[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-deploy-k8s-gpu-cluster":3,"mdc--rpn1ve-key":34,"related-org-nvidia-deploy-k8s-gpu-cluster":504,"related-repo-nvidia-deploy-k8s-gpu-cluster":661},{"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},"deploy-k8s-gpu-cluster","deploy Kubernetes GPU clusters","Deploy a Kubernetes GPU cluster with DeepOps (Kubespray + GPU Operator) and prove it schedules GPU pods. Use when asked to deploy or rebuild Kubernetes on GPU servers with this repository.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,17,20],{"name":13,"slug":14,"type":15},"Deployment","deployment","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"Kubernetes","kubernetes",{"name":21,"slug":22,"type":15},"Infrastructure","infrastructure",1460,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fdeepops","2026-07-17T06:05:02.11504",null,353,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"Tools for building GPU clusters","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fdeepops\u002Ftree\u002FHEAD\u002Fskills\u002Fdeploy-k8s-gpu-cluster","---\nname: deploy-k8s-gpu-cluster\ndescription: Deploy a Kubernetes GPU cluster with DeepOps (Kubespray + GPU Operator) and prove it schedules GPU pods. Use when asked to deploy or rebuild Kubernetes on GPU servers with this repository.\n---\n\n# Deploy a Kubernetes GPU cluster\n\n## Preconditions\n\n- Ubuntu 22.04\u002F24.04 or RHEL\u002FRocky 8\u002F9 hosts you may fully manage (driver\n  installs may reboot them; no active users or workloads).\n- SSH access from the provisioning machine to every host as a sudo-capable\n  user.\n- `submodules\u002Fkubespray` initialized — Kubernetes playbooks fail on missing\n  `kubespray_defaults` role imports without it.\n- Run everything from the repository root.\n\n## Procedure\n\n1. Prepare the environment and verify it:\n\n   ```bash\n   git submodule update --init --recursive\n   .\u002Fscripts\u002Fsetup.sh\n   cp -r config.example config\n   ```\n\n2. Edit `config\u002Finventory`: control plane nodes under\n   `[kube_control_plane]` and `[etcd]`, workers under `[kube_node]` (a\n   single machine can hold all three roles).\n\n3. Preflight — must pass before deploying:\n\n   ```bash\n   python3 scripts\u002Fvalidation\u002Fdeepops_doctor.py --remote --json\n   ```\n\n4. Deploy:\n\n   ```bash\n   ansible-playbook -l k8s_cluster playbooks\u002Fk8s-cluster.yml\n   ```\n\n   This runs Kubespray and installs the NVIDIA GPU Operator. Expect\n   roughly 45–90 minutes on a first run.\n\n5. Validate — the success signal is this, not the play recap:\n\n   ```bash\n   python3 scripts\u002Fvalidation\u002Fvalidate_k8s.py --json --cuda-smoke\n   ```\n\n   Require `\"ok\": true` with `nodes_ready == nodes_total`,\n   `gpus_allocatable > 0`, and `cuda_smoke_ok: true`.\n\n## Failure branches\n\n- **Playbook fails on a transient error**: rerun the same playbook;\n  Kubespray is rerun-safe. A converged rerun reports `changed=0`.\n- **Syntax\u002Fimport error mentioning `kubespray_defaults`**: the submodule is\n  not initialized; run `git submodule update --init --recursive`.\n- **`gpus_allocatable: 0`**: the GPU Operator stack is not ready. Check\n  `kubectl get pods -A | grep -i nvidia` — the driver DaemonSet can take\n  10+ minutes on first deploy; if pods are crash-looping, follow\n  `skills\u002Fdiagnose-driver-install\u002F`.\n- **CUDA smoke pod stuck `Pending`**: `kubectl -n deepops-validate\n  describe pod deepops-validate-cuda` — usually no allocatable GPU\n  (see above) or an image pull problem on airgapped networks (use\n  `--cuda-image` to point at a mirrored image).\n- **Single-node clusters**: control plane taints are handled by the\n  playbook for the single-node case; if pods stay Pending on a multi-role\n  node, check taints with `kubectl describe node \u003Cname> | grep -i taint`.\n",{"data":35,"body":36},{"name":4,"description":6},{"type":37,"children":38},"root",[39,48,55,95,101,371,377,498],{"type":40,"tag":41,"props":42,"children":44},"element","h1",{"id":43},"deploy-a-kubernetes-gpu-cluster",[45],{"type":46,"value":47},"text","Deploy a Kubernetes GPU cluster",{"type":40,"tag":49,"props":50,"children":52},"h2",{"id":51},"preconditions",[53],{"type":46,"value":54},"Preconditions",{"type":40,"tag":56,"props":57,"children":58},"ul",{},[59,65,70,90],{"type":40,"tag":60,"props":61,"children":62},"li",{},[63],{"type":46,"value":64},"Ubuntu 22.04\u002F24.04 or RHEL\u002FRocky 8\u002F9 hosts you may fully manage (driver\ninstalls may reboot them; no active users or workloads).",{"type":40,"tag":60,"props":66,"children":67},{},[68],{"type":46,"value":69},"SSH access from the provisioning machine to every host as a sudo-capable\nuser.",{"type":40,"tag":60,"props":71,"children":72},{},[73,80,82,88],{"type":40,"tag":74,"props":75,"children":77},"code",{"className":76},[],[78],{"type":46,"value":79},"submodules\u002Fkubespray",{"type":46,"value":81}," initialized — Kubernetes playbooks fail on missing\n",{"type":40,"tag":74,"props":83,"children":85},{"className":84},[],[86],{"type":46,"value":87},"kubespray_defaults",{"type":46,"value":89}," role imports without it.",{"type":40,"tag":60,"props":91,"children":92},{},[93],{"type":46,"value":94},"Run everything from the repository root.",{"type":40,"tag":49,"props":96,"children":98},{"id":97},"procedure",[99],{"type":46,"value":100},"Procedure",{"type":40,"tag":102,"props":103,"children":104},"ol",{},[105,187,224,259,300],{"type":40,"tag":60,"props":106,"children":107},{},[108,110],{"type":46,"value":109},"Prepare the environment and verify it:",{"type":40,"tag":111,"props":112,"children":117},"pre",{"className":113,"code":114,"language":115,"meta":116,"style":116},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","git submodule update --init --recursive\n.\u002Fscripts\u002Fsetup.sh\ncp -r config.example config\n","bash","",[118],{"type":40,"tag":74,"props":119,"children":120},{"__ignoreMap":116},[121,154,163],{"type":40,"tag":122,"props":123,"children":126},"span",{"class":124,"line":125},"line",1,[127,133,139,144,149],{"type":40,"tag":122,"props":128,"children":130},{"style":129},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[131],{"type":46,"value":132},"git",{"type":40,"tag":122,"props":134,"children":136},{"style":135},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[137],{"type":46,"value":138}," submodule",{"type":40,"tag":122,"props":140,"children":141},{"style":135},[142],{"type":46,"value":143}," update",{"type":40,"tag":122,"props":145,"children":146},{"style":135},[147],{"type":46,"value":148}," --init",{"type":40,"tag":122,"props":150,"children":151},{"style":135},[152],{"type":46,"value":153}," --recursive\n",{"type":40,"tag":122,"props":155,"children":157},{"class":124,"line":156},2,[158],{"type":40,"tag":122,"props":159,"children":160},{"style":129},[161],{"type":46,"value":162},".\u002Fscripts\u002Fsetup.sh\n",{"type":40,"tag":122,"props":164,"children":166},{"class":124,"line":165},3,[167,172,177,182],{"type":40,"tag":122,"props":168,"children":169},{"style":129},[170],{"type":46,"value":171},"cp",{"type":40,"tag":122,"props":173,"children":174},{"style":135},[175],{"type":46,"value":176}," -r",{"type":40,"tag":122,"props":178,"children":179},{"style":135},[180],{"type":46,"value":181}," config.example",{"type":40,"tag":122,"props":183,"children":184},{"style":135},[185],{"type":46,"value":186}," config\n",{"type":40,"tag":60,"props":188,"children":189},{},[190,192,198,200,206,208,214,216,222],{"type":46,"value":191},"Edit ",{"type":40,"tag":74,"props":193,"children":195},{"className":194},[],[196],{"type":46,"value":197},"config\u002Finventory",{"type":46,"value":199},": control plane nodes under\n",{"type":40,"tag":74,"props":201,"children":203},{"className":202},[],[204],{"type":46,"value":205},"[kube_control_plane]",{"type":46,"value":207}," and ",{"type":40,"tag":74,"props":209,"children":211},{"className":210},[],[212],{"type":46,"value":213},"[etcd]",{"type":46,"value":215},", workers under ",{"type":40,"tag":74,"props":217,"children":219},{"className":218},[],[220],{"type":46,"value":221},"[kube_node]",{"type":46,"value":223}," (a\nsingle machine can hold all three roles).",{"type":40,"tag":60,"props":225,"children":226},{},[227,229],{"type":46,"value":228},"Preflight — must pass before deploying:",{"type":40,"tag":111,"props":230,"children":232},{"className":113,"code":231,"language":115,"meta":116,"style":116},"python3 scripts\u002Fvalidation\u002Fdeepops_doctor.py --remote --json\n",[233],{"type":40,"tag":74,"props":234,"children":235},{"__ignoreMap":116},[236],{"type":40,"tag":122,"props":237,"children":238},{"class":124,"line":125},[239,244,249,254],{"type":40,"tag":122,"props":240,"children":241},{"style":129},[242],{"type":46,"value":243},"python3",{"type":40,"tag":122,"props":245,"children":246},{"style":135},[247],{"type":46,"value":248}," scripts\u002Fvalidation\u002Fdeepops_doctor.py",{"type":40,"tag":122,"props":250,"children":251},{"style":135},[252],{"type":46,"value":253}," --remote",{"type":40,"tag":122,"props":255,"children":256},{"style":135},[257],{"type":46,"value":258}," --json\n",{"type":40,"tag":60,"props":260,"children":261},{},[262,264,294,298],{"type":46,"value":263},"Deploy:",{"type":40,"tag":111,"props":265,"children":267},{"className":113,"code":266,"language":115,"meta":116,"style":116},"ansible-playbook -l k8s_cluster playbooks\u002Fk8s-cluster.yml\n",[268],{"type":40,"tag":74,"props":269,"children":270},{"__ignoreMap":116},[271],{"type":40,"tag":122,"props":272,"children":273},{"class":124,"line":125},[274,279,284,289],{"type":40,"tag":122,"props":275,"children":276},{"style":129},[277],{"type":46,"value":278},"ansible-playbook",{"type":40,"tag":122,"props":280,"children":281},{"style":135},[282],{"type":46,"value":283}," -l",{"type":40,"tag":122,"props":285,"children":286},{"style":135},[287],{"type":46,"value":288}," k8s_cluster",{"type":40,"tag":122,"props":290,"children":291},{"style":135},[292],{"type":46,"value":293}," playbooks\u002Fk8s-cluster.yml\n",{"type":40,"tag":295,"props":296,"children":297},"br",{},[],{"type":46,"value":299},"This runs Kubespray and installs the NVIDIA GPU Operator. Expect\nroughly 45–90 minutes on a first run.",{"type":40,"tag":60,"props":301,"children":302},{},[303,305,334,337,339,345,347,353,355,361,363,369],{"type":46,"value":304},"Validate — the success signal is this, not the play recap:",{"type":40,"tag":111,"props":306,"children":308},{"className":113,"code":307,"language":115,"meta":116,"style":116},"python3 scripts\u002Fvalidation\u002Fvalidate_k8s.py --json --cuda-smoke\n",[309],{"type":40,"tag":74,"props":310,"children":311},{"__ignoreMap":116},[312],{"type":40,"tag":122,"props":313,"children":314},{"class":124,"line":125},[315,319,324,329],{"type":40,"tag":122,"props":316,"children":317},{"style":129},[318],{"type":46,"value":243},{"type":40,"tag":122,"props":320,"children":321},{"style":135},[322],{"type":46,"value":323}," scripts\u002Fvalidation\u002Fvalidate_k8s.py",{"type":40,"tag":122,"props":325,"children":326},{"style":135},[327],{"type":46,"value":328}," --json",{"type":40,"tag":122,"props":330,"children":331},{"style":135},[332],{"type":46,"value":333}," --cuda-smoke\n",{"type":40,"tag":295,"props":335,"children":336},{},[],{"type":46,"value":338},"Require ",{"type":40,"tag":74,"props":340,"children":342},{"className":341},[],[343],{"type":46,"value":344},"\"ok\": true",{"type":46,"value":346}," with ",{"type":40,"tag":74,"props":348,"children":350},{"className":349},[],[351],{"type":46,"value":352},"nodes_ready == nodes_total",{"type":46,"value":354},",\n",{"type":40,"tag":74,"props":356,"children":358},{"className":357},[],[359],{"type":46,"value":360},"gpus_allocatable > 0",{"type":46,"value":362},", and ",{"type":40,"tag":74,"props":364,"children":366},{"className":365},[],[367],{"type":46,"value":368},"cuda_smoke_ok: true",{"type":46,"value":370},".",{"type":40,"tag":49,"props":372,"children":374},{"id":373},"failure-branches",[375],{"type":46,"value":376},"Failure branches",{"type":40,"tag":56,"props":378,"children":379},{},[380,398,420,449,481],{"type":40,"tag":60,"props":381,"children":382},{},[383,389,391,397],{"type":40,"tag":384,"props":385,"children":386},"strong",{},[387],{"type":46,"value":388},"Playbook fails on a transient error",{"type":46,"value":390},": rerun the same playbook;\nKubespray is rerun-safe. A converged rerun reports ",{"type":40,"tag":74,"props":392,"children":394},{"className":393},[],[395],{"type":46,"value":396},"changed=0",{"type":46,"value":370},{"type":40,"tag":60,"props":399,"children":400},{},[401,411,413,419],{"type":40,"tag":384,"props":402,"children":403},{},[404,406],{"type":46,"value":405},"Syntax\u002Fimport error mentioning ",{"type":40,"tag":74,"props":407,"children":409},{"className":408},[],[410],{"type":46,"value":87},{"type":46,"value":412},": the submodule is\nnot initialized; run ",{"type":40,"tag":74,"props":414,"children":416},{"className":415},[],[417],{"type":46,"value":418},"git submodule update --init --recursive",{"type":46,"value":370},{"type":40,"tag":60,"props":421,"children":422},{},[423,432,434,440,442,448],{"type":40,"tag":384,"props":424,"children":425},{},[426],{"type":40,"tag":74,"props":427,"children":429},{"className":428},[],[430],{"type":46,"value":431},"gpus_allocatable: 0",{"type":46,"value":433},": the GPU Operator stack is not ready. Check\n",{"type":40,"tag":74,"props":435,"children":437},{"className":436},[],[438],{"type":46,"value":439},"kubectl get pods -A | grep -i nvidia",{"type":46,"value":441}," — the driver DaemonSet can take\n10+ minutes on first deploy; if pods are crash-looping, follow\n",{"type":40,"tag":74,"props":443,"children":445},{"className":444},[],[446],{"type":46,"value":447},"skills\u002Fdiagnose-driver-install\u002F",{"type":46,"value":370},{"type":40,"tag":60,"props":450,"children":451},{},[452,463,465,471,473,479],{"type":40,"tag":384,"props":453,"children":454},{},[455,457],{"type":46,"value":456},"CUDA smoke pod stuck ",{"type":40,"tag":74,"props":458,"children":460},{"className":459},[],[461],{"type":46,"value":462},"Pending",{"type":46,"value":464},": ",{"type":40,"tag":74,"props":466,"children":468},{"className":467},[],[469],{"type":46,"value":470},"kubectl -n deepops-validate describe pod deepops-validate-cuda",{"type":46,"value":472}," — usually no allocatable GPU\n(see above) or an image pull problem on airgapped networks (use\n",{"type":40,"tag":74,"props":474,"children":476},{"className":475},[],[477],{"type":46,"value":478},"--cuda-image",{"type":46,"value":480}," to point at a mirrored image).",{"type":40,"tag":60,"props":482,"children":483},{},[484,489,491,497],{"type":40,"tag":384,"props":485,"children":486},{},[487],{"type":46,"value":488},"Single-node clusters",{"type":46,"value":490},": control plane taints are handled by the\nplaybook for the single-node case; if pods stay Pending on a multi-role\nnode, check taints with ",{"type":40,"tag":74,"props":492,"children":494},{"className":493},[],[495],{"type":46,"value":496},"kubectl describe node \u003Cname> | grep -i taint",{"type":46,"value":370},{"type":40,"tag":499,"props":500,"children":501},"style",{},[502],{"type":46,"value":503},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":505,"total":660},[506,524,540,551,563,577,590,604,617,626,640,649],{"slug":507,"name":507,"fn":508,"description":509,"org":510,"tags":511,"stars":521,"repoUrl":522,"updatedAt":523},"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},[512,515,518],{"name":513,"slug":514,"type":15},"Documentation","documentation",{"name":516,"slug":517,"type":15},"MCP","mcp",{"name":519,"slug":520,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":525,"name":525,"fn":526,"description":527,"org":528,"tags":529,"stars":537,"repoUrl":538,"updatedAt":539},"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},[530,533,534],{"name":531,"slug":532,"type":15},"Containers","containers",{"name":13,"slug":14,"type":15},{"name":535,"slug":536,"type":15},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":541,"name":541,"fn":542,"description":543,"org":544,"tags":545,"stars":537,"repoUrl":538,"updatedAt":550},"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},[546,549],{"name":547,"slug":548,"type":15},"CI\u002FCD","ci-cd",{"name":13,"slug":14,"type":15},"2026-07-14T05:25:59.97109",{"slug":552,"name":552,"fn":553,"description":554,"org":555,"tags":556,"stars":537,"repoUrl":538,"updatedAt":562},"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},[557,558,559],{"name":547,"slug":548,"type":15},{"name":13,"slug":14,"type":15},{"name":560,"slug":561,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":564,"name":564,"fn":565,"description":566,"org":567,"tags":568,"stars":537,"repoUrl":538,"updatedAt":576},"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},[569,572,573],{"name":570,"slug":571,"type":15},"Debugging","debugging",{"name":560,"slug":561,"type":15},{"name":574,"slug":575,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":578,"name":578,"fn":579,"description":580,"org":581,"tags":582,"stars":537,"repoUrl":538,"updatedAt":589},"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},[583,586],{"name":584,"slug":585,"type":15},"Best Practices","best-practices",{"name":587,"slug":588,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":591,"name":591,"fn":592,"description":593,"org":594,"tags":595,"stars":537,"repoUrl":538,"updatedAt":603},"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},[596,599,602],{"name":597,"slug":598,"type":15},"Machine Learning","machine-learning",{"name":600,"slug":601,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":605,"name":605,"fn":606,"description":607,"org":608,"tags":609,"stars":537,"repoUrl":538,"updatedAt":616},"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},[610,613],{"name":611,"slug":612,"type":15},"QA","qa",{"name":614,"slug":615,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":618,"name":618,"fn":619,"description":620,"org":621,"tags":622,"stars":537,"repoUrl":538,"updatedAt":625},"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},[623,624],{"name":13,"slug":14,"type":15},{"name":21,"slug":22,"type":15},"2026-07-14T05:25:49.362534",{"slug":627,"name":627,"fn":628,"description":629,"org":630,"tags":631,"stars":537,"repoUrl":538,"updatedAt":639},"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},[632,635,636],{"name":633,"slug":634,"type":15},"Code Review","code-review",{"name":560,"slug":561,"type":15},{"name":637,"slug":638,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":641,"name":641,"fn":642,"description":643,"org":644,"tags":645,"stars":537,"repoUrl":538,"updatedAt":648},"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},[646,647],{"name":611,"slug":612,"type":15},{"name":614,"slug":615,"type":15},"2026-07-14T05:25:54.928983",{"slug":650,"name":650,"fn":651,"description":652,"org":653,"tags":654,"stars":537,"repoUrl":538,"updatedAt":659},"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},[655,658],{"name":656,"slug":657,"type":15},"Automation","automation",{"name":547,"slug":548,"type":15},"2026-07-30T05:29:03.275638",496,{"items":662,"total":703},[663,670,680,690],{"slug":4,"name":4,"fn":5,"description":6,"org":664,"tags":665,"stars":23,"repoUrl":24,"updatedAt":25},{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[666,667,668,669],{"name":13,"slug":14,"type":15},{"name":21,"slug":22,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"slug":671,"name":671,"fn":672,"description":673,"org":674,"tags":675,"stars":23,"repoUrl":24,"updatedAt":679},"deploy-slurm-cluster","deploy Slurm GPU clusters","Deploy a Slurm GPU cluster with DeepOps and prove it works. Use when asked to deploy, install, or rebuild Slurm on one or more GPU servers with this repository.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[676,677,678],{"name":13,"slug":14,"type":15},{"name":21,"slug":22,"type":15},{"name":9,"slug":8,"type":15},"2026-07-17T06:05:02.462676",{"slug":681,"name":681,"fn":682,"description":683,"org":684,"tags":685,"stars":23,"repoUrl":24,"updatedAt":689},"diagnose-driver-install","diagnose NVIDIA driver installation failures","Diagnose NVIDIA driver installation failures on DeepOps-managed nodes — nvidia-smi errors, \"No devices were found\", DKMS build failures, or GPU pods crash-looping. Use before reinstalling anything.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[686,687,688],{"name":570,"slug":571,"type":15},{"name":21,"slug":22,"type":15},{"name":9,"slug":8,"type":15},"2026-07-17T06:05:02.811247",{"slug":691,"name":691,"fn":692,"description":693,"org":694,"tags":695,"stars":23,"repoUrl":24,"updatedAt":702},"validate-gpu-cluster","validate GPU cluster health","Check whether a DeepOps-deployed Slurm or Kubernetes GPU cluster is healthy and report a machine-readable verdict. Use for health checks, post-deploy verification, \"is the cluster working?\" questions, and after any node or driver change.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[696,697,700,701],{"name":21,"slug":22,"type":15},{"name":698,"slug":699,"type":15},"Monitoring","monitoring",{"name":9,"slug":8,"type":15},{"name":611,"slug":612,"type":15},"2026-07-17T06:07:15.184954",4]