Data science, career and other topics
09 June 2026
A review of Mike X Cohen 50 ML Projects To Understand LLMs, a hands-on book that uses code, statistics, and controlled experiments to investigate what happens inside GPT-2 — covering what it does well, who it is for, and a few small nitpicks.
01 June 2026
A review of Gamma-World, NVIDIA's generative multi-agent world model that produces shared, action-controllable video rollouts for multiple independently acting agents. It places agents at the vertices of a regular simplex in rotary space for permutation symmetry, routes cross-agent interaction through learnable hub tokens to keep attention linear in the number of agents, and distills a diffusion teacher into a causal student that streams at 24 FPS.
18 May 2026
An evaluation of MiniMax M2.7 used through Claude Code on three workflows I run regularly — writing code for a Kaggle competition submission, drafting and auditing Obsidian vault notes, and refactoring an old PyTorch project — with Claude Opus 4.7 as the comparison baseline.
24 April 2026
DeepSeek V4 pairs a hybrid sparse-attention stack with on-policy distillation across domain specialists to bring 1M-token inference to frontier quality at a fraction of the FLOPs and KV cache of its predecessor.
20 April 2026
FIPO - an RL algorithm that fixes one of the core limitations of RL for LLM reasoning - credit assignment. Instead of giving every token in a rollout the same outcome advantage, it re-weights tokens by a discounted future-KL signal, enabling longer and more effective reasoning chains.
09 April 2026
A review of Keith Bourne second edition of Unlocking Data with Generative AI and RAG, covering the running example that ties all 20 chapters together, explanations of agentic RAG, and where the book is most useful in practice.
Type at least 2 characters to search...