4 papers
Post-Hoc Merging is Not Enough: Many-Shot Model Merging with Loss-Gap Balancing
Kyungjin Im, Miru Kim, Chanin Eom +1
Model merging has become a practical post-training strategy for building a single multi-task large language model (LLM) by combining multiple task-specialized models. However, most…
Fed-ADE: Adaptive Learning Rate for Federated Post-adaptation under Distribution Shift
Heewon Park, Mugon Joe, Miru Kim +2
Federated learning (FL) in post-deployment settings must adapt to non-stationary data streams across heterogeneous clients without access to ground-truth labels. A major challenge…
OASIS: Open-world Adaptive Self-supervised and Imbalanced-aware System
Miru Kim, Mugon Joe, Minhae Kwon
The expansion of machine learning into dynamic environments presents challenges in handling open-world problems where label shift, covariate shift, and unknown classes emerge. Post…
ASAP: Unsupervised Post-training with Label Distribution Shift Adaptive Learning Rate
Heewon Park, Mugon Joe, Miru Kim +1
In real-world applications, machine learning models face online label shift, where label distributions change over time. Effective adaptation requires careful learning rate selecti…