Konstantin Schürholt

Researcher@Google·瑞士·OpenReview

研究方向

phase transitions · deep learning · Reasoning · Self Play · Program Synthesis · Representation Learning · Weight Space Learning · Hyper-Representations · Weight Generation · weight analysis

4.0

合作者 (20)

Konstantin Schürholt

Learning Neural Representations From Publicly Available Model Hubs

Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights

Scaling Up Parameter Generation: A Recurrent Diffusion Approach

Model Zoos for Benchmarking Phase Transitions in Neural Networks

Structure and Behavior in Weight Space Representation Learning