Imagen de Portada

Reasoning in Transformers -- Mitigating Spurious Correlations and Reasoning Shortcuts

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Enström, Daniel, Kjellberg, Viktor, Johansson, Moa
Formato:	Preprint
Publicado:	2024
Materias:	Machine Learning Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Understanding and Mitigating Spurious Signal Amplification in Test-Time Reinforcement Learning for Math Reasoning
por: Yu, Yongcan, et al.
Publicado: (2026)

Spurious Rewards Paradox: Mechanistically Understanding How RLVR Activates Memorization Shortcuts in LLMs
por: Yan, Lecheng, et al.
Publicado: (2026)

In-context Learning in Presence of Spurious Correlations
por: Harutyunyan, Hrayr, et al.
Publicado: (2024)

MMLU-Pro+: Evaluating Higher-Order Reasoning and Shortcut Learning in LLMs
por: Taghanaki, Saeid Asgari, et al.
Publicado: (2024)

Mitigating Spurious Correlations in NLI via LLM-Synthesized Counterfactuals and Dynamic Balanced Sampling
por: Jaimes, Christopher Román
Publicado: (2025)

Fighting Spurious Correlations in Text Classification via a Causal Learning Perspective
por: Zhou, Yuqing, et al.
Publicado: (2024)

Shortcut Mitigation via Spurious-Positive Samples
por: Le, Phuong Quynh, et al.
Publicado: (2026)

Navigating Shortcuts, Spurious Correlations, and Confounders: From Origins via Detection to Mitigation
por: Steinmann, David, et al.
Publicado: (2024)

Short-circuiting Shortcuts: Mechanistic Investigation of Shortcuts in Text Classification
por: Eshuijs, Leon, et al.
Publicado: (2025)

Transformers meet Neural Algorithmic Reasoners
por: Bounsi, Wilfried, et al.
Publicado: (2024)

Implicit Reasoning in Transformers is Reasoning through Shortcuts
por: Lin, Tianhe, et al.
Publicado: (2025)

Mitigating Catastrophic Forgetting in Mathematical Reasoning Finetuning through Mixed Training
por: Reynolds, John Graham
Publicado: (2025)

Removing Spurious Correlation from Neural Network Interpretations
por: Fotouhi, Milad, et al.
Publicado: (2024)

Do LLMs Overcome Shortcut Learning? An Evaluation of Shortcut Challenges in Large Language Models
por: Yuan, Yu, et al.
Publicado: (2024)

Barriers to Universal Reasoning With Transformers (And How to Overcome Them)
por: Kraus, Oliver, et al.
Publicado: (2026)

Generalization or Hallucination? Understanding Out-of-Context Reasoning in Transformers
por: Huang, Yixiao, et al.
Publicado: (2025)

Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in SDOH Extraction with Large Language Models
por: Sakib, Fardin Ahsan, et al.
Publicado: (2025)

Navigating the Shortcut Maze: A Comprehensive Analysis of Shortcut Learning in Text Classification by Language Models
por: Zhou, Yuqing, et al.
Publicado: (2024)

When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs
por: Wang, Shaowen, et al.
Publicado: (2025)

Transformer See, Transformer Do: Copying as an Intermediate Step in Learning Analogical Reasoning
por: Hellwig, Philipp, et al.
Publicado: (2026)

Reasoning Towards Fairness: Mitigating Bias in Language Models through Reasoning-Guided Fine-Tuning
por: Kabra, Sanchit, et al.
Publicado: (2025)

ThinkEdit: Interpretable Weight Editing to Mitigate Overly Short Thinking in Reasoning Models
por: Sun, Chung-En, et al.
Publicado: (2025)

Specify What? Enhancing Neural Specification Synthesis by Symbolic Methods
por: Granberry, George, et al.
Publicado: (2024)

Unsupervised Concept Discovery Mitigates Spurious Correlations
por: Arefin, Md Rifat, et al.
Publicado: (2024)

Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers
por: Zhang, Zhongwang, et al.
Publicado: (2025)

A Symbolic Framework for Evaluating Mathematical Reasoning and Generalisation with Transformers
por: Meadows, Jordan, et al.
Publicado: (2023)

Reason to Rote: Rethinking Memorization in Reasoning
por: Du, Yupei, et al.
Publicado: (2025)

CURE: Controlled Unlearning for Robust Embeddings -- Mitigating Conceptual Shortcuts in Pre-Trained Language Models
por: Kocak, Aysenur, et al.
Publicado: (2025)

Seq-VCR: Preventing Collapse in Intermediate Transformer Representations for Enhanced Reasoning
por: Arefin, Md Rifat, et al.
Publicado: (2024)

Feature Resemblance: Towards a Theoretical Understanding of Analogical Reasoning in Transformers
por: Xu, Ruichen, et al.
Publicado: (2026)

Mitigating Hallucinations in Large Language Models via Causal Reasoning
por: Li, Yuangang, et al.
Publicado: (2025)

Mitigating Overthinking in Large Reasoning Models via Manifold Steering
por: Huang, Yao, et al.
Publicado: (2025)

LYNX: Learning Dynamic Exits for Confidence-Controlled Reasoning
por: Akgül, Ömer Faruk, et al.
Publicado: (2025)

Reasoning with Latent Thoughts: On the Power of Looped Transformers
por: Saunshi, Nikunj, et al.
Publicado: (2025)

Compositional Reasoning with Transformers, RNNs, and Chain of Thought
por: Yehudai, Gilad, et al.
Publicado: (2025)

Reasoning Inconsistencies and How to Mitigate Them in Deep Learning
por: Arakelyan, Erik
Publicado: (2025)

Do Transformers Use their Depth Adaptively? Evidence from a Relational Reasoning Task
por: Curth, Alicia, et al.
Publicado: (2026)

Influence Functions for Efficient Data Selection in Reasoning
por: Humane, Prateek, et al.
Publicado: (2025)

GeoReasoner: Reasoning On Geospatially Grounded Context For Natural Language Understanding
por: Yan, Yibo, et al.
Publicado: (2024)

LLMs are not Zero-Shot Reasoners for Biomedical Information Extraction
por: Nagar, Aishik, et al.
Publicado: (2024)