⚡️ Reverse Engineering OpenAI's Training Data — Pratyush Maini, Datology

Training data composition drives model behavior. Work that infers what frontier models were trained on helps engineers reason about capability gaps and evaluation contamination.

• attention — The mechanism that lets a model weigh which earlier words matter for the word it's currently processing — the core operation of a transformer.
• eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.

• videoWhy Every Agent needs Open Source Cloud Sandboxes
• videoTerminal-Bench 2.0: the most impt coding agent benchmark of 2025 gets a v2! Launch + Q&A w/ founders
• video[State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI
• videoThe Agent-Native Cloud: 3M Users, 100K Signups/Wk, Data Centers, & Death PRs — Jake Cooper, Railway

• videoBetter Data is All You Need — Ari Morcos, DatologyLatent Space
• videoData Quality Is the Compute Multiplier — Ari Morcos, DatologyAIAI Engineer
• videoData and Environment Curation for Post-Training LLMs — Mahesh Sathiamoorthy, Bespoke LabsAI Engineer

Checking sign-in…

Loading comments…