Pith. sign in

REVIEW 12 cited by

Orca-Math: Unlocking the potential of SLMs in Grade School Math

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 2402.14830 v1 pith:R56UD5BH submitted 2024-02-16 cs.CL cs.AI

classification cs.CLcs.AI
keywords modelorca-mathmodelsachievesgsm8kperformanceproblemsslms
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Mathematical word problem-solving has long been recognized as a complex task for small language models (SLMs). A recent study hypothesized that the smallest model size, needed to achieve over 80% accuracy on the GSM8K benchmark, is 34 billion parameters. To reach this level of performance with smaller models, researcher often train SLMs to generate Python code or use tools to help avoid calculation errors. Additionally, they employ ensembling, where outputs of up to 100 model runs are combined to arrive at a more accurate result. Result selection is done using consensus, majority vote or a separate a verifier model used in conjunction with the SLM. Ensembling provides a substantial boost in accuracy but at a significant cost increase with multiple calls to the model (e.g., Phi-GSM uses top-48 to boost the performance from 68.2 to 81.5). In this work, we present Orca-Math, a 7-billion-parameter SLM based on the Mistral-7B, which achieves 86.81% on GSM8k without the need for multiple model calls or the use of verifiers, code execution or any other external tools. Our approach has the following key elements: (1) A high quality synthetic dataset of 200K math problems created using a multi-agent setup where agents collaborate to create the data, (2) An iterative learning techniques that enables the SLM to practice solving problems, receive feedback on its solutions and learn from preference pairs incorporating the SLM solutions and the feedback. When trained with Supervised Fine-Tuning alone, Orca-Math achieves 81.50% on GSM8k pass@1 metric. With iterative preference learning, Orca-Math achieves 86.81% pass@1. Orca-Math surpasses the performance of significantly larger models such as LLAMA-2-70B, WizardMath-70B, Gemini-Pro, ChatGPT-3.5. It also significantly outperforms other smaller models while using much smaller data (hundreds of thousands vs. millions of problems).

Discussion (0). Sign in to comment.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Multi-Turn On-Policy Distillation with Prefix Replay

    cs.LG 2026-07 conditional novelty 6.0 of 10

    ReOPD offline-distills multi-turn agentic LLMs via teacher-prefix replay plus step-decay sampling, matching online OPD accuracy at ≥4× speed with zero tool calls.

  2. StepHint: Multi-level Stepwise Hints Enhance Reinforcement Learning to Reason

    cs.AI 2025-07 conditional novelty 6.0 of 10

    StepHint improves RLVR math reasoning by giving the model multiple prefix-level hints drawn from correct chains generated by stronger models, beating several RLVR baselines on six math benchmarks and two out-of-domain sets.

  3. Thunder-LLM: Efficiently Adapting LLMs to Korean with Minimal Resources

    cs.CL 2025-06 conditional novelty 6.0 of 10

    A cost-effective recipe consisting of tokenizer extension, continual pretraining, FP8 training, and SFT/DPO post-training yields Korean-English bilingual 8B models with top Korean benchmark scores.

  4. Excessive Reasoning Attack on Reasoning LLMs

    cs.CR 2025-06 conditional novelty 6.0 of 10

    A 10-token adversarial suffix, optimized with three custom losses, raises reasoning length 3x to 9x on open reasoning LLMs without hurting accuracy, and partly transfers to closed models.

  5. Reinforce LLM Reasoning through Multi-Agent Reflection

    cs.LG 2025-06 conditional novelty 6.0 of 10

    DPSDP trains an actor-critic LLM pair with DPO-style preference learning on self-generated trajectories, improving iterative refinement accuracy on math benchmarks.

  6. Self-Route: Automatic Mode Switching via Capability Estimation for Efficient Reasoning

    cs.CL 2025-05 conditional novelty 6.0 of 10

    Self-Route uses hidden-layer representations from a brief pre-inference plan to route each question to either short or long chain-of-thought, cutting tokens by 30-55% with under 2% accuracy loss.

  7. EasyMath: A 0-shot Math Benchmark for SLMs

    cs.CL 2025-05 conditional novelty 6.0 of 10

    EasyMath, a new 0-shot math benchmark for small language models, shows accuracy rising with model size and training, modest chain-of-thought gains, and better consistency at larger scale.

  8. RTTC: Reward-Guided Collaborative Test-Time Compute

    cs.CL 2025-08 reject novelty 5.0 of 10

    RTTC uses a reward model to route each query to direct inference, retrieval-augmented generation, or test-time training, plus caching to cut redundant work.

  9. Boosting LLM Reasoning via Spontaneous Self-Correction

    cs.AI 2025-06 conditional novelty 5.0 of 10

    SPOC trains LLMs to interleave self-verification and solution attempts in a single pass, reporting gains on math benchmarks, though most gains come from stronger first attempts.

  10. MDPO: Multi-Granularity Direct Preference Optimization for Mathematical Reasoning

    cs.LG 2025-05 conditional novelty 5.0 of 10

    MDPO applies a SimPO-style length-normalized reward to preference pairs built at solution, inference, and step granularities, yielding small accuracy gains on math benchmarks.

  11. HOFT: Householder Orthogonal Fine-tuning

    cs.LG 2025-05 conditional novelty 5.0 of 10

    HOFT and SHOFT fine-tune foundation models with two Householder-built orthogonal matrices, matching or beating LoRA, DoRA, OFT, BOFT and HRA on reasoning, translation, image generation and math.

  12. MASTER: Enhancing Large Language Model via Multi-Agent Simulated Teaching

    cs.CL 2025-06 conditional novelty 4.0 of 10

    A multi-agent simulated teaching pipeline creates BOOST-QA, and fine-tuning on it lifts reported LLM benchmark scores by up to 31 points over the original data.

Pith tools