Imagen de Portada

Gemma Needs Help: Investigating and Mitigating Emotional Instability in LLMs

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Soligo, Anna, Mikulik, Vladimir, Saunders, William
Formato:	Preprint
Publicado:	2026
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

When Do LLMs Need Retrieval Augmentation? Mitigating LLMs' Overconfidence Helps Retrieval Augmentation
por: Ni, Shiyu, et al.
Publicado: (2024)

TxGemma: Efficient and Agentic LLMs for Therapeutics
por: Wang, Eric, et al.
Publicado: (2025)

Text Is Not All You Need: Multimodal Prompting Helps LLMs Understand Humor
por: Baluja, Ashwin
Publicado: (2024)

CodeGemma: Open Code Models Based on Gemma
por: CodeGemma Team, et al.
Publicado: (2024)

Emergent Misalignment is Easy, Narrow Misalignment is Hard
por: Soligo, Anna, et al.
Publicado: (2026)

ShieldGemma: Generative AI Content Moderation Based on Gemma
por: Zeng, Wenjun, et al.
Publicado: (2024)

Help! Need Advice on Identifying Advice
por: Govindarajan, Venkata Subrahmanyan, et al.
Publicado: (2020)

Emergent misalignment as prompt sensitivity: A research note
por: Wyse, Tim, et al.
Publicado: (2025)

Investigating Symbolic Triggers of Hallucination in Gemma Models Across HaluEval and TruthfulQA
por: Lamba, Naveen, et al.
Publicado: (2025)

TranslateGemma Technical Report
por: Finkelstein, Mara, et al.
Publicado: (2026)

Gemma 3 Technical Report
por: Gemma Team, et al.
Publicado: (2025)

MedGemma Technical Report
por: Sellergren, Andrew, et al.
Publicado: (2025)

Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2
por: Lieberum, Tom, et al.
Publicado: (2024)

T5Gemma 2: Seeing, Reading, and Understanding Longer
por: Zhang, Biao, et al.
Publicado: (2025)

White Men Lead, Black Women Help? Benchmarking and Mitigating Language Agency Social Biases in LLMs
por: Wan, Yixin, et al.
Publicado: (2024)

EmbeddingGemma: Powerful and Lightweight Text Representations
por: Vera, Henrique Schechter, et al.
Publicado: (2025)

Does Context Help Mitigate Gender Bias in Neural Machine Translation?
por: Gete, Harritxu, et al.
Publicado: (2024)

Gemma: Open Models Based on Gemini Research and Technology
por: Gemma Team, et al.
Publicado: (2024)

SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs
por: Zhao, Sihang, et al.
Publicado: (2026)

Investigating the Impact of Model Instability on Explanations and Uncertainty
por: Marjanović, Sara Vera, et al.
Publicado: (2024)

From Text to Emotion: Unveiling the Emotion Annotation Capabilities of LLMs
por: Niu, Minxue, et al.
Publicado: (2024)

Stress-Testing Emotional Support Models: Moving from Homogeneous to Diverse Help Seekers
por: Heo, Chaewon, et al.
Publicado: (2026)

From Syntax to Emotion: A Mechanistic Analysis of Emotion Inference in LLMs
por: Shu, Bangzhao, et al.
Publicado: (2026)

Emotion-o1: Adaptive Long Reasoning for Emotion Understanding in LLMs
por: Song, Changhao, et al.
Publicado: (2025)

Evaluating Prompting Strategies with MedGemma for Medical Order Extraction
por: Balachandran, Abhinand, et al.
Publicado: (2025)

Assessing RAG and HyDE on 1B vs. 4B-Parameter Gemma LLMs for Personal Assistants Integretion
por: Sorstkins, Andrejs
Publicado: (2025)

Unrequited Emotions: Investigating the Gaps in Motivation and Practice in Speech Emotion Recognition Research
por: Wong, Taryn, et al.
Publicado: (2026)

Inclusion-of-Thoughts: Mitigating Preference Instability via Purifying the Decision Space
por: Madani, Mohammad Reza Ghasemi, et al.
Publicado: (2026)

An LLM's Apology: Outsourcing Awkwardness in the Age of AI
por: Stone, Twm, et al.
Publicado: (2025)

Benchmarking Llama2, Mistral, Gemma and GPT for Factuality, Toxicity, Bias and Propensity for Hallucinations
por: Nadeau, David, et al.
Publicado: (2024)

When Seekers Are Hard to Help: Evaluating Emotional Support Dialogue Systems in Worst-Case Interactions
por: Yang, Jiajie, et al.
Publicado: (2026)

From "Help" to Helpful: A Hierarchical Assessment of LLMs in Mental e-Health Applications
por: Steigerwald, Philipp, et al.
Publicado: (2026)

Contrastive Distillation of Emotion Knowledge from LLMs for Zero-Shot Emotion Recognition
por: Niu, Minxue, et al.
Publicado: (2025)

CULEMO: Cultural Lenses on Emotion -- Benchmarking LLMs for Cross-Cultural Emotion Understanding
por: Belay, Tadesse Destaw, et al.
Publicado: (2025)

Emotionally Numb or Empathetic? Evaluating How LLMs Feel Using EmotionBench
por: Huang, Jen-tse, et al.
Publicado: (2023)

Between Help and Harm: An Evaluation of Mental Health Crisis Handling by LLMs
por: Arnaiz-Rodriguez, Adrian, et al.
Publicado: (2025)

Disclosure and Mitigation of Gender Bias in LLMs
por: Dong, Xiangjue, et al.
Publicado: (2024)

Understanding and Mitigating Language Confusion in LLMs
por: Marchisio, Kelly, et al.
Publicado: (2024)

Mitigating Unhelpfulness in Emotional Support Conversations with Multifaceted AI Feedback
por: Wang, Jiashuo, et al.
Publicado: (2024)

Can Large Language Models be Good Emotional Supporter? Mitigating Preference Bias on Emotional Support Conversation
por: Kang, Dongjin, et al.
Publicado: (2024)