Place to express my learnings
As of today, it’s not feasible to fully understand LLM. Therefore, scientists are hypothesizing that studying a toy model would help us to understand the big model. I came across such toy model released by ARC and I tried to understand their understanding.
The whole ego competition started when I received a text message saying, “Anyone here is trying this?” with a blog title: Designing AI-resistant technical tech evaluation. As I read the blog, I was excited to see how far AI agents have caught up with coding tasks, and it also created a subtle fear in me of not being valuable anymore. I bet some of you have already felt that way at some point.…
My initial aim was to build a document processing model. But the idea was far fetched for my skill at the time. So, I settled down for building a toy version of visual language model for better understanding of VLM. I’m documenting my intuition for the benefit of myself and others. My model will receive image as an input and return its caption as an output. Luckily, I found a dataset with image…
My Journey to Building Agentic Apps
Neural Networks learn to predict by backpropagation. This article aims to help you, build a solid intuition about the concept using a simple example. The ideas we learn here can be expanded for bigger nerual network. I assume that you already know how feed forward neural network works.
The first thing I do when I create a project is to create the debugger launch config at the .vscode folder. Debuggers help me to avoid putting print statements and building the program again. I always wondered how a debugger can stop the program on the line number I want and be able to inspect variables. Debugger workings have always been dark magic for me. At last, I managed to learn dark art by…
I’m a mediocre engineer who does systems work and never had experience in the typical user-facing software space. I’ve contributed to software that scales but never really had a chance to experience the vibe of serving millions of users.