July 2025 — page 2 of 2
- Study finds Flash Attention and specialized variants outperform standard models on energy efficiency· AI
- Researchers tackle pun translation using AI agents and phonetic embeddings· Computational Linguistics
- New Framework Shrinks Neural Network Models for RF Power Amplifiers to Just 450 Parameters· AI
- Can Model Internals Forecast Behavior on New Data? New Interpretability Approach Shows Promise· AI
- Hidden Commands in Academic Papers Manipulate AI-Powered Peer Review Systems· AI
- AI-Powered Interviews Boost Hiring Accuracy, Especially for Entry-Level Roles· AI
- GPTKB v1.5 Builds 100-Million-Triple Knowledge Base from GPT-4 to Map LLM Factual Knowledge· AI
- New Diffusion Model Recovers Missing Sensor Data in Emotion and Intent Recognition· AI
- New AI Jailbreak Exploits Dialogue History by Impersonating the Model's Own Responses· AI Security
- Researchers Create Synthetic Datasets to Build AI Systems That Answer Questions in Conversation While Citing Sources· AI
- Researchers propose benchmark to test memory capabilities in AI agents across multi-turn conversations· AI
- VOTE: Ensemble Voting Technique Cuts Robot Vision-Language Model Inference Time by 39x· Agentic AI
- New AI method tackles gigapixel medical images by focusing on what matters· AI
- New benchmark reveals gaps in AI vision-language models for assembly task understanding· Artificial Intelligence
- New technique boosts language model performance on few-shot tasks without retraining weights· AI
- Research framework tackles validity crisis in AI psychology studies· AI
- New Watermark-Free Method Traces AI-Generated Speech Without Sacrificing Audio Quality· AI
- UI Tweaks That Push Longer Prompts Strengthen Users' Sense of Ownership in AI Writing· Human-Computer Interaction
- New Dataset and Model Improve How AI Tackles Visual Math Problems with Multiple Solution Paths· AI
- New RL-Based Memory Agent Pushes LLMs to Handle 3.5M Tokens With Minimal Performance Drop· Agentic AI
- New Defense Method Shields AI Agents from Prompt Injection Without Sacrificing Performance· AI Security
- New Dataset Tests How Well AI Handles Real Computer Science Competitions· AI
- LLMs fail to spot their own mistakes even when they can spot identical errors from users· AI
- Study reveals mixed picture on AI coding assistants' impact on developer productivity· AI
- Researchers Boost LLM Speed by Pruning Upper-Layer Attention Heads with Adaptive Rescaling· AI
- Researchers enhance Python test generation by inferring types during code execution· Software Engineering
- Search algorithm boosts LLM patch generation, cuts repair costs in half· AI
- New Framework Weaves Emotion Into Task-Oriented AI Dialogue Systems· AI
- Reverse Language Models Learn to Reason Backwards, Unlocking New AI Capabilities· AI
- Faster AI Reasoning: New Framework Eliminates Extra Reward Models Without Losing Performance· Agentic AI
- New method enables fine-grained pronunciation editing in Finnish speech synthesis· AI
- New Framework Cuts LLM Training Time by 59% Through Asynchronous Processing· AI
- New method improves robot learning by combining prior knowledge with human feedback· AI
- WebArxiv Benchmark Tests AI Agents on Scholarly Research Tasks· AI