Apr

Deep Dive into Automatic Speech Recognition: Benchmarking Whisper JAX and PyTorch Implementations Across PlatformsIn the world of Automatic Speech Recognition (ASR), speed and accuracy are of great importance. The size of the data and models has been growing substantially recently, making it hard to be efficient. Nonetheless, the race is just starting, and we see new developments every week. In this article, we focus on Whisper JAX, a recent implementation of Whisper using a different backend framework that seems to run…

A tale of taming unruly documents to create the ultimate GPT-based chatbotPhoto by Rick Mason on UnsplashPicture this: you’re at a rapidly growing tech company, and you’ve been given the mission to create a state-of-the-art chatbot using the mind-blowing GPT technology. This chatbot is destined to become the company’s crown jewel, a virtual oracle that’ll answer questions based on the treasure trove of knowledge stored in your Confluence spaces. Sounds like a dream job, right?But, as you take a closer look at the…

Not very much, but we can improve it.Photo by Greg Rakozy on UnsplashThe F-test statistic for joint significance of the slope coefficients of a regression is routinely reported in regression outputs, along with other key statistics such as R² and t-ratio values.The question is whether it is useful or informative as a key statistic. Does it add any value to your regression results? While it is routinely reported, one may observe that the F-statistic almost always rejects H0 in practical applications. What does it tell us…

OpinionA sober perspective as to why boring is best, even for AIOver the last few weeks, we’ve had a trove of custom LLM requests from clients and partners. This excitement, although warranted, is based on tech news inundation, not on getting a fundamental corporate advantage.LLMs, even though they are not conceptually far off from most transformer-based training pipelines, require much more complex machinery to fine-tune and operate smoothly in a corporate setting. All the ones we already tested and deployed for clients…

A massive breakthrough in scene representation(Photo by nuddle on Unsplash)As we have seen with methods like DeepSDF and SRNs , encoding 3D objects and scenes within the weights of a feed-forward neural network is a memory-efficient, implicit representation of 3D data that is both accurate and high-resolution. However, the approaches we have seen so far are not quite capable of capturing realistic and complex scenes with sufficient fidelity. Rather, discrete representations (e.g., triangle meshes or voxel grids) produce…

A deep dive into the application of the transformer architecture and its self-attention operation for visionImage created by author using craiyon AIWhile self-attention is already widely adopted in NLP and significantly contributes to the performance of state-of-the-art models (e.g. , ), more and more work is being done to achieve similar results in vision.Even though, there are hybrid approaches that combine for example CNNs with attention or apply linear transformations on patches of the image , a pure attention-based…

Donut versus Pix2Struct on custom dataImage by author (with)Donut and Pix2Struct are image-to-text models that combine the simplicity of pure pixel inputs with visual language understanding tasks. Simply put: an image goes in and extracted indexes come out as JSON.Recently I released a Donut model finetuned on invoices. Ever so often I get the question how to train with a custom dataset. Also, a similar model was released: Pix2Struct, it claims to be significantly better. But is that so?Time to roll up my sleeves. I will…

Lesson 1: Batch Serving. Feature Stores. Feature Engineering Pipelines.Photo by Hassan Pasha on UnsplashThis tutorial represents lesson 1 out of a 7-lesson course that will walk you step-by-step through how to design, implement, and deploy an ML system using MLOps good practices. During the course, you will build a production-ready model forecasting energy consumption for the next 24 hours across multiple consumer types from Denmark.By the end of this course, you will understand all the fundamentals of designing, coding…

Grab the low-hanging fruits with open-sourceMaster Data Management, or MDM, is commercial vendors' buzzword for an entity resolution framework. I talked to several vendors, most offering SaaS and priced by the total number of records ingested from sources. That totals in the 6- to 7-digit $ range per year for larger enterprises.The target audience for this articleAre you planning to implement MDM soon? Have you asked vendors for a quote? Or did your company already invest in an MDM SaaS? For sure, it is not a small…

Basic recommenders that are easy to understand and implement, as well as fast to trainCreated by me on dreamstudio.ai.Recommender systems are algorithms designed to provide user recommendations based on their past behavior, preferences, and interactions. Becoming integral to various industries, including e-commerce, entertainment, and advertising, recommender systems improve user experience, increase customer retention, and drive sales.While various advanced recommender systems exist, today I want to show you one of the…