Scale AI reposted this
Excited to share that we have 5 papers accepted to ICLR 2026 from Scale AI! 🎉 These works span Agentic RL, LLM evaluation, and AI safety, as we continue pushing the frontier of eval-driven RL, rubric-based rewards, and agentic capabilities. 📄 Accepted papers: 1. Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains: https://lnkd.in/gUshMQNu 2. Chasing the Tail: Effective Rubric-based Reward Modeling for Large Language Model Post-Training: https://lnkd.in/gntq9YZx 3. ResearchRubrics: A Benchmark of Prompts and Rubrics For Evaluating Deep Research Agents: https://lnkd.in/gNsZVkaE 4. MoReBench: Evaluating Procedural and Pluralistic Moral Reasoning in Language Models, More than Outcomes: https://lnkd.in/gSMcXHpy 5. PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approach: https://lnkd.in/gfNWidnv 🚀 We’re also hiring Research Leads If you’re excited about shaping the next generation of: • Evaluation for frontier models • Agentic RL for long-horizon tasks • ML systems for RL training and eval Feel free to DM me or comment below! #ICLR2026 #AIresearch #LLMEval #AgenticRL