REVIEW 7 cited by
Flow of Reasoning: Training LLMs for Divergent Reasoning with Minimal Examples
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
The ability to generate diverse solutions to a given problem is a hallmark of human creativity. This divergent reasoning is also crucial for machines, enhancing their robustness and enabling them to assist humans in many applications such as scientific discovery. However, existing approaches to multi-step reasoning with large language models (LLMs) have mostly focused only on reasoning accuracy, without further discovering more diverse valid solutions. For example, supervised fine-tuning improves reasoning quality but requires vast labeled data, while reward-maximizing reinforcement learning finds top-reward solutions while neglecting the solution diversity. To fill this gap, we propose Flow of Reasoning (FoR), an efficient diversity-seeking LLM finetuning method aimed at improving reasoning quality and diversity with minimal data. FoR formulates multi-step LLM reasoning as a Markovian flow on a DAG-structured reasoning graph. This formulation allows us to incorporate and adapt principled GFlowNet approaches, for finetuning LLMs to sample divergent paths with probabilities proportional to the (unnormalized) reward of target problems. Extensive experiments show that, with limited training examples (e.g., 15 examples), FoR enables the discovery of diverse, creative, high-quality solutions, greatly outperforming a wide range of existing inference and training methods across six challenging reasoning tasks, including BlocksWorld (embodied reasoning), Game24 (math puzzle solving), Rubik's Cube (spatial reasoning), 1D-ARC (abstraction reasoning), GSM8k (math reasoning), and ProntoQA (logical reasoning). Code is available at https://github.com/Yu-Fangxu/FoR.
Forward citations
Cited by 7 Pith papers
-
TSRouter: Dynamic Modality-Model Selection for Time Series Reasoning
A heterogeneous-graph router jointly selects the optimal modality (text, vision, or both) and model per time series query, beating prior routing baselines and generalizing to unseen models and tasks.
-
Weak-to-Strong On-Policy Distillation
A strong LLM is improved by distilling from the logit difference of two weaker models instead of from a stronger teacher.
-
GFlowGR: Fine-tuning Generative Recommendation Frameworks with Generative Flow Networks
GFlowGR fine-tunes generative recommender LLMs with GFlowNet losses and multi-signal rewards, beating SFT, DPO, and GRPO baselines on three datasets and in production.
-
Beyond the Proxy: Trajectory-Distilled Guidance for Offline GFlowNet Training
TD-GFN uses IRL-derived edge rewards to prune the environment DAG and sample backward trajectories, training offline GFlowNets directly from ground-truth terminal rewards without a proxy reward model.
-
BAR: A Backward Reasoning based Agent for Complex Minecraft Tasks
A backward-reasoning LLM agent with recursive goal decomposition, state-consistency repair, and stage memory outperforms forward-planning baselines on Minecraft task planning.
-
Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space
A training-free method that feeds probability-weighted token embeddings back into LLMs during reasoning, improving accuracy and token efficiency on math and coding tasks.
-
SCOUT: Teaching Pre-trained Language Models to Enhance Reasoning via Flow Chain-of-Thought
SCOUT combines progressive distillation with a cross-attention module to make recursive latent reasoning work through fine-tuning, yielding up to 1.8% accuracy gains over standard fine-tuning.
Discussion (0). Continue with ORCID to comment.