https://amks.me/ Recent writing by Arushi Somani en Sun, 11 Oct 2026 12:00:00 -0700 https://amks.me/writing/bouldering/ Sun, 11 Oct 2026 12:00:00 -0700 https://amks.me/writing/bouldering/ https://amks.me/writing/movies25/ Mon, 29 Dec 2025 12:00:00 -0800 https://amks.me/writing/movies25/ https://amks.me/teaching/algos/ Sat, 22 Nov 2025 00:07:27 -0800 https://amks.me/teaching/algos/ Modern RL algorithms are a chain of fixes where each one solves the most painful problem of the last. This article explores that story, with some math involved. https://amks.me/teaching/kl/ Sat, 08 Nov 2025 01:04:46 -0800 https://amks.me/teaching/kl/ An intuitive, example-driven introduction to KL divergence, explaining its connection to surprise, entropy, log-likelihood, forward vs. reverse KL behavior, and practical KL estimation in RLHF. https://amks.me/writing/hadestown/ Sun, 19 Jan 2025 10:50:07 -0800 https://amks.me/writing/hadestown/ https://amks.me/writing/books24/ Sat, 28 Dec 2024 21:00:20 -0800 https://amks.me/writing/books24/ Fantasy, science fiction, romantic comedies, some roasts and recommendations along the way. https://amks.me/writing/optimize/ Fri, 22 Nov 2024 14:59:36 -0600 https://amks.me/writing/optimize/ The curious case of the robot that, in a fit of existential crisis, decided that sweeping corridors was far too mundane a task to be done without a touch of flair. https://amks.me/teaching/taxonomy/ Tue, 29 Oct 2024 15:55:00 -0700 https://amks.me/teaching/taxonomy/ A guide to understanding and categorizing the many flavors of reinforcement learning algorithms, from value iteration to PPO. https://amks.me/teaching/rlsteve/ Sun, 20 Oct 2024 10:00:00 -0700 https://amks.me/teaching/rlsteve/ Notes on reinforcement learning from Steve Brunton's Data-Driven Science and Engineering book, covering core RL concepts, mathematical formalism, and key ideas https://amks.me/teaching/pp/ Sat, 20 Jul 2024 17:35:00 -0700 https://amks.me/teaching/pp/ Pipeline parallelism is a technique for training large ML models, showing how to efficiently partition model layers across devices to optimize distributed training and manage memory constraints. https://amks.me/teaching/mcts/ Sun, 07 Jul 2024 17:35:00 -0700 https://amks.me/teaching/mcts/ Monte Carlo Tree Search (MCTS) is AI algorithm that makes decisions by strategically sampling possible futures. It builds search trees incrementally, balancing exploration of new paths with exploitation of promising ones, and uses random simulations to tackle problems too complex for exhaustive analysis https://amks.me/writing/intl/ Sat, 25 May 2024 00:00:00 -0700 https://amks.me/writing/intl/ https://amks.me/teaching/tp/ Sun, 24 Mar 2024 17:35:00 -0700 https://amks.me/teaching/tp/ Tensor Parallelism is a technique for training large ML models by splitting individual tensors across multiple devices, enabling efficient distributed training of models too large to fit on a single accelerator. https://amks.me/teaching/dp/ Tue, 05 Mar 2024 18:37:00 -0600 https://amks.me/teaching/dp/ Data Parallelism is a technique for training large ML models by distributing data across multiple devices, enabling parallel processing while maintaining model consistency through gradient synchronization. https://amks.me/teaching/moe/ Sat, 03 Feb 2024 15:55:00 -0700 https://amks.me/teaching/moe/ A deep dive into Mixture of Expert (MoE) models, exploring how they work, their benefits and challenges, and their role in modern language models like Mixtral. https://amks.me/teaching/archive/ Fri, 07 Jul 2023 17:35:00 -0700 https://amks.me/teaching/archive/ The Daily Ink is a discontinued newsletter featuring bi-weekly breakdowns of research papers in the domain of machine learning and ML systems. This is an archive of all the articles. https://amks.me/writing/synthesis/ Thu, 23 Mar 2023 18:37:00 -0600 https://amks.me/writing/synthesis/ An analysis of current program synthesis tools and what we need to build better automated programmers https://amks.me/teaching/eecs126/ Tue, 11 May 2021 10:30:30 -0700 https://amks.me/teaching/eecs126/ A comprehensive overview of key concepts and theorems in Probability and Random Processes, covering essential topics such as conditional probability, Bayes' theorem, independence, and counting principles, serving as a valuable reference for students in EECS126.