pipelines by scattering retrieved logits into a sparse full-vocabulary tensor. On CPU inference with Gemma 3, Llama 3.2, and Qwen 3 models, our method substantially accelerates the output projection and improves end-to-end batch-size-one decoding throughput by up to 82% for Gemma 3 270M, while preserving generation quality under AlpacaEval evaluation. These results suggest approximate retrieval is a practical alternative to dense output projections in latency-sensitive small-batch decoding.

cs.CL↗
arXiv2026

On Subquadratic Architectures: From Applications to Principles

Transformers dominate modern sequence modeling, but their quadratic attention incurs substantial computational cost. Subquadratic architectures offer a scalable alternative. However, it remains unclear which designs yield the most effective sequence models. We compare three leading approaches: xLSTM, Mamba-2, and Gated DeltaNet. We evaluate these models on tasks with complex dependencies: (1) code-model pre-training, (2) distillation of code models from large language models, and (3) pre-training of time-series foundation models. Across these settings, xLSTM delivers the strongest overall performance. To explain xLSTM's advantage, we present a unified formulation and analyze the underlying architectural mechanisms, focusing on state tracking and memory dynamics. Our results show that xLSTM enables more flexible and stable memory correction via its gating scheme. We corroborate these findings on controlled synthetic length-generalization tasks. Overall, our findings indicate that xLSTM's gains on complex tasks stem from robust state tracking and accumulation.

cs.LG↗
arXiv2026

RREDCoT: Segment-Level Reward Redistribution for Reasoning Models

Recent advancements in reasoning language models have been driven by Reinforcement Learning (RL) fine-tuning. Most often, these rely on the Group Relative Policy Optimization (GRPO) algorithm or modifications thereof to steer the models to produce Chain-of-Thought (CoT) traces. The final answer can only be verified, and the reward assigned, after the CoT trace is complete, making it a delayed reward problem. GRPO and its modifications correspond to Monte Carlo methods in standard RL, which are known to suffer from high variance. A possible solution to this problem is the redistribution of rewards through credit assignment, where segments of the CoT trace that are important for arriving at the desirable solution are emphasized by assigning a higher reward. While Monte Carlo sampling can be used to provide an unbiased estimate of intermediate state values, its computational overhead makes it unsuitable for train-time credit assignment in long contexts at high granularity. We introduce RREDCoT (Reward REDistribution for Chain of Thoughts), which utilizes the model itself to approximate the optimal reward redistribution without additional generation. We investigate the advantages of our method compared to MC sampling and several attribution methods. We further analyze several aspects relevant to the construction of the redistribution such as segmentation of CoT traces and state value estimation.

cs.LG↗
arXiv2026

Unlocking the Working Memory of Large Language Models for Latent Reasoning

To improve the reasoning capabilities of large language models, test-time compute is typically scaled by generating intermediate tokens before the final answer. However, this couples reasoning to autoregressive generation and thereby conflates internal computation with external communication. In contrast, human cognition can use working memory to hold and manipulate information internally without the need to externalize intermediate thoughts. Drawing on this principle, we introduce Reasoning in Memory (RiM), a latent reasoning method that replaces the autoregressive generation of reasoning steps with memory blocks. These memory blocks are fixed sequences of special tokens that unlock the working-memory capacity of large language models. Since they are fixed rather than generated, they can be processed in a single forward pass, enabling compute-efficient latent reasoning. To operationalize these memory blocks, we employ a two-stage curriculum. First, we ground them by predicting explicit reasoning steps after each memory block. Second, we discard this step-level supervision and iteratively refine the final answer after each memory block. Our experiments on reasoning benchmarks show that, across language models of different families and sizes, RiM matches or exceeds existing latent reasoning methods while avoiding the autoregressive generation of thoughts. These results demonstrate that large language models can be trained to use working memory as an effective mechanism for latent reasoning.

cs.CL↗
arXiv2026

The Offline-Frontier Shift: Diagnosing Distributional Limits in Generative Multi-Objective Optimization

Offline multi-objective optimization (MOO) aims to recover Pareto-optimal designs given a finite, static dataset. Recent generative approaches, including diffusion models, show strong performance under hypervolume, yet their behavior under other established MOO metrics is less understood. We show that generative methods systematically underperform evolutionary alternatives with respect to other metrics, such as generational distance. We relate this failure mode to the offline-frontier shift, i.e., the displacement of the offline dataset from the Pareto front, which acts as a fundamental limitation in offline MOO. We argue that overcoming this limitation requires out-of-distribution sampling in objective space (via an integral probability metric) and empirically observe that generative methods remain conservatively close to the offline objective distribution. Our results position offline MOO as a distribution-shift--limited problem and provide a diagnostic lens for understanding when and why generative optimization methods fail.

cs.LG↗
arXiv2026

Rethinking Uncertainty Estimation in LLMs: A Principled Single-Sequence Measure

Large Language Models (LLMs) are increasingly employed in real-world applications, driving the need to evaluate the trustworthiness of their generated text. To this end, reliable uncertainty estimation is essential. Leading uncertainty estimation methods generate and analyze multiple output sequences, which is computationally expensive and impractical at scale. In this work, we inspect the theoretical foundations of these methods and explore new directions to enhance computational efficiency. Building on the framework of proper scoring rules, we find that the negative log-likelihood of the most likely output sequence constitutes a theoretically principled uncertainty measure. To approximate this alternative measure, we propose G-NLL, obtained using a single output sequence from greedy decoding. This approach streamlines uncertainty estimation while preserving theoretical rigor. Empirical results demonstrate that G-NLL achieves state-of-the-art performance across various scenarios. Our work lays the theoretical foundation for efficient and reliable uncertainty estimation in natural language generation, challenging the necessity of the prevalent methods that are more complex and resource-intensive.

cs.LG↗
arXiv2026

Effective Distillation to Hybrid xLSTM Architectures

There have been numerous attempts to distill quadratic attention-based large language models (LLMs) into sub-quadratic linearized architectures. However, despite extensive research, such distilled models often fail to match the performance of their teacher LLMs on various downstream tasks. We set out the goal of lossless distillation, which we define in terms of tolerance-corrected Win-and-Tie rates between student and teacher on sets of tasks. To this end, we introduce an effective distillation pipeline for xLSTM-based students. We propose an additional merging stage, where individually linearized experts are combined into a single model. We show the effectiveness of this pipeline by distilling base and instruction-tuned models from the Llama, Qwen, and Olmo families. In many settings, our xLSTM-based students recover most of the teacher's performance, and even exceed it on some downstream tasks. Our contributions are an important step towards more energy-efficient and cost-effective replacements for transformer-based LLMs.

cs.LG↗
arXiv2026

xLSTM Scaling Laws: Competitive Performance with Linear Time-Complexity

Scaling laws play a central role in the success of Large Language Models (LLMs), enabling the prediction of model performance relative to compute budgets prior to training. While Transformers have been the dominant architecture, recent alternatives such as xLSTM offer linear complexity with respect to context length while remaining competitive in the billion-parameter regime. We conduct a comparative investigation on the scaling behavior of Transformers and xLSTM along the following lines, providing insights to guide future model design and deployment. First, we study the scaling behavior for xLSTM in compute-optimal and over-training regimes using both IsoFLOP and parametric fit approaches on a wide range of model sizes (80M-7B) and number of training tokens (2B-2T). Second, we examine the dependence of optimal model sizes on context length, a pivotal aspect that was largely ignored in previous work. Finally, we analyze inference-time scaling characteristics. Our findings reveal that in typical LLM training and inference scenarios, xLSTM scales favorably compared to Transformers. Notably, xLSTM models consistently Pareto-dominate Transformer models, delivering lower cross-entropy loss for the same compute budget.

cs.LG↗
arXiv2026

Adaptive Retrieval helps Reasoning in LLMs -- but mostly if it's not used

Large Language Models (LLMs) often falter in complex reasoning tasks due to their static, parametric knowledge, leading to hallucinations and poor performance in specialized domains like mathematics. This work explores a fundamental principle for enhancing generative models: treating retrieval as a form of dynamic in-context learning. We test an adaptive retrieval-augmented architecture where an LLM agent actively decides when to query an external knowledge base during its reasoning process. We compare this adaptive strategy against a standard Chain-of-Thought (CoT) baseline and a static retrieval approach on the GSM8K and MATH-500 benchmarks. Although our experiments show that static retrieval is inferior to CoT, the adaptive retrieval shows interesting behavior: While traces including retrieved results show slightly worse performance compared to CoT, traces that do not include retrieval actually perform better compared to CoT. This suggests that: (a) retrieval only rarely helps reasoning (we show a few counterexamples, e.g. using useful theorems) and (b) actively not using retrieval is indicative of good model performance. Furthermore, we find that the model scales its retrieval frequency with the difficulty of the problem, reinforcing that the decision to retrieve is a crucial metacognitive signal. The agent's ability to self-assess its knowledge and selectively engage with external information represents a key principle for building more robust and reliable generative models.

cs.LG↗
arXiv2026

AP-OOD: Attention Pooling for Out-of-Distribution Detection

Out-of-distribution (OOD) detection, which maps high-dimensional data into a scalar OOD score, is critical for the reliable deployment of machine learning models. A key challenge in recent research is how to effectively leverage and aggregate token embeddings from language models to obtain the OOD score. In this work, we propose AP-OOD, a novel OOD detection method for natural language that goes beyond simple average-based aggregation by exploiting token-level information. AP-OOD is a semi-supervised approach that flexibly interpolates between unsupervised and supervised settings, enabling the use of limited auxiliary outlier data. Empirically, AP-OOD sets a new state of the art in OOD detection for text: in the unsupervised setting, it reduces the FPR95 (false positive rate at 95% true positives) from 27.84% to 4.67% on XSUM summarization, and from 77.08% to 70.37% on WMT15 En-Fr translation.

cs.LG↗
arXiv2025

Tiled Flash Linear Attention: More Efficient Linear RNN and xLSTM Kernels

Linear RNNs with gating recently demonstrated competitive performance compared to Transformers in language modeling. Although their linear compute scaling in sequence length offers theoretical runtime advantages over Transformers, realizing these benefits in practice requires optimized custom kernels, as Transformers rely on the highly efficient Flash Attention kernels (Dao, 2024). Leveraging the chunkwise-parallel formulation of linear RNNs, Flash Linear Attention (FLA) (Yang & Zhang, 2024) shows that linear RNN kernels are faster than Flash Attention, by parallelizing over chunks of the input sequence. However, since the chunk size of FLA is limited, many intermediate states must be materialized in GPU memory. This leads to low arithmetic intensity and causes high memory consumption and IO cost, especially for long-context pre-training. In this work, we present Tiled Flash Linear Attention (TFLA), a novel kernel algorithm for linear RNNs, that enables arbitrary large chunk sizes and high arithmetic intensity by introducing an additional level of sequence parallelization within each chunk. First, we apply TFLA to the xLSTM with matrix memory, the mLSTM (Beck et al., 2024). Second, we propose an mLSTM variant with sigmoid input gate and reduced computation for even faster kernel runtimes at equal language modeling performance. In our speed benchmarks, we show that our new mLSTM kernels based on TFLA outperform highly optimized Flash Attention, Linear Attention and Mamba kernels, setting a new state of the art for efficient long-context sequence modeling primitives.

cs.LG↗
arXiv2025

Rethinking Losses for Diffusion Bridge Samplers

Diffusion bridges are a promising class of deep-learning methods for sampling from unnormalized distributions. Recent works show that the Log Variance (LV) loss consistently outperforms the reverse Kullback-Leibler (rKL) loss when using the reparametrization trick to compute rKL-gradients. While the on-policy LV loss yields identical gradients to the rKL loss when combined with the log-derivative trick for diffusion samplers with non-learnable forward processes, this equivalence does not hold for diffusion bridges or when diffusion coefficients are learned. Based on this insight we argue that for diffusion bridges the LV loss does not represent an optimization objective that can be motivated like the rKL loss via the data processing inequality. Our analysis shows that employing the rKL loss with the log-derivative trick (rKL-LD) does not only avoid these conceptual problems but also consistently outperforms the LV loss. Experimental results with different types of diffusion bridges on challenging benchmarks show that samplers trained with the rKL-LD loss achieve better performance. From a practical perspective we find that rKL-LD requires significantly less hyperparameter optimization and yields more stable training behavior.

cs.LG↗
arXiv2025

Improving Uncertainty Estimation through Semantically Diverse Language Generation

Large language models (LLMs) can suffer from hallucinations when generating text. These hallucinations impede various applications in society and industry by making LLMs untrustworthy. Current LLMs generate text in an autoregressive fashion by predicting and appending text tokens. When an LLM is uncertain about the semantic meaning of the next tokens to generate, it is likely to start hallucinating. Thus, it has been suggested that predictive uncertainty is one of the main causes of hallucinations. We introduce Semantically Diverse Language Generation (SDLG) to quantify predictive uncertainty in LLMs. SDLG steers the LLM to generate semantically diverse yet likely alternatives for an initially generated text. This approach provides a precise measure of aleatoric semantic uncertainty, detecting whether the initial text is likely to be hallucinated. Experiments on question-answering tasks demonstrate that SDLG consistently outperforms existing methods while being the most computationally efficient, setting a new standard for uncertainty estimation in LLMs.

cs.LG↗
arXiv2025

TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learning

In-context learning, the ability of large language models to perform tasks using only examples provided in the prompt, has recently been adapted for time series forecasting. This paradigm enables zero-shot prediction, where past values serve as context for forecasting future values, making powerful forecasting tools accessible to non-experts and increasing the performance when training data are scarce. Most existing zero-shot forecasting approaches rely on transformer architectures, which, despite their success in language, often fall short of expectations in time series forecasting, where recurrent models like LSTMs frequently have the edge. Conversely, while LSTMs are well-suited for time series modeling due to their state-tracking capabilities, they lack strong in-context learning abilities. We introduce TiRex that closes this gap by leveraging xLSTM, an enhanced LSTM with competitive in-context learning skills. Unlike transformers, state-space models, or parallelizable RNNs such as RWKV, TiRex retains state-tracking, a critical property for long-horizon forecasting. To further facilitate its state-tracking ability, we propose a training-time masking strategy called CPM. TiRex sets a new state of the art in zero-shot time series forecasting on the HuggingFace benchmarks GiftEval and Chronos-ZS, outperforming significantly larger models including TabPFN-TS (Prior Labs), Chronos Bolt (Amazon), TimesFM (Google), and Moirai (Salesforce) across both short- and long-term forecasts.

cs.LG↗
arXiv2025

Pre-trained Forecasting Models: Strong Zero-Shot Feature Extractors for Time Series Classification

Recent research on time series foundation models has primarily focused on forecasting, leaving it unclear how generalizable their learned representations are. In this study, we examine whether frozen pre-trained forecasting models can provide effective representations for classification. To this end, we compare different representation extraction strategies and introduce two model-agnostic embedding augmentations. Our experiments show that the best forecasting models achieve classification accuracy that matches or even surpasses that of state-of-the-art models pre-trained specifically for classification. Moreover, we observe a positive correlation between forecasting and classification performance. These findings challenge the assumption that task-specific pre-training is necessary, and suggest that learning to forecast may provide a powerful route toward constructing general-purpose time series foundation models.

cs.LG↗