{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/openmathinstruct-1-a-1-8-million-math","title":"OpenMathInstruct-1: A 1.8 Million Math Instruction Tuning Dataset","arxiv_id":"2402.10176","date":"2024-02-15","proceeding":null,"authors":["Shubham Toshniwal","Ivan Moshkov","Sean Narenthiran","Daria Gitman","Fei Jia","Igor Gitman"],"abstract":"Recent work has shown the immense potential of synthetically generated datasets for training large language models (LLMs), especially for acquiring targeted skills. Current large-scale math instruction tuning datasets such as MetaMathQA (Yu et al., 2024) and MAmmoTH (Yue et al., 2024) are constructed using outputs from closed-source LLMs with commercially restrictive licenses. A key reason limiting the use of open-source LLMs in these data generation pipelines has been the wide gap between the mathematical skills of the best closed-source LLMs, such as GPT-4, and the best open-source LLMs. Building on the recent progress in open-source LLMs, our proposed prompting novelty, and some brute-force scaling, we construct OpenMathInstruct-1, a math instruction tuning dataset with 1.8M problem-solution pairs. The dataset is constructed by synthesizing code-interpreter solutions for GSM8K and MATH, two popular math reasoning benchmarks, using the recently released and permissively licensed Mixtral model. Our best model, OpenMath-CodeLlama-70B, trained on a subset of OpenMathInstruct-1, achieves a score of 84.6% on GSM8K and 50.7% on MATH, which is competitive with the best gpt-distilled models. We release our code, models, and the OpenMathInstruct-1 dataset under a commercially permissive license.","url_abs":"https://arxiv.org/abs/2402.10176v2","url_pdf":"https://arxiv.org/pdf/2402.10176v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"openmathinstruct-1-a-1-8-million-math","repo_url":"https://github.com/kipok/nemo-skills","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"arithmetic-reasoning","task_name":"Arithmetic Reasoning"},{"task_slug":"gsm8k","task_name":"GSM8K"},{"task_slug":"math","task_name":"Math"},{"task_slug":"math-word-problem-solving","task_name":"Math Word Problem Solving"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"gpt-4","method_name":"GPT-4"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-70B (w/ code, SC, k=50)","rank_in_archive_order":20,"of":164,"metrics":{"Accuracy":"90.8","Parameters (Billion)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-Llama2-70B (w/ code, SC, k=50)","rank_in_archive_order":22,"of":164,"metrics":{"Accuracy":"90.1","Parameters (Billion)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-34B (w/ code, SC, k=50)","rank_in_archive_order":31,"of":164,"metrics":{"Accuracy":"88.0","Parameters (Billion)":"34"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-Mistral-7B (w/ code, SC, k=50)","rank_in_archive_order":38,"of":164,"metrics":{"Accuracy":"86.9","Parameters (Billion)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-13B (w/ code, SC, k=50)","rank_in_archive_order":41,"of":164,"metrics":{"Accuracy":"86.8","Parameters (Billion)":"13"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-7B (w/ code, SC, k=50)","rank_in_archive_order":46,"of":164,"metrics":{"Accuracy":"84.8","Parameters (Billion)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-Llama2-70B (w/ code)","rank_in_archive_order":48,"of":164,"metrics":{"Accuracy":"84.7","Parameters (Billion)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-70B (w/ code)","rank_in_archive_order":49,"of":164,"metrics":{"Accuracy":"84.6","Parameters (Billion)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-34B (w/ code)","rank_in_archive_order":70,"of":164,"metrics":{"Accuracy":"80.7","Parameters (Billion)":"34"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-Mistral-7B (w/ code)","rank_in_archive_order":74,"of":164,"metrics":{"Accuracy":"80.2","Parameters (Billion)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-13B (w/ code)","rank_in_archive_order":76,"of":164,"metrics":{"Accuracy":"78.8","Parameters (Billion)":"13"},"uses_additional_data":true},{"leaderboard":"/sota/arithmetic-reasoning-on-gsm8k","task":"Arithmetic Reasoning","dataset":"GSM8K","model":"OpenMath-CodeLlama-7B (w/ code)","rank_in_archive_order":84,"of":164,"metrics":{"Accuracy":"75.9","Parameters (Billion)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-asdiv-a","task":"Math Word Problem Solving","dataset":"ASDiv-A","model":"OpenMath-CodeLlama-70B (w/ code)","rank_in_archive_order":5,"of":9,"metrics":{"Execution Accuracy":"84.7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-70B (w/ code, SC, k=50)","rank_in_archive_order":25,"of":135,"metrics":{"Accuracy":"60.4","Parameters (Billions)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-34B (w/ code, SC, k=50)","rank_in_archive_order":26,"of":135,"metrics":{"Accuracy":"60.2","Parameters (Billions)":"34"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-Llama2-70B (w/ code, SC, k=50)","rank_in_archive_order":29,"of":135,"metrics":{"Accuracy":"58.3","Parameters (Billions)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-13B (w/ code, SC, k=50)","rank_in_archive_order":31,"of":135,"metrics":{"Accuracy":"57.6","Parameters (Billions)":"13"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-Mistral-7B (w/ code, SC, k=50)","rank_in_archive_order":32,"of":135,"metrics":{"Accuracy":"57.2","Parameters (Billions)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-7B (w/ code, SC, k=50)","rank_in_archive_order":36,"of":135,"metrics":{"Accuracy":"55.6","Parameters (Billions)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-70B (w/ code)","rank_in_archive_order":47,"of":135,"metrics":{"Accuracy":"50.7","Parameters (Billions)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-34B (w/ code)","rank_in_archive_order":54,"of":135,"metrics":{"Accuracy":"48.3","Parameters (Billions)":"34"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-Llama2-70B (w/ code)","rank_in_archive_order":60,"of":135,"metrics":{"Accuracy":"46.3","Parameters (Billions)":"70"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-13B (w/ code)","rank_in_archive_order":61,"of":135,"metrics":{"Accuracy":"45.5","Parameters (Billions)":"13"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-Mistral-7B (w/ code)","rank_in_archive_order":68,"of":135,"metrics":{"Accuracy":"44.5","Parameters (Billions)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-math","task":"Math Word Problem Solving","dataset":"MATH","model":"OpenMath-CodeLlama-7B (w/ code)","rank_in_archive_order":70,"of":135,"metrics":{"Accuracy":"43.6","Parameters (Billions)":"7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-mawps","task":"Math Word Problem Solving","dataset":"MAWPS","model":"OpenMath-CodeLlama-70B (w/ code)","rank_in_archive_order":1,"of":25,"metrics":{"Accuracy (%)":"95.7"},"uses_additional_data":true},{"leaderboard":"/sota/math-word-problem-solving-on-svamp","task":"Math Word Problem Solving","dataset":"SVAMP","model":"OpenMath-CodeLlama-70B (w/ code)","rank_in_archive_order":5,"of":26,"metrics":{"Execution Accuracy":"87.8"},"uses_additional_data":true}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2402.10176","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}