Wei Liu
7 papers · Latest:
AuditRepairBench: A Paired-Execution Trace Corpus for Evaluator-Channel Ranking Instability in Agent Repair
AuditRepairBench introduces a corpus and screening method to identify and mitigate evaluator-channel ranking instability in agent-repair leaderboards.
LLM-Augmented Semantic Steering of Text Embedding Projection Spaces
This paper introduces LLM-augmented semantic steering to reorganize text embedding projections based on user intent, improving alignment without retraining.
Time-Interval-Aware Disentangled Expert Modeling for Next-Basket Recommendation
TIDE is a novel next-basket recommendation model that disentangles user habits from exploration and incorporates time-interval awareness for improved predictions.
AROMA: Augmented Reasoning Over a Multimodal Architecture for Virtual Cell Genetic Perturbation Modeling
AROMA is a multimodal AI that accurately and interpretably models genetic perturbations in virtual cells, outperforming prior methods.
Photometry and physical characterization of near-Earth asteroid 2025 FA$_{22}$ from one apparition
Researchers characterized Potentially Hazardous Asteroid 2025 FA$_{22}$ during a close approach, determining its spin, shape, and contact-binary nature.
Seedance 2.0: Advancing Video Generation for World Complexity
Seedance 2.0 is a new multi-modal audio-video generation model with a unified architecture, offering advanced capabilities and improved performance.
SSD: Single Shot MultiBox Detector
SSD is a fast and accurate single-shot object detection method that eliminates the need for proposal generation by predicting bounding boxes and class scores directly from multiple feature maps.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.