Cover Image

Intrinsic Bias is Predicted by Pretraining Data and Correlates with Downstream Performance in Vision-Language Encoders

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Ghate, Kshitish, Slaughter, Isaac, Wilson, Kyra, Diab, Mona, Caliskan, Aylin
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Biases Propagate in Encoder-based Vision-Language Models: A Systematic Analysis From Intrinsic Measures to Zero-shot Retrieval Outcomes
by: Ghate, Kshitish, et al.
Published: (2025)

Personal Information Parroting in Language Models
by: Subramani, Nishant, et al.
Published: (2026)

Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval
by: Wilson, Kyra, et al.
Published: (2024)

Bias Amplification in Stable Diffusion's Representation of Stigma Through Skin Tones and Their Homogeneity
by: Wilson, Kyra, et al.
Published: (2025)

Beyond Cooperative Simulators: Generating Realistic User Personas for Robust Evaluation of LLM Agents
by: Chopra, Harshita, et al.
Published: (2026)

Generative Value Conflicts Reveal LLM Priorities
by: Liu, Andy, et al.
Published: (2025)

Pre-Calc: Learning to Use the Calculator Improves Numeracy in Language Models
by: Veerendranath, Vishruth, et al.
Published: (2024)

ChatGPT Perpetuates Gender Bias in Machine Translation and Ignores Non-Gendered Pronouns: Findings across Bengali and Five other Low-Resource Languages
by: Ghosh, Sourojit, et al.
Published: (2023)

Identifying Features Associated with Bias Against 93 Stigmatized Groups in Language Models and Guardrail Model Safety Mitigation
by: Gueorguieva, Anna-Maria, et al.
Published: (2025)

No Thoughts Just AI: Biased LLM Hiring Recommendations Alter Human Decision Making and Limit Human Autonomy
by: Wilson, Kyra, et al.
Published: (2025)

EVALUESTEER: Measuring Reward Model Steerability Towards Values and Preferences
by: Ghate, Kshitish, et al.
Published: (2025)

Deep Reasoning in General Purpose Agents via Structured Meta-Cognition
by: Light, Dean, et al.
Published: (2026)

A Taxonomy of Stereotype Content in Large Language Models
by: Nicolas, Gandalf, et al.
Published: (2024)

Finetune-Informed Pretraining Boosts Downstream Performance
by: Faysal, Atik, et al.
Published: (2026)

Persona-Assigned Large Language Models Exhibit Human-Like Motivated Reasoning
by: Dash, Saloni, et al.
Published: (2025)

Renaissance: Investigating the Pretraining of Vision-Language Encoders
by: Fields, Clayton, et al.
Published: (2024)

SimBA: Simplifying Benchmark Analysis Using Performance Matrices Alone
by: Subramani, Nishant, et al.
Published: (2025)

Emotion Classification in Low and Moderate Resource Languages
by: Tafreshi, Shabnam, et al.
Published: (2024)

Do Generative AI Models Output Harm while Representing Non-Western Cultures: Evidence from A Community-Centered Approach
by: Ghosh, Sourojit, et al.
Published: (2024)

REALM: A Dataset of Real-World LLM Use Cases
by: Cheng, Jingwen, et al.
Published: (2025)

StressRoBERTa: Cross-Condition Transfer Learning from Depression, Anxiety, and PTSD to Stress Detection
by: Alqahtani, Amal, et al.
Published: (2025)

Can Large Language Models Infer Causation from Correlation?
by: Jin, Zhijing, et al.
Published: (2023)

Vision-Language Model Selection and Reuse for Downstream Adaptation
by: Tan, Hao-Zhe, et al.
Published: (2025)

VEAT Quantifies Implicit Associations in Text-to-Video Generator Sora and Reveals Challenges in Bias Mitigation
by: Sun, Yongxu, et al.
Published: (2026)

DSPA: Dynamic SAE Steering for Data-Efficient Preference Alignment
by: Wedgwood, James, et al.
Published: (2026)

Scaling Laws for Predicting Downstream Performance in LLMs
by: Chen, Yangyi, et al.
Published: (2024)

Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
by: Muhamed, Aashiq, et al.
Published: (2024)

A Causal Language Modeling Detour Improves Encoder Continued Pretraining
by: Touchent, Rian, et al.
Published: (2026)

Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks
by: Li, Miaomiao, et al.
Published: (2025)

How Early Adopters Used Generative AI Worldwide: Variation by Country Income and Language
by: Daepp, Madeleine I. G., et al.
Published: (2026)

CoRAG: Collaborative Retrieval-Augmented Generation
by: Muhamed, Aashiq, et al.
Published: (2025)

Harnessing the Intrinsic Knowledge of Pretrained Language Models for Challenging Text Classification Settings
by: Gao, Lingyu
Published: (2024)

When Do LLM Preferences Predict Downstream Behavior?
by: Slama, Katarina, et al.
Published: (2026)

Quantifying the Importance of Data Alignment in Downstream Model Performance
by: Chawla, Krrish, et al.
Published: (2025)

Downstream Transfer Attack: Adversarial Attacks on Downstream Models with Pre-trained Vision Transformers
by: Zheng, Weijie, et al.
Published: (2024)

Locating Demographic Bias at the Attention-Head Level in CLIP's Vision Encoder
by: Yasser, Alaa, et al.
Published: (2026)

Piano Transcription by Hierarchical Language Modeling with Pretrained Roll-based Encoders
by: Li, Dichucheng, et al.
Published: (2025)

Blind to Position, Biased in Language: Probing Mid-Layer Representational Bias in Vision-Language Encoders for Zero-Shot Language-Grounded Spatial Understanding
by: An, Na Min, et al.
Published: (2025)

RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models
by: Muhamed, Aashiq, et al.
Published: (2025)

Modeling Caption Diversity in Contrastive Vision-Language Pretraining
by: Lavoie, Samuel, et al.
Published: (2024)