A framework for optimizing text components of language-model systems using natural-language feedback as gradients.

A generative framework for efficient, one-step real-world video deblurring.

A molecular representation learning framework using SENet-calibration and Law of Cosines constraints for property prediction.

Shadcn-style UI components for Go

A benchmark for measuring Coerced Fabrication Rate and Escape Utilization Rate in LLM form-filling.

A Theta-Gamma hierarchical oscillatory reasoning framework for multi-hop QA.

A multi-tenant service implementing the five primitives of Agentic Context Management.

RL algorithm using Q-CVAE to generate expert policy priors from online replay buffers.

Foundation model for EEG representation learning using Mamba and ViT encoders.

Multi-scale self-supervised foundation model for hierarchical EEG dynamics learning.

A framework for Inductive Logic Programming capable of learning complex, non-monotonic hypotheses.

A high-throughput LLM annotation agent using in-context batching for feature extraction.

A program caching framework using small models to reduce inference latency and cost.

A reliability-guided preference optimization framework for LLM alignment.

A benchmark for evaluating information seeking on academic knowledge graphs using LLM agents.

A non-invasive architectural intervention to achieve spectral balance in LLMs and prevent representation collapse.

Protocol for tracing the shared subspace of LLM decode-time decisions.

A cross-platform development library designed to provide low-level access to hardware.

An AI-driven platform for developers to optimize and deploy AI models on AMD hardware.

Efficient one-step speech enhancement integrating Schrödinger Bridge training and Mamba architecture.