Quantization
1046 papers with code • 10 benchmarks • 18 datasets
Quantization is a promising technique to reduce the computation cost of neural network training, which can replace high-cost floating-point numbers (e.g., float32) with low-cost fixed-point numbers (e.g., int8/int16).
Source: Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers
Benchmarks
These leaderboards are used to track progress in Quantization
| Trend | Dataset | Best Model | Paper | Code | Compare |
|---|---|---|---|---|---|
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
Libraries
Use these libraries to find Quantization models and implementationsDatasets
CIFAR-10
ImageNet
MS COCO
LFW
Reddit
IJB-C
IJB-B
Word Sense Disambiguation: a Unified Evaluation Framework and Empirical Comparison
Tiny Images
Visual Wake Words
Most implemented papers
Q8BERT: Quantized 8Bit BERT
NervanaSystems/nlp-architect
•
•
Recently, pre-trained Transformer based language models such as BERT and GPT, have shown great improvement in many Natural Language Processing (NLP) tasks.
ConveRT: Efficient and Accurate Conversational Representations from Transformers
golsun/dialogrpt
•
•
Findings of the Association for Computational Linguistics 2020
General-purpose pretrained sentence encoders such as BERT are not ideal for real-world conversational AI applications; they are computationally heavy, slow, and expensive to train.
Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond
KaidiXu/auto_LiRPA
•
•
NeurIPS 2020
Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes provable linear bounds of output neurons given a certain amount of input perturbation, has become a core component in robustness verification and certified defense.
TernaryBERT: Distillation-aware Ultra-low Bit BERT
huawei-noah/Pretrained-Language-Model
•
•
EMNLP 2020
Transformer-based pre-training models like BERT have achieved remarkable performance in many natural language processing tasks. However, these models are both computation and memory expensive, hindering their deployment to resource-constrained devices.
Jointly Optimizing Query Encoder and Product Quantization to Improve Retrieval Performance
jingtaozhan/JPQ
•
•
Compared with previous DR models that use brute-force search, JPQ almost matches the best retrieval performance with 30x compression on index size.
AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
We then propose to search for the optimal per-channel scaling that protects the salient weights by observing the activation, not weights.
End-to-end Learning of Deep Visual Representations for Image Retrieval
almazan/deep-image-retrieval
•
•
Second, we build on the recent R-MAC descriptor, show that it can be interpreted as a deep and differentiable architecture, and present improvements to enhance it.
Trained Ternary Quantization
To solve this problem, we propose Trained Ternary Quantization (TTQ), a method that can reduce the precision of weights in neural networks to ternary values.
Quantizing deep convolutional networks for efficient inference: A whitepaper
Per-channel quantization of weights and per-layer quantization of activations to 8-bits of precision post-training produces classification accuracies within 2% of floating point networks for a wide variety of CNN architectures.
Fast Adjustable Threshold For Uniform Neural Network Quantization (Winning solution of LPIRC-II)
It can be performed without fine-tuning using calibration procedure (calculation of parameters necessary for quantization), or it is possible to train the network with quantization from scratch.
