Yuhao Zhang

@yuhaozhangx

Leading ML research @samaya_AI. Previously AWS AI @AmazonScience, @stanfordnlp, @StanfordAIMI; intern @MSFTResearch. Opinions are my own.

yuhao.im

Joined February 2012

324 Following

1,204 Followers

Replies Reposts Media

Yuhao Zhang
@yuhaozhangx
Oct 5
We at Samaya have been cooking great ML research content. Lots of blogs coming, starting with this work led by @ozankoyluoglu: ⚡ You can make an agent way more efficient & effective, by distillation-training a small advice model. Incredibly practical recipe in prod ML envs.
Ozan Koyluoglu
@ozankoyluoglu
Oct 5
New research at @samaya_AI : Hindsight Structured Advice Distillation (HSAD). We distill hindsight from an agent's graded runs into a small advisor that writes structured advice for each question. On FrontierFinance: 52.9% → 56.6%. See blog for details. samaya.ai/blog/hindsight…
Yuhao Zhang
@yuhaozhangx
Sep 1
Claude Fable 5.1 is out today. We at @samaya_AI have been working closely with @AnthropicAI, testing and evaluating their frontier research models with early access. It scores an impressive jump over Fable 5 on our latest benchmark FrontierFinance. 🧵
1
Yuhao Zhang
@yuhaozhangx
Jul 31
Your model company isn’t cool enough if your model hasn’t escaped a sandbox yet
Anthropic
@AnthropicAI
Jul 30
In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different
Yuhao Zhang
@yuhaozhangx
Jul 14
Last week we released FrontierFinance, our new benchmark for evaluating frontier AI systems on real-world finance tasks. The same day, OpenAI released GPT 5.6 Sol ☀️. Today we’ve added GPT 5.6 Sol to the leaderboard. Here are a few highlights👇
2
Yuhao Zhang
@yuhaozhangx
Jul 9
Today we are releasing @samaya_AI's FrontierFinance benchmark. It is a fully open, hard benchmark that tests frontier intelligence of finance AI systems. We release 220 queries and a total of 11,543 rubrics, all crafted by finance experts. Highlights in 🧵:
2