Recent Research Signals

Found 117 results

Filtering papers...

Paper	Possible Impact	Features	Possible Business Area
Distilling the Essence: Efficient Reasoning Distillation via Sequence Truncation Computation and Language	Makes AI learn faster by focusing on thinking steps.	LinkedIn Code Data 2 Countries	Natural Language Processing (79%)	Score: 4 6 days ago
PROFASR-BENCH: A Benchmark for Context-Conditioned ASR in High-Stakes Professional Speech Computation and Language	Helps computers understand important words better.	Code Data	Speech Recognition (75%)	Score: 3 1 day ago
SWE-RM: Execution-free Feedback For Software Engineering Agents Computation and Language	Helps computers write better code by learning from mistakes.	Alibaba 2 Countries	Simulation (78%)	Score: 3 4 days ago
AI Meets Brain: Memory Systems from Cognitive Neuroscience to Autonomous Agents Computation and Language	Helps AI remember like humans do.	Code 2 Countries	Artificial Intelligence (81%)	Score: 2 1 day ago
CricBench: A Multilingual Benchmark for Evaluating LLMs in Cricket Analytics Computation and Language	Helps computers answer tough cricket questions.	Code	Cricket (81%)	Score: 2 4 days ago
The Effect of Gender Diversity on Scientific Team Impact: A Team Roles Perspective Computation and Language	Mixed-gender teams make science more successful.	Code 2 Countries	Social Impact (76%)	Score: 2 1 day ago
Gamayun's Path to Multilingual Mastery: Cost-Efficient Training of a 1.5B-Parameter LLM Computation and Language	Helps computers understand many languages, especially Russian.	Code	Language Learning (77%)	Score: 2 5 days ago
Heaven-Sent or Hell-Bent? Benchmarking the Intelligence and Defectiveness of LLM Hallucinations Computation and Language	Helps AI be creative and correct at same time.	Code 2 Countries	Human Computer Interaction (76%)	Score: 2 5 days ago
Lie to Me: Knowledge Graphs for Robust Hallucination Self-Detection in LLMs Computation and Language	Makes AI tell the truth, not make things up.	Code	Visual Search (76%)	Score: 2 1 day ago
AI4Reading: Chinese Audiobook Interpretation System Based on Multi-Agent Collaboration Computation and Language	AI makes book summaries that sound like podcasts.	Code 2 Countries	Natural Language Processing (82%)	Score: 2 1 day ago
M2G-Eval: Enhancing and Evaluating Multi-granularity Multilingual Code Generation Computation and Language	Tests how well computers write code in many ways.	Code	Natural Language Processing (76%)	Score: 2 3 days ago
Bridging the Copyright Gap: Do Large Vision-Language Models Recognize and Respect Copyrighted Content? Computation and Language	Helps AI avoid stealing copyrighted work.	Code	Image Recognition (78%)	Score: 2 4 days ago
How important is Recall for Measuring Retrieval Quality? Computation and Language	Finds best answers even when you don't know all the facts.	Code Data	Semantic Search (78%)	Score: 2 6 days ago
Chain-of-thought Reviewing and Correction for Time Series Question Answering Computation and Language	Fixes computer answers about numbers.	Code	Natural Language Processing (80%)	Score: 2 3 days ago
WeDLM: Reconciling Diffusion Language Models with Standard Causal Attention for Fast Inference Computation and Language	Makes AI write much faster by changing how it thinks.	Tencent Code	Natural Language Processing (75%)	Score: 2 2 days ago
SmartSnap: Proactive Evidence Seeking for Self-Verifying Agents Computation and Language	Agent proves it finished tasks with snapshots.	Tencent Code	Natural Language Processing (77%)	Score: 2 4 days ago
Oogiri-Master: Benchmarking Humor Understanding via Oogiri Computation and Language	Teaches computers to tell funnier jokes.	Code	Natural Language Processing (76%)	Score: 2 5 days ago
Gamayun's Path to Multilingual Mastery: Cost-Efficient Training of a 1.5B-Parameter LLM Computation and Language	Helps computers understand many languages, especially Russian.	Code	Language Learning (77%)	Score: 2 5 days ago
Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning Computation and Language	Makes AI smarter and faster, understanding more.	Code Data	Natural Language Processing (77%)	Score: 2 6 days ago
Nested Browser-Use Learning for Agentic Information Seeking Computation and Language	Lets computers explore websites like people do.	Alibaba Code	Semantic Search (81%)	Score: 2 1 day ago
MoRAgent: Parameter Efficient Agent Tuning with Mixture-of-Roles Computation and Language	Teaches AI to do jobs by splitting tasks.	Huawei Data	Natural Language Processing (75%)	Score: 2 5 days ago
Introducing TrGLUE and SentiTurca: A Comprehensive Benchmark for Turkish General Language Understanding and Sentiment Analysis Computation and Language	Tests computer understanding of Turkish words.	Code Data	Natural Language Processing (77%)	Score: 2 4 days ago
Accelerating Language Model Workflows with Prompt Choreography Computation and Language	Makes AI talk faster and smarter together.	Johns Hopkins University Code	Natural Language Processing (77%)	Score: 2 2 days ago
Fine-Tuning LLMs with Fine-Grained Human Feedback on Text Spans Computation and Language	Teaches computers to write better with helpful edits.	Johns Hopkins University Microsoft	Natural Language Processing (78%)	Score: 2 1 day ago
Automatic Replication of LLM Mistakes in Medical Conversations Computation and Language	Finds AI mistakes in doctor-patient talks.	Data	Natural Language Processing (77%)	Score: 1 6 days ago
Accelerate Speculative Decoding with Sparse Computation in Verification Computation and Language	Makes AI write faster without losing quality.	Meituan	Semantic Search (75%)	Score: 1 4 days ago
Rethinking Supervised Fine-Tuning: Emphasizing Key Answer Tokens for Improved LLM Accuracy Computation and Language	Improves AI answers by focusing on the final solution.	Data	Natural Language Processing (77%)	Score: 1 6 days ago
Semi-Supervised Learning for Large Language Models Safety and Content Moderation Computation and Language	Teaches AI to be safer with less data.		Natural Language Processing (78%)	Score: 1 6 days ago
Explainable Statute Prediction via Attention-based Model and LLM Prompting Computation and Language	Helps lawyers find the right laws for cases.	Code	Natural Language Processing (78%)	Score: 1 4 days ago
ClarifyMT-Bench: Benchmarking and Improving Multi-Turn Clarification for Conversational Large Language Models Computation and Language	Helps computers ask questions before answering.	Code	Natural Language Processing (78%)	Score: 1 6 days ago
Beyond Heuristics: A Decision-Theoretic Framework for Agent Memory Management Computation and Language	Helps AI remember things better for longer.	Code	Artificial Intelligence (77%)	Score: 1 5 days ago
A Unified Definition of Hallucination, Or: It's the World Model, Stupid Computation and Language	Makes AI tell the truth, not make things up.	Stanford University	Virtual Reality (77%)	Score: 1 5 days ago
Optimizing Decoding Paths in Masked Diffusion Models by Quantifying Uncertainty Computation and Language	Finds better answers by managing guessing.	Code	Darknet (74%)	Score: 1 6 days ago
Chinese Morph Resolution in E-commerce Live Streaming Scenarios Computation and Language	Cleans up online shopping streams from fake ads.	Code	Natural Language Processing (77%)	Score: 1 1 day ago
SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation Computation and Language	Teaches computers new languages with very little talking.	Meta	Semantic Web (76%)	Score: 1 6 days ago
Automatic Detection of Complex Quotation Patterns in Aggadic Literature Computation and Language	Finds old Bible quotes in other old writings.		Text Analytics (79%)	Score: 1 1 day ago
A Dataset and Benchmark for Consumer Healthcare Question Summarization Computation and Language	Helps computers understand health questions better.		Text Analytics (78%)	Score: 1 1 day ago
CATCH: A Controllable Theme Detection Framework with Contextualized Clustering and Hierarchical Generation Computation and Language	Helps chatbots understand what you're talking about.	Code	Semantic Search (78%)	Score: 1 5 days ago
NVIDIA Nemotron 3: Efficient and Open Intelligence Computation and Language	Makes computers smarter and faster for many tasks.		Natural Language Processing (79%)	Score: 1 6 days ago
Parallel Token Prediction for Language Models Computation and Language	Makes computers write sentences much faster.		Natural Language Processing (76%)	Score: 1 6 days ago
Coupling Experts and Routers in Mixture-of-Experts via an Auxiliary Loss Computation and Language	Teaches AI models to use their parts better.	ByteDance	RFID (74%)	Score: 1 1 day ago
LLM-Guided Exemplar Selection for Few-Shot Wearable-Sensor Human Activity Recognition Computation and Language	Helps computers learn from fewer examples.		Image Recognition (77%)	Score: 1 4 days ago
On The Conceptualization and Societal Impact of Cross-Cultural Bias Computation and Language	Helps AI understand different cultures better.	University of Washington	Natural Language Processing (82%)	Score: 1 4 days ago
Enabling Conversational Behavior Reasoning Capabilities in Full-Duplex Speech Computation and Language	Helps computers understand conversations like people.	University of California, Berkeley	Natural Language Processing (79%)	Score: 1 5 days ago
Not too long do read: Evaluating LLM-generated extreme scientific summaries Computation and Language	Helps computers write short summaries of science papers.	Code	Text Analytics (78%)	Score: 1 1 day ago
Semantic Tree Inference on Text Corpa using a Nested Density Approach together with Large Language Model Embeddings Computation and Language	Organizes texts into a family tree of ideas.	2 Countries	Semantic Search (80%)	Score: 1 1 day ago
Reflection Pretraining Enables Token-Level Self-Correction in Biological Sequence Models Computation and Language	Helps computers "think" to solve biology problems.	2 Countries	Natural Language Processing (79%)	Score: 1 6 days ago
TabiBERT: A Large-Scale ModernBERT Foundation Model and Unified Benchmarking Framework for Turkish Computation and Language	Helps computers understand Turkish text much faster.		Text Analytics (76%)	Score: 1 2 days ago
AlignAR: Generative Sentence Alignment for Arabic-English Parallel Corpora of Legal and Literary Texts Computation and Language	Helps computers translate hard Arabic and English texts.	Code	Text Analytics (79%)	Score: 1 4 days ago
HeartBench: Probing Core Dimensions of Anthropomorphic Intelligence in LLMs Computation and Language	Tests AI's feelings and ethics in Chinese.	Code	Natural Language Processing (78%)	Score: 1 4 days ago