Twitxy Saved

AfterQuery

Verified account This account is verified because it's an official organization on X. Learn more

@AfterQuery

Applied research lab curating data solutions to accelerate foundation model development.
afterquery.com Joined February 2025
11 Following 4.1K Followers 111 Posts

Latest posts 5

AfterQuery Verified account @AfterQuery
We added support for SDPO in SkyRL alongside UC Berkeley SkyLab researchers. Our new blog explores self-distillation across math, document, and chart-to-code tasks.

On MATH-500, Qwen3-8B improved from 85.6% to 91.2% (+5.6 points), while using 8× fewer rollouts than GRPO. afterquery.com/blog/on-policy…
www.afterquery.com Building Support for On-Policy Self-Distillation in SkyRL - AfterQuery With one sampled answer per prompt and a reward from the verifier, Qwen3-8B trained with SDPO reached 91.2% on MATH-500 after 6,400 responses, compared with 91.8% after 51,200 responses in our GRPO...
9 replies 9 reposts 55 likes 9.3K views View on X
AfterQuery Verified account @AfterQuery
Training data is increasingly produced by systems where humans and models work together. That means data labs must be able to show exactly what produced their data and why it should be trusted.

Today we're publishing our closed-model policy, how we enforce it, and a call for responsible training data production. afterquery.com/blog/a-call-fo…
www.afterquery.com A Call for Responsible Training Data Production - AfterQuery The way we produce training data is changing. Instead of authoring examples end-to-end, experts increasingly contribute to systems in which models perform some of the work.
10 replies 11 reposts 101 likes 19K views View on X
AfterQuery Verified account @AfterQuery
Today we're releasing CrossSearch, a new benchmark testing whether AI agents can do something people do at work every day: look through their files, search the internet, and reason through ambiguity.

Unlike benchmarks that test web search or file understanding separately, CrossSearch tests whether agents can do both together to solve real-world problems.

Across 100 tasks and 14 models, Gemini 4 Argon leads at just 64% pass³.

Full blog, leaderboard, and dataset in the replies.
0:08
38 replies 8 reposts 84 likes 4.3K views View on X
AfterQuery Verified account @AfterQuery
Can AI agents run a business as conditions change?

Introducing StoreBench, a high-fidelity RL environment where agents manage a simulated online apparel store, with safeguards against reward hacking.

StoreBench includes 12 evaluation tasks spanning 30–365 simulated days. Across the 11 tasks lasting 30–45 days, all seven models score lower on average than human participants and smart-triage, a rule-based program that prioritizes urgent work.

Late restocking, unresolved refunds, and decisions based on outdated information show that sustained operations remain challenging for AI agents.

Check out the comments to read more.
0:12