2026 09 Neurips Accepted

Three papers accepted by NeurIPS 2026. Congratulations to everyone!

• “Powering Up Zeroth-Order Training via Subspace Gradient Orthogonalization”, Y. Lang, C. Wang, Y. Zhang, M. Hong, Z. Zhang, W. Yin, S. Liu. See here
• “Revisiting the Adam-SGD Gap in LLM Pre-Training: The Role of Large Effective Learning Rates”, A. Glentis, D. Li, C.-Y. Yau, M. Hong. See here
• “EMA-Nesterov: Stabilizing Nesterov’s Lookahead for Accelerated Deep Learning Optimization”, C.-Y. Yau, D. Li, A. Glentis, V. Boreiko, H.-T. Wai, M. Hong. See here