AI.info
Latest Research — Page 52
Browse Latest Research on AI.info.
- Exploring the Effects of Alignment on Numerical Bias in Large Language Models
- Finding the Sweet Spot: Trading Quality, Cost, and Speed During Inference-Time LLM Reflection
- Pursuing Minimal Sufficiency in Spatial Reasoning
- Code as Worlds: Agentic Discovery of Executable World Representations for Physical Reasoning
- Learning Self-Correction in Vision-Language Models via Rollout Augmentation
- ANTE: Anticipate, Then Execute -- Learning Internal World Models for Coding Agents from Verified Beliefs
- CAPRMIL: Context-Aware Patch Representations for Multiple Instance Learning
- SolveEdit: Benchmarking Visual Problem Solving in Generative Models
- MARS: Multi-Specialist LLM Relay System for Competitive Programming
- Ambient Dataloops: Generative Models for Dataset Refinement
- AraLingBench A Human-Annotated Benchmark for Evaluating Arabic Linguistic Capabilities of Large Language Models
- AgentSWE: Can Coding Agents Build the Agent You Actually Want?
- Learning Ordinal Probabilistic Reward from Preferences
- Stop Training Against Soft Users: Mentalized Adversarial Self-Play for Proactive Dialogue
- Synthesis of Ovarian Cancer Multi-Omics Datasets: A Biology-Informed Evaluation Framework
- Reasoning Model Is Superior LLM-Judge, Yet Suffers from Biases
- Monitor-Generate-Verify (MGV): Formalising Metacognitive Theory for Language Model Reasoning
- PromptRL: Prompt Matters in RL for Flow-Based Image Generation
- Perceptive Humanoid Parkour: Chaining Dynamic Human Skills via Motion Matching
- Travel Time Prediction from Sparse Open Data
- Multimodal Classification via Total Correlation Maximization
- RV-HATE: Reinforced Multi-Module Voting for Implicit Hate Speech Detection
- FlowCast: Trajectory Forecasting for Scalable Zero-Cost Speculative Flow Matching
- Retrieval Heads are Dynamic
- KAPSO: A Knowledge-grounded framework for Autonomous Program Synthesis and Optimization
- How Far are Modern Trackers from UAV-Anti-UAV? A Million-Scale Benchmark and New Baseline
- IdeaLens: Detecting AI Ideas in Long-form Writing
- Mitigating the Safety Alignment Tax with Null-Space Constrained Policy Optimization
- Towards Physically Grounded Articulated 3D Assets: A Multi-Agent Pipeline and Benchmark
- GUDA: Counterfactual Group-wise Training Data Attribution for Diffusion Models via Unlearning
- Regional Attention-Enhanced Swin Transformer for Clinically Relevant Medical Image Captioning
- COLD-Steer: Steering Large Language Models via In-Context One-step Learning Dynamics
- LongWeave: A Long-Form Generation Benchmark Bridging Real-World Relevance and Verifiability
- NavQ: Learning a Q-Model for Foresighted Vision-and-Language Navigation
- Tagging-Augmented Generation: Assisting Language Models in Finding Intricate Knowledge In Long Contexts
- ENACT: Evaluating Embodied Cognition with World Modeling of Egocentric Interaction
- TAO: Tool-Aware Joint Scheduling for LLM-Based Agentic Serving
- Reinforcement-Learned Unequal Error Protection for Quantized Semantic Embeddings
- Omega-S: A Functional Resilience Index for LLM Fine-Tuning
- Video-KTR: Reinforcing Video Reasoning via Key Token Attribution