Imagen de Portada

When Mean CE Fails: Median CE Can Better Track Language Model Quality

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Guo, Hao, Dennis, Simon, Patil, Rivaan, Shabahang, Kevin
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Compiling Agentic Workflows into LLM Weights: Near-Frontier Quality at Two Orders of Magnitude Less Cost
por: Dennis, Simon, et al.
Publicado: (2026)

In-Context Prompting Obsoletes Agent Orchestration for Procedural Tasks
por: Dennis, Simon, et al.
Publicado: (2026)

Beyond Inference-Only Deployment: Comparing Weight-Based Consolidation Against Cascading Compaction
por: Dennis, Simon, et al.
Publicado: (2026)

When Sensors Fail: Temporal Sequence Models for Robust PPO under Sensor Drift
por: Vogt-Lowell, Kevin, et al.
Publicado: (2026)

PaCE: Parsimonious Concept Engineering for Large Language Models
por: Luo, Jinqi, et al.
Publicado: (2024)

MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies
por: Zhang, Shiyue, et al.
Publicado: (2023)

Better World Models Can Lead to Better Post-Training Performance
por: Gupta, Prakhar, et al.
Publicado: (2025)

Frequency Matters: When Time Series Foundation Models Fail Under Spectral Shift
por: Wang, Tianze, et al.
Publicado: (2025)

When Counterfactual Reasoning Fails: Chaos and Real-World Complexity
por: Aalaila, Yahya, et al.
Publicado: (2025)

DyCE: Dynamically Configurable Exiting for Deep Learning Compression and Real-time Scaling
por: Wang, Qingyuan, et al.
Publicado: (2024)

Recitation over Reasoning: How Cutting-Edge Language Models Can Fail on Elementary School-Level Reasoning Problems?
por: Yan, Kai, et al.
Publicado: (2025)

CE-SSL: Computation-Efficient Semi-Supervised Learning for ECG-based Cardiovascular Diseases Detection
por: Zhou, Rushuang, et al.
Publicado: (2024)

When Will It Fail?: Anomaly to Prompt for Forecasting Future Anomalies in Time Series
por: Park, Min-Yeong, et al.
Publicado: (2025)

DeduCE: Deductive Consistency as a Framework to Evaluate LLM Reasoning
por: Pandey, Atharva, et al.
Publicado: (2025)

When Validation Fails: Cross-Institutional Blood Pressure Prediction and the Limits of Electronic Health Record-Based Models
por: Azam, Md Basit, et al.
Publicado: (2025)

Synthetic Error Injection Fails to Elicit Self-Correction In Language Models
por: Wu, David X., et al.
Publicado: (2025)

Learnware of Language Models: Specialized Small Language Models Can Do Big
por: Tan, Zhi-Hao, et al.
Publicado: (2025)

When Softmax Fails at the Top: Extreme Value Corrections for InfoNCE
por: Erol, Melihcan, et al.
Publicado: (2026)

When Continue Learning Meets Multimodal Large Language Model: A Survey
por: Huo, Yukang, et al.
Publicado: (2025)

When Chain-of-Thought Fails, the Solution Hides in the Hidden States
por: Mehrafarin, Houman, et al.
Publicado: (2026)

When Planning Fails Despite Correct Execution: On Epistemic Calibration for LLM-Based Multi-Agent Systems
por: Wang, Zehao, et al.
Publicado: (2026)

Logistics Hub Location Optimization: A K-Means and P-Median Model Hybrid Approach Using Road Network Distances
por: Rahman, Muhammad Abdul, et al.
Publicado: (2023)

When Less is More: 8-bit Quantization Improves Continual Learning in Large Language Models
por: Zhang, Michael S., et al.
Publicado: (2025)

Ask Again, Then Fail: Large Language Models' Vacillations in Judgment
por: Xie, Qiming, et al.
Publicado: (2023)

When Does Multimodality Lead to Better Time Series Forecasting?
por: Zhang, Xiyuan, et al.
Publicado: (2025)

Ascent Fails to Forget
por: Mavrothalassitis, Ioannis, et al.
Publicado: (2025)

When LLM Judge Scores Look Good but Best-of-N Decisions Fail
por: Landesberg, Eddie
Publicado: (2026)

When Refusals Fail: Unstable Safety Mechanisms in Long-Context LLM Agents
por: Hadeliya, Tsimur, et al.
Publicado: (2025)

When Intelligence Fails: An Empirical Study on Why LLMs Struggle with Password Cracking
por: Rehman, Mohammad Abdul, et al.
Publicado: (2025)

Geometric Median (GM) Matching for Robust Data Pruning
por: Acharya, Anish, et al.
Publicado: (2024)

When Models Know When They Do Not Know: Calibration, Cascading, and Cleaning
por: Hao, Chenjie, et al.
Publicado: (2026)

Theoretical Analysis of Relative Errors in Gradient Computations for Adversarial Attacks with CE Loss
por: Yu, Yunrui, et al.
Publicado: (2025)

(How) Do Language Models Track State?
por: Li, Belinda Z., et al.
Publicado: (2025)

When Can LLMs Learn to Reason with Weak Supervision?
por: Rahman, Salman, et al.
Publicado: (2026)

When LLM Reward Design Fails: Diagnostic-Driven Refinement for Sparse Structured RL
por: Wang, Youting, et al.
Publicado: (2026)

When Can Proxies Improve the Sample Complexity of Preference Learning?
por: Zhu, Yuchen, et al.
Publicado: (2024)

Lost in Transmission: When and Why LLMs Fail to Reason Globally
por: Schnabel, Tobias, et al.
Publicado: (2025)

Do Large Language Models Reason Causally Like Us? Even Better?
por: Dettki, Hanna M., et al.
Publicado: (2025)

ReasonIF: Large Reasoning Models Fail to Follow Instructions During Reasoning
por: Kwon, Yongchan, et al.
Publicado: (2025)

Can Language Models Use Forecasting Strategies?
por: Pratt, Sarah, et al.
Publicado: (2024)