Twitxy Saved

Kyle Montgomery

Verified account This account is verified. Learn more

@kylepmont

PhD student at UC Santa Cruz
Santa Cruz, CA Joined January 2016
49 Following 61 Followers 25 Posts

Latest posts 4

Kyle Montgomery Verified account @kylepmont
🚨 New preprint: Budget-aware Test-time Scaling via Discriminative Verification 👉 arxiv.org/pdf/2510.14913

We show that discriminative verification is the best option for test-time scaling under 25.5 minutes, outperforming state-of-the-art generative verification in both accuracy and efficiency, for example achieving up to +15.3% and +2.8% higher accuracy on AIME2025 at latency budgets of 13.8 min and 15.7 min, respectively.

🧠 Blog: cedar-baryonyx-84b.notion.site/Budget-aware-T…
💻 Code: github.com/wang-research-…
🤗 HuggingFace (data/models): huggingface.co/collections/Wa…
3 replies 3 reposts 9 likes 2.4K views View on X
Kyle Montgomery Verified account @kylepmont
Thrilled to have been a part of this release — looking forward to what’s coming next with rLLM!
rLLM Verified account @rllm_project Oct 17, 2025
🚀 Introducing rLLM v0.2 - train arbitrary agentic programs with RL, with minimal code changes.

Most RL training systems adopt the agent-environment abstraction. But what about complex workflows? Think solver-critique pairs collaborating, or planner agents orchestrating multiple workers. These are hard to express with traditional RL abstractions.

v0.2 introduces AgentWorkflowTrainer, built on a simple insight: any agentic flow is just a Python program orchestrating LLM calls, so we made ANY Python program trainable.

Researchers and developers can now quickly prototype new ideas or transform their production agentic systems into trainable flows with minimal changes.

rLLM now uses official @verl_project ==0.5.0 as our backend (no more custom verl forks!). Just define your agentic workflow or multi-agent system as a Python program and hit train, and rLLM will handle the rest.

Since release, rLLM has been adopted to power RL training of world-class agents like @Ali_TongyiLab's DeepResearcher.

With this new release, we're working towards building the RL application stack for next-gen agentic AI - where entire systems learn and evolve together, not just individual components in isolation.

📖 Blog post: rllm-project.com/post.html?post…
👨‍💻 GitHub: github.com/rllm-org/rllm

What agentic program will you train first? 👀
View quoted post on X
Kyle Montgomery Verified account @kylepmont
Excited to share our work at #ICLR2025! JudgeBench ⚖️ tests the reliability of LLM-based judges with a focus on objective correctness. JudgeBench converts tough 🧠 datasets in knowledge, reasoning, math & code into labeled response pairs, forcing objective grading over vibes. Even strong models like GPT-4o barely beat random guessing 🎲. Swing by📍Poster #227 (Session 4) to see how your favorite model fares. Huge thanks 🙏 to @sijun_tan @SiyuanZhuang3 William Tang @ChenguangWang @ralucaadapopa & Ion Stoica!