Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
arXiv:2603.02221v3 Announce Type: replace-cross Abstract: In clinical tabular prediction, classical machine learning models with feature engineering often outperform neural methods. LLMs are increasi...
ReForge is a bilevel optimization framework that refines merged models by treating module-wise refinement as Bayesian linear regression with an anchor-centered prior. The inner level produces a closed‑form MAP estimate from unlabeled calibration activations, while the outer level employs Bayesian optimization to jointly select regularization strengths and assembly scales using validation data. A data‑free variant replaces activation statistics with task‑vector Grams, enabling refinement without calibration examples, and across extensive vision and language benchmarks ReForge consistently outperforms existing plug‑and‑play anchor baselines, achieving significant accuracy gains on large‑scale tasks such as 20‑task ViT‑B/32 and eight‑task ViT‑L/14.
arXiv:2606.00341v2 Announce Type: replace-cross Abstract: As AI agents are increasingly deployed in real personal and corporate settings (email accounts, development workflows, company databases, etc...
arXiv:2606.09030v2 Announce Type: replace-cross Abstract: Clinical early warning systems built on irregularly sampled medical time series (ISMTS) from electronic health records must deliver continuou...
arXiv:2609.35763v3 Announce Type: replace Abstract: Distributional training provides collective supervision for one-step visual generation by matching real and generated features in frozen representa...
arXiv:2609.39644v2 Announce Type: new Abstract: Ribosome profiling (Ribo-seq) measures ribosome distributions along mRNAs, but observed occupancy profiles also contain experiment-specific distortions...
arXiv:2609.39792v2 Announce Type: new Abstract: Distinguishing periodic from chaotic dynamics in a time series is a fundamental challenge in both physics and engineering. Yet, end-to-end learned arch...
arXiv:2601.22169v2 Announce Type: replace-cross Abstract: Humans are susceptible to undesirable behaviours and privacy leaks under the influence of alcohol. This paper investigates drunk language, i....
arXiv:2604.12102v3 Announce Type: replace Abstract: We describe compute-grounded reasoning (CGR), a design pattern in which code computes selected sub-problems from explicit intermediate representati...
arXiv:2605.12491v2 Announce Type: replace Abstract: Vision Transformers (ViTs) learn rich visual-semantic representations through all-to-all self-attention among patch tokens. However, this design im...
arXiv:2605.22875v2 Announce Type: replace Abstract: Long-horizon mathematical reasoning fails less often because a model cannot produce a valid next step than because an agent fails to maintain and e...
arXiv:2605.28009v2 Announce Type: replace-cross Abstract: Memory-augmented large language models extend reasoning beyond a fixed context window by maintaining long-term memory across interactions. Ho...
arXiv:2609.32353v2 Announce Type: replace Abstract: Visual token reduction is an effective way to accelerate multimodal large language models (MLLMs), but performance deteriorates rapidly under extre...
Selection-Based Structured Reasoning (SSR) is a framework that replaces free-form reasoning with the selection of pre-specified natural‑language reasoning candidates, allowing multimodal agents to choose from a set of reusable high‑level reasoning traces. By scoring these candidates in parallel using a shared context KV cache, SSR eliminates the need for an auxiliary task head and reduces inference cost. Experiments on seven multimodal search benchmarks with 2B and 4B models show that SSR maintains competitive success rates while cutting per‑turn reasoning latency by over 90% and overall inference latency by 28–54%.
arXiv:2609.40265v1 Announce Type: new Abstract: Real-world time-series applications increasingly require models that can handle time series forecasting, context-conditioned prediction, and language-b...
arXiv:2609.40245v2 Announce Type: cross Abstract: Robot navigation in dynamic, human-centered environments requires socially-compliant decisions grounded in robust scene understanding. Recent Vision-...
PickMoment is a continuous‑time model that learns to predict the interval‑mean blur over arbitrary sub‑intervals of a camera exposure, unifying single‑image deblurring, blur‑to‑video generation, and continuous‑time pick‑a‑moment recovery. It is trained with three supervisions derived from the blur integral: an empirical reconstruction loss, an additivity loss for self‑consistency, and a sharp‑frame loss at zero interval. The model achieves state‑of‑the‑art performance on GoPro and HIDE for generative deblurring, competitive results on RealBlur, and the highest per‑frame fidelity on GoPro‑7 blur‑to‑video, all in a single forward pass.
arXiv:2604.01151v3 Announce Type: replace Abstract: As LLM agents are increasingly deployed in multi-agent systems, they introduce risks of covert coordination that may evade standard forms of human...
The paper introduces Hierarchical Hash Retrieval (HHR), a coarse‑to‑fine framework designed to improve hash‑based retrieval for large language models. HHR combines Geometry‑Aware Key Routing (GKR) to redistribute feature magnitudes and prune low‑logit keys, with Learned Hash Projection (LHP) to align Hamming distance with true query‑key relevance for fine‑grained retrieval. Experiments on diverse LLMs and benchmarks show that HHR outperforms existing methods, boosting LongBench scores by 1.10 points and achieving up to 3.30× decoding speedup at 128K context length for Llama‑3.1‑8B‑Instruct.
MoLE introduces a Mixture of Latent Experts framework that encourages distinct latent tokens to extract complementary visual information, rather than sharing visual evidence. By isolating latent visual experts during evidence extraction and aggregating their outputs with dedicated summary experts, MoLE achieves higher performance on visual reasoning benchmarks compared to existing latent reasoning methods. The approach uses a two-stage training pipeline that first forces visual evidence through the latent pathway and then restores direct visual access, without requiring predefined expert roles or intermediate visual targets.