About
Courses by Jonathan A.
-
GPT-4: The New GPT Release and What You Need to Know31m
GPT-4: The New GPT Release and What You Need to Know
By: Jonathan Fernandes
Contributions
-
What are the best practices for training and fine-tuning vision transformers on large-scale datasets?
You can also use data augmentation techniques with Vision transformers. One of the benefits of working with computer vision is that you can quickly visually confirm if the data augmentation techniques are working as expected. Here is a short example of preprocessing images: https://www.linkedin.com/learning/advanced-ai-transformers-for-computer-vision/preprocessing-images?autoplay=true&resume=false&u=2125562
-
What are the best practices for training and fine-tuning vision transformers on large-scale datasets?
If you are looking at an image classification task, then accuracy is typically your go-to metric. I talk a little more about this here: https://www.linkedin.com/learning/advanced-ai-transformers-for-computer-vision/training-arguments?autoplay=true&resume=false&u=2125562
-
What are the best practices for training and fine-tuning vision transformers on large-scale datasets?
Using a well-known pre-trained model is always a great starting point. This is because most of these models have been trained on a large dataset like ImageNet, with has over 1000 different classes of images and in excess of 1 million images. Using a pre-trained models allows you to you this model as your starting point and you can then fine-tune this model based on your dataset. I get into the details of this here: https://www.linkedin.com/learning/advanced-ai-transformers-for-computer-vision/using-a-pretrained-model-without-fine-tuning?autoplay=true&resume=false&u=2125562
-
What are the best practices for training and fine-tuning vision transformers on large-scale datasets?
Great question. Researchers had excellent results with the BERT for NLP. It was only a matter of time before they decided to try and use an architecture very similar to BERT for Image data. So instead of words, you are working with patches of images. This video gets into the details: https://www.linkedin.com/learning/advanced-ai-transformers-for-computer-vision/comparing-vision-transformers-to-bert?autoSkip=true&autoplay=true&resume=false&u=2125562
Activity
-
Its early days for this new way of measuring the performance of Large Language Models. Based on which primary domain your business is in, its still…
Its early days for this new way of measuring the performance of Large Language Models. Based on which primary domain your business is in, its still…
Shared by Jonathan A. Fernandes
-
Many of the AI community are losing it over the 'o' in GPT-4o. That's nothing. The real drama is how GPT-4o even has a "GPT-4" in its name. Its…
Many of the AI community are losing it over the 'o' in GPT-4o. That's nothing. The real drama is how GPT-4o even has a "GPT-4" in its name. Its…
Posted by Jonathan A. Fernandes
-
Ever been frustrated by a test where the questions didn’t make sense? Turns out, our AI models have the same problem! The industry has been using a…
Ever been frustrated by a test where the questions didn’t make sense? Turns out, our AI models have the same problem! The industry has been using a…
Shared by Jonathan A. Fernandes
Experience & Education
Licenses & Certifications
Volunteer Experience
-
Church small group leader and musician
Citygate church (http://www.citygatechurch.net)
Being involved in my local church is important to me. My wife and I lead one of the mid-week church groups and I am involved as a musician in the Sunday worship team.
-
Board Trustee
The Sheiling Trust
- Present 7 years 3 months
Education
The Sheiling Trust has been set up to aid Christian education overseas and acts as a channel for gifts to the schools and those connected with them.
Publications
-
Python Pandas Essential Training
LinkedIn Learning
I was invited by LinkedIn Learning to create a course on the essential's of Python's Pandas. This course includes basic data analysis and plotting, indexing, groupby and reshaping. The entire course is based on the Olympics dataset and each section builds on knowledge covered in previous sections.
Anyone who engages with the course content should have intermediate-level Pandas skills at the end of the course. -
Apache PySpark by Example
-
Want to get up and running with Apache Spark as soon as possible? If you're well versed in Python, the Spark Python API (PySpark) is your ticket to accessing the power of this hugely popular big data platform. This practical, hands-on course helps you get comfortable with PySpark, explaining what it has to offer and how it can enhance your data science work. To begin, instructor Jonathan Fernandes digs into the Spark ecosystem, detailing its advantages over other data science platforms, APIs…
Want to get up and running with Apache Spark as soon as possible? If you're well versed in Python, the Spark Python API (PySpark) is your ticket to accessing the power of this hugely popular big data platform. This practical, hands-on course helps you get comfortable with PySpark, explaining what it has to offer and how it can enhance your data science work. To begin, instructor Jonathan Fernandes digs into the Spark ecosystem, detailing its advantages over other data science platforms, APIs, and tool sets. Next, he looks at the DataFrame API and how it's the platform's answer to many big data challenges. Finally, he goes over Resilient Distributed Datasets (RDDs), the building blocks of Spark.
-
Docker for Data Scientists
-
In a field where reproducible results are essential, Docker is rapidly emerging as one of the top tools for bringing efficiency to the work that data science teams—particularly those working in machine learning (ML)—are doing. Creating and developing ML models is often messy. Seasoned data scientists know that different versions of the same software can produce different results. With Docker, you can include the right versions of each needed dependency and library, so no one ever has to do any…
In a field where reproducible results are essential, Docker is rapidly emerging as one of the top tools for bringing efficiency to the work that data science teams—particularly those working in machine learning (ML)—are doing. Creating and developing ML models is often messy. Seasoned data scientists know that different versions of the same software can produce different results. With Docker, you can include the right versions of each needed dependency and library, so no one ever has to do any configuration. After the Dockerfile is built, you'll have exactly what you need. In this course, Jonathan Fernandes helps data scientists get up and running with Docker, demonstrating how to build a Dockerized ML application that can easily be shared. Along the way, he shares common use cases for the tool. Upon wrapping up this course, you'll be prepared to leverage the power of containers in your other ML projects.
-
Introducing AI to your organisation
-
Artificial intelligence (AI) is taking the world by storm. Manufacturing, healthcare, and a host of other industries are steadily adopting this technology to streamline processes, enhance predictability, and generally keep ahead of the curve. In this course, discover what it takes to successfully introduce AI to your organization. Instructor Jonathan Fernandes steps through how to determine whether your organization is ready for AI, as well as how to develop and present a compelling business…
Artificial intelligence (AI) is taking the world by storm. Manufacturing, healthcare, and a host of other industries are steadily adopting this technology to streamline processes, enhance predictability, and generally keep ahead of the curve. In this course, discover what it takes to successfully introduce AI to your organization. Instructor Jonathan Fernandes steps through how to determine whether your organization is ready for AI, as well as how to develop and present a compelling business case for adopting the technology. Plus, he shares how to successfully implement AI—including how to do so using the scrum methodology—how to handle data collection and AI modeling, deploy, and finally monitor AI models once in production.
-
Introduction to Deep Learning with OpenCV
-
Deep learning is a fairly recent and hugely popular branch of artificial intelligence (AI) that finds patterns and insights in data, including images and video. Its layering and abstraction give deep learning models almost human-like abilities—including advanced image recognition. Using OpenCV—a widely adopted computer vision software—you can run previously trained deep learning models on inexpensive hardware and generate powerful insights from digital images and video. In this course…
Deep learning is a fairly recent and hugely popular branch of artificial intelligence (AI) that finds patterns and insights in data, including images and video. Its layering and abstraction give deep learning models almost human-like abilities—including advanced image recognition. Using OpenCV—a widely adopted computer vision software—you can run previously trained deep learning models on inexpensive hardware and generate powerful insights from digital images and video. In this course, instructor Jonathan Fernandes introduces you to the world of deep learning via inference, using the OpenCV Deep Neural Networks (dnn) module. You can get an overview of deep learning concepts and architecture, and then discover how to view and load images and videos using OpenCV and Python. Jonathan also shows how to provide classification for both images and videos, use blobs (the equivalent of tensors in other frameworks), and leverage YOLOv3 for custom object detection.
-
PyTorch essential training - Deep Learning
-
PyTorch is quickly becoming one of the most popular deep learning frameworks around, as well as a must-have skill in your artificial intelligence tool kit. It's gained admiration from industry leaders due to its deep integration with Python; its integration with top cloud platforms, including Amazon SageMaker and Google Cloud Platform; and its computational graphs that can be defined on the fly. In this course, join Jonathan Fernandes as he dives into the basics of deep learning using PyTorch…
PyTorch is quickly becoming one of the most popular deep learning frameworks around, as well as a must-have skill in your artificial intelligence tool kit. It's gained admiration from industry leaders due to its deep integration with Python; its integration with top cloud platforms, including Amazon SageMaker and Google Cloud Platform; and its computational graphs that can be defined on the fly. In this course, join Jonathan Fernandes as he dives into the basics of deep learning using PyTorch. Starting with a working image recognition model, he shows how the different components fit and work in tandem—from tensors, loss functions, and autograd all the way to troubleshooting a PyTorch network.
-
Transfer Learning for images using PyTorch
-
After its debut in 2017, PyTorch quickly became the tool of choice for many deep learning researchers. In this course, Jonathan Fernandes shows you how to leverage this popular machine learning framework for a similarly buzzworthy technique: transfer learning. Using a hands-on approach, Jonathan explains the basics of transfer learning, which enables you to leverage the pretrained parameters of an existing deep-learning model for other tasks. He then shows how to implement transfer learning for…
After its debut in 2017, PyTorch quickly became the tool of choice for many deep learning researchers. In this course, Jonathan Fernandes shows you how to leverage this popular machine learning framework for a similarly buzzworthy technique: transfer learning. Using a hands-on approach, Jonathan explains the basics of transfer learning, which enables you to leverage the pretrained parameters of an existing deep-learning model for other tasks. He then shows how to implement transfer learning for images using PyTorch, including how to create a fixed feature extractor and freeze neural network layers. Plus, find out about using learning rates and differential learning rates.
Projects
-
Pycon UK 2018 - PySpark with Machine Learning
I provided an overview of Apache Spark ecosystem, what makes Spark tick, RDDs and Spark's Dataframes API. We then looked at an application of Spark ML using the pipeline API with logistic regression.
-
Exploratory data analysis - London Data Science summit
I delivered a 1.5 hour exploratory data analysis workshop during the London Data Science summit.
-
PyCon 2017 - Introduction to Convolution Neural Networks
I conducted a 3-hour workshop providing attendees an introduction to Neural networks. We then determined why convolution neural networks produce better results. We examined examples of how Dropout, batch normalization and image augmentation improved accuracy. We also looked at varying different hyper-parameters including the type of activation function, the learning rate and regularization to see the impact this had on the neural networks. We concluded with a look at VGG16 - one of the top…
I conducted a 3-hour workshop providing attendees an introduction to Neural networks. We then determined why convolution neural networks produce better results. We examined examples of how Dropout, batch normalization and image augmentation improved accuracy. We also looked at varying different hyper-parameters including the type of activation function, the learning rate and regularization to see the impact this had on the neural networks. We concluded with a look at VGG16 - one of the top models from Imagenet 's ILSVR 2014 competition.
-
Data analysis - Cambridge Data Science Summit
I conducted a 1.5 hour data analysis workshop at the Cambridge Data Science summit.
-
PyData 2017 - Introduction to Pandas
I conducted a 1.5 workshop to get Python users from beginners to intermediate level in Python's Pandas. This included basic data analysis, indexing, groupby, plotting, stacking and regular expressions.
-
EuroPython 2021 Docker for ML Engineers
-
In this workshop, I reviewed the importance of Docker for ML. Understanding the components of Docker, which linux flavours are particularly helpful to ML engineers using Python. I ended the workshop by providing a couple of examples of ML projects that I have been involved in at work.
-
PyData London 2019 - Fundamentals of image classification using PyTorch
-
Pytorch is quickly gaining in popularity as a deep learning framework. If you have ever wondered, why bother with Pytorch when there are several other frameworks out there, then this is for you. This will be a hands-on tutorial quickly getting to speed with image classification using PyTorch, starting with the autograd function, CNN fundamentals and ending with the benefits of transfer learning.
Honors & Awards
-
SIA/NOL Scholarship for Undergraduate Studies
Singapore Airlines / Neptune Orient Lines
Full scholarship covering all tuition fees and a maintenance allowance for the duration of undergraduate studies.
Languages
-
German (intermediate)
Limited working proficiency
Recommendations received
-
LinkedIn User
38 people have recommended Jonathan A.
Join now to viewMore activity by Jonathan A.
-
Cast your mind back to last June—remember the buzz when the 40B Falcon models hit the scene? They were the hot topic, released under the open Apache…
Cast your mind back to last June—remember the buzz when the 40B Falcon models hit the scene? They were the hot topic, released under the open Apache…
Posted by Jonathan A. Fernandes
-
One of the questions I get frequently asked is for a language model that you can just run on your desktop. I've used several different tools, but a…
One of the questions I get frequently asked is for a language model that you can just run on your desktop. I've used several different tools, but a…
Shared by Jonathan A. Fernandes
-
If you want a hint about the future of AI, it is worth trying Gemini 1.5 with the 1M token context window, now available to everyone, apparently. It…
If you want a hint about the future of AI, it is worth trying Gemini 1.5 with the 1M token context window, now available to everyone, apparently. It…
Liked by Jonathan A. Fernandes
-
Don't want to spend the $20? Just drink regular coffee for a month. The only way to figure out how to use Large Language Models for the tasks you…
Don't want to spend the $20? Just drink regular coffee for a month. The only way to figure out how to use Large Language Models for the tasks you…
Shared by Jonathan A. Fernandes
-
So Elon Musk did what he said he would and released Grok-1 as an open model. Grok-1 is: - An Open model (Apache 2.0 license) so you can download the…
So Elon Musk did what he said he would and released Grok-1 as an open model. Grok-1 is: - An Open model (Apache 2.0 license) so you can download the…
Posted by Jonathan A. Fernandes
-
Wow. Super impressed. I needed some T-shirts printed for work. Contacted Mark B. from Envista Branding and they were able to turn it around in a…
Wow. Super impressed. I needed some T-shirts printed for work. Contacted Mark B. from Envista Branding and they were able to turn it around in a…
Posted by Jonathan A. Fernandes
-
While Gemini still doesn't generate images of people, you can (occasionally) get away with generating pictures of pixilated people. I managed to get…
While Gemini still doesn't generate images of people, you can (occasionally) get away with generating pictures of pixilated people. I managed to get…
Liked by Jonathan A. Fernandes
-
Guess what? Mistral isn't exactly rolling in cash like Meta, so they couldn't keep dishing out free models forever. It was bound to happen -…
Guess what? Mistral isn't exactly rolling in cash like Meta, so they couldn't keep dishing out free models forever. It was bound to happen -…
Shared by Jonathan A. Fernandes
-
Here's a little AI model training secret that could save you hours (and possibly prevent a few grey hairs): start small. Before unleashing the full…
Here's a little AI model training secret that could save you hours (and possibly prevent a few grey hairs): start small. Before unleashing the full…
Posted by Jonathan A. Fernandes
-
What do the following have in common: GPT-4, GPT-4-Turbo, Mixtral, Gemini and Gemini 1.5 (all AI models and Large Lanugage Models). Under the hood…
What do the following have in common: GPT-4, GPT-4-Turbo, Mixtral, Gemini and Gemini 1.5 (all AI models and Large Lanugage Models). Under the hood…
Posted by Jonathan A. Fernandes
Other similar profiles
-
Xavier (Xavi) Amatriain
Leading AI Products at Google
Connect -
Dave Birss
Connect -
Alexandra Manthey
Connect -
Souvik Ghosh
Connect -
Tomer Cohen
Connect -
Marvin Purtorab
Connect -
Denys Linkov
Connect -
Patrick Lewis
Connect -
Sandra Kublik
I study large language models and how they are used • Author of GPT-3 (Packt) • Developer Relations @ Cohere • YouTuber • Podcaster
Connect -
Robin Andrews
Connect
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More