We develop methods, benchmarks, and training systems that turn expert data into frontier AI

Curating training data is among the most consequential yet labor-intensive parts of modern AI development: practitioners iteratively propose, implement, evaluate, and revise data policies against noisy benchmark feedback. We ask whether generalist coding agents can automate this data-curation loop. We introduce CURATION-BENCH, an agent-centric benchmark that fixes the model, training recipe, and evaluation suite while giving agents commandline access to…

Recent AI systems have achieved strong results on a wide range of benchmarks, yetthese gains have not translated into economically meaningful deployment acrossmany professional domains. We argue that this gap is largely an evaluation problem:widely used benchmarks lack sustained performance measurement on real andeconomically valuable workflows. This paper introduces Agents’ Last Exam(ALE), a benchmark designed to evaluate AI agents on…

At our latest Snorkel AI Reading Group, Yijia Shao (Stanford NLP) stopped by our San Francisco office to present Collaborative Gym: A Framework for Enabling and Evaluating Human-Agent Collaboration. As LLM agents get better at automating tasks on their own, a large class of real-world problems still needs a human in the loop – for their preferences, their domain expertise, or simply for control….

For our second Benchtalks, the series dedicated to the researchers building the measurement toolkits that frontier labs hill-climb on, Snorkel AI co-founder Vincent Sunn Chen sat down with John Yang, a Stanford PhD student and creator of the SWE-bench franchise, SWE-smith, CodeClash, and most recently ProgramBench. Highlights More on ProgramBench: See the benchmark and the upcoming leaderboard at programbench.com. More from John Yang: Publications and writing at john-b-yang.github.io. Snorkel…

Two methodologies dominate current practices of benchmarking: rubric-based scoring evaluates items against predefined criteria, whereas comparative judgment elicits pairwise preferences between outputs. Although both methodologies are widely used, the choice between them is rarely justified. We release JudgmentBench, a benchmark of 30 real-world legal tasks, paired with 1,539 rubric scores and 1,530 pairwise preference judgments collected from practicing attorneys–including at…

At our latest Snorkel AI Reading Group, Carter Wendelken of Google DeepMind walked us through two related papers he presented at ICLR: Code World Models for General Game Playing and AutoHarness: Improving LLM Agents by Automatically Synthesizing a Code Harness. Both ask the same question from opposite ends: when you want an LLM to act reliably in a complex, possibly…

Coding agents have moved from tab-complete to teammate. They autonomously inspect repositories, edit files, run commands, diagnose failures, and work through multi-step engineering tasks. That creates a harder reliability problem. A model that only suggests code is easy for a human to evaluate. A coding agent refactoring your repository and testing its own changes is much harder to supervise –…

At our latest Snorkel AI Reading Group, Mayee Chen (Stanford, Hazy Research) stopped by our San Francisco office to walk us through Olmix: A Framework for Data Mixing Throughout LM Development — work she contributed to during her internship at Ai2 on OLMo 3. Olmix tackles one of the messiest, least-documented levers in LLM pre-training: how to set the ratios…

Fine-tuning Large Language Models (LLMs) typically relies on large quantities of high-quality annotated data, or questions with well-defined ground truth answers in the case of Reinforcement Learning with Verifiable Rewards (RLVR). While previous work has explored the benefits to model reasoning capabilities by scaling both data and compute used for RLVR, these results lack applicability in many real-world settings where…