[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-train-single-step":3,"mdc-ihw8q5-key":34,"related-repo-nvidia-tao-train-single-step":410,"related-org-nvidia-tao-train-single-step":512},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"tao-train-single-step","train and export TAO models","Standard single-step train\u002Feval\u002Fexport workflow for any TAO model. Use when training a TAO model on a dataset without iterative data augmentation, AutoML, or DEFT loops. Trigger phrases include \"single train run\", \"train then evaluate then export\", \"plain TAO training\", \"normal training\", \"no AutoML\", \"skip the loop\". Routes through the per-model SKILL.md for action specifics and through `tao-launch-workflow` for platform\u002Fcredentials\u002Fdataset intake.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,22],{"name":13,"slug":14,"type":15},"Automation","automation","tag",{"name":17,"slug":18,"type":15},"Machine Learning","machine-learning",{"name":20,"slug":21,"type":15},"Deployment","deployment",{"name":9,"slug":8,"type":15},2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:28:49.335607","Apache-2.0",281,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftao-train-single-step","---\nname: tao-train-single-step\ndescription: Standard single-step train\u002Feval\u002Fexport workflow for any TAO model. Use when training a TAO model on a dataset\n  without iterative data augmentation, AutoML, or DEFT loops. Trigger phrases include \"single train run\", \"train then evaluate\n  then export\", \"plain TAO training\", \"normal training\", \"no AutoML\", \"skip the loop\". Routes through the per-model SKILL.md\n  for action specifics and through `tao-launch-workflow` for platform\u002Fcredentials\u002Fdataset intake.\nlicense: Apache-2.0\ncompatibility: Requires docker + nvidia-container-toolkit. Workflows declare additional requirements.\nmetadata:\n  author: NVIDIA Corporation\n  version: \"0.1.0\"\nallowed-tools: Read Bash Write\ntags:\n- training\n- single-step\n- generic\n---\n\n# Normal Train\n\nStandard supervised fine-tuning: train a model on a labeled dataset, optionally evaluate, then optionally export. The most common TAO workflow for adapting a pretrained model to a new dataset.\n\n## Steps\n\n1. **train** — executed through AutoML when the selected model has\n   `automl_enabled: true` and `automl_policy` is `on`; set\n   `automl_policy=off` for a plain single training run\n2. **eval** — executed if `eval_dataset_uri` is resolved\n3. **export** — optional, on user request after training\n\n## Prerequisites\n\n### Required\n- **model**: A compatible TAO model (e.g., clip, nvdinov2, grounding_dino)\n- **train_dataset_uri**: URI of the training dataset (e.g., `s3:\u002F\u002Fbucket\u002Ftrain\u002F`)\n- **platform**: Ask from the generated supported-platform list:\n  `${TAO_SKILL_BANK_PATH:-~\u002Ftao-skills-external}\u002Fscripts\u002Flist_tao_platforms.py --format text`\n- **container image confirmation**: resolve the default image from the selected\n  model\u002Faction config, show it to the user, and require confirmation or\n  `image=\u003Coverride>` before creating runner files or submitting training.\n\n### Optional\n- **eval_dataset_uri**: Some model skills mark this as required — check the resolved model skill before treating it as optional.\n- **base_checkpoint**: If not provided, defaults to the NGC pretrained checkpoint listed in the model skill, or trains from scratch if no NGC checkpoint exists.\n- **automl_policy**: `on` by default; set `off` to bypass model-level AutoML for this run while leaving model metadata unchanged. Use only `on` \u002F `off` in new launch settings.\n- **image override**: Use `image=\u003Coverride>` to pin a specific TAO toolkit build\n  after reviewing the resolved default.\n\n## Launch Intake\n\nAfter the user confirms they want this standard train\u002Feval\u002Fexport workflow,\nask which supported platform they intend to run on. Generate the choices with\n`scripts\u002Flist_tao_platforms.py --format text`; do not scan platform docs or\nfolders.\n\nBefore creating a plain train runner, inspect the selected model's metadata\nwith `scripts\u002Flist_tao_models.py --scope automl --format json` or read\n`skills\u002Fmodels\u002F\u003Cnetwork>\u002Freferences\u002Fskill_info.yaml`. If `automl_enabled` is true and\nthe helper reports a valid train schema for that model, route the train stage\nthrough `skills\u002Fapplications\u002Ftao-run-automl` by default. Only stay on the plain train path\nwhen `automl_policy=off`, the user explicitly asks for no HPO\u002FAutoML, or AutoML\nis enabled but not runnable because the model's train schema is not packaged\nyet.\n\nAlso ask whether long-running monitoring should stay enabled and how many\nminutes between status updates. Defaults: enabled, 5 minutes.\n\nAfter the model\u002Faction are known, run `scripts\u002Fresolve_tao_image.py --model\n\u003Cnetwork> --action train --format text` and ask whether to use the resolved\nimage or an `image=\u003Coverride>`. Do not create the tao-train-single-step runner until the\nimage is confirmed.\n\nAfter platform selection, run\n`scripts\u002Flist_tao_platforms.py --platform \u003Cplatform> --format text` and ask\nonly for credentials relevant to that platform, plus any selected-model\ncredentials. Do not ask for unrelated platform credentials.\n",{"data":35,"body":45},{"name":4,"description":6,"license":26,"compatibility":36,"metadata":37,"allowed-tools":40,"tags":41},"Requires docker + nvidia-container-toolkit. Workflows declare additional requirements.",{"author":38,"version":39},"NVIDIA Corporation","0.1.0","Read Bash Write",[42,43,44],"training","single-step","generic",{"type":46,"children":47},"root",[48,57,63,70,147,153,160,226,232,309,315,328,372,377,397],{"type":49,"tag":50,"props":51,"children":53},"element","h1",{"id":52},"normal-train",[54],{"type":55,"value":56},"text","Normal Train",{"type":49,"tag":58,"props":59,"children":60},"p",{},[61],{"type":55,"value":62},"Standard supervised fine-tuning: train a model on a labeled dataset, optionally evaluate, then optionally export. The most common TAO workflow for adapting a pretrained model to a new dataset.",{"type":49,"tag":64,"props":65,"children":67},"h2",{"id":66},"steps",[68],{"type":55,"value":69},"Steps",{"type":49,"tag":71,"props":72,"children":73},"ol",{},[74,119,137],{"type":49,"tag":75,"props":76,"children":77},"li",{},[78,84,86,93,95,101,103,109,111,117],{"type":49,"tag":79,"props":80,"children":81},"strong",{},[82],{"type":55,"value":83},"train",{"type":55,"value":85}," — executed through AutoML when the selected model has\n",{"type":49,"tag":87,"props":88,"children":90},"code",{"className":89},[],[91],{"type":55,"value":92},"automl_enabled: true",{"type":55,"value":94}," and ",{"type":49,"tag":87,"props":96,"children":98},{"className":97},[],[99],{"type":55,"value":100},"automl_policy",{"type":55,"value":102}," is ",{"type":49,"tag":87,"props":104,"children":106},{"className":105},[],[107],{"type":55,"value":108},"on",{"type":55,"value":110},"; set\n",{"type":49,"tag":87,"props":112,"children":114},{"className":113},[],[115],{"type":55,"value":116},"automl_policy=off",{"type":55,"value":118}," for a plain single training run",{"type":49,"tag":75,"props":120,"children":121},{},[122,127,129,135],{"type":49,"tag":79,"props":123,"children":124},{},[125],{"type":55,"value":126},"eval",{"type":55,"value":128}," — executed if ",{"type":49,"tag":87,"props":130,"children":132},{"className":131},[],[133],{"type":55,"value":134},"eval_dataset_uri",{"type":55,"value":136}," is resolved",{"type":49,"tag":75,"props":138,"children":139},{},[140,145],{"type":49,"tag":79,"props":141,"children":142},{},[143],{"type":55,"value":144},"export",{"type":55,"value":146}," — optional, on user request after training",{"type":49,"tag":64,"props":148,"children":150},{"id":149},"prerequisites",[151],{"type":55,"value":152},"Prerequisites",{"type":49,"tag":154,"props":155,"children":157},"h3",{"id":156},"required",[158],{"type":55,"value":159},"Required",{"type":49,"tag":161,"props":162,"children":163},"ul",{},[164,174,192,208],{"type":49,"tag":75,"props":165,"children":166},{},[167,172],{"type":49,"tag":79,"props":168,"children":169},{},[170],{"type":55,"value":171},"model",{"type":55,"value":173},": A compatible TAO model (e.g., clip, nvdinov2, grounding_dino)",{"type":49,"tag":75,"props":175,"children":176},{},[177,182,184,190],{"type":49,"tag":79,"props":178,"children":179},{},[180],{"type":55,"value":181},"train_dataset_uri",{"type":55,"value":183},": URI of the training dataset (e.g., ",{"type":49,"tag":87,"props":185,"children":187},{"className":186},[],[188],{"type":55,"value":189},"s3:\u002F\u002Fbucket\u002Ftrain\u002F",{"type":55,"value":191},")",{"type":49,"tag":75,"props":193,"children":194},{},[195,200,202],{"type":49,"tag":79,"props":196,"children":197},{},[198],{"type":55,"value":199},"platform",{"type":55,"value":201},": Ask from the generated supported-platform list:\n",{"type":49,"tag":87,"props":203,"children":205},{"className":204},[],[206],{"type":55,"value":207},"${TAO_SKILL_BANK_PATH:-~\u002Ftao-skills-external}\u002Fscripts\u002Flist_tao_platforms.py --format text",{"type":49,"tag":75,"props":209,"children":210},{},[211,216,218,224],{"type":49,"tag":79,"props":212,"children":213},{},[214],{"type":55,"value":215},"container image confirmation",{"type":55,"value":217},": resolve the default image from the selected\nmodel\u002Faction config, show it to the user, and require confirmation or\n",{"type":49,"tag":87,"props":219,"children":221},{"className":220},[],[222],{"type":55,"value":223},"image=\u003Coverride>",{"type":55,"value":225}," before creating runner files or submitting training.",{"type":49,"tag":154,"props":227,"children":229},{"id":228},"optional",[230],{"type":55,"value":231},"Optional",{"type":49,"tag":161,"props":233,"children":234},{},[235,244,254,292],{"type":49,"tag":75,"props":236,"children":237},{},[238,242],{"type":49,"tag":79,"props":239,"children":240},{},[241],{"type":55,"value":134},{"type":55,"value":243},": Some model skills mark this as required — check the resolved model skill before treating it as optional.",{"type":49,"tag":75,"props":245,"children":246},{},[247,252],{"type":49,"tag":79,"props":248,"children":249},{},[250],{"type":55,"value":251},"base_checkpoint",{"type":55,"value":253},": If not provided, defaults to the NGC pretrained checkpoint listed in the model skill, or trains from scratch if no NGC checkpoint exists.",{"type":49,"tag":75,"props":255,"children":256},{},[257,261,263,268,270,276,278,283,285,290],{"type":49,"tag":79,"props":258,"children":259},{},[260],{"type":55,"value":100},{"type":55,"value":262},": ",{"type":49,"tag":87,"props":264,"children":266},{"className":265},[],[267],{"type":55,"value":108},{"type":55,"value":269}," by default; set ",{"type":49,"tag":87,"props":271,"children":273},{"className":272},[],[274],{"type":55,"value":275},"off",{"type":55,"value":277}," to bypass model-level AutoML for this run while leaving model metadata unchanged. Use only ",{"type":49,"tag":87,"props":279,"children":281},{"className":280},[],[282],{"type":55,"value":108},{"type":55,"value":284}," \u002F ",{"type":49,"tag":87,"props":286,"children":288},{"className":287},[],[289],{"type":55,"value":275},{"type":55,"value":291}," in new launch settings.",{"type":49,"tag":75,"props":293,"children":294},{},[295,300,302,307],{"type":49,"tag":79,"props":296,"children":297},{},[298],{"type":55,"value":299},"image override",{"type":55,"value":301},": Use ",{"type":49,"tag":87,"props":303,"children":305},{"className":304},[],[306],{"type":55,"value":223},{"type":55,"value":308}," to pin a specific TAO toolkit build\nafter reviewing the resolved default.",{"type":49,"tag":64,"props":310,"children":312},{"id":311},"launch-intake",[313],{"type":55,"value":314},"Launch Intake",{"type":49,"tag":58,"props":316,"children":317},{},[318,320,326],{"type":55,"value":319},"After the user confirms they want this standard train\u002Feval\u002Fexport workflow,\nask which supported platform they intend to run on. Generate the choices with\n",{"type":49,"tag":87,"props":321,"children":323},{"className":322},[],[324],{"type":55,"value":325},"scripts\u002Flist_tao_platforms.py --format text",{"type":55,"value":327},"; do not scan platform docs or\nfolders.",{"type":49,"tag":58,"props":329,"children":330},{},[331,333,339,341,347,349,355,357,363,365,370],{"type":55,"value":332},"Before creating a plain train runner, inspect the selected model's metadata\nwith ",{"type":49,"tag":87,"props":334,"children":336},{"className":335},[],[337],{"type":55,"value":338},"scripts\u002Flist_tao_models.py --scope automl --format json",{"type":55,"value":340}," or read\n",{"type":49,"tag":87,"props":342,"children":344},{"className":343},[],[345],{"type":55,"value":346},"skills\u002Fmodels\u002F\u003Cnetwork>\u002Freferences\u002Fskill_info.yaml",{"type":55,"value":348},". If ",{"type":49,"tag":87,"props":350,"children":352},{"className":351},[],[353],{"type":55,"value":354},"automl_enabled",{"type":55,"value":356}," is true and\nthe helper reports a valid train schema for that model, route the train stage\nthrough ",{"type":49,"tag":87,"props":358,"children":360},{"className":359},[],[361],{"type":55,"value":362},"skills\u002Fapplications\u002Ftao-run-automl",{"type":55,"value":364}," by default. Only stay on the plain train path\nwhen ",{"type":49,"tag":87,"props":366,"children":368},{"className":367},[],[369],{"type":55,"value":116},{"type":55,"value":371},", the user explicitly asks for no HPO\u002FAutoML, or AutoML\nis enabled but not runnable because the model's train schema is not packaged\nyet.",{"type":49,"tag":58,"props":373,"children":374},{},[375],{"type":55,"value":376},"Also ask whether long-running monitoring should stay enabled and how many\nminutes between status updates. Defaults: enabled, 5 minutes.",{"type":49,"tag":58,"props":378,"children":379},{},[380,382,388,390,395],{"type":55,"value":381},"After the model\u002Faction are known, run ",{"type":49,"tag":87,"props":383,"children":385},{"className":384},[],[386],{"type":55,"value":387},"scripts\u002Fresolve_tao_image.py --model \u003Cnetwork> --action train --format text",{"type":55,"value":389}," and ask whether to use the resolved\nimage or an ",{"type":49,"tag":87,"props":391,"children":393},{"className":392},[],[394],{"type":55,"value":223},{"type":55,"value":396},". Do not create the tao-train-single-step runner until the\nimage is confirmed.",{"type":49,"tag":58,"props":398,"children":399},{},[400,402,408],{"type":55,"value":401},"After platform selection, run\n",{"type":49,"tag":87,"props":403,"children":405},{"className":404},[],[406],{"type":55,"value":407},"scripts\u002Flist_tao_platforms.py --platform \u003Cplatform> --format text",{"type":55,"value":409}," and ask\nonly for credentials relevant to that platform, plus any selected-model\ncredentials. Do not ask for unrelated platform credentials.",{"items":411,"total":511},[412,429,441,455,467,482,497],{"slug":413,"name":413,"fn":414,"description":415,"org":416,"tags":417,"stars":23,"repoUrl":24,"updatedAt":428},"accelerated-computing-cudf","accelerate data processing with cuDF","Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV\u002FParquet I\u002FO, nullable semantics, and multi-GPU DataFrame workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[418,421,424,425],{"name":419,"slug":420,"type":15},"Data Analysis","data-analysis",{"name":422,"slug":423,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":426,"slug":427,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":430,"name":430,"fn":431,"description":432,"org":433,"tags":434,"stars":23,"repoUrl":24,"updatedAt":440},"aiq-deploy","deploy and manage NVIDIA AI-Q infrastructure","Use when asked to install, deploy, run, validate, troubleshoot, or stop NVIDIA AI-Q Blueprint infrastructure.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[435,436,439],{"name":20,"slug":21,"type":15},{"name":437,"slug":438,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":442,"name":442,"fn":443,"description":444,"org":445,"tags":446,"stars":23,"repoUrl":24,"updatedAt":454},"aiq-research","conduct deep research with AI-Q","Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[447,450,451],{"name":448,"slug":449,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":452,"slug":453,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":456,"name":456,"fn":457,"description":458,"org":459,"tags":460,"stars":23,"repoUrl":24,"updatedAt":466},"amc-run-sample-calibration","run AMC sample dataset calibration","Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[461,462,463],{"name":419,"slug":420,"type":15},{"name":9,"slug":8,"type":15},{"name":464,"slug":465,"type":15},"Testing","testing","2026-07-17T05:29:03.913266",{"slug":468,"name":468,"fn":469,"description":470,"org":471,"tags":472,"stars":23,"repoUrl":24,"updatedAt":481},"amc-run-video-calibration","calibrate video datasets with AutoMagicCalib","Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API. Use when user has local MP4s and says 'calibrate my videos', 'run AMC on these videos', or similar. For RTSP\u002Flive streams, use amc-run-rtsp-calibration instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[473,474,477,478],{"name":13,"slug":14,"type":15},{"name":475,"slug":476,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":479,"slug":480,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":483,"name":483,"fn":484,"description":485,"org":486,"tags":487,"stars":23,"repoUrl":24,"updatedAt":496},"amc-setup-calibration-stack","deploy AutoMagicCalib microservice with Docker","Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[488,489,492,493],{"name":20,"slug":21,"type":15},{"name":490,"slug":491,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":494,"slug":495,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":498,"name":498,"fn":499,"description":500,"org":501,"tags":502,"stars":23,"repoUrl":24,"updatedAt":510},"cudaq-guide","develop quantum applications with CUDA-Q","CUDA-Q onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[503,504,507],{"name":9,"slug":8,"type":15},{"name":505,"slug":506,"type":15},"Quantum Computing","quantum-computing",{"name":508,"slug":509,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305,{"items":513,"total":662},[514,532,548,559,571,585,598,610,621,630,644,653],{"slug":515,"name":515,"fn":516,"description":517,"org":518,"tags":519,"stars":529,"repoUrl":530,"updatedAt":531},"nemoclaw-user-guide","retrieve NemoClaw documentation and configuration","Guides human users' AI agents to the NemoClaw docs MCP server and canonical Fern documentation in Markdown form. Use when users ask how to install, configure, operate, troubleshoot, secure, or learn NemoClaw with an AI coding assistant. Trigger keywords - nemoclaw docs, use nemoclaw with ai agent, nemoclaw mcp docs, nemoclaw install help, nemoclaw quickstart, nemoclaw markdown docs, llms.txt, agent skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[520,523,526],{"name":521,"slug":522,"type":15},"Documentation","documentation",{"name":524,"slug":525,"type":15},"MCP","mcp",{"name":527,"slug":528,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":533,"name":533,"fn":534,"description":535,"org":536,"tags":537,"stars":545,"repoUrl":546,"updatedAt":547},"mcore-build-and-dependency","manage Megatron-LM development environments","Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[538,541,542],{"name":539,"slug":540,"type":15},"Containers","containers",{"name":20,"slug":21,"type":15},{"name":543,"slug":544,"type":15},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":549,"name":549,"fn":550,"description":551,"org":552,"tags":553,"stars":545,"repoUrl":546,"updatedAt":558},"mcore-bump-base-image","update NVIDIA PyTorch base images","Bump the NVIDIA PyTorch base image (`nvcr.io\u002Fnvidia\u002Fpytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker\u002F.ngc_version.dev` and GitLab CI in `.gitlab\u002Fstages\u002F01.build.yml`), the post-bump CI loop (re-run functional tests, refresh golden values, mark broken tests), and the gotchas that bit PRs",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[554,557],{"name":555,"slug":556,"type":15},"CI\u002FCD","ci-cd",{"name":20,"slug":21,"type":15},"2026-07-14T05:25:59.97109",{"slug":560,"name":560,"fn":561,"description":562,"org":563,"tags":564,"stars":545,"repoUrl":546,"updatedAt":570},"mcore-cicd","manage CI\u002FCD pipelines for Megatron-LM","CI\u002FCD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request\u002FBRANCH ref — always dry-run and verify the destination first; never run against shared or protected branches), and CI failure investigation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[565,566,567],{"name":555,"slug":556,"type":15},{"name":20,"slug":21,"type":15},{"name":568,"slug":569,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":572,"name":572,"fn":573,"description":574,"org":575,"tags":576,"stars":545,"repoUrl":546,"updatedAt":584},"mcore-create-issue","investigate CI failures and create issues","Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[577,580,581],{"name":578,"slug":579,"type":15},"Debugging","debugging",{"name":568,"slug":569,"type":15},{"name":582,"slug":583,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":586,"name":586,"fn":587,"description":588,"org":589,"tags":590,"stars":545,"repoUrl":546,"updatedAt":597},"mcore-linting-and-formatting","lint and format Megatron-LM code","Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[591,594],{"name":592,"slug":593,"type":15},"Best Practices","best-practices",{"name":595,"slug":596,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":599,"name":599,"fn":600,"description":601,"org":602,"tags":603,"stars":545,"repoUrl":546,"updatedAt":609},"mcore-migrate-gpt-to-hybrid","migrate Megatron-LM models to HybridModel","Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[604,605,608],{"name":17,"slug":18,"type":15},{"name":606,"slug":607,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":611,"name":611,"fn":612,"description":613,"org":614,"tags":615,"stars":545,"repoUrl":546,"updatedAt":620},"mcore-onboard-gb200-1node-tests","onboard functional tests for GB200","Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[616,619],{"name":617,"slug":618,"type":15},"QA","qa",{"name":464,"slug":465,"type":15},"2026-07-14T05:25:53.673039",{"slug":622,"name":622,"fn":623,"description":624,"org":625,"tags":626,"stars":545,"repoUrl":546,"updatedAt":629},"mcore-run-on-slurm","launch distributed training jobs on SLURM","How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[627,628],{"name":20,"slug":21,"type":15},{"name":437,"slug":438,"type":15},"2026-07-14T05:25:49.362534",{"slug":631,"name":631,"fn":632,"description":633,"org":634,"tags":635,"stars":545,"repoUrl":546,"updatedAt":643},"mcore-split-pr","split pull requests to reduce review load","Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[636,639,640],{"name":637,"slug":638,"type":15},"Code Review","code-review",{"name":568,"slug":569,"type":15},{"name":641,"slug":642,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":645,"name":645,"fn":646,"description":647,"org":648,"tags":649,"stars":545,"repoUrl":546,"updatedAt":652},"mcore-testing","run and manage Megatron-LM tests","Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[650,651],{"name":617,"slug":618,"type":15},{"name":464,"slug":465,"type":15},"2026-07-14T05:25:54.928983",{"slug":654,"name":654,"fn":655,"description":656,"org":657,"tags":658,"stars":545,"repoUrl":546,"updatedAt":661},"nightly-sync","manage nightly main-to-dev sync workflows","Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[659,660],{"name":13,"slug":14,"type":15},{"name":555,"slug":556,"type":15},"2026-07-30T05:29:03.275638",496]