Miguel Mendez personal website. Posts about Machine Learning and Deep Learning.
Learn how to manage multiple GitHub accounts on the same computer using SSH keys and Git configuration. This step-by-step guide covers SSH key setup, Git configuration, and automating repository access, making it easy to switch between work and personal accounts without conflicts.
This post covers LLM preference alignment, focusing on key techniques like PPO, DPO, and ORPO. Understand how large language models (LLMs) can be aligned with human preferences. Learn about the latest advancements in LLM alignment and how these methods optimize models to follow specific rules and preferences.
A profiling comparison between CPU and GPU performance when normalizing images in PyTorch. There are several factors to consider when optimizing preprocessing pipelines, such as data types, data transfer, and parallel processing capabilities. This posts explores these factors and provides insights on how to optimize your data pipeline.
In high-concurrency networking situations, handling a large number of requests can lead to different behaviors between httpx and aiohttp. While httpx may fail under heavy load, switching to aiohttp offers a more reliable solution for managing high traffic in asynchronous Python applications. This post explores how httpx struggles with concurrency and how aiohttp outperforms it in such scenarios
Explore the essential Object Tracking metrics with this comprehensive guide. We break down and compare key metrics like MOTA, IDF1, and HOTA, while also covering DetA and AssA. Understand how these metrics assess spatial accuracy and temporal consistency in object tracking.
Discover how Kubernetes' Horizontal Pod Autoscaler (HPA) functions using a real use case. Learn to manage and observe HPA in action with practical examples. This post will show you how to optimize resource usage, streamline pod scaling, and enhance application performance using a simple visualization tool.
Get a detailed overview of the transformative impact of transformer models on video instance segmentation. Learn about the evolution of VIS methods, the role of datasets like YouTube-VIS, and the latest research trends that are defining the future of video processing technologies. This includes detailed comparisons among leading models like VisTR, IFC, and TeViT.
Learn to create a Docker image for your Pytorch projects. Discover how to manage dependencies with Poetry and Python 3.11. We'll walk you through using Docker Buildx, handling Torch versions, and optimizing your build. Ideal for developers ready to quickly start their deep learning projects.
Explore computer vision techniques for mapping NFL field images to real-world coordinates. Learn how to develop a web app that uses homography to transform video feed data into precise spatial insights for sports analytics
Follow this step-by-step guide on deploying MLFlow in AWS using Terraform. Learn how to effectively manage your machine learning lifecycle, set up a Postgres database, create a secure S3 bucket, and customize a MLFlow Docker image. Improve you Machine Learning experiment tracking and model management in the cloud.
Dive into the complexities of object tracking in computer vision with this detailed overview of tracking by detection. Discover the intricacies and advancements of key algorithms like SORT, DeepSORT, and ByteTrack, and learn to select the right one for your project. From foundational concepts to cutting-edge techniques, the performance and applications of multi-object tracking for real-world…
Revolutionizing Sports Analytics with Artificial Intelligence: Discover how StatsBomb leverages cutting-edge AI and deep learning to redefine sports analytics using homography estimation, providing unparalleled insights into player performance and game strategies. Dive into the innovative world of data-driven sports analysis with StatsBomb's latest breakthroughs.
This post compares the performance of Torchserve's REST and gRPC communication protocols for transmitting images to a computer vision deep learning model. We conducted benchmarks for structured data, Base64 encoded images, and binary image transmission. The goal is to help practitioners make informed decisions when choosing the right communication protocol for their specific use case, taking into…
A Hugging Face Space to compare ResNet Class Activation Map to Vit Attention Rollout
The best way for monitoring your server is through your own code,