Pith. sign in

REVIEW 40 cited by

STaR: Bootstrapping Reasoning With Reasoning

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

arxiv 2203.14465 v2 pith:4RY3WEJC submitted 2022-03-28 cs.LG cs.AIcs.CL

classification cs.LGcs.AIcs.CL
keywords modelrationalereasoningrationalesstaranswerslanguageanswer
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Generating step-by-step "chain-of-thought" rationales improves language model performance on complex reasoning tasks like mathematics or commonsense question-answering. However, inducing language model rationale generation currently requires either constructing massive rationale datasets or sacrificing accuracy by using only few-shot inference. We propose a technique to iteratively leverage a small number of rationale examples and a large dataset without rationales, to bootstrap the ability to perform successively more complex reasoning. This technique, the "Self-Taught Reasoner" (STaR), relies on a simple loop: generate rationales to answer many questions, prompted with a few rationale examples; if the generated answers are wrong, try again to generate a rationale given the correct answer; fine-tune on all the rationales that ultimately yielded correct answers; repeat. We show that STaR significantly improves performance on multiple datasets compared to a model fine-tuned to directly predict final answers, and performs comparably to fine-tuning a 30$\times$ larger state-of-the-art language model on CommensenseQA. Thus, STaR lets a model improve itself by learning from its own generated reasoning.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 40 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 116 citations worldwide. Full citation record

  1. The Calibration Floor: Format Repair Can Masquerade as Self-Correction at Small-to-Mid Scale

    cs.CL 2026-08 conditional novelty 7.0 of 10

    Across 0.8B-12B models plus a frontier arm, apparent self-correction effects are dominated by format-recovery and format-loss artifacts, with near-zero content-level change at capable scale.

  2. RL Post-Training Builds Compositional Reasoning Strategies

    cs.AI 2026-07 conditional novelty 7.0 of 10

    RL post-training composes primitive rewrite skills into reusable macro and parallel contraction strategies that solve problems inaccessible to the base model under large sampling budgets.

  3. TRIAGE: Role-Typed Credit Assignment for Agentic Reinforcement Learning

    cs.LG 2026-06 unverdicted novelty 7.0 of 10

    TRIAGE augments GRPO with role-typed segment rewards derived from a judge that detects regression and exploration, yielding higher success rates and fewer turns on ALFWorld, Search-QA, and WebShop.

  4. The Verifier is the Curriculum: Execution-Gated Self-Distillation for Cross-Family Game Generation

    cs.AI 2026-06 conditional novelty 6.5 of 10

    Strict-launch-gated rejection-sampling self-distillation compounds cross-family clean Godot generation from 8.8% to 42.2% and full best-of-K coverage, while gold duplication and a lenient BUILD filter erase the gain.

  5. On-Policy Self-Distillation without Any Supervision

    cs.LG 2026-08 conditional novelty 6.0 of 10

    A model self-improves on math reasoning by majority-voting its own rollouts, using the winning answer as a pseudo-solution, and distilling itself on the disagreeing rollouts, without any labeled data.

  6. Cybersecurity Detection Classification with Reasoning-enabled Language Models

    cs.LG 2026-07 conditional novelty 6.0 of 10

    CoT-trained triage plus a separate reasoning calibrator reaches 82.6% accuracy and large high-confidence recall gains over direct-label LLM classifiers on real SOC detections.

  7. Search Hardness-Aware LLM-Based Problem Formulation for Expensive Simulation-Driven Design

    cs.NE 2026-07 conditional novelty 6.0 of 10

    SHA-PF uses initial simulation data to select a 'hard but promising' anchor satisfaction state and evolves LLM-generated formulations that prioritize it, reaching target designs with fewer expensive simulations on ant...

  8. Cura 1T: Specialized Model for Agentic Healthcare

    cs.AI 2026-07 conditional novelty 6.0 of 10

    A healthcare LLM trained with an agent-driven loop that converts benchmark failures into new training data outperforms frontier baselines on most healthcare tests while staying competitive on general reasoning.

  9. Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops

    cs.AI 2026-07 conditional novelty 6.0 of 10

    A survey of 1,250 papers organizes AI self-improvement along two axes—what is improved and loop closure—finding that demonstrated self-improvement strength tracks a verification hierarchy from formal verifiers down to...

  10. Dr. Boot: Bootstrapping Program Synthesis Language Models to Perform Repairing

    cs.SE 2025-07 conditional novelty 6.0 of 10

    A bootstrapping algorithm that trains program synthesis models on their own successful outputs and on repaired failures modestly improves pass@k over regular fine-tuning on MBPP, with mixed results on APPS.

  11. Intent Factored Generation: Unleashing the Diversity in Your Language Model

    cs.AI 2025-06 conditional novelty 6.0 of 10

    Intent Factored Generation samples a high-temperature intent, such as keywords or a summary, and then samples the final response at lower temperature conditioned on that intent, increasing semantic diversity while kee...

  12. How Much Backtracking is Enough? Exploring the Interplay of SFT and RL in Enhancing LLM Reasoning

    cs.AI 2025-05 conditional novelty 6.0 of 10

    On three controlled tasks, the optimal number of backtracks in SFT warm-up data grows with task difficulty, while trajectory correctness has little effect on final RL performance.

  13. Decomposing Elements of Problem Solving: What "Math" Does RL Teach?

    cs.AI 2025-05 conditional novelty 6.0 of 10

    Reinforcement learning (GRPO) on math LLMs primarily increases execution robustness on already-solvable problems, not planning or coverage of new problems.

  14. REACT: Representation Extraction And Controllable Tuning to Overcome Overfitting in LLM Knowledge Editing

    cs.AI 2025-05 conditional novelty 6.0 of 10

    REACT edits LLM facts by adding a learned belief-shift vector to hidden states only when a classifier decides the context is edit-relevant, reducing overfitting on EVOKE while keeping balanced editing metrics.

  15. BARE: Leveraging Base Language Models for Few-Shot Synthetic Data Generation

    cs.CL 2025-02 conditional novelty 6.0 of 10

    BARE generates diverse, high-quality synthetic training data from only three seed examples by having a base model draft and an instruction-tuned model refine, improving downstream fine-tuning accuracy in few-shot settings.

  16. T1: Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling

    cs.LG 2025-01 conditional novelty 6.0 of 10

    A training pipeline called T1 scales RL for LLM reasoning via diverse sampling and rewards, and claims that longer allowed generations directly improve math accuracy without verifiers.

  17. TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimization

    cs.SD 2024-12 conditional novelty 6.0 of 10

    A fast flow-matching text-to-audio model aligned via CLAP-ranked self-generated preference pairs reports state-of-the-art AudioCaps and human-evaluation scores.

  18. Refining Answer Distributions for Improved Large Language Model Reasoning

    cs.CL 2024-12 conditional novelty 6.0 of 10

    RAD iteratively refines a distribution over answers by marginalizing over refinement samples, improving accuracy on six arithmetic benchmarks over self-consistency and hint-based prompting.

  19. HARP: A challenging human-annotated math reasoning benchmark

    cs.LG 2024-12 conditional novelty 6.0 of 10

    HARP, a new benchmark of 5,409 US math competition problems with human solutions and choices, keeps frontier LLMs far from saturation: the best model scores 75.9% overall and only 41.1% on the hardest 197 problems.

  20. AlphaVerus: Bootstrapping Formally Verified Code Generation through Self-Improving Translation and Treefinement

    cs.LG 2024-12 reject novelty 6.0 of 10

    AlphaVerus bootstraps a Llama-70B model to generate Rust code that passes the Verus verifier by iteratively translating Dafny programs and refining candidates with tree search.

  21. Open Security Benchmark: Towards Autonomous Enterprise Cyber Defense

    cs.CR 2026-07 conditional novelty 5.0 of 10

    OSB proposes frozen synthetic-enterprise snapshots with gold posture answers so AI agents can be benchmarked on security investigation via SQL or native vendor APIs.

  22. Sentence Splitter: Uncovering Latent Factual Structure for Self-Supervised Learning

    cs.CL 2026-07 conditional novelty 5.0 of 10

    A T5-based sentence splitter trained on verbalized knowledge-graph triples extracts prefix–tail pairs from raw text, and the resulting structure-aware supervision raises few-shot CommonsenseQA and ATOMIC completion scores.

  23. A Survey on Evaluating Quality and Trustworthiness in LLM-Generated Data

    cs.AI 2026-01 conditional novelty 5.0 of 10

    A metric-oriented survey that classifies intrinsic quality and trustworthiness metrics for LLM-generated data across six modalities and documents systematic evaluation gaps in the current literature.

  24. Who Gets the Reward & Who Gets the Blame? Evaluation-Aligned Training Signals for Multi-LLM Agents

    cs.MA 2025-11 conditional novelty 5.0 of 10

    A game-theoretic framework converts global success/failure evaluations of multi-LLM agents into signed, credit-conserving per-message training signals.

  25. Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting

    cs.LG 2025-10 conditional novelty 5.0 of 10

    RL post-training forgets less than SFT because it trains on on-policy data; refreshing SFT data each epoch also reduces forgetting.

  26. Reasoning Vectors: Transferring Chain-of-Thought Capabilities via Task Arithmetic

    cs.CL 2025-09 conditional novelty 5.0 of 10

    A vector computed as the weight difference between GRPO-trained and SFT-trained models, when added to a base model, transfers chain-of-thought reasoning improvements across benchmarks.

  27. Hermes 4 Technical Report

    cs.AI 2025-08 conditional novelty 5.0 of 10

    Hermes 4 releases three open-weight reasoning models (14B, 70B, 405B) trained with synthetic data and a length-control SFT stage, evaluated on mathematics, code, knowledge, and alignment benchmarks.

  28. EvoCurr: Self-evolving Curriculum with Behavior Code Generation for Complex Decision-making

    cs.AI 2025-08 reject novelty 5.0 of 10

    EvoCurr couples an LLM curriculum designer with an LLM code-generating solver, but its only reported success is 1 of 5 runs and no direct baseline is shown.

  29. Pheromone-based Learning of Optimal Reasoning Paths

    cs.CL 2025-01 reject novelty 5.0 of 10

    An ant-colony-optimization-guided tree-of-thought method with multiple fine-tuned LLM experts reports accuracy gains on GSM8K, ARC-Challenge, and MATH over CoT, ToT, and IRPO baselines.

  30. In-situ graph reasoning and knowledge expansion using Graph-PReFLexOR

    cs.AI 2025-01 reject novelty 5.0 of 10

    Graph-PReFLexOR trains a small LLM to construct knowledge graphs and symbolic abstractions during reasoning, with demonstrations in materials design and creative tasks but only qualitative, self-referential evaluation.

  31. Graph-Aware Isomorphic Attention for Adaptive Dynamics in Transformers

    cs.LG 2025-01 conditional novelty 5.0 of 10

    GIN-style graph aggregation inside Transformer attention and during fine-tuning improves validation perplexity and generalization compared with standard attention and LoRA in the tested settings.

  32. Compute-Accuracy Pareto Frontiers for Open-Source Reasoning Large Language Models

    cs.CL 2025-12 conditional novelty 4.0 of 10

    Across five reasoning benchmarks, sparse MoE models dominate the accuracy-vs-FLOPs Pareto frontier, inference-compute gains saturate, and wrong answers systematically consume more compute than correct ones.

  33. AI Reasoning for Wireless Communications and Networking: A Survey and Perspectives

    cs.NI 2025-09 conditional novelty 4.0 of 10

    A survey that organizes LLM and AI reasoning methods into a taxonomy and maps them onto the physical, link, network, transport, and application layers of wireless networks.

  34. ReasonBridge: Efficient Reasoning Transfer from Closed to Open-Source Language Models

    cs.AI 2025-06 conditional novelty 4.0 of 10

    ReasonBridge distills 1,000 curated reasoning traces from a closed model into open-source models via hierarchical adapters and guided inference, claiming up to 23% benchmark gains.

  35. Enhancing Generalization in Chain of Thought Reasoning for Smaller Models

    cs.LG 2025-01 reject novelty 4.0 of 10

    PRADA combines P-Tuning and domain-adversarial training with CoT distillation and claims improved cross-domain reasoning in small models, though the evaluation is confounded by target-data access.

  36. The Race to Efficiency: A New Perspective on AI Scaling Laws

    cs.LG 2025-01 conditional novelty 4.0 of 10

    A simple formula generalizes static AI scaling laws to time-varying efficiency, showing that efficiency-doubling rates around 2 per year keep loss reduction on a near-exponential path.

  37. SmolTulu: Higher Learning Rate to Batch Size Ratios Can Lead to Better Reasoning in SLMs

    cs.CL 2024-12 conditional novelty 4.0 of 10

    Higher learning-rate-to-batch-size ratios improve reasoning-task performance in small language models, and the resulting SmolTulu-1.7B reports top sub-2B scores on IFEval and GSM8K.

  38. Token-Operations-Oriented Inference Optimization Techniques for Large Models

    cs.SE 2026-06 unverdicted novelty 3.0 of 10

    The paper introduces a four-layer technical architecture for token-operations-oriented inference optimization in large models and reviews key technologies and industry status at each layer.

  39. Industrial brain: a human-like autonomous neuro-symbolic cognitive decision-making system

    cs.AI 2025-06 reject novelty 3.0 of 10

    The paper presents an 'industrial brain' framework for supply-chain resilience prediction, but the reported improvements are not supported by the paper's own inconsistent tables and missing baselines.

  40. The Hitchhiker's Guide to Agentic AI: From Foundations to Systems

    cs.AI 2026-06 unverdicted novelty 2.0 of 10

    A survey-style reference book mapping the full agentic-AI stack from transformer internals to production deployment, with no new research result.

Pith tools