Rewarding Reasoning, Not Answers: Fixing and Bounding Test-Time Reinforcement Learning on Medical QA

arXiv:2609.16660 · cs.CL · Submitted 2026-09-15 · Read on arXiv

cs.CL

Submitted: 2026-09-15

Updated: 2026-09-15

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Accuracy — 30× in this paper · explained in 1 episode
• Adaptation — 22× in this paper · explained in 1 episode
• Aggregation — 20× in this paper · explained in 1 episode
• Reinforcement Learning — 8× in this paper · explained in 4 episodes
• GRPO — 7× in this paper · explained in 6 episodes
• Reward Hacking — 5× in this paper · explained in 8 episodes
• Generalization — 4× in this paper · explained in 13 episodes
• Verifier — 4× in this paper · explained in 2 episodes
• Consistency — 3× in this paper · explained in 3 episodes
• Pseudo-labels — 3× in this paper · explained in 2 episodes
• Self-consistency — 3× in this paper · explained in 2 episodes
• Test-time Scaling — 3× in this paper · explained in 2 episodes
• Trace — 3× in this paper · explained in 2 episodes
• Ablation — 2× in this paper · explained in 1 episode
• Benchmark — 2× in this paper · explained in 2 episodes
• Verification — 2× in this paper · explained in 2 episodes
• Grounding — 1× in this paper · explained in 3 episodes
• Process Reward Model (PRM) — 1× in this paper · explained in 3 episodes
• RAG — 1× in this paper · explained in 3 episodes
• Reasoning Traces — 1× in this paper · explained in 3 episodes
• Scaling Laws — 1× in this paper · explained in 4 episodes
• Sparsity — 1× in this paper · explained in 4 episodes

Sources

• Concrete Problems in AI Safety
• HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs
• The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
• The Llama 3 Herd of Models
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• m1: Unleash the Potential of Test-Time Scaling for Medical Reasoning with Large Language Models
• OpenAI o1 System Card
• Rethinking Reward Models for Multi-Domain Test-Time Scaling
• Let's Verify Step by Step
• ETTRL: Balancing Exploration and Exploitation in LLM Test-Time Reinforcement Learning Via Entropy Mechanism
• Continuous Self-Improvement of Large Language Models by Test-time Training with Verifier-Driven Sample Selection
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Defining and Characterizing Reward Hacking
• Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
• Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
• Solving math word problems with process- and outcome-based feedback
• Self-Consistency Improves Chain of Thought Reasoning in Language Models
• Qwen3 Technical Report
• TTCS: Test-Time Curriculum Synthesis for Self-Evolving
• Curing Miracle Steps in LLM Mathematical Reasoning with Rubric Rewards

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering