Tag：Prune

Time：2021723
preface Wechat search【Java3y】Paying attention to this man with dreams and praising attention is my greatest support! The text has been included in my GitHub：https://github.com/ZhongFuCheng3y/3y, there are more than 300 original articles, recently in seriesInterviews and projectsSeries! I’d like to start with you todaykylin(kylin). Due to work needs, some time agokylinNow let’s take notes. (my words […]

Time：2021629
The characteristic of condensenet is that it can learn the grouping convolution and prune in combination with the training process. It can not only prune accurately, but also continue to train and make the network weight smoother, which is a very good work Source: Xiaofei’s algorithm Engineering Notes official account Thesis: neural architecture search with […]

Time：2021322
Link to the original text:http://tecdat.cn/?p=9859 Overview This article is aboutTree basedRegression and classification methods. Tree methods are easy to understand, but they are very useful for interpretation. However, in terms of prediction accuracy, they usually cannot compete with the best supervised learning method. Therefore, we also introduce bagging, random forest and enhanced tree. Each of […]

Time：2021314
In the last article, we briefly introduced how to get started with optuna, including its concise API design and rich storage / analysis / visualization suite. In fact, optuna is more than that. As the main framework of deep learning oriented hyper parameter tuning development, optuna takes into account various practical situations of largescale model […]

Time：2021310
Fast context switching of DL tasks on GPU PipeSwitch: Fast Pipelined Context Switching for Deep Learning Applications https://www.usenix.org/confer… （Johns Hopkins University & ByteDance） brief introduction Background, motivation DL task: throughput sensitive training task, delay sensitive reasoning task. In order to ensure the reasoning SLOS, the mainstream design is to deploy them separately on different GPU […]

Time：202124
In this paper, the DBTD method is used to calculate the filtering threshold, and then the random pruning algorithm is used to prune the eigenvalue gradient. The sparse eigenvalue gradient can reduce the amount of calculation in the return phase. The training on CPU and arm has 3.99 times and 5.92 times acceleration effect respectively […]

Time：20201115
In this paper, DBTD method is used to calculate the filtering threshold, and then the random pruning algorithm is combined to prune the eigenvalue gradient, which can reduce the calculation amount in the backhaul phase. The training on CPU and arm has 3.99 times and 5.92 times acceleration effect respectively Source: Xiaofei’s algorithm Engineering Notes […]

Time：2020929
Reading notes of hanjiawai: Chapter 5 (data cube technology) BUC 1. BUC (bottom up construction) concept Calculation of iceberg cube from vertex cube downAlgorithm for computing sparse iceberg cubePruning based on a priori property###2. Algorithm calculation Cases a (A1, A2, A3), B (B1, B2), C (C1, C2) (a1,,);(a1,b1,*);(a1,b1,c1);(a1,b1,c2)…… This process uses a priori property to […]

Time：2020928
BUC 1. BUC (bottom up construction) concept Calculation of iceberg cube from vertex cube downAlgorithm for computing sparse iceberg cubePruning based on a priori property###2. Algorithm calculation Cases a (A1, A2, A3), B (B1, B2), C (C1, C2) (a1,,);(a1,b1,*);(a1,b1,c1);(a1,b1,c2)…… This process uses a priori property to save time. If (A1, B1, *) does not meet […]

Time：202086
Python code repository Python incorporated this part of the pruning code in November, 19. Python provides some directly available APIs. The user only needs to pass in the module instance to be pruned and the parameter name to be pruned. The system automatically helps complete the pruning operation. It seems that the interface is very […]

Time：2020729
Introduction to decision tree This part briefly introduces the principle of decision tree and the call method of sklearn. import warnings warnings.filterwarnings(‘ignore’) import pandas as pd import numpy as np import matplotlib.pyplot as plt %matplotlib inline import seaborn as sns import mglearn decision tree is essentially a series of if / else statements, which […]

Time：2020720
Condensenet is characterized by the proposition of learnable grouping convolution. Combining with the training process, pruning can not only accurately prune, but also continue to train, making the network weight more smooth, which is a very good work Source: Xiaofei’s algorithm Engineering Notes official account Paper: neural architecture search with reinforcement learning Thesis address: https://arxiv.org/abs/1711.09224 […]