Tagged articles

EMNLP 2026

17 articles · Page 1 of 1

Oct 18, 2026 · Artificial Intelligence

SPADER: Step-wise GRPO with Diversity-Aware Rewards Boosts Long-Tail Answer Discovery in Multi-Answer QA

SPADER introduces a step-wise GRPO framework with diversity-aware exploration rewards to improve long-tail answer coverage in multi-answer question answering, achieving state-of-the-art results across four benchmarks and two base models by refining credit assignment and incentivizing rare entity discovery.

Diversity-Aware RewardsEMNLP 2026GRPO

0 likes · 16 min read

SPADER: Step-wise GRPO with Diversity-Aware Rewards Boosts Long-Tail Answer Discovery in Multi-Answer QA

Machine Learning Algorithms & Natural Language Processing

Oct 13, 2026 · Artificial Intelligence

LightRSI TokenPilot: Slash LLM Agent Context Costs Up to 87% with Smart Cleanup

Zhejiang University's LightRSI project introduces TokenPilot, an EMNLP 2026 accepted system for cache-efficient context management in LLM agents, combining prefix cache stabilization, tool output summarization, and task-level context cleaning to reduce inference costs by up to 87% on Claw-Eval while integrating with six coding agents and providing a visualization dashboard.

EMNLP 2026LLM agentsLightRSI

0 likes · 13 min read

LightRSI TokenPilot: Slash LLM Agent Context Costs Up to 87% with Smart Cleanup

Bilibili Tech

Oct 10, 2026 · Artificial Intelligence

HOMURA: Reinforcement Learning for Time-Constrained LLM Translation (EMNLP 2026 Oral)

HOMURA introduces a reinforcement learning framework with KL regularization and dynamic syllable-ratio rewards to achieve precise length control in LLM translation for subtitles and dubbing, outperforming baselines on the Sand-Glass benchmark and deployed at scale on Bilibili.

EMNLP 2026GRPOHOMURA

0 likes · 21 min read

HOMURA: Reinforcement Learning for Time-Constrained LLM Translation (EMNLP 2026 Oral)

Machine Learning Algorithms & Natural Language Processing

Sep 30, 2026 · Artificial Intelligence

How AI Finds Evidence in Hundred-Page Documents: Multimodal Retrieval Heads in Long-Context VLMs

Researchers identify multimodal retrieval heads in long-context vision-language models that locate relevant evidence across text and images, showing causal impact on QA performance and enabling training-free document retrieval with state-of-the-art results on MMDocIR.

EMNLP 2026Gemma3MMDocIR

0 likes · 7 min read

How AI Finds Evidence in Hundred-Page Documents: Multimodal Retrieval Heads in Long-Context VLMs

Machine Heart

Sep 30, 2026 · Artificial Intelligence

HaluMem: Evaluating Hallucinations in AI Agent Memory Systems

HaluMem introduces an operation-level benchmark for AI agent memory systems, evaluating extraction, update, and QA stages to uncover hallucinations; experiments on six systems show high recall often accompanies false memories, while low update hallucination rates mask massive omissions, and upstream memory errors propagate to final answers.

AI agent memoryEMNLP 2026HaluMem

0 likes · 15 min read

HaluMem: Evaluating Hallucinations in AI Agent Memory Systems

Machine Heart

Sep 30, 2026 · Artificial Intelligence

ME-Decoding: Reframing LLM Decoding as Ensemble Pruning with Semantic Redundancy Control

Renmin University researchers propose ME-Decoding, a method that selects candidate tokens by jointly optimizing probability confidence and semantic diversity via Mahalanobis distance, outperforming probability-only truncation on reasoning and open-ended generation across multiple models and temperatures with only 3% latency overhead.

EMNLP 2026LLM decodingMahalanobis distance

0 likes · 11 min read

ME-Decoding: Reframing LLM Decoding as Ensemble Pruning with Semantic Redundancy Control

Machine Heart

Sep 25, 2026 · Artificial Intelligence

V-Rubrics: 350K Fine-Grained Rubrics Fix Credit Assignment in Multimodal RL

V-Rubrics decomposes 50,248 visual samples into 352,938 atomic rubrics across visual faithfulness, reasoning consistency, and instruction following, enabling fine-grained credit assignment in GRPO training that lifts Overall Avg. to 68.04 on general benchmarks and 62.45 on visual math and logic tasks.

Credit AssignmentEMNLP 2026GRPO

0 likes · 13 min read

V-Rubrics: 350K Fine-Grained Rubrics Fix Credit Assignment in Multimodal RL

Amap Tech

Sep 22, 2026 · Artificial Intelligence

DIANOIA: Multi-Agent Diagnosis & Repair for Navigation Tool Trajectories

Amap introduces DIANOIA, a multi-agent system that diagnoses and repairs low-confidence tool-call trajectories for navigation Live mode by generating diverse candidates, executing them in real tool environments, cross-reviewing failures, and synthesizing corrected sequences, boosting data quality for LLM training while reducing compute costs.

DIANOIAEMNLP 2026LLM agents

0 likes · 28 min read

DIANOIA: Multi-Agent Diagnosis & Repair for Navigation Tool Trajectories

Machine Learning Algorithms & Natural Language Processing

Sep 21, 2026 · Artificial Intelligence

LongDS v1.1 Benchmark Released: GPT-6 Astra Tops Long-Horizon Data Analysis Lite Leaderboard

Zhejiang University and Ant Group release LongDS v1.1, a benchmark for long-horizon multi-turn data analysis agents built from real Kaggle workflows; the Lite subset of 24 tasks and 777 turns shows GPT-6 Astra leading at 78.17, with detailed error analysis revealing cascading state errors as the primary failure mode.

BenchmarkEMNLP 2026GPT-6 Astra

0 likes · 24 min read

LongDS v1.1 Benchmark Released: GPT-6 Astra Tops Long-Horizon Data Analysis Lite Leaderboard

Machine Heart

Sep 18, 2026 · Artificial Intelligence

Readout Bottleneck: LLMs Know Answers But Fail to Output Them

A Peking University and Yixin AI Lab paper accepted at EMNLP 2026 reveals that large language models often encode correct reasoning in hidden states but suffer a 'readout bottleneck' where output-layer biases collapse final predictions; a two-parameter calibration recovers up to 30+ accuracy points without retraining.

EMNLP 2026LLM evaluationLinear Probe

0 likes · 14 min read

Readout Bottleneck: LLMs Know Answers But Fail to Output Them

Machine Learning Algorithms & Natural Language Processing

Sep 15, 2026 · Artificial Intelligence

DynHD: Detecting Hallucinations in Diffusion LLMs by Tracking Denoising Dynamics

This paper introduces DynHD, a hallucination detection method for diffusion large language models that analyzes denoising dynamics across token and temporal dimensions, achieving 84% AUROC on LLaDA-8B and Dream-7B without repeated sampling.

Denoising DynamicsDreamDynHD

0 likes · 8 min read

DynHD: Detecting Hallucinations in Diffusion LLMs by Tracking Denoising Dynamics

Machine Heart

Sep 14, 2026 · Artificial Intelligence

T-Mem: Teaching AI Associative Recall Beyond Similarity Search

Tencent's T-Mem introduces associative recall for LLM agents by pre-storing trigger cues during memory writing, enabling retrieval via semantic associations rather than surface similarity, achieving SOTA on LoCoMo and LoCoMo-Plus benchmarks with minimal performance drop on associative tasks.

EMNLP 2026LLM agentsLOCOMO benchmark

0 likes · 13 min read

T-Mem: Teaching AI Associative Recall Beyond Similarity Search

Machine Heart

Sep 10, 2026 · Artificial Intelligence

DiffuTester: Accelerating Diffusion LLM Unit Test Generation via Structural Pattern Mining

DiffuTester is a training-free framework that accelerates diffusion language model unit test generation by mining shared structural patterns across test cases via AST, enabling more tokens per denoising step and achieving 2–3× speedup while maintaining coverage across Python, C++, and Java on DiffuCoder and Dream models.

ASTDiffusion LLMEMNLP 2026

0 likes · 9 min read

DiffuTester: Accelerating Diffusion LLM Unit Test Generation via Structural Pattern Mining

DataFunTalk

Sep 10, 2026 · Artificial Intelligence

CAPO-SLT: Confidence-Aware RL Fixes Fluent-but-Wrong Sign Language Translation

vivo AI Lab introduces CAPO-SLT, a confidence-aware policy optimization method that stabilizes sign language translation by giving each token a dynamic clipping bound based on old-policy confidence, achieving state-of-the-art results on Chinese and American sign language benchmarks using only pose input.

CSL-DailyEMNLP 2026How2Sign

0 likes · 14 min read

CAPO-SLT: Confidence-Aware RL Fixes Fluent-but-Wrong Sign Language Translation

Machine Heart

Sep 8, 2026 · Artificial Intelligence

Behavior Consistency Beats State Consistency in Text World Models for Agents

The paper introduces BehR, a behavior consistency reward for training text-based world models, showing that optimizing for agent decision alignment rather than text fidelity improves trajectory-level consistency across 16 configurations, reduces false positives in offline evaluation from 42.5% to 9.5%, and enhances lookahead planning for weaker agents.

Behavior ConsistencyEMNLP 2026GRPO

0 likes · 11 min read

Behavior Consistency Beats State Consistency in Text World Models for Agents

Machine Heart

Sep 3, 2026 · Artificial Intelligence

EASE: Teaching Multimodal RL Where to Look, Not Just What to Answer

EASE introduces evidence-anchored spatial attention supervision to multimodal reinforcement learning with verifiable rewards, using annotated evidence bounding boxes to guide model attention toward relevant image regions during training, improving accuracy on visual reasoning benchmarks without inference overhead.

Attention MechanismEMNLP 2026Evidence Grounding

0 likes · 13 min read

EASE: Teaching Multimodal RL Where to Look, Not Just What to Answer

Xiaomi Tech

Aug 28, 2026 · Artificial Intelligence

Xiaomi's 8 EMNLP 2026 Papers: Mobile Agents, Efficient LLMs, Proactive AI

Xiaomi had eight papers accepted at EMNLP 2026, covering mobile GUI agents with self-evolving rewards and process reward agents, diffusion LLM training with informed masking, speculative decoding via early-layer collaboration, token-budgeted reasoning, neural architecture search with code semantics, and proactive mobile agents with pre-reasoning perception.

EMNLP 2026LLM efficiencyXiaomi

0 likes · 16 min read

Xiaomi's 8 EMNLP 2026 Papers: Mobile Agents, Efficient LLMs, Proactive AI