You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Presented at the 2023 International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision (WSCG 2023). Lightweight mirror segmentation CNN that uses an EfficientNet backbone, employs parallel convolutional layers to capture edge features, and applies filter pruning for model compression
The original experiments code for AAAI 2020 paper, "AutoCompress: An Automatic DNN Str
8000
uctured Pruning Framework for Ultra-High Compression Rates"
Dropping MoE Experts and Measuring Energy Per Intelligence: Where Is the Efficient Operating Point? Expert pruning on Qwen3-30B-A3B measured by EPI on Pi 5 cluster.
Removing Attention Heads: Does the Energy Actually Drop or Do Remaining Heads Compensate? Dual-stream correlation of PyTorch timing hooks and epi-meter power traces on Pi 5 cluster.