🏢 ETH Zurich
UIP2P: Unsupervised Instruction-based Image Editing via Cycle Edit Consistency
·3351 words·16 mins·
loading
·
loading
AI Generated
🤗 Daily Papers
Computer Vision
Image Generation
🏢 ETH Zurich
UIP2P: Unsupervised instruction-based image editing achieves high-fidelity edits by enforcing Cycle Edit Consistency, eliminating the need for ground-truth data.
LoRACLR: Contrastive Adaptation for Customization of Diffusion Models
·2785 words·14 mins·
loading
·
loading
AI Generated
🤗 Daily Papers
Computer Vision
Image Generation
🏢 ETH Zurich
LoRACLR merges multiple LoRA models for high-fidelity multi-concept image generation, using a contrastive objective to ensure concept distinctiveness and prevent interference.
Do I Know This Entity? Knowledge Awareness and Hallucinations in Language Models
·5261 words·25 mins·
loading
·
loading
AI Generated
🤗 Daily Papers
Natural Language Processing
Large Language Models
🏢 ETH Zurich
LLMs’ hallucinations stem from entity recognition: SAEs reveal model ‘self-knowledge’, causally affecting whether it hallucinates or refuses to answer. This mechanism is even repurposed by chat finet…