accuracy preservation
Accuracy preservation refers to the ability of a model or algorithm to maintain its predictive accuracy during transformations or adaptations, such as model compression or transfer learning, ensuring performance does not degrade significantly.
- A Private Approximation of the 2nd-Moment Matrix of Any Subsamplable Input
- Adaptive Fission: Post-training Encoding for Low-latency Spike Neural Networks
- FairDD: Fair Dataset Distillation
- Inference-Time Hyper-Scaling with KV Cache Compression
- SNAP: Low-Latency Test-Time Adaptation with Sparse Updates
- Training Language Models to Reason Efficiently