{"id":50436624,"url":"https://github.com/krishnaura45/aimo-problem-solving","last_synced_at":"2026-05-31T17:02:40.865Z","repository":{"id":356384372,"uuid":"1232301175","full_name":"krishnaura45/aimo-problem-solving","owner":"krishnaura45","description":"🧮Olympiad Math Reasoning 🤖LLMs for LaTeX Problem Solving ⚙️GPT-OSS-120B","archived":false,"fork":false,"pushed_at":"2026-05-15T21:24:50.000Z","size":131,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-16T00:04:46.132Z","etag":null,"topics":["aimo","artificial-intelligence","custom-metrics","deepseek-r1","gpt-oss-120b","gpt-oss-20b","kaggle-competition","latex","llm","mathematical-olympiad","qwen-3","qwen-3-5"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/krishnaura45.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-07T19:46:02.000Z","updated_at":"2026-05-15T21:24:54.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/krishnaura45/aimo-problem-solving","commit_stats":null,"previous_names":["krishnaura45/aimo-problem-solving"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/krishnaura45/aimo-problem-solving","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/krishnaura45%2Faimo-problem-solving","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/krishnaura45%2Faimo-problem-solving/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/krishnaura45%2Faimo-problem-solving/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/krishnaura45%2Faimo-problem-solving/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/krishnaura45","download_url":"https://codeload.github.com/krishnaura45/aimo-problem-solving/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/krishnaura45%2Faimo-problem-solving/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33739861,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-05-31T02:00:06.040Z","response_time":95,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["aimo","artificial-intelligence","custom-metrics","deepseek-r1","gpt-oss-120b","gpt-oss-20b","kaggle-competition","latex","llm","mathematical-olympiad","qwen-3","qwen-3-5"],"created_at":"2026-05-31T17:02:39.863Z","updated_at":"2026-05-31T17:02:40.860Z","avatar_url":"https://github.com/krishnaura45.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# AI Mathematical Olympiad - Progress Prize 3 ~ `91st Place Solution`\n\u003e Solving olympiad-level mathematical reasoning problems using open-source LLM inference pipeline and competitive reasoning workflow.\n\n![Python](https://img.shields.io/badge/Python-3776AB?style=for-the-badge\u0026logo=python\u0026logoColor=white)\n![Kaggle](https://img.shields.io/badge/Kaggle-20BEFF?style=for-the-badge\u0026logo=kaggle\u0026logoColor=white)\n![LLM](https://img.shields.io/badge/Open--Source-LLMs-purple?style=for-the-badge)\n![Math Reasoning](https://img.shields.io/badge/AI-Math%20Reasoning-blueviolet?style=for-the-badge)\n![Best Score](https://img.shields.io/badge/Best%20Score-43.0-2ECC71?style=for-the-badge)\n![Rank](https://img.shields.io/badge/Rank-92%20of%204138-brightgreen?style=for-the-badge)\n![Solo](https://img.shields.io/badge/Submission-Solo-orange?style=for-the-badge)\n\n### **Project Duration**: April 01, 2026 - April 15, 2026\n\n---\n\n## File Structure\n\n```bash\n├── INFO.md                      # Competition details, evaluation, rules and timeline\n├── README.md\n├── aimo-inference.ipynb         # Main competition inference pipeline\n├── dependency-install.ipynb     # Dependency installation and environment setup\n├── inference.py                 # Standalone inference script\n├── utils.ipynb                  # Utility experiments and helper workflows\n├── certificate.png              # Certificate of Achievement from Kaggle\n├── assets\n│   └── eval_info.png            # helper image\n└── output\n    └── submission.parquet       # Best submission file\n```\n\n### Installation\n\n```bash\ngit clone https://github.com/krishnaura45/aimo-problem-solving.git\ncd aimo-problem-solving\n```\n\n### Usage\n\n```bash\n# Open notebook workflow\njupyter notebook aimo-inference.ipynb\n\n# or run standalone inference\npython inference.py\n```\n\n---\n\n## Problem Statement\n\nThe goal of the **[AI Mathematical Olympiad - Progress Prize 3](https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-3)** Kaggle competition was to create open-source algorithms capable of solving olympiad-level mathematical reasoning problems written entirely in LaTeX notation.\n\nThe competition featured highly challenging problems spanning:\n\n- Algebra\n- Combinatorics\n- Geometry\n- Number Theory\n\nwith difficulty ranging from national olympiad level up to IMO-standard mathematics.\n\nHosted on Kaggle, the challenge evaluated submissions using a specialized penalized accuracy framework across public and hidden private reruns. Each problem required predicting a non-negative integer answer between `0` and `99999`.\n\nThe competition emphasized genuine mathematical reasoning capabilities and robust inference under constrained notebook environments.\n\n---\n\n## Approach\n\n### Inference Pipeline\n\nThe primary workflow was implemented in:\n\n- `aimo-inference.ipynb`\n- `inference.py`\n\nThe overall pipeline focused on efficient large language model inference under strict Kaggle notebook constraints.\n\n---\n\n### Problem Processing\n\n- Processed olympiad-style mathematical problems written in LaTeX\n- Structured prompts for reasoning-oriented inference\n\n---\n\n### Reasoning Workflow\n\nThe inference workflow emphasized:\n\n- Multi-step mathematical reasoning\n- Symbolic interpretation of problem statements\n- Integer answer extraction and normalization\n\n---\n\n## Competition Results\n\n- Announced on: May 12, 2026.\n\n- **Public/Private Leaderboard Scores**:\n  - `37`\n  - `38`\n  - `39`\n  - `40`\n  - `42`\n  - `43`\n\n- **Performance**:\n  - **Best Private Score**: `43.0`\n  - Placed **91st out of 4066 participants** and **4138 teams** as a **solo participant**.\n\n---\n\n## References\n\n- Dataset: [AIMO - PP3](https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-3/data)\n- Competition Info: `INFO.md`\n- Main Inference Notebook: `aimo-inference.ipynb`\n- Official Models Page: https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-3/models\n\n---\n\n## Tech Stack\n\n- **Language**: Python\n- **Libraries / Frameworks**:\n  - `transformers`\n  - `torch`\n  - `pandas`\n  - `numpy`\n- **Techniques**:\n  - LLM Inference\n  - Mathematical Reasoning\n  - Sequential API-based Evaluation\n  - Prompt-based Solving\n- **Tools**:\n  - Jupyter Notebook\n  - Kaggle Notebooks\n  - GPU-based inference environments\n\n---\n\n📌 *This project demonstrates the growing capability of open-source reasoning systems in solving olympiad-level mathematical problems under competitive inference constraints.*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkrishnaura45%2Faimo-problem-solving","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkrishnaura45%2Faimo-problem-solving","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkrishnaura45%2Faimo-problem-solving/lists"}