Language Models as Zero-shot Lossless Gradient Compressors: Towards General Neural Parameter Prior Modelsby Hui-Po Wang,…
RmGPT: Rotating Machinery Generative Pretrained Modelby Yilin Wang, Yifei Yu, Kong Sun, Peixuan Lei, Yuxuan…
On the Optimal Memorization Capacity of Transformersby Tokio Kajitsuka, Issei SatoFirst submitted to arxiv on:…
Non-asymptotic Convergence of Training Transformers for Next-token Predictionby Ruiquan Huang, Yingbin Liang, Jing YangFirst submitted…
Characterizing stable regions in the residual stream of LLMsby Jett Janiak, Jacek Karwowski, Chatrik Singh…
Counterfactual Token Generation in Large Language Modelsby Ivi Chatzi, Nina Corvelo Benz, Eleni Straitouri, Stratis…
Layout-Corrector: Alleviating Layout Sticking Phenomenon in Discrete Diffusion Modelby Shoma Iwai, Atsuki Osanai, Shunsuke Kitada,…
The Credibility Transformerby Ronald Richman, Salvatore Scognamiglio, Mario V. WüthrichFirst submitted to arxiv on: 25…
WeatherFormer: Empowering Global Numerical Weather Forecasting with Space-Time Transformerby Junchao Gong, Tao Han, Kang Chen,…
Adversarial Attacks on Parts of Speech: An Empirical Study in Text-to-Image Generationby G M Shahariar,…