PyTorch 101: Understanding Hooks
We cover debugging and visualization in PyTorch. We explore PyTorch hooks, how to use them, visualize activations and modify gradients.
A Deeper Dive into Graph Neural Networks
Learn the fundamentals of Graph Neural Networks, how they work, and how to implement them using PyTorch. Explore key concepts and examples.
Understanding Reasoning in Large Language Models: Overview of the paper “Towards Reasoning in Large Language Models: A Survey”
In this article, we break down the paper “Towards Reasoning in Large Language Models: A Survey” in an attempt to explain relevant reasoning concepts used by LLMs.
Scaling Gradient with GPU Droplets and the cloud provider Networking
Build Scalable and Secure Gradient with the cloud provider’s GPUs, VPC, and Global Load Balancers.
MobiLlama: Your Compact Language Companion
Explore MobiLlama, SLM essentially a scaled-down versions of Llama, featuring 0.5 billion parameters, in contrast to LLMs that boast hundreds of billions or even trillions of parameters.
Text Labeling and Image Resolution with the Monkey Chat Vision Model
The Monkey Vision model, excels in generating detailed image captions and analyzing images through the Monkey Chat Vision model.
Solving Question-Answering on Tabular Data: A Comparison
Tapas & TableQA are libraries that allow users to input questions directly, as if using regular speech, to enact SQL-like queries on tabular data.
Everything you need to know about Few-Shot Learning
Learn everything you need to know about Few-Shot Learning, including key techniques, use cases, and how it enables machine learning models to perform with minimal training data.
Training YOLOv8 on Custom Data
This blog post explores YOLOv8, comparing its architectural changes to YOLOv5. We’ll also demonstrate the new model’s Python API functionality by testing its detection capabilities on a Basketball dataset.
Optimizing AI Models with Quanto
In this article discover Quanto a powerful quantization technique designed to optimize deep learning models without compromising the performance of the model.
