Search papers, labs, and topics across Lattice.
University of Southern Queensland
3
0
5
Task arithmetic reveals that not all fine-tuning combinations yield predictable model behaviors, challenging assumptions about parameter addition's reliability.
PriorTR reveals that ignoring model-induced priors can lead to the loss of critical task-specific information, enhancing MLLM efficiency without sacrificing accuracy.
By dynamically swapping state-space parameters between visual and temporal streams, M3S-Net achieves deep cross-modal coupling for PV power forecasting with linear complexity, outperforming shallow concatenation methods.
