paper

Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning

paperactiveprovisionallearning-to-reason-by-analogy-via-retrieval-augmented-reinforcement-fine-tuning-27c27d10·1 events·first seen 5d ago

Aliases: Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning

Co-occurring entities

RA-RFT GRPO Qwen3-4B AIME 2025 Qwen3-1.7B

More like this (12)

Reasoning Enhancement Reference-Augmented Training Reasoning Imitation reinforcement fine-tuning Reasoning Language Models curiosity-driven reinforcement learning Long-context Reasoning Benchmarks When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models Chain-of-Thought Fine-Tuning Predicting Future Behaviors in Reasoning Models Enables Better Steering Adaptive Parallel Reasoning Difference-Aware Retrieval Policies for Imitation Learning

Recent events (1)

6arXiv · cs.AI·5d ago·source ↗

RA-RFT: Retrieval-Augmented Reinforcement Fine-Tuning teaches LLMs to reason by analogy

Researchers propose Retrieval-Augmented Reinforcement Fine-Tuning (RA-RFT), a post-training framework that trains a retriever to rank contexts by expected reasoning benefit rather than semantic similarity, then fine-tunes a policy model via reinforcement learning using retrieved analogous demonstrations. The key insight is that reasoning-relevant retrieval surfaces complementary solution strategies rather than superficially similar problems. On mathematical reasoning benchmarks, RA-RFT improves AIME 2025 average@32 accuracy by 7.1 and 2.8 points over GRPO for Qwen3-1.7B and Qwen3-4B respectively, suggesting reasoning-aware retrieval is orthogonal to reward design and training curriculum improvements.

Evaluation and Benchmarking Alignment and RLHF RA-RFT Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning GRPO +3 more