Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
arXiv:2609.37786v1 Announce Type: new Abstract: Concept Bottleneck Models (CBMs) built on vision-language models such as CLIP represent a latent space as human-understandable concepts. These represen...
arXiv:2609.37605v1 Announce Type: cross Abstract: Supervised deep learning has advanced sparse-view tomographic reconstruction. However, conventional models, which typically map filtered back-project...
arXiv:2609.36956v1 Announce Type: cross Abstract: Manipulating next-token probabilities during generation can bypass the safety alignment of large language models. Existing approaches, however, rely...
arXiv:2609.37647v1 Announce Type: cross Abstract: Jev is a commercial System One model from TypeSafe AI that does not generate text: given a state and typed questions, it returns a choice from fixed...
arXiv:2605.06223v4 Announce Type: replace Abstract: Natural-language instance navigation becomes challenging when the initial user request does not uniquely specify the target instance. A practical a...
The paper introduces a decision‑theoretic framework that splits a large language model’s decision loss into belief formation and action selection components. Using a synthetic benchmark, it evaluates how reinforcement‑learning interventions on beliefs, decisions, or both affect these components across three domains. The study finds that targeting a single component improves that part but may not transfer to others, while jointly targeting both improves both only when training and evaluation formats match.
arXiv:2609.35694v2 Announce Type: replace Abstract: Continuous diffusion generates complete reasoning solutions through iterative refinement in latent space. We introduce the Continuous Embedding Dif...
arXiv:2609.36572v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has been extended to Large Vision-Language Models (LVLMs), and perception-aware methods further e...
arXiv:2609.35942v1 Announce Type: new Abstract: Recent work in visual question answering has shown that vision-language models can exhibit strong reasoning capabilities by translating visual inputs i...
MeshSplatBench is the first benchmark designed to evaluate triangle- and mesh-based neural rendering methods from native rendering to deployment in graphics engines such as Unity and Blender. It introduces a hierarchical deployment protocol with standard and dedicated options, and a structural audit for mesh splatting to assess topological and geometric integrity. The benchmark’s evaluations show that graphics engine deployment degrades quality, dedicated deployment preserves fidelity at a significant speed cost, and current mesh splatting methods lack sufficient connectivity for manifoldness.
arXiv:2609.36755v1 Announce Type: new Abstract: Modern image editors excel at semantic manipulation and visual synthesis, yet remain limited in precise spatial control, motivating the development of...
arXiv:2609.32352v2 Announce Type: replace-cross Abstract: Vision-language models (VLMs) have shown increasing potential for medical image understanding, yet their capabilities in ophthalmic imaging r...
arXiv:2609.35954v1 Announce Type: cross Abstract: Large language model post-training generates self-generated rollouts through reinforcement learning and on-policy distillation, yet this experience i...
The paper introduces FastRL, a reinforcement learning framework designed to enhance the efficiency of Group Relative Policy Optimization (GRPO) and its variants. FastRL employs an advantage-aware pruning strategy that retains high-advantage trajectories while preserving gradient diversity, and an adaptive rollout sampling mechanism that adjusts sampling scale during training based on historical pruning data. Experiments show that FastRL can be integrated into GRPO, DAPO, and GSPO, yielding a 2.07× speedup on Geometry3K and GeoQA8K-R1V and a 1.64% accuracy improvement on visual reasoning benchmarks.
arXiv:2609.37029v1 Announce Type: cross Abstract: Speculative decoding accelerates autoregressive inference by verifying multiple draft tokens in a single target forward pass. However, as the context...
arXiv:2605.16981v3 Announce Type: replace Abstract: Streaming 3D reconstruction under a strict constant-memory budget hinges on how the recurrent state is updated as the stream evolves. We profile TT...
As preferences, goals, and facts change, LLM agents must use the current state while earlier versions remain in context. Yet they can answer with an old value of the same variable, a failure that we c...
Embodied tasks demand accurate, flexible, and semantically rich 3D scene representations. 3D semantic occupancy is well suited to this requirement, as it can model holistic 3D spaces by encoding geome...
As vision language models are increasingly deployed in clinical diagnosis, understanding how they internally resolve competing visual and textual signals becomes a safety imperative. Existing mechanis...