Minimal Implementation of Elastic Training in Pytorch

Minimal Implementation of Elastic Training in Pytorch

Click the above “Getting Started with Vision” to add a Star or “Pin” Important content delivered immediately Scan the QR code below to join the cutting-edge academic paper exchange group!You can get the latest top conference/journal paper idea interpretations and the interpretation PDFs and materials from beginner to advanced in CV, as well as the … Read more

Understanding PyTorch Memory Management Mechanism

Understanding PyTorch Memory Management Mechanism

Author丨Mia Luo @ Zhihu (Authorized Reprint) Source丨https://zhuanlan.zhihu.com/p/486360176Editor丨Xiaoshutong, Jizhi Shutong 1. Background Introduction The analysis of the PyTorch memory management mechanism mainly aims to reduce the impact of “memory fragmentation”. A simple example is: As shown in the figure above, suppose we want to allocate 800MB of memory. Although the total free memory is 1000MB, the … Read more

Common Issues When Setting Up PyTorch Environment

Common Issues When Setting Up PyTorch Environment

1 Issues 1. Always stuck on solving environment and can’t get out. Network solutions include: configuring Tsinghua source, updating conda, etc., but nothing worked.2. After downloading, there are 3 things that need to be done, the last one starting with exe (maybe), always reports an error. Network solutions include: opening prompt with administrator privileges, etc., … Read more

Settings for Reproducible Experiments in PyTorch

Click on the above “Beginner’s Guide to Vision” to choose to add “Star” or “Pin“ Important content delivered promptly Author: Alxander@Zhihu (authorized) Source: https://zhuanlan.zhihu.com/p/448284000 Editor: Jishi Platform Jishi Guide During the training process in deep learning, due to random initialization and the randomness of sample reading, repeated experimental results may differ, with some variations being … Read more

Understanding PyTorch Memory Management Mechanism

Understanding PyTorch Memory Management Mechanism

Author丨Mialo@Zhihu Source丨https://zhuanlan.zhihu.com/p/486360176 1. Background Introduction Analyzing the PyTorch memory management mechanism primarily aims to reduce the impact of “memory fragmentation”. A simple example is as follows: As shown in the figure above, suppose we want to allocate 800MB of memory. Although the total free memory is 1000MB, the free memory shown in the upper figure … Read more

Common Pitfalls in PyTorch

Common Pitfalls in PyTorch

Click the “CVer” above to select “Star” or “Pin”. Heavyweight content delivered at the first time. Author: Bi Ji Ji https://zhuanlan.zhihu.com/p/59271905 This article is authorized, and no secondary reproduction is allowed without permission. 1. The Differences Between nn.Module.cuda() and Tensor.cuda() Both the cuda() function can achieve memory migration from CPU to GPU for models and … Read more

Running Deekseek-R1 Distillation Model with Llama Edge

Running Deekseek-R1 Distillation Model with Llama Edge

DeepSeek-R1 uses reinforcement learning to significantly enhance the model’s inference capabilities. In tasks such as mathematics, coding, and natural language reasoning, its performance rivals that of OpenAI’s official version o1.The small model distilled from DeepSeek-R1 effectively inherits the reasoning patterns learned by the large model.This article primarily tests DeepSeek-R1-Distill-Llama-8B-GGUF using Llama Edge. Welcome to experiment … Read more

DeepSeek: Unraveling the AGI Black Box

DeepSeek: Unraveling the AGI Black Box

As tech giants erect parameter monuments in the desert of computing power, a squad of engineers adorned with dynamic routing badges is cutting open the metal abdomen of large models with algorithm welding guns. The latest leaked battle map from the DeepSeek laboratory shows that their open-source model is rewriting the underlying game theory of … Read more

Faster R-CNN Model and Deep Learning Environment Setup

Faster R-CNN Model and Deep Learning Environment Setup

1. Faster R-CNN Model The R-CNN series networks are the most classic networks in the field of object detection, and their model update ideas are easy to understand. The object detection process is divided into three stages: candidate box generation, feature extraction, classification, and regression. R-CNN is a detection network assembled from many modules, where … Read more