Cover Image

Pretraining Frequency Predicts Compositional Generalization of CLIP on Real-World Tasks

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wiedemer, Thaddäus, Sharma, Yash, Prabhu, Ameya, Bethge, Matthias, Brendel, Wieland
Format:	Preprint
Published:	2025
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?
by: Mayilvahanan, Prasanna, et al.
Published: (2023)

Provable Compositional Generalization for Object-Centric Learning
by: Wiedemer, Thaddäus, et al.
Published: (2023)

LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws
by: Mayilvahanan, Prasanna, et al.
Published: (2025)

VGGSounder: Audio-Visual Evaluations for Foundation Models
by: Zverev, Daniil, et al.
Published: (2025)

MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
by: Mayilvahanan, Prasanna, et al.
Published: (2025)

MentisOculi: Revealing the Limits of Reasoning with Mental Imagery
by: Zeller, Jana, et al.
Published: (2026)

No "Zero-Shot" Without Exponential Data: Pretraining Concept Frequency Determines Multimodal Model Performance
by: Udandarao, Vishaal, et al.
Published: (2024)

In Search of Forgotten Domain Generalization
by: Mayilvahanan, Prasanna, et al.
Published: (2024)

Mapping Post-Training Forgetting in Language Models at Scale
by: Harmon, Jackson, et al.
Published: (2025)

Are We Done with Object-Centric Learning?
by: Rubinstein, Alexander, et al.
Published: (2025)

Wu's Method can Boost Symbolic AI to Rival Silver Medalists and AlphaGeometry to Outperform Gold Medalists at IMO Geometry
by: Sinha, Shiven, et al.
Published: (2024)

LLM generation novelty through the lens of semantic similarity
by: Davydov, Philipp, et al.
Published: (2025)

A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility
by: Hochlehnert, Andreas, et al.
Published: (2025)

Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample Creation
by: Sinha, Shiven, et al.
Published: (2025)

A Practitioner's Guide to Continual Multimodal Pretraining
by: Roth, Karsten, et al.
Published: (2024)

Solving Spatial Supersensing Without Spatial Supersensing
by: Udandarao, Vishaal, et al.
Published: (2025)

Efficient Lifelong Model Evaluation in an Era of Rapid Progress
by: Prabhu, Ameya, et al.
Published: (2024)

Rule Extrapolation in Language Models: A Study of Compositional Generalization on OOD Prompts
by: Mészáros, Anna, et al.
Published: (2024)

ONEBench to Test Them All: Sample-Level Benchmarking Over Open-Ended Capabilities
by: Ghosh, Adhiraj, et al.
Published: (2024)

Intrinsic Credit Assignment for Long Horizon Interaction
by: Auzina, Ilze Amanda, et al.
Published: (2026)

PostTrainBench: Can LLM Agents Automate LLM Post-Training?
by: Rank, Ben, et al.
Published: (2026)

How to Merge Your Multimodal Models Over Time?
by: Dziadzio, Sebastian, et al.
Published: (2024)

Round-Trip Translation Reveals What Frontier Multilingual Benchmarks Miss
by: Skorobogat, Ronald, et al.
Published: (2026)

Position: Understanding LLMs Requires More Than Statistical Generalization
by: Reizinger, Patrik, et al.
Published: (2024)

Position: An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research
by: Reizinger, Patrik, et al.
Published: (2025)

Estimating Treatment Effects with Independent Component Analysis
by: Reizinger, Patrik, et al.
Published: (2025)

Generation is Required for Data-Efficient Perception
by: Brady, Jack, et al.
Published: (2025)

Strategic Dishonesty Can Undermine AI Safety Evaluations of Frontier LLMs
by: Panfilov, Alexander, et al.
Published: (2025)

Reflecting on the State of Rehearsal-free Continual Learning with Pretrained Models
by: Thede, Lukas, et al.
Published: (2024)

Scaling Open-Ended Reasoning to Predict the Future
by: Chandak, Nikhil, et al.
Published: (2025)

CLIP with Quality Captions: A Strong Pretraining for Vision Tasks
by: Vasu, Pavan Kumar Anasosalu, et al.
Published: (2024)

Great Models Think Alike and this Undermines AI Oversight
by: Goel, Shashwat, et al.
Published: (2025)

Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
by: Reizinger, Patrik, et al.
Published: (2024)

kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies
by: Gui, Zhongrui, et al.
Published: (2024)

Equivariance by Contrast: Identifiable Equivariant Embeddings from Unlabeled Finite Group Actions
by: Schmidt, Tobias, et al.
Published: (2025)

Adaptation Odyssey in LLMs: Why Does Additional Pretraining Sometimes Fail to Improve?
by: Öncel, Fırat, et al.
Published: (2024)

An Interventional Perspective on Identifiability in Gaussian LTI Systems with Independent Component Analysis
by: Rajendran, Goutham, et al.
Published: (2023)

Who Guards the Guardians? The Challenges of Evaluating Identifiability of Learned Representations
by: Joshi, Shruti, et al.
Published: (2026)

Causality is Key for Interpretability Claims to Generalise
by: Joshi, Shruti, et al.
Published: (2026)

Interaction Asymmetry: A General Principle for Learning Composable Abstractions
by: Brady, Jack, et al.
Published: (2024)