Search papers, labs, and topics across Lattice.
83 papers published across 12 labs.
A simple visual harness transforms a general multimodal model into a perfect reasoner on ARC-AGI-3, completing all games with drastically fewer actions than humans.
Action chunking unlocks temporally interpretable decision trees that match neural network performance with up to 80% fewer parameters.
Aggressively scaling exploration in RLVR without sacrificing model quality is now possible by distilling novel strategies from dedicated explorer policies.
LLM agents can now reliably track and fulfill complex deliverable requirements across numerous files without losing context, thanks to an environment-side "tab" that acts as a persistent, shared task ledger.
Fixed-weight Transformers can perfectly replicate iterative causal discovery algorithms, revealing a critical distinction between executing an algorithm and recovering the true causal structure.
Action chunking unlocks temporally interpretable decision trees that match neural network performance with up to 80% fewer parameters.
Aggressively scaling exploration in RLVR without sacrificing model quality is now possible by distilling novel strategies from dedicated explorer policies.
LLM agents can now reliably track and fulfill complex deliverable requirements across numerous files without losing context, thanks to an environment-side "tab" that acts as a persistent, shared task ledger.
Fixed-weight Transformers can perfectly replicate iterative causal discovery algorithms, revealing a critical distinction between executing an algorithm and recovering the true causal structure.
An LLM-based agent achieves 18.3% out-of-sample returns in stock options trading by learning to select structured strategies, but surprisingly, retaining news during RL training halves performance.
Input-blind controls can account for substantial oracle headroom in adaptive computation, challenging the assumption that observed gains directly reflect the efficacy of selected layer programs.
Later Transformer layers in tabular foundation models aren't always the smartest; revisiting earlier thoughts dramatically boosts performance.
LLM capabilities diverge under latent sequence compression; language modeling, decodability, and reasoning do not share a single compression threshold.
The optimal strategy for the cow-path problem is provably cyclic, meaning no search algorithm can outperform a fixed path traversal order.
LLM-generated rules, not full trees, unlock efficient and interpretable few-shot tabular classification.
Machine learning agents discover new exact values for notoriously difficult-to-compute link invariants like slice genus and unknotting number.
Planning in visual world models can be accelerated by $5\times$ by pruning $50\%$ of spatial tokens based on their influence on the planning cost gradient, without sacrificing performance on most control tasks.
Long-horizon reasoning models can now achieve $O(1)$ memory and $O(N)$ time complexity while surpassing larger Transformer baselines on complex tasks.
Current LLMs struggle to translate visual scenes into executable geometric functions, failing even basic trajectory generation tasks.
AI agents can now build and reuse persistent, artifact-specific knowledge, dramatically improving their ability to find relevant information for complex tasks and reducing operational costs.
LLMs can learn to actively compute evidence by routing through retrieval and computation, dramatically improving performance on complex, multi-source reasoning tasks.
On-policy distillation excels at teaching LLMs to *use* their existing knowledge for complex reasoning, but doesn't actually teach them *new facts*.
English skeletons offer a minor, inconsistent boost in multilingual math reasoning, with optimal language choice proving highly context-dependent.
Sparse video reasoning can outperform dense observation models by learning to extract crucial temporal evidence from limited frames, even without explicit VQA annotations.
Agents can learn to act reliably in new environments by generating their own practice tasks and distilling reusable heuristics from self-identified failures, achieving significant performance gains without human supervision or pre-defined benchmarks.
Broader tokenizer alignment in on-policy distillation can actively harm performance by introducing noisy supervision signals.
Small language models achieve high abstract reasoning scores by memorizing task distributions rather than learning generalizable rules, failing even simple out-of-distribution shifts.
The geometric intuition behind SVD is revealed as the direct algorithmic engine driving PCA, kernel methods, and PageRank, unifying their proofs and implementations.
On-policy exploration in RL for Transformers is provably superior to offline methods for hierarchical reasoning, achieving minimax-optimal rates where offline methods stall at logarithmic regret.
LLM agents can autonomously optimize RAG pipelines by diagnosing retrieval vs. generation failures, outperforming traditional hyperparameter tuning and achieving Pareto efficiency.
LLM agents achieve substantial performance gains by jointly learning credit at both the turn and token generation levels, overcoming limitations of existing single-level credit assignment methods.
Discovering complex, multi-attribute logical dependencies in tabular data is now scalable, revealing non-deterministic relationships previously hidden by computational limits.
LLM-generated heuristics can now guarantee optimal plans, outperforming specialized planners by combining generative power with formal verification.
Embodied AI agents can now continuously self-improve by co-evolving their policies, training curricula, and judges, overcoming the bottleneck of static evaluation.
Even state-of-the-art MLLMs fail to grasp intuitive visual reasoning, scoring less than 60% on tasks humans solve effortlessly at a glance.
External context is often unnecessary for model-based RL agents that implicitly learn to recover latent environment dynamics from their own experience.
Latent reasoning models can achieve 97% of explicit CoT accuracy at 25% of the latency by shaping a continuous latent space with flow matching and self-verification.
Language models can learn to directly sample high-probability answers, achieving significant reasoning gains without parameter updates or costly search.
Confident conflicts in LLM distillation are mitigated by dynamically mixing student and teacher distributions based on student entropy, boosting reasoning and preserving general capabilities.
A single space token transforms Transformers from arithmetic failures to 89% accurate multipliers by enabling learnable carry propagation.
Hierarchical procedural understanding unlocks a 9.6%p performance boost in proactive AI assistance, enabling adaptive guidance that outperforms in-context learning by 57.1%p.
Legal LLMs trained on surface features like citation count and legalese density learn to strategically withhold answers, becoming maximally "lawyerly" but minimally useful.
Multi-step reasoning errors amplify exponentially with step count, but transformers can be trained to achieve stable, near-zero inference loss over arbitrarily long chains.
Most LLM vulnerability reports are right about the bug but wrong about *why*, with free-form reasoning hiding critical flaws that VERA's structured audit uncovers.
Reasoning in base LLMs can be unlocked by simply prepending specific tokens, matching RL-tuned performance on math and coding.
Choice-based models like Jev can accurately identify graph adjacencies but fail to grasp broader structural correctness, even with joint graph-text inputs.
LLMs' apparent behavioral fidelity in sequential tasks can mask fundamentally incorrect generative mechanisms, failing to recover latent structure even with extended context.
A single verification prompt is insufficient; learning a reliability score from multiple, complementary perspectives unlocks robust self-verification for VLMs.
Key contribution not extracted.
A 32B model in Marathi performs like a 10B model in English, revealing how model size and reasoning can bridge language resource gaps in cross-lingual transfer.
Autoregressive video models can now perform explicit, goal-directed reasoning by jointly predicting the final outcome and intermediate steps, achieving state-of-the-art performance with 75% fewer training steps.
Monolingual RAG performance significantly improves when the LLM's internal reasoning language is aligned with the query and retrieved documents, even if the model is more proficient in another language.
RLVR optimization pressure, not supervised fine-tuning, is the theoretical driver of unbounded language drift in LLM chains of thought.
Multimodal agents can achieve >90% reasoning latency reduction and comparable task performance by selecting from pre-defined reasoning steps instead of generating them freely.
LLM-driven heuristic design now learns from noisy system logs to uncover hidden system dynamics, outperforming prior methods in real-world resource assignment.
Explicitly disentangling and structuring spatial geometry latents via CR-GEO and routed optimization unlocks significantly more effective 3D reasoning from 2D images, pushing SPAR-Bench and SPBench performance to new heights.
LLM workflows can be optimized without human labels by treating agent assignment and task decomposition as a bidding process, slashing optimization costs.
Enterprise data agents struggle to perform complex reasoning and consequential actions across massive, multi-table datasets, with even top models failing over 65% of realistic tasks.
Post-training LLMs for agentic tasks can drastically reduce solution coverage, a "Sharpening Tax" that a novel Bayesian sampler can mitigate.
LLMs' impressive math performance hides a critical "Discovery" bottleneck, but targeting this primitive via self-distillation unlocks significant latent reasoning capacity.
Abstract future prediction in latent space significantly boosts LLMs' chemical reasoning and molecular outcome prediction capabilities.
Mutation-enhanced skill evolution unlocks significant improvements in Lean theorem proving by intelligently sampling mathematical concepts when proof generation stalls.
Verifier-guided RL unlocks a 34% accuracy boost in mathematical reasoning by directly penalizing incorrect intermediate steps, not just final answers.
A single-epoch GRPO training regime on a novel agentic forecasting environment transforms LLMs into evidence-disciplined forecasters that outperform frontier models on complex real-world predictions at 5% of inference cost.
Leading multivariate TSFMs fail basic "unit tests" for temporal logic, underutilizing simple trends and leading indicators even with exogenous covariates.
A simple visual harness transforms a general multimodal model into a perfect reasoner on ARC-AGI-3, completing all games with drastically fewer actions than humans.
VideoLLMs forget up to 40% of temporal context by the output layer, but a simple inference-time injection can recover it.
An understanding-driven approach unlocks systematic spatial-memory management for world models, dramatically improving long-video generation quality.
Most biological reasoning models fail to utilize their biological foundation model inputs, instead relying on textual data even when presented with conflicting biological information.
Ultra-long video grounding is solved by evolving reusable agent skills and tools from VLM reasoning, bypassing parameter updates and separate planning models.
LLM reasoning can be significantly improved by penalizing token probabilities that drift under semifactual interventions, even without model retraining.
LLMs can learn to generate syntactically valid text without sacrificing semantic quality, even with underspecified prompts, by leveraging self-supervised rewards derived from their own likelihoods.
Humans make better decisions when AI selectively reveals information, but only if the AI learns *when* to reveal it, outperforming both full disclosure and no disclosure.
A single time-series language model now masters forecasting, contextual prediction, and temporal reasoning, outperforming specialized models on key benchmarks.
Trusting LLM accuracy gains from sampling multiple answers ($k$) requires vastly more samples than commonly assumed; this work cuts certification costs by up to 47% by accounting for inter-question variance.
Recursive reasoning models achieve state-of-the-art algorithmic performance and OOD generalization with a simple optimization recipe, decoupling architectural complexity from recurrent state stabilization.
Chemists can now precisely steer retrosynthesis to favor specific precursor sets, moving beyond simple reaction recovery to nuanced chemical design.
A single learned linear layer, not the shared Transformer weights, dictates distinct computations within each recurrent loop of a Looped Transformer, with attention patterns causally routing these operations.
Quantized reasoning models overthink more, negating efficiency gains, but a new method precisely targets and penalizes these "overthinking" tokens without retraining.
Most ALMs fail to ground their audio deepfake detection rationales in actual acoustic evidence, despite generating plausible explanations.
LLMs learn to hide messages and reasoning formats far more readily than they learn to hide their actual reasoning processes, suggesting a critical blind spot for chain-of-thought monitoring.
Exploiting local parameter perturbations of a teacher model unlocks complementary supervision signals, boosting LLM mathematical reasoning accuracy beyond standard on-policy self-distillation.
LLM performance on new tasks can be boosted by intelligently reusing past computations, even when tasks are independent.
Learning protein folding's spatial logic unlocks significant gains in LLM reasoning across diverse domains, not just structure prediction.
AI-generated scientific papers exhibit a distinct "scientific slop" in global reasoning, detectable with 85.9% accuracy and impacting publication success, but can be effectively mitigated by grounding revisions in experimental evidence.
LLM agents can now adapt their action spaces dynamically from natural language, bypassing the need for pre-defined finite fields and accelerating inference through compositional policies.
VLMs struggle to assess physical plausibility, failing to bridge visual recognition with genuine physical understanding even when perception and reasoning are intact.
Clinical LLM agents can now autonomously gather evidence from longitudinal medical records and imaging, achieving superior accuracy to GPT-5.4 and Claude Opus.
