The first benchmark that combines agent-designed experiments, a deliberately flawed simulator, and calibrated interval scoring against ground truth.
What is new.
What it means for your work.
Discover recent technology and evidence. Browse older material in All notes.
Changes to start with
Newly added notes across different industries
AgentRoboCritic · Action-Level Analysis for Evaluating Robot Control Capabilities
TideCache: Transparent Cache Management and Layer-wise Streaming for Hybrid LLM Serving
Virus-like particles enable efficient gene editing in pigs and poultry models
New technology notes
Originals from the past 14 days, ordered by when added to Blackdeck.
A model-agnostic framework that provides time-resolved explanations for dynamic survival predictions with longitudinal data.
A learnable query per degree of freedom enables context-adaptive fusion of visual and EMG data.
LLMs directly program statistical computations in tabular foundation models without gradient-based optimization.
A software-hardware co-designed prefetcher that reduces HBF latency by tracking page demand with demand-paced staging
Introduces Global Regression Pretraining (GREP) to learn temporal patterns from unlabeled sensor signals.
An open-weight 27B reasoning model is trained to natively use real robots through simulation-based reinforcement learning.
Removes geometric constraints of existing template-based mapping by enabling flexible computation packing
Large language model agents automatically derive CDE schemas from sarcoma case report abstracts
A multi-objective approach explicitly enhances interactions with specific hotspots, improving binding mode diversity
The phenomenon of agents continuing execution after reaching a correct solution is quantitatively analyzed.
By leveraging confidence scores from short prefixes, computation is reallocated to boundary-adjacent questions for improved efficiency and accuracy
A new backdoor attack implants a dormant backdoor into a pretrained encoder and activates it during fine-tuning on clean downstream data.
A prediction-market rule rewards progress per submitted step, preventing splitting and merging of identities.
Introduces event-based anchors to separate activity-purpose mobility patterns into distinct semantic channels.