Picture for Karsten Roth

Karsten Roth

Understanding the Limits of Lifelong Knowledge Editing in LLMs

Add code
Mar 07, 2025
Viaarxiv icon

How to Merge Your Multimodal Models Over Time?

Add code
Dec 09, 2024
Figure 1 for How to Merge Your Multimodal Models Over Time?
Figure 2 for How to Merge Your Multimodal Models Over Time?
Figure 3 for How to Merge Your Multimodal Models Over Time?
Figure 4 for How to Merge Your Multimodal Models Over Time?
Viaarxiv icon

Context-Aware Multimodal Pretraining

Add code
Nov 22, 2024
Viaarxiv icon

A Practitioner's Guide to Continual Multimodal Pretraining

Add code
Aug 26, 2024
Viaarxiv icon

Disentangled Representation Learning through Geometry Preservation with the Gromov-Monge Gap

Add code
Jul 10, 2024
Figure 1 for Disentangled Representation Learning through Geometry Preservation with the Gromov-Monge Gap
Figure 2 for Disentangled Representation Learning through Geometry Preservation with the Gromov-Monge Gap
Figure 3 for Disentangled Representation Learning through Geometry Preservation with the Gromov-Monge Gap
Figure 4 for Disentangled Representation Learning through Geometry Preservation with the Gromov-Monge Gap
Viaarxiv icon

Reflecting on the State of Rehearsal-free Continual Learning with Pretrained Models

Add code
Jun 13, 2024
Viaarxiv icon

ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization

Add code
Jun 06, 2024
Figure 1 for ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization
Figure 2 for ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization
Figure 3 for ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization
Figure 4 for ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization
Viaarxiv icon

ETHER: Efficient Finetuning of Large-Scale Models with Hyperplane Reflections

Add code
May 30, 2024
Viaarxiv icon

Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models

Add code
May 02, 2024
Figure 1 for Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models
Figure 2 for Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models
Figure 3 for Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models
Figure 4 for Improving Intervention Efficacy via Concept Realignment in Concept Bottleneck Models
Viaarxiv icon

kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies

Add code
Apr 15, 2024
Figure 1 for kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies
Figure 2 for kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies
Figure 3 for kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies
Figure 4 for kNN-CLIP: Retrieval Enables Training-Free Segmentation on Continually Expanding Large Vocabularies
Viaarxiv icon