Hard Negatives Reveal What Easy Negatives Hide: Cross-Lingual Harmfulness Representations Degrade with Resource Tier Under Hard Negatives
Research
Research News
New papers, breakthroughs and academic AI research
Same Scores, Different Decisions: Evaluating JEV and Language Models for Legal Document Understanding
Research
Counterfactual Constraint-Conditioned On-Policy Distillation for Multi-Constraint Instruction Following
Research
Active Learning for Biodiversity Monitoring: From Label Efficiency to Reliable Ecological Inference
Research
Forecast Workflow Bench: Evaluating Language-Model Decisions with Budgeted Forecast Tools
Research
RAIN: Region-Aware Inversion Network for Semantic Watermark Extraction
Research
Anomaly-Free Self-Optimization via AUC Bounds
Research
Quantization-Robust Unlearning through the Lens of Retain-Forget Loss Landscapes Interaction
Research
Discrete Diffusion Models via Evolving Variational Autoregressive Networks
Research
Information Capacity of Generative Video Compression: Quantifying the Rate-Compute Exchange at Identical Quality
Research
DeltaS: Reading the Gated Linear Attention State for KV Cache Eviction in Streaming Video
Research
HYDRO: Towards Non-Reversible Face De-Identification Using a High-Fidelity Hybrid Diffusion and Target-Oriented Approach
Research