←
Ran Xu58 posts
Ran Xu avatar

Ran Xu

@ritaranx

Research Scientist @GoogleDeepMind | Prev: CS PhD @EmoryUniversity | Gemini Agentic Post-training | CyberSecurity

◷ Joined September 2022▧ 15 Photos and videos
Ran Xu avatar
Ran Xu@ritaranx·Sep 2

So excited to finally share what our team has been building! 🚀 Meet Gemini 3.8 Flash and Gemini 3.8 Flash Cyber: ⚡Massive leaps in long-horizon coding & agentic tasks 🛡️Frontier-level automated vulnerability detection & patching with Flash speed/pricing

Introducing Gemini 3.8, our best reasoning & coding model yet. By leveraging long-running agentic loops, we’re building on the momentum of 3.7 Flash from just three weeks ago to release two new 3.8 variants: Meet Gemini 3.8 Flash and Gemini 3.8 Flash Cyber 🧵

Post attachment
Yue Yu avatar
⟳ Ran Xu retweeted
Yue Yu@yue___yu·Oct 12

Wonder is here, and it's looking Wonder Full! 🍉 We've worked hard to push its intelligence and real-world usefulness, especially for coding, agentic and personal superintelligence. Proud to be part of this journey. More wonders ahead!

Introducing Muse Wonder, our new family of frontier models with state of the art performance for personal and professional agentic work.

Post attachment
Raluca Ada Popa avatar
⟳ Ran Xu retweeted
Raluca Ada Popa@ralucaadapopa·Sep 30

Today, we announced Gemini 4 Argon, our next generation model and a cybersecurity powerhouse equipping defenders for the next era of cyberattack. On the Gray Swan IPI Benchmark, Gemini 4 Argon leads in prompt injection robustness. @GoogleDeepMind

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·Sep 30

Excited to bring Gemini 4 Argon! Incredibly strong across general reasoning, agents, coding, and cybersecurity. Great time to be working on cyber defense and turning frontier capabilities into real-world vulnerability discovery and remediation. 🚀

Introducing Gemini 4 Argon – our new frontier model. It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.

Post attachment
koray kavukcuoglu avatar
⟳ Ran Xu retweeted
koray kavukcuoglu@koraykv·Sep 2

Today we’re launching Gemini 3.8 Flash Cyber and Gemini 3.8 Flash. ⚡️🛡️ 3.8 Flash Cyber is our most capable cybersecurity model for finding and fixing vulnerabilities. It sits on the Pareto Frontier on CWE-Bench for patching. Available to trusted defenders through our new Fairwind Program.

Post attachment
Demis Hassabis avatar
⟳ Ran Xu retweeted
Demis Hassabis@demishassabis·Sep 2

Introducing Gemini 3.8 Flash, another upgrade in under a month! And the new 3.8 Flash Cyber pushes the frontier of cyber defense. Relentless progress 🚀 More on the models & our cyber security work in the blog - happy building! blog.google/innovation-and-a…

Introducing Gemini 3.8 Flash, another jump in Gemini's agentic + coding capabilities, and our 3rd updated Flash model in only 6 weeks... This model has been a ton of fun to work with, excited to see what you all think!

Post attachmentPost attachmentPost attachmentPost attachment
Raluca Ada Popa avatar
⟳ Ran Xu retweeted
Raluca Ada Popa@ralucaadapopa·Sep 2

Excited to announce Gemini 3.8 Flash and 3.8 Flash Cyber @google blog.google/innovation-and-a… Our cyber model has frontier capabilities being competitive on finding and patching benchmarks with much larger models, at a cheaper cost. @GoogleDeepMind

Yue Yu avatar
⟳ Ran Xu retweeted
Yue Yu@yue___yu·Aug 5

Muse Spark 1.2 and MuseCode are out—less than a month after Muse Spark 1.1! 🚀 Excited to join the journey and see the pace of progress. Give it a try and let us know what you build! curl -fsS dev.meta.ai/install.sh | bash

muse code in beta is live. first coding agent from msl, built on muse spark 1.2. install: curl -fsS dev.meta.ai/install.sh | bash here's what you should know:

Post attachmentPost attachment
Yue Yu avatar
⟳ Ran Xu retweeted
Yue Yu@yue___yu·Jul 9

Muse Spark 1.1 🥑 is out! with lot of improvements on agents / reasoning / coding. Scaling RL has been tiring, rewarding, and full of lessons. Try it on dev.meta.ai and please send feedback + failure cases. They help make the model better. More to come!

1/ muse spark 1.1 is an industry-competitive agentic and coding model. across many agentic evals it rivals gpt-5.5 and opus-4.8. available now through the new meta model api and in meta ai. 🧵

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·3 Dec 2025

Excited to be at #NeurIPS through Dec 8 — happy to connect! I’ll be presenting our Spotlight paper on complex QA and reasoning with search: 🗓️ Dec 5, 11:00–2:00pm PST 📍 Exhibit C/D/E — Poster #1908 Also exploring full-time opportunities—DMs open if you’d like to chat!

Ran Xu@ritaranx2 Oct 2025

🚨 Happy to share AceSearcher accepted to #NeurIPS2025 #Spotlight! 🔹 One LLM, two roles: Decomposer (split queries) + Solver (combine context) 🔹 +7.6% on QA & fact verification 🔹 32B ≈ DeepSeek-V3 on DocMath 📂 Code: github.com/ritaranx/AceSearc… 📑 arXiv: arxiv.org/abs/2509.24193

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·10 Nov 2025

Thanks for featuring our work! 🙌

DAIR.AI@dair_ai9 Nov 2025

4. TIR-Judge Google and collaborators introduce TIR-Judge, an end-to-end reinforcement learning framework that trains LLM judges to integrate code execution for precise evaluation.

Ran Xu avatar
Ran Xu@ritaranx·6 Nov 2025

Thanks for sharing our work on improving LLM judges with agentic RL!

New Google paper trains LLM judges to use small bits of code alongside reasoning, so their decisions become precise. So judging stops being guesswork and becomes checkable. Text only judges often miscount, miss structure rules, or accept shaky logic that a simple program would catch. TIR-Judge makes the judge think step by step, write code to check claims, run it in a sandbox, then update the verdict. Training mixes tasks where code can verify answers and tasks where it cannot, so the judge learns when to call tools and when to rely on reasoning. One prompt schema covers pointwise scoring, pairwise choices, and listwise selection, so it plugs into many workflows. Reinforcement learning rewards being correct, following strict output tags, and using at most 3 tool calls. A variant called TIR-Judge-Zero skips teacher distillation and still improves by alternating reinforcement learning, rejection sampling, and supervised fine tuning. Across public judge benchmarks it beats text only judges, and with 8B it reaches 96% of Claude Opus 4 on listwise ranking. The core idea, give the judge verifiable checks plus rewards that favor careful tool use. ---- Paper – arxiv. org/abs/2510.23038 Paper Title: "Incentivizing Agentic Reasoning in LLM Judges via Tool-Integrated Reinforcement Learning"

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·5 Nov 2025

6/n 📊Best-of-N on Policy Models TIR-Judge is not only a better judge — it makes other models better. When selecting responses in best-of-N inference, TIR-Judge improves policy accuracy by +3.9~6.7% on AIME, BigCodeBench, IFEval, etc. → Better downstream reasoning too🎯

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·5 Nov 2025

5/n 📊 Results across 7 benchmarks ✅ +6.4 % (pointwise) / +7.7 % (pairwise) ✅ 96 % of Claude-Opus-4 (listwise) ✅ Only 8 B params, 26 K pairs Tool-integrated RL → smarter verifiable judges.

Post attachment
Ran Xu avatar
Ran Xu@ritaranx·5 Nov 2025

4/n 🧬 Two variants 🔹 TIR-Judge-Distill: starts from Gemini-2.5-Flash-Tool 🔹 TIR-Judge-Zero: trained from initial checkpoint via iterative RL → Zero matches or surpasses Distill — RL alone can teach tool-use! 🤯