Research Feed Page 5

Semantic Overlays add an out-of-band annotation layer to input streams to help models distinguish between trusted instructions and untrusted content, significantly reducing prompt injection success rates.

This paper shows that while models can retrieve relevant financial data from long documents, they fail to act on that information unless the data is restated specifically for the decision prompt.

The paper introduces WeMM-Embedding, a series of multimodal models based on Qwen3.5 that achieve state-of-the-art performance on retrieval benchmarks and demonstrate consistent gains in production applications.

StarHarness evolves environment-specific agent harnesses to improve task success rates and lower inference costs in enterprise environments.

Researchers built a large-scale web navigation dataset by orchestrating parallel browser sandboxes to generate diverse, high-quality interaction trajectories.

StepGuard provides a 4B guard model that validates agent tool actions before execution to reduce unsafe behaviors while maintaining performance.

The paper introduces CAFE, a system that improves search agent performance and reduces hallucinations by alternatingly updating the agent and its critic through co-evolving feedback.

The OODA-Tool method improves reliability in complex multi-turn tool interactions by decoupling state management from the process of executing actions.

The paper formalizes construct validity for LLM-as-a-judge evaluation and reveals that current judges are significantly more sensitive to scope edits than strength edits.

The paper introduces cflow, a model architecture and runtime co-design that optimizes autoregressive decoding on CPUs by prioritizing memory bandwidth constraints.

The SMITH framework uses reinforcement learning to train a single language model to both synthesize reusable Python tools and apply them effectively to solve complex tasks.

Selective Regenerative Decoding optimizes LLM inference by identifying and regenerating only the degraded portions of a reasoning sequence rather than discarding the entire trajectory.

The paper investigates the efficiency and accuracy trade-offs when transitioning tasks between low-cost and high-cost language models in agentic coding workflows.

The paper introduces PeakBench to evaluate how effectively AI agents manage system resources when executing parallel tool workflows.

The paper investigates whether providing on-demand AI assistance for logic tasks reduces the user's ability to develop independent problem-solving skills.

The paper introduces K-Bench 01 to evaluate how frontier models perform on real-world scientific agent requests sampled from live user traffic.

The paper introduces IntentQA and the X-CaVIR framework to help models infer latent human intentions in video content through cognitive context reasoning.

OptiSight combines semantic object identification with geometric control to enable efficient robot navigation while minimizing reliance on high-frequency language model inference.

The researchers introduced a multimodal benchmark and an agentic framework that outperforms commercial models in complex basketball reasoning tasks.

The researchers introduce a penalty method to prevent language models from sacrificing safety protocols when they are fine-tuned for improved reasoning tasks.