4 accepted papers
Gradient-based optimization is fundamental to deep learning, yet standard backpropagation (BP) is inherently limited by the requirement of differentiability, rendering it brittle when encountering piecewise-constant objectives with vanishing gradients (e.g., hard 0-1 loss) or black-box feedback. Whi…
Reinforcement learning with verifiable reward has recently emerged as a central paradigm for post-training large language models (LLMs); however, prevailing mean-based methods, such as Group Relative Policy Optimization (GRPO), suffer from entropy collapse and limited reasoning gains. We argue that…
Though current long-context large language models (LLMs) have demonstrated impressive capacities in answering various questions based on extensive text, the lack of citations in their responses makes user verification difficult, leading to concerns about their trustworthiness due to the potential ha…
Current Graph Contrastive Learning (GCL) methods primarily focus on adapting data augmentation techniques from Computer Vision (CV) or Natural Language Processing (NLP) domains. These techniques typically involve modifying input data via node sampling, edge perturbation, or graph structure perturbat…