Natural Language Processing

Large language models (LLMs) often prematurely abandon promising reasoning paths, a phenomenon called ‘underthinking’. This paper introduces a novel metric to quantify this issue and proposes a decodi…

GuardReasoner enhances LLM safety with reasoning-based guardrails, improving performance, explainability, and generalization on various benchmarks.

Virus: A new attack method easily bypasses LLM guardrails, highlighting the inadequacy of current safety measures and urging for more robust solutions.

Large language models underperform humans in open-ended exploration due to prioritizing immediate choices over long-term strategic thinking, but innovative models show promise.

Critique Fine-Tuning (CFT) outperforms traditional supervised fine-tuning (SFT) in training language models, achieving comparable results with significantly less data and opening new avenues in AI.

Reinforcement learning (RL) surpasses supervised fine-tuning (SFT) in fostering generalization in foundation models, while SFT aids RL’s stability; a comparative study across text and visual domains r…

SafeRAG: A new benchmark exposes critical security vulnerabilities in Retrieval-Augmented Generation (RAG) systems by introducing four novel attack types and a comprehensive dataset for evaluation, re…

Boosting Large Language Model (LLM) performance, researchers introduce Over-Tokenized Transformers, decoupling input/output vocabularies to improve language modeling. Scaling input vocabularies improv…

First-ever FP4 training framework for LLMs achieves accuracy comparable to BF16 and FP8, enabling efficient ultra-low precision training.

HISTOIRESMORALES: a new French dataset tackles the crucial issue of aligning language models with human moral values, providing valuable resources for ethical AI research in a previously underserved l…

IndicMMLU-Pro: a new benchmark rigorously evaluates large language models’ multi-task language understanding capabilities across nine major Indian languages, pushing Indic language AI research forward…

Atla Selene Mini: A state-of-the-art small LLM judge surpassing larger models in benchmark performance!

ARWKV: A novel RNN-attention-based language model, distilled from a larger model, achieves strong performance using significantly fewer resources, opening a new path in efficient language model develo…

RealCritic: A new benchmark effectively evaluates language models’ critique abilities using a closed-loop methodology, showcasing advanced reasoning models’ superiority in self and iterative critique.

Humanity’s Last Exam (HLE): a groundbreaking multi-modal benchmark pushing the boundaries of large language model (LLM) capabilities, revealing a significant gap between current LLMs and human experts…

CoRAG, a novel Chain-of-Retrieval Augmented Generation model, dynamically refines queries for improved accuracy in multi-hop question answering, achieving state-of-the-art performance.

SIGMA, a novel large language model, achieves up to 33.36% faster inference speeds by using DiffQKV attention, which differentially optimizes query, key, and value components in the attention mech…

Low-rank adapters combined with neural architecture search revolutionize LLM compression, enabling efficient fine-tuning and significantly reduced memory footprint.

Large language models (LLMs) are rapidly evolving, yet often struggle to adapt to human preferences quickly. This paper introduces Test-Time Preference Optimization (TPO), an innovative framework that…

Pairwise RM, a novel reward model with knockout tournaments, significantly boosts large language model accuracy in test-time scaling by comparing solution pairs, eliminating arbitrary scoring inconsis…