REVIEW 3 cited by
Training with Quantization Noise for Extreme Model Compression
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
We tackle the problem of producing compact models, maximizing their accuracy for a given model size. A standard solution is to train networks with Quantization Aware Training, where the weights are quantized during training and the gradients approximated with the Straight-Through Estimator. In this paper, we extend this approach to work beyond int8 fixed-point quantization with extreme compression methods where the approximations introduced by STE are severe, such as Product Quantization. Our proposal is to only quantize a different random subset of weights during each forward, allowing for unbiased gradients to flow through the other weights. Controlling the amount of noise and its form allows for extreme compression rates while maintaining the performance of the original model. As a result we establish new state-of-the-art compromises between accuracy and model size both in natural language processing and image classification. For example, applying our method to state-of-the-art Transformer and ConvNet architectures, we can achieve 82.5% accuracy on MNLI by compressing RoBERTa to 14MB and 80.0 top-1 accuracy on ImageNet by compressing an EfficientNet-B3 to 3.3MB.
Forward citations
Cited by 3 Pith papers
-
Multi-Modal Learning meets Genetic Programming: Analyzing Alignment in Latent Space Optimization
SNIP's symbolic-numeric alignment stays coarse and does not improve during optimization, so multi-modal LSO does not yet deliver effective bi-modal search for symbolic regression.
-
BiVM: Accurate Binarized Neural Network for Efficient Video Matting
BiVM is a 1-bit binarized video matting network that beats prior binarized methods on accuracy and efficiency, with 11.82 MAD on VideoMatte240K versus 28.49 for ReActNet-binarized RVM.
-
Resource-Efficient Automatic Software Vulnerability Assessment via Knowledge Distillation and Particle Swarm Optimization
A PSO-guided knowledge distillation framework compresses a CodeBERT vulnerability assessor to 0.6% of its original size while retaining 89.3% of its accuracy.
Discussion (0). Sign in to comment.