Datasetpublic

Desertsmogtech/finemath-filtered-v1

sourceHugging Faceupdated 1y agoView on Hugging Face

0likes1downloads

dirREADME.md--- annotations_creators: [no-annotation] language: [en] license: apache-2.0 multilinguality: [monolingual] size_categories: [n<1K] source_datasets: [original] task_categories: [question-answering, reasoning] task_ids: [math-question-answering] pretty_name: FineMath Filtered v1 --- # Dataset Card for FineMath Filtered v1 ## 🧠 Dataset Summary This dataset contains a curated subset of math reasoning prompts filtered from Common Crawl sources. Each entry includes a math problem, source URL, crawl ID, token count, score, and math type. It is designed for fine-tuning models on math comprehension, symbolic reasoning, and prompt engineering. ## 📚 Supported Tasks - **Math Question Answering**: Solve or interpret math problems from natural language prompts. - **Reasoning & Prompt Engineering**: Evaluate model performance on structured math inputs. ## 🌐 Languages - English (monolingual) ## 📁 Dataset Structure Each row is a JSON object with the following fields: - `text`: Math problem prompt - `url`: Source URL - `crawl`: Crawl ID (e.g. CC-MAIN-2020-05) - `token_count`: Number of tokens in the prompt - `score`: Heuristic relevance score - `math_type`: Category (e.g. Algebra, Calculus) ## 🛠️ Usage Notes This dataset is ideal for: - Fine-tuning LLMs on math reasoning - Filtering or scoring math prompts - Benchmarking symbolic comprehension ## ⚖️ Licensing Apache-2.0 ## ✍️ Author Created by [Desertsmogtech](https:/

file.gitattributes2 KBdownload

filefiltered-math-data.jsonl450 Bdownload

fileREADME. 📜 License This dataset is released under a dual-license model: MIT License — Applies to the dataset structure, enhancement logic, filtering methodology, and documentation authored by Jason Vanzile. Source Data Notice — Original data may include public-domain or third-party content subject to separate terms. Users are responsible for verifying compliance when remixing or redistributing derived works. By using this dataset, you agree to: Attribute the enhancement logic and structure to Jason Vanzile. Respect any original data licenses if applicable. Avoid deploying this dataset in systems that violate ethical or regulatory boundaries. For compliance-sensitive use cases or commercial licensing inquiries, contact via Jason’s Hugging Face profile or official channels.download

Desertsmogtech/finemath-filtered-v1 · main · files are served by the source, never re-hosted here