A Cheap Verifier is Good Enough: LLM Post-training is Robust to Erroneous Rewards
cs.LG, cs.AI, cs.CL
Submitted: 2026-09-27
Updated: 2026-09-27
Code: https://github.com/THUDM/slime
Terminology
Sources
- PRBench: Large-Scale Expert Rubrics for Evaluating High-Stakes Professional Reasoning
- HealthBench: Evaluating Large Language Models Towards Improved Human Health
- Program Synthesis with Large Language Models
- Reinforcement Learning with Verifiable yet Noisy Rewards under Imperfect Verifiers
- Exploration vs Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
- Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains
- AdvancedIF: Rubric-Based Benchmarking and Reinforcement Learning for Advancing LLM Instruction Following
- Kimi K3: Open Frontier Intelligence
- BankerToolBench: Evaluating AI Agents in End-to-End Investment Banking Workflows
- Examining Reasoning LLMs-as-Judges in Non-Verifiable LLM Post-Training
- Noise-corrected GRPO: From Noisy Rewards to Unbiased Gradients
- RubricEval: A Rubric-Level Meta-Evaluation Benchmark for LLM Judges in Instruction Following
- GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks
- An Imperfect Verifier is Good Enough: Learning with Noisy Rewards
- Rate or Fate? RLV$^\varepsilon$R: Reinforcement Learning with Verifiable Noisy Rewards
- Spurious Rewards: Rethinking Training Signals in RLVR
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Crossing the Reward Bridge: Expanding RL with Verifiable Rewards Across Diverse Domains
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks