{"data":{"slug":"huggingface-nanotron","name":"nanotron","tagline":"Minimalistic large language model 3D-parallelism training","github_url":"https://github.com/huggingface/nanotron","owner":"huggingface","repo":"nanotron","owner_avatar_url":"https://avatars.githubusercontent.com/u/25720743?v=4","primary_language":"Python","stars":2775,"forks":329,"topics":[],"archived":false,"github_pushed_at":"2026-05-26T10:32:37+00:00","maintenance_label":"Steady","url":"https://www.graphcanon.com/tools/huggingface-nanotron","markdown_url":"https://www.graphcanon.com/tools/huggingface-nanotron.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/huggingface-nanotron","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=huggingface-nanotron","description":"Minimalistic large language model 3D-parallelism training","homepage_url":null,"license":"Apache-2.0","open_issues":149,"watchers":52,"ai_summary":"A minimalistic repository for 3D-parallelism in large language model training, focused on efficient distributed computing.","readme_excerpt":"## Installation\n\nTo run the code in this project, first create a Python virtual environment using e.g. `uv`:\n\n\n```shell\nuv venv nanotron --python 3.11 && source nanotron/bin/activate && uv pip install --upgrade pip\n```\n\n> [!TIP]\n> For Hugging Face cluster users, add `export UV_LINK_MODE=copy` to your `.bashrc` to suppress cache warnings from `uv`\n\nNext, install Pytorch:\n\n```shell\nuv pip install torch --index-url https://download.pytorch.org/whl/cu124\n```\n\nThen install the core dependencies with:\n\n```shell\nuv pip install -e .\n```\n\nTo run the example scripts, install the remaining dependencies as follows:\n\n```shell\nuv pip install datasets transformers datatrove[io] numba wandb","github_created_at":"2023-09-11T14:40:28+00:00","created_at":"2026-07-11T10:35:31.458323+00:00","updated_at":"2026-08-07T00:02:01.454012+00:00","categories":[{"slug":"model-training","name":"Model Training","url":"https://www.graphcanon.com/categories/model-training","markdown_url":"https://www.graphcanon.com/categories/model-training.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/model-training"}],"tags":[{"slug":"3d-parallelism","name":"3d_parallelism"},{"slug":"distributed-training","name":"distributed-training"},{"slug":"llm","name":"llm"},{"slug":"pytorch","name":"pytorch"}],"trust":{"provenance":{"is_fork":false,"github_id":690106318,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-07T00:02:00.668Z","maintenance":{"label":"Steady","score":60,"methodology":"github_public_v1","releases_90d":0,"days_since_push":72,"last_release_at":"2024-03-04T17:21:17Z"},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T10:35:32.894Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-08-07T00:02:01.134Z"},"languages":{"value":["python"],"source":"github.language+pyproject.toml","observed_at":"2026-08-07T00:02:01.134Z"},"license_spdx":{"value":"Apache-2.0","source":"github.license","observed_at":"2026-08-07T00:02:01.134Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["You aim to implement 3D-parallelism for large language models with minimal code complexity and high efficiency.","Your project is centered around Python and you are looking to optimize distributed training processes without the extra bloat."],"when_not_to_use":["You require robust integration capabilities that come with larger, more feature-rich training frameworks.","Need extensive out-of-the-box solutions for common data processing tasks as Nanotron focuses narrowly on parallelism and efficient computing, potentially missing broader functionalities."],"source":"enrich:decision_facts","observed_at":"2026-07-14T20:35:51.951Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"Nanotron specializes in minimalistic large language model 3D-parallelism training via efficient distributed computing techniques."}]}}