Welcome to Ando's Logs
RSS Feed👋 Hi - I'm Alex Andonian (aka Ando) and these are my personal "training" logs.
As a deep learning researcher, I spend a lot of time inspecting data and debugging training logs. Recently, I wondered why I don't do the same for my own neural network. So here we are!
Jokes aside, I'm hoping this will become a growing collection of articles, tutorials, and guides about deep learning, AI, programming, and technology. By documenting my own learning trajectories, I hope it can help others (human or otherwise) on their fine-tuning journeys.
Featured
Multistep Reasoning Agents (with GRPO & RLEF) - Project Euler Edition
Published: at 02:38 AM in 21 min readConverting chat LLMs into reasoning agents through GRPO and reinforcement learning from execution feedback — achieving performance improvements on Project Euler's algorithmic challenges with multi-step reasoning, tool use, and code execution.
Bridging the Knowledge Gap in Multimodal AI with DeepResearch
Published: at 04:27 AM in 19 min readHow OpenAI's DeepResearch reveals what industry labs aren't sharing about next-gen AI assistants
o3 and the Future of Science
Published: at 10:04 PM in 23 min reado3 and Human-AI scientific collaboration
Back to Backprop
Published: at 06:42 AM in 29 min readA review of backpropagation, the workhorse of deep learning.
Can AI Write about AI Personality With Personality?
Published: at 07:33 AM in 31 min readShould AI have "Personality" or will "Bland and Mid" suffice?
o1 and Reasoning
Published: at 12:53 AM in 62 min readOrganizing research presumably related to OpenAI's o1 reasoning model.
All you need (to know) about attention
Published: at 02:35 AM in 35 min readA highly compressed guide to quickly refresh (mostly) all you need to know about attention mechanisms in deep learning.
Recent Posts
Swirling Thoughts on AI, LLMs, Novel Hypothesis Generation and Crypto?
Published: at 09:45 PM in 14 min readIn this post, I swirl my own thoughts on the very topic of swirling thoughts in the context of AI, LLMs and their potential for novel hypothesis generation. I ground this discussion in some of the other ongoing parallel lines of thoughts/investigation including GRPO, reinforcement learning, the future of science and scientific peer review and the notion of being able to "purchase" scientific innovation.
Pandas Primer
Published: at 01:15 AM in 5 min readA rapid review of the Pandas library for the impatient data scientist.
Python and Pandas and P-values, Oh My! A Statistical Journey
Published: at 01:15 AM in 26 min readA rapid refresher on the basics of statistics and how to apply them in Python.
A Straight Line Through Linear Algebra
Published: at 07:27 PM in 9 min readA brief review of core concepts in linear algebra, with a focus on applications in machine learning.