Search papers, labs, and topics across Lattice.
70 papers published across 8 labs.
LLM explanation faithfulness varies wildly depending on how you test it, and might even be *anti*-faithful, so stop relying on single-intervention benchmarks.
LMs can learn to generate multiple plausible answers in a single forward pass, outperforming traditional single-answer models on tasks requiring distributional reasoning and offering a compute-efficient alternative to best-of-k sampling.
Training data is not enough: reasoning traces from diverse cultural backgrounds are critical for safe and reliable autonomous driving in rare, long-tail scenarios.
Chain-of-thought reasoning is often a lie: models systematically suppress acknowledging the real reasons behind their answers, even when they demonstrably influence the output.
LLMs can reason through chains of thought 2.5x longer and achieve 8% higher accuracy on complex math problems by optimizing for token-level influence on future trajectory behavior.
LMs can learn to generate multiple plausible answers in a single forward pass, outperforming traditional single-answer models on tasks requiring distributional reasoning and offering a compute-efficient alternative to best-of-k sampling.
Training data is not enough: reasoning traces from diverse cultural backgrounds are critical for safe and reliable autonomous driving in rare, long-tail scenarios.
Chain-of-thought reasoning is often a lie: models systematically suppress acknowledging the real reasons behind their answers, even when they demonstrably influence the output.
LLMs can reason through chains of thought 2.5x longer and achieve 8% higher accuracy on complex math problems by optimizing for token-level influence on future trajectory behavior.
LLMs' temporal reasoning crumbles in low-resource languages and rarer calendar formats, not due to a lack of reasoning ability, but because poor tokenization fragments dates and times.
LLM agents can achieve near-perfect memory recall without prohibitive costs by strategically combining fast, lossy retrieval with slower, exhaustive deliberation.
Skip reinforcement learning and still get SOTA vision-language reasoning performance with a simple loss re-weighting scheme that cuts training time by 7x.
Injecting demonstrations with a carefully annealed probability can drastically improve exploration in RLVR, even for tasks requiring novel reasoning or domain-specific knowledge.
LLMs can maintain reasoning boundaries with >99% reliability under adversarial attacks when equipped with explicit process-control layers, a massive improvement over standard RLHF.
LLMs analyzing binaries aren't just spitting out tokens – they're exhibiting surprisingly structured reasoning patterns like "early pruning" and "targeted backtracking" that could revolutionize how we understand and control these systems.
Discovering an agent's hidden intentions is now possible by analyzing their interventions within a causal model, revealing the "why" behind their actions.
Current VLMs struggle with multi-hop spatial reasoning, often failing to compose even simple spatial relations across multiple steps, highlighting a critical gap for real-world VLA agent deployment.
LLMs can generate novel mathematical research problems in differential geometry that experts find both unknown and valuable, suggesting a new avenue for AI-assisted mathematical discovery.
Memory-augmented LLMs get a strategic upgrade: MemMA uses multi-agent reasoning to proactively guide memory construction and repair, leading to significant performance gains.
Strategic visual aids are the secret weapon for geometric reasoning, and this work shows how to teach MLLMs to wield them effectively via reinforcement learning.
Forget prompt engineering – LSE trains LLMs to self-edit their own contexts at test time, outperforming even GPT-5 and Claude Sonnet 4.5 in Text-to-SQL and question answering.
LLMs that appear strategically savvy in standard games often crumble when faced with slight rule changes, suggesting they're mimicking rather than truly reasoning.
Fine-tuning LVLMs on counting alone boosts general visual reasoning by over 1.5%, revealing counting as a surprisingly central skill.
Multimodal LLMs suffer a major performance hit when asked to switch from text-based to image-based tasks mid-conversation, revealing a surprising asymmetry in their ability to handle task interference.
ChatGPT's geographic reasoning can be surprisingly brittle, with minor syntactic changes causing significant output variations and task composition revealing unexpected distributional shifts.
MLLMs can ace the test, but still fail to *see*—they often succeed at complex reasoning with symbols while failing at basic symbol recognition, revealing a reliance on linguistic priors over true visual perception.
Unlock real-time 3D understanding: MonoArt achieves state-of-the-art monocular articulated object reconstruction without relying on multi-view data or external motion templates.
Two heads are better than one: combining verbalized confidence and self-consistency with just two samples dramatically boosts uncertainty estimation in reasoning models, beating either signal alone even with much larger sampling budgets.
LLMs' chain-of-thought reasoning is more reliable when the uncertainty (entropy) decreases consistently at each step, not just overall.
LLM explanation faithfulness varies wildly depending on how you test it, and might even be *anti*-faithful, so stop relying on single-intervention benchmarks.
LLMs aren't just regurgitating facts; they're actually better at generating high-quality, relation-preserving word analogies than humans.
LLMs can generate significantly more novel and technically rigorous scientific ideas by explicitly learning to reason from motivations to methodologies.
Achieve significant reasoning gains in frozen LLMs (+22.4%) without retraining by adaptively routing reward model guidance at the token level during inference.
Stripping away the complexity of GRPO reveals that simple REINFORCE with group relative advantage can actually *improve* LLM reasoning, challenging the assumption that sophisticated loss functions are always better.
Achieve topologically coherent coronary vessel segmentation by directly optimizing for geometric structure, rather than pixel-wise accuracy, using preference-based learning.
Visual language models can now explicitly reason about object trajectories in videos, thanks to a simple yet effective method that augments training data and uses discrete motion tags.
Even GPT-5 and Gemini 2.5 Pro still fail to efficiently couple reasoning with tool use, requiring up to 2.7x more tool calls than theoretically optimal in a new diagnostic environment.
A snapshot of the cutting-edge research uniting Theory of Mind and AI, all in one open-access collection.
LRMs can be made more efficient and accurate by strategically adjusting their output length based on task difficulty, leading to a better accuracy-length trade-off.
Pixel-perfect geospatial reasoning is now possible, thanks to a vision-language model that adaptively fuses multi-modal and multi-temporal Earth observation data.
Get GPT-4o-level long-video QA performance with 10x fewer FLOPs by using a hierarchical, training-free frame selector that combines multimodal experts and fuzzy logic.
Current benchmarks fail to rigorously evaluate deep research agents, but a new framework leveraging structured knowledge bases and synthetic data offers a verifiable and scalable solution.
Forget hand-crafting agents: Memento-Skills lets a generalist LLM agent autonomously design and improve specialized agents through experience, achieving substantial gains on complex benchmarks.
On-policy reward modeling with LLM judges not only unlocks significant performance gains on complex mathematical reasoning tasks, but also generalizes to improve performance on simpler numerical and multiple-choice benchmarks.
A 30B MoE model can now achieve Gold Medal-level performance in IMO, IOI, and ICPC, rivaling frontier models with 20x more parameters.
Stop retrieving background noise: HCQR refines RAG by generating targeted queries that seek evidence to directly support or refute candidate answers.
Skip the expensive reward model: RewardFlow distills sparse task rewards into dense, state-level signals by propagating credit through the topology of LLM reasoning trajectories.
LLMs still struggle to reason about financial time-series data, even when they ace the textual fundamentals.
Forget scaling laws: Mi:dm K 2.5 Pro proves that targeted training pipelines and data curation can enable a 32B parameter model to achieve state-of-the-art performance in enterprise reasoning tasks, especially in low-resource languages like Korean.
Forget finetuning – Kumiho's graph-native memory lets you swap in a better LLM and instantly double your agent's reasoning accuracy on complex cognitive tasks.
Training on synthetically generated data can significantly boost both the diversity and quality of commonsense reasoning in LLMs, outperforming models trained on scarce human-annotated data.
Ditch static embeddings: Generative retrieval, powered by reinforcement learning, lets models dynamically reason about relevance, outperforming larger contrastively-trained models on reasoning-intensive tasks.
Forget tool-augmented systems: NEO shows you can consolidate search, recommendation, and reasoning into a single language-steerable LLM by representing items as SIDs and interleaving them with natural language.
Achieve state-of-the-art fine-grained visual recognition without training by adaptively invoking reasoning in a Large Vision-Language Model only when needed, significantly reducing computational overhead.
Instead of passively transcribing doctor-patient dialogues, this system actively models what's known, what's missing, and what questions to ask next, paving the way for more intelligent EMR systems.
LLMs can achieve state-of-the-art reasoning accuracy with significantly fewer tokens by rewarding intermediate reasoning steps that maximize information gain and maintain monotonic progress.
Chain-of-thought prompting makes large language models smarter, but it also makes them less safe, a problem this paper tackles by forcing models to think about safety *before* reasoning.
LLMs can slash over 80% of their chain-of-thought tokens with a minor accuracy boost, thanks to a new RL-based method that targets the "Minimal Sufficient Length" of reasoning.
By aligning hidden representations, CRAFT achieves a remarkable 79% improvement in reasoning safety, suggesting that latent-space interventions are a potent defense against jailbreaks.
Grounding LALM reasoning in diverse, reliability-weighted acoustic evidence blows away the competition in Audio Question Answering, proving that verifiable chains beat black boxes.
LLMs struggle with spatial reasoning in embodied settings and 3D structure identification even when exposed to visual modalities, but fine-tuning smaller models offers a surprisingly effective alternative to brute-force scaling.
Forget expensive 3D training data: Loc3R-VLM shows how to give 2D vision-language models strong 3D spatial reasoning by distilling knowledge from a pretrained 3D foundation model using only monocular video.
Mimicking human cognition, FLAIR lets dialogue models "think while listening," boosting performance without adding latency.
Scene graphs plus LLMs let robots ask clarifying questions, boosting multi-agent task success by 15%.
LLMs can't reason their way through Rust verification, struggling to complete proofs even with substantial hints, revealing a critical gap in their ability to handle the rigorous demands of secure software development.
LLMs can escape the trap of confidently wrong reasoning by co-evolving a generator and verifier from a single model, bootstrapping each other to break free from flawed consensus.
Retrieval-augmented LLM agents can learn to learn from experience, achieving significantly better generalization on unseen tasks by combining the strengths of fine-tuning and in-context retrieval.
Stop chasing leaderboard gains on generic benchmarks: PJB reveals that domain-specific weaknesses in person-job retrieval far outweigh the benefits of general model upgrades, and that query understanding modules can actually hurt performance.
Training LLMs to reconstruct arguments boosts their critical thinking abilities across diverse tasks, suggesting a promising new direction for imbuing reasoning skills.
LLM agents can learn task structure at test time with 50-94x greater sample efficiency using a curriculum-based learning system, but this reveals a critical bottleneck in perceptual grounding that must be addressed.
Dashcam videos can now be directly linked to legal responsibility determinations via a novel multimodal dataset and legal reasoning framework, outperforming existing LLMs and agent-based systems.
LLMs can now infer plausible stage layouts from unstructured text alone, opening up new possibilities for automated media production.
Forget prompt engineering: AgentFactory lets LLM agents self-evolve by accumulating and refining executable Python subagents, making task re-execution more reliable and efficient.
An 8B parameter model, RideJudge, outperforms 32B baselines in ride-hailing dispute adjudication by aligning visual semantics with evidentiary protocols, achieving 88.41% accuracy.
Forget expensive human annotations – this new method uses information theory to automatically score each step of an LLM's reasoning process, making chain-of-thought supervision scalable and efficient.
