phmatthaus的专栏

Anthropic发布的MCP，彻底打开了企业级 AI 的想象空间

OpenAI Gym入门与实操（3）

OpenAI Gym入门与实操（2）

强化学习基础知识

OpenAI Gym入门与实操（1）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 6

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 5（5）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 5（4）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 5（3）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 5（2）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 5（1）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 4（3）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 4（2）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 4（1）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 3（3）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 3（2）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters 3（1）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters（2）

论文译读 —— STUN: Reinforcement-Learning-Based Optimization of Kernel Scheduler Parameters for Static Workload Performance（1）

ResidualAttentionModelCovid_92( (conv1): Sequential( (0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (2): ReLU(inplace=True...

conda安装cv2库conda install opencv-python或者pip install opencv-python（不过好像是这个比较有效）

数据在CPU与GPU之间来回切换的pytorch方法：数据从CPU放到GPU，即数据从CPU到GPU的迁移，使用以下语句：data.to("cuda")数据 从GPU到CPU，使用以下语句：data.to("cpu")data通常会有两种数据类型：1. Tensor2. Moduleto函数：转换数据类型/设备1. tensor.to(*args, **kwargs)2. module.to(*args, **kwargs)举例：...

参考：https://blog.youkuaiyun.com/LEELOVESTUDY/article/details/106521130?utm_medium=distribute.pc_relevant_t0.none-task-blog-BlogCommendFromMachineLearnPai2-1.channel_param&depth_1-utm_source=distribute.pc_relevant_t0.none-task-blog-BlogCommendFromMachineLear

基于UCSD-AI4H/COVID-CT进行训练时，出现以下错误：RuntimeError: CUDA out of memory. Tried to allocate 98.00 MiB (GPU 0; 2.00 GiB total capacity; 1.53 GiB already allocated; 9.93 MiB free; 1.55 GiB reserved in total by PyTorch)解决方法：...

本机环境：Pycharm-2019.3.3 torchvision-0.5.0 Python-3.7 pytorch-1.4.0 numpy、sklearn、Visdom、torchxrayvision GTX1050安装visdom，命令为：pip install visdom安装过程log如下：Looking in indexes: https://mirrors.aliyun.com/pypi/simpleCollecting visdom Downloadi..

项目中执行 import torchxrayvision as xrv，解决了No module named ‘skimage‘、No module named ‘tqdm‘、No module named ‘pandas‘、No module named ‘pydicom‘的问题后出现以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 14, in <m.

项目中执行 import torchxrayvision as xrv，解决了No module named ‘skimage‘、No module named ‘tqdm‘、No module named ‘pandas‘的问题后出现以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 14, in <module> import torch...

项目中执行 import torchxrayvision as xrv，解决了No module named ‘skimage‘和No module named ‘tqdm‘的问题后出现以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 14, in <module> import torchxrayvision as xrv File ...

项目中执行 import torchxrayvision as xrv，解决了No module named ‘skimage‘的问题后，紧跟着出现以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 14, in <module> import torchxrayvision as xrv File "D:\研究生\毕业设计\COVID1...

项目中执行 import torchxrayvision as xrv时出现以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 14, in <module> import torchxrayvision as xrv File "D:\研究生\毕业设计\COVID19\COVID19_CT\torchxrayvision\__init_...

1. 创建虚拟环境执行 conda create -n pytorch_1.4_gpu python=3.7 命令，创建虚拟环境（具体版本根据自己实际情况设定）：/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////Collecting package metadata (current_repodata

在执行from torchvision import transforms时出现了以下错误：Traceback (most recent call last): File "D:/研究生/毕业设计/COVID19/COVID19_CT/conduct.py", line 8, in <module> from torchvision import transforms File "D:\Anaconda3\envs\pytorch_1.2_gpu\lib\site-pack...

(pytorch_1.2_gpu) D:\研究生\毕业设计\COVID19\COVID19_CT>pip install sklearnLooking in indexes: https://mirrors.aliyun.com/pypi/simpleCollecting sklearn Downloading https://mirrors.aliyun.com/pypi/packages/1e/7a/dbb3be0ce9bd5c8b7e3d87328e79063f8b263b2b1bfa.

pycharm官网下载地址：https://www.jetbrains.com/pycharm/download/根据自己的PC，选择不同平台：Windows、Mac、Linux。如果是历史版本，可以选择其中的“Other versions”。网址如下：https://www.jetbrains.com/pycharm/download/other.html...

anaconda下载地址:官网:http://anaconda.com/products/individual清华镜像源:https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=N&O=D

pytorch安装安装步骤：1. 检查是否有合适GPU，如果有，则需要安装CUDA和CuDNN。2. CUDA和CuDNN下载和安装（非必须）。3. 下载pytorch的whl文件。4. 下载torchvision的whl文件。5. 通过pip进行安装。6. 通过命令查看pytorch版本，验证pytorch是否安装成功。登录网址：https://download.pytorch.org/whl/torch_stable.html。选择合适的版本进行下载。笔者下载的.

train_tf.py源码如下:from __future__ import print_functionimport tensorflow as tfimport os, argparse, pathlibfrom eval import evalfrom data import BalanceCovidDatasetparser = argparse.ArgumentParser(description='COVID-Net Training Script')parser.add

import numpy as npimport pandas as pdimport osimport randomfrom shutil import copyfileimport pydicom as dicomimport cv2# set parameters heresavepath = 'data'seed = 0np.random.seed(seed) # Reset the seed so all runs are the same.random.seed(...

data.py源码如下:import tensorflow as tffrom tensorflow import kerasimport numpy as npimport osimport cv2from tensorflow.keras.preprocessing.image import ImageDataGeneratordef _process_csv_file(file): with open(file, 'r') as fr: files =

docs/train_eval_inference.md内容如下：# Training, Evaluation and InferenceThe network takes as input an image of shape (N, 224, 224, 3) and outputs the softmax probabilities as (N, 3), where N is the number of batches.If using the TF checkpoints, here are