Represents agent tool intent as executable tool programs for flexible web services

A benchmark to measure the stamina of coding agents over 100+ interaction turns.

Liquid Network architecture for efficient and robust audio deepfake detection.

Secondary-structure-aware graph neural network for protein representation learning.

Source code and videos for human-like autonomy training via self-play and limited human data.

Counterfactual reasoning for neural probabilistic logic programs via quotient-WMC.

Framework for modular-sparsity synchronization in conflict-averse training for generalized PINNs.

A harness framework for coding agents to automate real-world robotic policy self-improvement.

An agentic system for DeFi risk supervision using forecast-grounded LLM decisions.

A machine unlearning method for LLMs that removes private information without needing unlearning targets.

Code for evaluating token-optimized formats (TOON, TRON) in agentic AI

Information-Driven Policy Optimization for User-Centric LLM Agents.

A pipeline and benchmark for capturing ripple effects of targeted interventions on language models.

An architecture for accountable behavior in LLM agents through modular separation of cognition and control.

Benchmark for Task-completion and Resistance to Active Privacy-extraction in AI agents.

Pipeline for synthesizing simulation-ready URDFs from RGB-D sequences.

A tool for managing dotfiles across multiple machines securely and efficiently.

A compact EEG classification architecture for low-power BCI deployment.

A manually annotated benchmark for evaluating PII redaction in LLMs across 11 domains.

Developer-first, multi-model AI creative platform