empirical evidence
Data collected through observation or experimentation that supports or refutes a hypothesis or claims about AI algorithms. It serves as the foundation for validating theoretical frameworks and models.
- Ascent Fails to Forget
- Assessing the quality of denoising diffusion models in Wasserstein distance: noisy score and optimal bounds
- Can MLLMs Absorb Math Reasoning Abilities from LLMs as Free Lunch?
- Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning
- DualOptim: Enhancing Efficacy and Stability in Machine Unlearning with Dual Optimizers
- Dynamical Properties of Tokens in Self-Attention and Effects of Positional Encoding
- Each Complexity Deserves a Pruning Policy
- From Condensation to Rank Collapse: A Two-Stage Analysis of Transformer Training Dynamics
- From Condensation to Rank Collapse: A Two-Stage Analysis of Transformer Training Dynamics
- HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
- Is PRM Necessary? Problem-Solving RL Implicitly Induces PRM Capability in LLMs
- Language Models Are Capable of Metacognitive Monitoring and Control of Their Internal Activations
- NeuroGenPoisoning: Neuron-Guided Attacks on Retrieval-Augmented Generation of LLM via Genetic Optimization of External Knowledge
- On the Surprising Effectiveness of Large Learning Rates under Standard Width Scaling
- RULE: Reinforcement UnLEarning Achieves Forget-retain Pareto Optimality
- Restricted Global-Aware Graph Filters Bridging GNNs and Transformer for Node Classification
- Revisiting Logit Distributions for Reliable Out-of-Distribution Detection
- The Graphon Limit Hypothesis: Understanding Neural Network Pruning via Infinite Width Analysis
- The Non-Linear Representation Dilemma: Is Causal Abstraction Enough for Mechanistic Interpretability?
- When majority rules, minority loses: bias amplification of gradient descent
- Who Reasons in the Large Language Models?