Deep Learning
Robustness Guarantees for Adversarially Trained Neural Networks
This allows us to give a convergence guarantee for the inner-loop PGD attack.
Language Models can Solve Computer Tasks
However, previous approaches to this problem require large amounts of expert demonstrations and task-specific reward functions, both of which are impractical for new tasks.
MARBLE: Music Audio Representation Benchmark for Universal Evaluation
The authors contributed equally to this work.
MIMONets: Multiple-Input-Multiple-Output Neural Networks Exploiting Computation in Superposition
Finally, we provide mathematical bounds on the interference between superposition channels in MIMOFormer.
Birder: Communication-Efficient 1-bit Adaptive Optimizer for Practical Distributed DNN Training
Therefore, from a system-level perspective, the design ethos of a system-efficient communication-compression algorithm is that we should guarantee that the compression/decompression of the algorithm is computationally light and takes less time, and it should also be friendly to efficient collective communication primitives.
The Surprising Effectiveness of Diffusion Models for Optical Flow and Monocular Depth Estimation
Extensive experiments focus on quantitative performance against benchmarks, ablations, and the model's ability to capture uncertainty and multimodality, and impute missing values.
Appendix for You Only Condense Once: Two Rules for Pruning Condensed Datasets
The augmentations include: Color: adjusts the brightness, saturation, and contrast of images. Flip: flips the images horizontally with a probability of 0.5. It happens at a probability of 0.5. The first part is the update of the synthetic images. The second part is the update of the network.
You Only Condense Once: Two Rules for Pruning Condensed Datasets
Condense Once (YOCO) to overcome these limitations.