Featured
One Scheduler, Many Workers: Running Luigi Across Multiple Servers with Zero Downtime
Every data team I know has the same origin story. One cron job, on one VM, running one Luigi pipeline. It works beautifully — right...
Leetcode-2140. Solving Questions With Brainpower
Intuition We need to split the binary string s into two non-empty parts—left and right—so that the score, defined as “number of zeros in the...
All Stories
LongLLMLingua Model: A Solution for LLMs in Long Context Scenarios
Core Challenges 1. Question-Context Relevance Problem Traditional prompt compression methods face several critical issues when dealing with long contexts: Traditional Approach: Input: [Document1, Document2, ..., DocumentN] + Question Process: Compress...
LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models
The Challenges of of LLMs Large language models (LLMs) have revolutionized various applications due to their remarkable capabilities. Advancements in techniques like chain-of-thought prompting and in-context learning have significantly enhanced...
LLaMA: Open and Efficient Foundation Language Models
Exploring the Architecture, Training Data, and Training Process of LLaMA The sources provide a detailed explanation of the LLaMA model, covering its architecture, the data it was trained on, and...
Understanding Anchor Boxes in Object Detection
Anchor boxes play a crucial role in overcoming the limitation of traditional object detection approaches, where each grid cell can detect only one object. By allowing multiple objects to be...
Data Augmentation
Deep Learning for Computer Vision: Navigating the Landscape Deep learning has made significant advancements in various domains such as computer vision, natural language processing, speech recognition, online advertising, and logistics....
The Importance of Data Augmentation in Computer Vision
Data augmentation is a crucial technique used to improve the performance of computer vision systems. In the realm of computer vision, where the input is an image composed of countless...
Transfer Learning: Accelerating Computer Vision Applications with Pre-Trained Models
When building computer vision applications, leveraging pre-trained models can significantly speed up the development process and yield faster progress. Instead of training a neural network from scratch, you can download...
EfficientNet: Scaling Neural Networks for Optimal Performance
In today’s rapidly evolving technological landscape, developing neural networks that can efficiently adapt to different devices and computational resources is crucial. This is where EfficientNet comes into play. EfficientNet is...
MobileNet Architecture
In this blog post, we will explore the MobileNet architecture, which leverages depthwise separable convolutions to create computationally efficient neural networks. MobileNet v1 introduced the concept of depthwise separable convolutions,...
Understanding MobileNets: A Blog on Efficient Convolutional Neural Network Architecture
In the field of computer vision, convolutional neural networks (CNNs) play a vital role in various applications. However, many CNN architectures can be computationally expensive, making it challenging to deploy...
Understanding the Inception Network: A Powerful Architecture for Image Classification
The Inception network, developed by researchers at Google, is a convolutional neural network architecture that has revolutionized image classification tasks. In this blog, we will delve into the key concepts...
Convolutional Neural Networks: A Simple Example
Convolutional Neural Networks (ConvNets) are widely used for image classification and recognition tasks. In this blog post, we’ll walk through a simple example of a ConvNet to understand its architecture...
Unleashing the Power of Convolutional Neural Networks: A Breakdown of their Advantages and Training Process
Introduction Convolutional Neural Networks (CNNs) have revolutionized the field of computer vision, enabling remarkable achievements in image recognition and analysis. In this blog article, we will delve into the advantages...
Understanding Strided Convolutions in Convolutional Neural Networks
Convolutional Neural Networks (CNNs) are a key component in various computer vision tasks, such as image classification, object detection, and image segmentation. One fundamental building block of CNNs is convolutions,...
Summary: Pooling Layers in Convolutional Neural Networks
In Convolutional Neural Networks (ConvNets), pooling layers are often used alongside convolutional layers to reduce the size of the representation and enhance certain features’ robustness. This blog post summarizes the...
Summary: Understanding a Layer in Convolutional Neural Networks
In this blog post, we will dive into the details of one layer in Convolutional Neural Networks (CNNs). CNNs are widely used for image classification and recognition tasks, and understanding...
Convolutions Over Volume: Building Blocks of Convolutional Neural Networks
Convolutional Neural Networks (CNNs) have revolutionized the field of computer vision by enabling machines to extract meaningful features from images. At the core of CNNs lies the concept of convolutions,...
Deep Learning Computer Vision: Advancements and Exciting Applications
Introduction Deep learning has propelled significant advancements in computer vision, revolutionizing various applications such as self-driving cars, face recognition, image recommendation, and even art generation. In this blog article, we...
Deep-Q-Learning(DQN)
Why DQN? in an enviroment with a continue state space, it is impossible to go through all the possible states and actions repeatedly, since there are an infinite number of...
Introduction to Federated Learning
Motivation Privacy-preserving Machine Learning had always been exciting for me. Since my B.Tech. thesis involving PPML (SMPC + Computer Vision), I didn’t get a chance to work on it after...