We have 18 of 114 papers
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
GPU kernels generated by large language model (LLM) agents can remain less efficient than expert implementations, but runtime alone does not reveal how the gap relates to design discovery and implementation. We introduce D2K-Bench, a diagnostic benchmark of 26 tasks and 85 workloads that measures how effectively agents...
Language-model agents increasingly face long-horizon tasks with evolving state, interdependent decisions, and delayed outcomes, and scale-matched training retains gains as mechanism size and horizon grow, indicating the potential for an evolving training substrate.
SWE-Universe, a scalable and efficient framework for automatically constructing real-world software engineering (SWE) verifiable environments from GitHub pull requests (PRs), utilizes a building agent powered by an efficient custom-trained model to ensure the reliable generation of high-fidelity, verifiable tasks.
This work introduces RecreationWorld, a five-platform framework built around recreation: given a running reference, an agent must discover its behavior and build a faithful implementation with no prescribed workflow, and introduces RecreationBench, comprising 250 diverse tasks across domains and platforms.
The proposed MIMIC framework fundamentally transforms algorithms into verifiable reasoning trajectories through narrative fusion, code-guided test synthesis, and dynamic code instrumentation, demonstrating that the procedural rigor of executable code can effectively unlock and enhance the generalized reasoning capabili...
Experiments demonstrate strong 3D perception and driving scene understanding while largely preserving general vision-language capability and comprehensive evaluations across open-loop, pseudo-closed-loop, and closed-loop settings further show highly competitive motion-planning performance.
Aligning over 210k synthetic preference samples, BridgeAlign enables Qwen3-8B to achieve the best average across 17 benchmarks against 11 strong baselines, leading on both human-preference and knowledge-based capabilities at once, with no trade-off between them.
Terminal-Universe is introduced, a framework which turns each trajectory into a reusable environment and explores it for synthesizing new tasks and continued interactions, and produces 37.3k task-sufficient environments.
COPD is proposed, a contrastive OPD framework that substantially reduces reasoning length without compromising model performance and consistently improves efficiency across different tasks and model scales.
It is demonstrated that the hidden states of probed answers more effectively differentiate distinct solution paths than semantic embeddings, and the perplexity of probed answers serves as a practical proxy for reasoning correctness.
FinIndices is a large-scale benchmark evaluating data-processing fidelity over uncropped financial statements (up to 32K tokens) and yields substantial zero-hint gains, validating that structured logic can be partially restored via data-centric alignment.
Quantile-Guided Density Estimation (QGDE), which approximates this distribution with multiple quantile trends and uses local density weighting to produce token-level estimates and suggests that released tokenizer vocabularies provide a useful signal for fine-grained corpus estimation beyond coarse composition inference...
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
