[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-nemotron-speech":3,"mdc--3z7yuy-key":34,"related-repo-nvidia-nemotron-speech":1203,"related-org-nvidia-nemotron-speech":1307},{"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},"nemotron-speech","deploy and run NVIDIA Nemotron Speech NIMs","Routes NVIDIA Nemotron Speech (Riva) NIM tasks — deploys, runs, and tests ASR, TTS, and NMT NIMs on build.nvidia.com or self-hosted.",{"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},"Audio","audio",{"name":21,"slug":22,"type":15},"Speech","speech",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:30:39.114079","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\u002Fnemotron-speech","---\nname: \"nemotron-speech\"\ndescription: Routes NVIDIA Nemotron Speech (Riva) NIM tasks — deploys, runs, and tests ASR, TTS, and NMT NIMs on build.nvidia.com or self-hosted.\ntriggers:\n  - Nemotron Speech\n  - deploy Riva NIM\n  - deploy ASR\u002FTTS\u002FNMT NIM\n  - Riva ASR\n  - Riva TTS\n  - Riva translation\n  - Parakeet\n  - Canary\n  - Whisper\n  - Nemotron ASR Streaming\n  - Magpie TTS\n  - DNT tag\n  - nemo2riva\n  - riva-build\n  - riva-deploy\n  - RMIR\n  - Riva NIM setup\n  - NGC API key\n  - force_eou\n  - Silero VAD\n  - Sortformer diarization\n  - chunk size Riva\n  - Riva HTTP\n  - Riva WebSocket\n  - grpc.nvcf.nvidia.com\n  - build.nvidia.com Riva\nversion: \"1.0.0\"\nlicense: Apache-2.0\nmetadata:\n  author: \"Nemotron Speech Team\"\n  team: riva\n  tags:\n    - nvidia\n    - nemotron-speech\n    - riva\n    - nim\n    - asr\n    - tts\n    - nmt\n    - speech\n    - speech-to-text\n    - text-to-speech\n    - translation\n    - parakeet\n    - canary\n    - whisper\n    - magpie\n    - nemotron\n    - grpc\n    - http\n    - websocket\n    - cloud\n    - nvcf\n  domain: ml\n---\n\n# Nemotron Speech Skills\n\n> **Note:** \"Nemotron Speech\" is the public-facing name for what NVIDIA documents today as **Riva** \u002F **Riva NIM**. All commands, container images, gRPC APIs, Python imports, and documentation URLs still use **\"Riva\"** — the rename is brand-only. Do not rename commands, images, or doc URLs.\n>\n> **Agent:** When walking the user through a multi-step workflow, announce each step before presenting it: **Step N\u002FM — Step Title** (e.g., \"**Step 1\u002F4 — Deploy the Container**\").\n\n## Purpose\n\nSingle entry point for all NVIDIA Nemotron Speech (Riva) NIM workflows: ASR (speech-to-text), TTS (text-to-speech), and NMT (translation). Covers cloud-hosted inference via build.nvidia.com, self-hosted Docker deployment, client-protocol choice for ASR (gRPC, HTTP, WebSocket), custom NeMo model deployment via `riva-build`, ASR pipeline tuning (VAD, diarization, language models), and the prerequisite Docker \u002F NGC \u002F driver setup.\n\n## When to Use This Skill\n\nUse this skill for any Nemotron Speech \u002F Riva NIM task — deployment, testing, custom model build, system requirements check, or model selection across ASR \u002F TTS \u002F NMT modalities.\n\n## Workflow\n\nIdentify the user's task type, then load the corresponding reference file from `references\u002F`. The reference files contain the detailed per-workflow content; this SKILL.md is a routing surface. Load only the reference relevant to the task at hand.\n\n## Prerequisites\n\n- For **self-hosted deployment**: NVIDIA AI Enterprise (NVAIE) entitlement, then complete the environment setup — NVIDIA drivers, Docker, Container Toolkit, NGC API key, Riva Python client. See [`references\u002Fsetup.md`](references\u002Fsetup.md).\n- For **cloud-hosted inference**: `pip install -U nvidia-riva-client` and a valid `NVIDIA_API_KEY` from https:\u002F\u002Fbuild.nvidia.com.\n- Treat `NVIDIA_API_KEY` and `NGC_API_KEY` as secrets: never print, paste, commit, or log real key values. Prefer `--password-stdin` for Docker login and store persistent keys in a credential manager or a `chmod 600` env file rather than world-readable shell startup files.\n- For **self-hosted Docker model caching**: host directories mounted at `\u002Fopt\u002Fnim\u002F.cache` must be writable by the container user (the NIM container runs as `nvs:1000` internally), not just the host user. Run `sudo chown 1000:1000 $LOCAL_NIM_CACHE` after creating the directory so the container can write to it. Avoid world-writable modes — they let any local user replace cached model artifacts. Also avoid `-u \"$(id -u):$(id -g)\"` on the docker run — `\u002Fopt\u002Fnim\u002Fworkspace` inside the container isn't writable to arbitrary UIDs. If you see `I\u002FO error Permission denied (os error 13)` during model download, the host directory ownership is the issue.\n\n## Instructions\n\n- Match the user's task to one reference file and load only that file; the references are detailed, so progressive disclosure keeps context tight.\n- Route setup requests for drivers, Docker, Container Toolkit, and NGC to [`references\u002Fsetup.md`](references\u002Fsetup.md).\n- Route GPU compatibility, deployment readiness, and container health checks to [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md).\n- Route model choice across ASR, TTS, and NMT to [`references\u002Fmodel-selection.md`](references\u002Fmodel-selection.md).\n- Route ASR deployment or inference for Parakeet, Canary, Whisper, and Nemotron ASR Streaming to [`references\u002Fasr.md`](references\u002Fasr.md).\n- Route custom-trained NeMo ASR deployment (`.nemo` → RMIR → NIM) to [`references\u002Fasr-custom.md`](references\u002Fasr-custom.md).\n- Route ASR pipeline configuration for VAD, diarization, language models, and chunk size to [`references\u002Fpipelines.md`](references\u002Fpipelines.md).\n- Route TTS deployment or inference for Magpie to [`references\u002Ftts.md`](references\u002Ftts.md).\n- Route NMT deployment or inference for Riva Translate, language pairs, and DNT tags to [`references\u002Fnmt.md`](references\u002Fnmt.md).\n\n## Source of truth\n\nFor per-release detail — current model catalog, container IDs, function IDs, voice lists, VRAM minimums, per-model feature support — **fetch or open the canonical NVIDIA doc** rather than relying on text in this SKILL.md or the references. Each reference file includes its own routing table to the relevant doc pages.\n\nTop-level landing pages:\n\n| Topic | URL |\n|---|---|\n| ASR support matrix | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Fasr.html |\n| TTS support matrix | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Ftts.html |\n| NMT support matrix | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Fnmt.html |\n| Prerequisites (driver \u002F GPU \u002F OS) | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fget-started\u002Fprerequisites.html |\n| ASR pipeline configuration | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fasr\u002Fcustomization\u002Fpipeline-configuration.html |\n| ASR runtime customization | https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fasr\u002Fcustomization\u002Fcustomization.html |\n| Cloud function IDs (per model) | `https:\u002F\u002Fbuild.nvidia.com\u002F\u003Corg>\u002F\u003Cmodel>\u002Fapi` |\n| NGC catalog | https:\u002F\u002Fcatalog.ngc.nvidia.com\u002Forgs\u002Fnim\u002Fteams\u002Fnvidia\u002Fmodels |\n\n## Examples\n\n**\"Deploy a Parakeet ASR NIM\"** → load [`references\u002Fasr.md`](references\u002Fasr.md), follow Option B (self-hosted), Steps 1–4.\n\n**\"Synthesize speech with Magpie\"** → load [`references\u002Ftts.md`](references\u002Ftts.md), follow Option A (cloud) or Option B (self-hosted).\n\n**\"Translate English to German\"** → load [`references\u002Fnmt.md`](references\u002Fnmt.md), follow the 4-step flow.\n\n**\"Convert my fine-tuned `.nemo` to a NIM\"** → load [`references\u002Fasr-custom.md`](references\u002Fasr-custom.md) for the 4-phase pipeline and [`references\u002Fpipelines.md`](references\u002Fpipelines.md) for build-time config.\n\n**\"Can my GPU run this?\"** → load [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md) and run the 6-step system check.\n\n**\"Which Riva model should I use?\"** → load [`references\u002Fmodel-selection.md`](references\u002Fmodel-selection.md), apply the decision framework, then fetch the support matrix for the specific current model name.\n\n## Naming & Terminology\n\n- **Skill brand**: Nemotron Speech (public-facing name).\n- **Internal naming preserved**: commands (`riva-build`, `riva-deploy`, `riva_streaming_asr_client`), Python client (`riva.client`), gRPC namespace (`nvidia.riva.asr.*`), container registry (`nvcr.io\u002Fnim\u002Fnvidia\u002F*`), and all NVIDIA documentation URLs still use **\"Riva\"**. Do not rename these in code, commands, or docs.\n\n## Troubleshooting\n\nFor task-specific runtime or modality issues, use the relevant reference file (`references\u002F\u003Ctask>.md`). Cross-cutting readiness checks:\n\n- **Container does not become ready** → [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md) (system check + health check table)\n- **Health check fails** → [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md)\n- **`docker pull` from `nvcr.io` returns 403** → [`references\u002Fsetup.md`](references\u002Fsetup.md) (Step 5 — Docker login)\n- **Wrong base image \u002F model architecture mismatch** → [`references\u002Fasr-custom.md`](references\u002Fasr-custom.md) (Phase 2 base image)\n- **VRAM \u002F GPU compatibility** → [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md), then verify on the support matrix\n\n## Limitations\n\n- x86_64 architecture only — WSL2 on Windows requires Podman and supports a subset of NIMs (see [`references\u002Fsetup.md`](references\u002Fsetup.md))\n- Self-hosted deployment requires an NVIDIA AI Enterprise license\n- Cloud-hosted inference requires an active `NVIDIA_API_KEY` and internet access\n- Public skill branding is **\"Nemotron Speech\"**; commands, container images, Python imports (`riva.client`), gRPC services (`nvidia.riva.*`), and NVIDIA documentation URLs still use **\"Riva\"** — follow official docs and catalogs for naming, do not rename these in commands or code\n\n## Next Steps\n\n- Verify hardware compatibility: [`references\u002Fdeployment-readiness-checks.md`](references\u002Fdeployment-readiness-checks.md)\n- Set up the environment: [`references\u002Fsetup.md`](references\u002Fsetup.md)\n- Pick a model: [`references\u002Fmodel-selection.md`](references\u002Fmodel-selection.md)\n- Deploy: [`references\u002Fasr.md`](references\u002Fasr.md), [`references\u002Ftts.md`](references\u002Ftts.md), or [`references\u002Fnmt.md`](references\u002Fnmt.md)\n",{"data":35,"body":86},{"name":4,"description":6,"triggers":36,"version":63,"license":26,"metadata":64},[37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62],"Nemotron Speech","deploy Riva NIM","deploy ASR\u002FTTS\u002FNMT NIM","Riva ASR","Riva TTS","Riva translation","Parakeet","Canary","Whisper","Nemotron ASR Streaming","Magpie TTS","DNT tag","nemo2riva","riva-build","riva-deploy","RMIR","Riva NIM setup","NGC API key","force_eou","Silero VAD","Sortformer diarization","chunk size Riva","Riva HTTP","Riva WebSocket","grpc.nvcf.nvidia.com","build.nvidia.com Riva","1.0.0",{"author":65,"team":66,"tags":67,"domain":85},"Nemotron Speech Team","riva",[8,4,66,68,69,70,71,22,72,73,74,75,76,77,78,79,80,81,82,83,84],"nim","asr","tts","nmt","speech-to-text","text-to-speech","translation","parakeet","canary","whisper","magpie","nemotron","grpc","http","websocket","cloud","nvcf","ml",{"type":87,"children":88},"root",[89,98,159,166,179,185,190,196,209,215,374,380,515,521,533,538,702,708,728,747,766,802,821,840,846,920,926,939,1049,1055,1123,1129],{"type":90,"tag":91,"props":92,"children":94},"element","h1",{"id":93},"nemotron-speech-skills",[95],{"type":96,"value":97},"text","Nemotron Speech Skills",{"type":90,"tag":99,"props":100,"children":101},"blockquote",{},[102,135],{"type":90,"tag":103,"props":104,"children":105},"p",{},[106,112,114,119,121,126,128,133],{"type":90,"tag":107,"props":108,"children":109},"strong",{},[110],{"type":96,"value":111},"Note:",{"type":96,"value":113}," \"Nemotron Speech\" is the public-facing name for what NVIDIA documents today as ",{"type":90,"tag":107,"props":115,"children":116},{},[117],{"type":96,"value":118},"Riva",{"type":96,"value":120}," \u002F ",{"type":90,"tag":107,"props":122,"children":123},{},[124],{"type":96,"value":125},"Riva NIM",{"type":96,"value":127},". All commands, container images, gRPC APIs, Python imports, and documentation URLs still use ",{"type":90,"tag":107,"props":129,"children":130},{},[131],{"type":96,"value":132},"\"Riva\"",{"type":96,"value":134}," — the rename is brand-only. Do not rename commands, images, or doc URLs.",{"type":90,"tag":103,"props":136,"children":137},{},[138,143,145,150,152,157],{"type":90,"tag":107,"props":139,"children":140},{},[141],{"type":96,"value":142},"Agent:",{"type":96,"value":144}," When walking the user through a multi-step workflow, announce each step before presenting it: ",{"type":90,"tag":107,"props":146,"children":147},{},[148],{"type":96,"value":149},"Step N\u002FM — Step Title",{"type":96,"value":151}," (e.g., \"",{"type":90,"tag":107,"props":153,"children":154},{},[155],{"type":96,"value":156},"Step 1\u002F4 — Deploy the Container",{"type":96,"value":158},"\").",{"type":90,"tag":160,"props":161,"children":163},"h2",{"id":162},"purpose",[164],{"type":96,"value":165},"Purpose",{"type":90,"tag":103,"props":167,"children":168},{},[169,171,177],{"type":96,"value":170},"Single entry point for all NVIDIA Nemotron Speech (Riva) NIM workflows: ASR (speech-to-text), TTS (text-to-speech), and NMT (translation). Covers cloud-hosted inference via build.nvidia.com, self-hosted Docker deployment, client-protocol choice for ASR (gRPC, HTTP, WebSocket), custom NeMo model deployment via ",{"type":90,"tag":172,"props":173,"children":175},"code",{"className":174},[],[176],{"type":96,"value":50},{"type":96,"value":178},", ASR pipeline tuning (VAD, diarization, language models), and the prerequisite Docker \u002F NGC \u002F driver setup.",{"type":90,"tag":160,"props":180,"children":182},{"id":181},"when-to-use-this-skill",[183],{"type":96,"value":184},"When to Use This Skill",{"type":90,"tag":103,"props":186,"children":187},{},[188],{"type":96,"value":189},"Use this skill for any Nemotron Speech \u002F Riva NIM task — deployment, testing, custom model build, system requirements check, or model selection across ASR \u002F TTS \u002F NMT modalities.",{"type":90,"tag":160,"props":191,"children":193},{"id":192},"workflow",[194],{"type":96,"value":195},"Workflow",{"type":90,"tag":103,"props":197,"children":198},{},[199,201,207],{"type":96,"value":200},"Identify the user's task type, then load the corresponding reference file from ",{"type":90,"tag":172,"props":202,"children":204},{"className":203},[],[205],{"type":96,"value":206},"references\u002F",{"type":96,"value":208},". The reference files contain the detailed per-workflow content; this SKILL.md is a routing surface. Load only the reference relevant to the task at hand.",{"type":90,"tag":160,"props":210,"children":212},{"id":211},"prerequisites",[213],{"type":96,"value":214},"Prerequisites",{"type":90,"tag":216,"props":217,"children":218},"ul",{},[219,244,279,315],{"type":90,"tag":220,"props":221,"children":222},"li",{},[223,225,230,232,242],{"type":96,"value":224},"For ",{"type":90,"tag":107,"props":226,"children":227},{},[228],{"type":96,"value":229},"self-hosted deployment",{"type":96,"value":231},": NVIDIA AI Enterprise (NVAIE) entitlement, then complete the environment setup — NVIDIA drivers, Docker, Container Toolkit, NGC API key, Riva Python client. See ",{"type":90,"tag":233,"props":234,"children":236},"a",{"href":235},"references\u002Fsetup.md",[237],{"type":90,"tag":172,"props":238,"children":240},{"className":239},[],[241],{"type":96,"value":235},{"type":96,"value":243},".",{"type":90,"tag":220,"props":245,"children":246},{},[247,248,253,255,261,263,269,271,278],{"type":96,"value":224},{"type":90,"tag":107,"props":249,"children":250},{},[251],{"type":96,"value":252},"cloud-hosted inference",{"type":96,"value":254},": ",{"type":90,"tag":172,"props":256,"children":258},{"className":257},[],[259],{"type":96,"value":260},"pip install -U nvidia-riva-client",{"type":96,"value":262}," and a valid ",{"type":90,"tag":172,"props":264,"children":266},{"className":265},[],[267],{"type":96,"value":268},"NVIDIA_API_KEY",{"type":96,"value":270}," from ",{"type":90,"tag":233,"props":272,"children":276},{"href":273,"rel":274},"https:\u002F\u002Fbuild.nvidia.com",[275],"nofollow",[277],{"type":96,"value":273},{"type":96,"value":243},{"type":90,"tag":220,"props":280,"children":281},{},[282,284,289,291,297,299,305,307,313],{"type":96,"value":283},"Treat ",{"type":90,"tag":172,"props":285,"children":287},{"className":286},[],[288],{"type":96,"value":268},{"type":96,"value":290}," and ",{"type":90,"tag":172,"props":292,"children":294},{"className":293},[],[295],{"type":96,"value":296},"NGC_API_KEY",{"type":96,"value":298}," as secrets: never print, paste, commit, or log real key values. Prefer ",{"type":90,"tag":172,"props":300,"children":302},{"className":301},[],[303],{"type":96,"value":304},"--password-stdin",{"type":96,"value":306}," for Docker login and store persistent keys in a credential manager or a ",{"type":90,"tag":172,"props":308,"children":310},{"className":309},[],[311],{"type":96,"value":312},"chmod 600",{"type":96,"value":314}," env file rather than world-readable shell startup files.",{"type":90,"tag":220,"props":316,"children":317},{},[318,319,324,326,332,334,340,342,348,350,356,358,364,366,372],{"type":96,"value":224},{"type":90,"tag":107,"props":320,"children":321},{},[322],{"type":96,"value":323},"self-hosted Docker model caching",{"type":96,"value":325},": host directories mounted at ",{"type":90,"tag":172,"props":327,"children":329},{"className":328},[],[330],{"type":96,"value":331},"\u002Fopt\u002Fnim\u002F.cache",{"type":96,"value":333}," must be writable by the container user (the NIM container runs as ",{"type":90,"tag":172,"props":335,"children":337},{"className":336},[],[338],{"type":96,"value":339},"nvs:1000",{"type":96,"value":341}," internally), not just the host user. Run ",{"type":90,"tag":172,"props":343,"children":345},{"className":344},[],[346],{"type":96,"value":347},"sudo chown 1000:1000 $LOCAL_NIM_CACHE",{"type":96,"value":349}," after creating the directory so the container can write to it. Avoid world-writable modes — they let any local user replace cached model artifacts. Also avoid ",{"type":90,"tag":172,"props":351,"children":353},{"className":352},[],[354],{"type":96,"value":355},"-u \"$(id -u):$(id -g)\"",{"type":96,"value":357}," on the docker run — ",{"type":90,"tag":172,"props":359,"children":361},{"className":360},[],[362],{"type":96,"value":363},"\u002Fopt\u002Fnim\u002Fworkspace",{"type":96,"value":365}," inside the container isn't writable to arbitrary UIDs. If you see ",{"type":90,"tag":172,"props":367,"children":369},{"className":368},[],[370],{"type":96,"value":371},"I\u002FO error Permission denied (os error 13)",{"type":96,"value":373}," during model download, the host directory ownership is the issue.",{"type":90,"tag":160,"props":375,"children":377},{"id":376},"instructions",[378],{"type":96,"value":379},"Instructions",{"type":90,"tag":216,"props":381,"children":382},{},[383,388,402,417,432,447,470,485,500],{"type":90,"tag":220,"props":384,"children":385},{},[386],{"type":96,"value":387},"Match the user's task to one reference file and load only that file; the references are detailed, so progressive disclosure keeps context tight.",{"type":90,"tag":220,"props":389,"children":390},{},[391,393,401],{"type":96,"value":392},"Route setup requests for drivers, Docker, Container Toolkit, and NGC to ",{"type":90,"tag":233,"props":394,"children":395},{"href":235},[396],{"type":90,"tag":172,"props":397,"children":399},{"className":398},[],[400],{"type":96,"value":235},{"type":96,"value":243},{"type":90,"tag":220,"props":403,"children":404},{},[405,407,416],{"type":96,"value":406},"Route GPU compatibility, deployment readiness, and container health checks to ",{"type":90,"tag":233,"props":408,"children":410},{"href":409},"references\u002Fdeployment-readiness-checks.md",[411],{"type":90,"tag":172,"props":412,"children":414},{"className":413},[],[415],{"type":96,"value":409},{"type":96,"value":243},{"type":90,"tag":220,"props":418,"children":419},{},[420,422,431],{"type":96,"value":421},"Route model choice across ASR, TTS, and NMT to ",{"type":90,"tag":233,"props":423,"children":425},{"href":424},"references\u002Fmodel-selection.md",[426],{"type":90,"tag":172,"props":427,"children":429},{"className":428},[],[430],{"type":96,"value":424},{"type":96,"value":243},{"type":90,"tag":220,"props":433,"children":434},{},[435,437,446],{"type":96,"value":436},"Route ASR deployment or inference for Parakeet, Canary, Whisper, and Nemotron ASR Streaming to ",{"type":90,"tag":233,"props":438,"children":440},{"href":439},"references\u002Fasr.md",[441],{"type":90,"tag":172,"props":442,"children":444},{"className":443},[],[445],{"type":96,"value":439},{"type":96,"value":243},{"type":90,"tag":220,"props":448,"children":449},{},[450,452,458,460,469],{"type":96,"value":451},"Route custom-trained NeMo ASR deployment (",{"type":90,"tag":172,"props":453,"children":455},{"className":454},[],[456],{"type":96,"value":457},".nemo",{"type":96,"value":459}," → RMIR → NIM) to ",{"type":90,"tag":233,"props":461,"children":463},{"href":462},"references\u002Fasr-custom.md",[464],{"type":90,"tag":172,"props":465,"children":467},{"className":466},[],[468],{"type":96,"value":462},{"type":96,"value":243},{"type":90,"tag":220,"props":471,"children":472},{},[473,475,484],{"type":96,"value":474},"Route ASR pipeline configuration for VAD, diarization, language models, and chunk size to ",{"type":90,"tag":233,"props":476,"children":478},{"href":477},"references\u002Fpipelines.md",[479],{"type":90,"tag":172,"props":480,"children":482},{"className":481},[],[483],{"type":96,"value":477},{"type":96,"value":243},{"type":90,"tag":220,"props":486,"children":487},{},[488,490,499],{"type":96,"value":489},"Route TTS deployment or inference for Magpie to ",{"type":90,"tag":233,"props":491,"children":493},{"href":492},"references\u002Ftts.md",[494],{"type":90,"tag":172,"props":495,"children":497},{"className":496},[],[498],{"type":96,"value":492},{"type":96,"value":243},{"type":90,"tag":220,"props":501,"children":502},{},[503,505,514],{"type":96,"value":504},"Route NMT deployment or inference for Riva Translate, language pairs, and DNT tags to ",{"type":90,"tag":233,"props":506,"children":508},{"href":507},"references\u002Fnmt.md",[509],{"type":90,"tag":172,"props":510,"children":512},{"className":511},[],[513],{"type":96,"value":507},{"type":96,"value":243},{"type":90,"tag":160,"props":516,"children":518},{"id":517},"source-of-truth",[519],{"type":96,"value":520},"Source of truth",{"type":90,"tag":103,"props":522,"children":523},{},[524,526,531],{"type":96,"value":525},"For per-release detail — current model catalog, container IDs, function IDs, voice lists, VRAM minimums, per-model feature support — ",{"type":90,"tag":107,"props":527,"children":528},{},[529],{"type":96,"value":530},"fetch or open the canonical NVIDIA doc",{"type":96,"value":532}," rather than relying on text in this SKILL.md or the references. Each reference file includes its own routing table to the relevant doc pages.",{"type":90,"tag":103,"props":534,"children":535},{},[536],{"type":96,"value":537},"Top-level landing pages:",{"type":90,"tag":539,"props":540,"children":541},"table",{},[542,561],{"type":90,"tag":543,"props":544,"children":545},"thead",{},[546],{"type":90,"tag":547,"props":548,"children":549},"tr",{},[550,556],{"type":90,"tag":551,"props":552,"children":553},"th",{},[554],{"type":96,"value":555},"Topic",{"type":90,"tag":551,"props":557,"children":558},{},[559],{"type":96,"value":560},"URL",{"type":90,"tag":562,"props":563,"children":564},"tbody",{},[565,583,600,617,634,651,668,685],{"type":90,"tag":547,"props":566,"children":567},{},[568,574],{"type":90,"tag":569,"props":570,"children":571},"td",{},[572],{"type":96,"value":573},"ASR support matrix",{"type":90,"tag":569,"props":575,"children":576},{},[577],{"type":90,"tag":233,"props":578,"children":581},{"href":579,"rel":580},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Fasr.html",[275],[582],{"type":96,"value":579},{"type":90,"tag":547,"props":584,"children":585},{},[586,591],{"type":90,"tag":569,"props":587,"children":588},{},[589],{"type":96,"value":590},"TTS support matrix",{"type":90,"tag":569,"props":592,"children":593},{},[594],{"type":90,"tag":233,"props":595,"children":598},{"href":596,"rel":597},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Ftts.html",[275],[599],{"type":96,"value":596},{"type":90,"tag":547,"props":601,"children":602},{},[603,608],{"type":90,"tag":569,"props":604,"children":605},{},[606],{"type":96,"value":607},"NMT support matrix",{"type":90,"tag":569,"props":609,"children":610},{},[611],{"type":90,"tag":233,"props":612,"children":615},{"href":613,"rel":614},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Freference\u002Fsupport-matrix\u002Fnmt.html",[275],[616],{"type":96,"value":613},{"type":90,"tag":547,"props":618,"children":619},{},[620,625],{"type":90,"tag":569,"props":621,"children":622},{},[623],{"type":96,"value":624},"Prerequisites (driver \u002F GPU \u002F OS)",{"type":90,"tag":569,"props":626,"children":627},{},[628],{"type":90,"tag":233,"props":629,"children":632},{"href":630,"rel":631},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fget-started\u002Fprerequisites.html",[275],[633],{"type":96,"value":630},{"type":90,"tag":547,"props":635,"children":636},{},[637,642],{"type":90,"tag":569,"props":638,"children":639},{},[640],{"type":96,"value":641},"ASR pipeline configuration",{"type":90,"tag":569,"props":643,"children":644},{},[645],{"type":90,"tag":233,"props":646,"children":649},{"href":647,"rel":648},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fasr\u002Fcustomization\u002Fpipeline-configuration.html",[275],[650],{"type":96,"value":647},{"type":90,"tag":547,"props":652,"children":653},{},[654,659],{"type":90,"tag":569,"props":655,"children":656},{},[657],{"type":96,"value":658},"ASR runtime customization",{"type":90,"tag":569,"props":660,"children":661},{},[662],{"type":90,"tag":233,"props":663,"children":666},{"href":664,"rel":665},"https:\u002F\u002Fdocs.nvidia.com\u002Fnim\u002Fspeech\u002Flatest\u002Fasr\u002Fcustomization\u002Fcustomization.html",[275],[667],{"type":96,"value":664},{"type":90,"tag":547,"props":669,"children":670},{},[671,676],{"type":90,"tag":569,"props":672,"children":673},{},[674],{"type":96,"value":675},"Cloud function IDs (per model)",{"type":90,"tag":569,"props":677,"children":678},{},[679],{"type":90,"tag":172,"props":680,"children":682},{"className":681},[],[683],{"type":96,"value":684},"https:\u002F\u002Fbuild.nvidia.com\u002F\u003Corg>\u002F\u003Cmodel>\u002Fapi",{"type":90,"tag":547,"props":686,"children":687},{},[688,693],{"type":90,"tag":569,"props":689,"children":690},{},[691],{"type":96,"value":692},"NGC catalog",{"type":90,"tag":569,"props":694,"children":695},{},[696],{"type":90,"tag":233,"props":697,"children":700},{"href":698,"rel":699},"https:\u002F\u002Fcatalog.ngc.nvidia.com\u002Forgs\u002Fnim\u002Fteams\u002Fnvidia\u002Fmodels",[275],[701],{"type":96,"value":698},{"type":90,"tag":160,"props":703,"children":705},{"id":704},"examples",[706],{"type":96,"value":707},"Examples",{"type":90,"tag":103,"props":709,"children":710},{},[711,716,718,726],{"type":90,"tag":107,"props":712,"children":713},{},[714],{"type":96,"value":715},"\"Deploy a Parakeet ASR NIM\"",{"type":96,"value":717}," → load ",{"type":90,"tag":233,"props":719,"children":720},{"href":439},[721],{"type":90,"tag":172,"props":722,"children":724},{"className":723},[],[725],{"type":96,"value":439},{"type":96,"value":727},", follow Option B (self-hosted), Steps 1–4.",{"type":90,"tag":103,"props":729,"children":730},{},[731,736,737,745],{"type":90,"tag":107,"props":732,"children":733},{},[734],{"type":96,"value":735},"\"Synthesize speech with Magpie\"",{"type":96,"value":717},{"type":90,"tag":233,"props":738,"children":739},{"href":492},[740],{"type":90,"tag":172,"props":741,"children":743},{"className":742},[],[744],{"type":96,"value":492},{"type":96,"value":746},", follow Option A (cloud) or Option B (self-hosted).",{"type":90,"tag":103,"props":748,"children":749},{},[750,755,756,764],{"type":90,"tag":107,"props":751,"children":752},{},[753],{"type":96,"value":754},"\"Translate English to German\"",{"type":96,"value":717},{"type":90,"tag":233,"props":757,"children":758},{"href":507},[759],{"type":90,"tag":172,"props":760,"children":762},{"className":761},[],[763],{"type":96,"value":507},{"type":96,"value":765},", follow the 4-step flow.",{"type":90,"tag":103,"props":767,"children":768},{},[769,781,782,790,792,800],{"type":90,"tag":107,"props":770,"children":771},{},[772,774,779],{"type":96,"value":773},"\"Convert my fine-tuned ",{"type":90,"tag":172,"props":775,"children":777},{"className":776},[],[778],{"type":96,"value":457},{"type":96,"value":780}," to a NIM\"",{"type":96,"value":717},{"type":90,"tag":233,"props":783,"children":784},{"href":462},[785],{"type":90,"tag":172,"props":786,"children":788},{"className":787},[],[789],{"type":96,"value":462},{"type":96,"value":791}," for the 4-phase pipeline and ",{"type":90,"tag":233,"props":793,"children":794},{"href":477},[795],{"type":90,"tag":172,"props":796,"children":798},{"className":797},[],[799],{"type":96,"value":477},{"type":96,"value":801}," for build-time config.",{"type":90,"tag":103,"props":803,"children":804},{},[805,810,811,819],{"type":90,"tag":107,"props":806,"children":807},{},[808],{"type":96,"value":809},"\"Can my GPU run this?\"",{"type":96,"value":717},{"type":90,"tag":233,"props":812,"children":813},{"href":409},[814],{"type":90,"tag":172,"props":815,"children":817},{"className":816},[],[818],{"type":96,"value":409},{"type":96,"value":820}," and run the 6-step system check.",{"type":90,"tag":103,"props":822,"children":823},{},[824,829,830,838],{"type":90,"tag":107,"props":825,"children":826},{},[827],{"type":96,"value":828},"\"Which Riva model should I use?\"",{"type":96,"value":717},{"type":90,"tag":233,"props":831,"children":832},{"href":424},[833],{"type":90,"tag":172,"props":834,"children":836},{"className":835},[],[837],{"type":96,"value":424},{"type":96,"value":839},", apply the decision framework, then fetch the support matrix for the specific current model name.",{"type":90,"tag":160,"props":841,"children":843},{"id":842},"naming-terminology",[844],{"type":96,"value":845},"Naming & Terminology",{"type":90,"tag":216,"props":847,"children":848},{},[849,859],{"type":90,"tag":220,"props":850,"children":851},{},[852,857],{"type":90,"tag":107,"props":853,"children":854},{},[855],{"type":96,"value":856},"Skill brand",{"type":96,"value":858},": Nemotron Speech (public-facing name).",{"type":90,"tag":220,"props":860,"children":861},{},[862,867,869,874,876,881,882,888,890,896,898,904,906,912,914,918],{"type":90,"tag":107,"props":863,"children":864},{},[865],{"type":96,"value":866},"Internal naming preserved",{"type":96,"value":868},": commands (",{"type":90,"tag":172,"props":870,"children":872},{"className":871},[],[873],{"type":96,"value":50},{"type":96,"value":875},", ",{"type":90,"tag":172,"props":877,"children":879},{"className":878},[],[880],{"type":96,"value":51},{"type":96,"value":875},{"type":90,"tag":172,"props":883,"children":885},{"className":884},[],[886],{"type":96,"value":887},"riva_streaming_asr_client",{"type":96,"value":889},"), Python client (",{"type":90,"tag":172,"props":891,"children":893},{"className":892},[],[894],{"type":96,"value":895},"riva.client",{"type":96,"value":897},"), gRPC namespace (",{"type":90,"tag":172,"props":899,"children":901},{"className":900},[],[902],{"type":96,"value":903},"nvidia.riva.asr.*",{"type":96,"value":905},"), container registry (",{"type":90,"tag":172,"props":907,"children":909},{"className":908},[],[910],{"type":96,"value":911},"nvcr.io\u002Fnim\u002Fnvidia\u002F*",{"type":96,"value":913},"), and all NVIDIA documentation URLs still use ",{"type":90,"tag":107,"props":915,"children":916},{},[917],{"type":96,"value":132},{"type":96,"value":919},". Do not rename these in code, commands, or docs.",{"type":90,"tag":160,"props":921,"children":923},{"id":922},"troubleshooting",[924],{"type":96,"value":925},"Troubleshooting",{"type":90,"tag":103,"props":927,"children":928},{},[929,931,937],{"type":96,"value":930},"For task-specific runtime or modality issues, use the relevant reference file (",{"type":90,"tag":172,"props":932,"children":934},{"className":933},[],[935],{"type":96,"value":936},"references\u002F\u003Ctask>.md",{"type":96,"value":938},"). Cross-cutting readiness checks:",{"type":90,"tag":216,"props":940,"children":941},{},[942,962,979,1011,1030],{"type":90,"tag":220,"props":943,"children":944},{},[945,950,952,960],{"type":90,"tag":107,"props":946,"children":947},{},[948],{"type":96,"value":949},"Container does not become ready",{"type":96,"value":951}," → ",{"type":90,"tag":233,"props":953,"children":954},{"href":409},[955],{"type":90,"tag":172,"props":956,"children":958},{"className":957},[],[959],{"type":96,"value":409},{"type":96,"value":961}," (system check + health check table)",{"type":90,"tag":220,"props":963,"children":964},{},[965,970,971],{"type":90,"tag":107,"props":966,"children":967},{},[968],{"type":96,"value":969},"Health check fails",{"type":96,"value":951},{"type":90,"tag":233,"props":972,"children":973},{"href":409},[974],{"type":90,"tag":172,"props":975,"children":977},{"className":976},[],[978],{"type":96,"value":409},{"type":90,"tag":220,"props":980,"children":981},{},[982,1000,1001,1009],{"type":90,"tag":107,"props":983,"children":984},{},[985,991,992,998],{"type":90,"tag":172,"props":986,"children":988},{"className":987},[],[989],{"type":96,"value":990},"docker pull",{"type":96,"value":270},{"type":90,"tag":172,"props":993,"children":995},{"className":994},[],[996],{"type":96,"value":997},"nvcr.io",{"type":96,"value":999}," returns 403",{"type":96,"value":951},{"type":90,"tag":233,"props":1002,"children":1003},{"href":235},[1004],{"type":90,"tag":172,"props":1005,"children":1007},{"className":1006},[],[1008],{"type":96,"value":235},{"type":96,"value":1010}," (Step 5 — Docker login)",{"type":90,"tag":220,"props":1012,"children":1013},{},[1014,1019,1020,1028],{"type":90,"tag":107,"props":1015,"children":1016},{},[1017],{"type":96,"value":1018},"Wrong base image \u002F model architecture mismatch",{"type":96,"value":951},{"type":90,"tag":233,"props":1021,"children":1022},{"href":462},[1023],{"type":90,"tag":172,"props":1024,"children":1026},{"className":1025},[],[1027],{"type":96,"value":462},{"type":96,"value":1029}," (Phase 2 base image)",{"type":90,"tag":220,"props":1031,"children":1032},{},[1033,1038,1039,1047],{"type":90,"tag":107,"props":1034,"children":1035},{},[1036],{"type":96,"value":1037},"VRAM \u002F GPU compatibility",{"type":96,"value":951},{"type":90,"tag":233,"props":1040,"children":1041},{"href":409},[1042],{"type":90,"tag":172,"props":1043,"children":1045},{"className":1044},[],[1046],{"type":96,"value":409},{"type":96,"value":1048},", then verify on the support matrix",{"type":90,"tag":160,"props":1050,"children":1052},{"id":1051},"limitations",[1053],{"type":96,"value":1054},"Limitations",{"type":90,"tag":216,"props":1056,"children":1057},{},[1058,1073,1078,1090],{"type":90,"tag":220,"props":1059,"children":1060},{},[1061,1063,1071],{"type":96,"value":1062},"x86_64 architecture only — WSL2 on Windows requires Podman and supports a subset of NIMs (see ",{"type":90,"tag":233,"props":1064,"children":1065},{"href":235},[1066],{"type":90,"tag":172,"props":1067,"children":1069},{"className":1068},[],[1070],{"type":96,"value":235},{"type":96,"value":1072},")",{"type":90,"tag":220,"props":1074,"children":1075},{},[1076],{"type":96,"value":1077},"Self-hosted deployment requires an NVIDIA AI Enterprise license",{"type":90,"tag":220,"props":1079,"children":1080},{},[1081,1083,1088],{"type":96,"value":1082},"Cloud-hosted inference requires an active ",{"type":90,"tag":172,"props":1084,"children":1086},{"className":1085},[],[1087],{"type":96,"value":268},{"type":96,"value":1089}," and internet access",{"type":90,"tag":220,"props":1091,"children":1092},{},[1093,1095,1100,1102,1107,1109,1115,1117,1121],{"type":96,"value":1094},"Public skill branding is ",{"type":90,"tag":107,"props":1096,"children":1097},{},[1098],{"type":96,"value":1099},"\"Nemotron Speech\"",{"type":96,"value":1101},"; commands, container images, Python imports (",{"type":90,"tag":172,"props":1103,"children":1105},{"className":1104},[],[1106],{"type":96,"value":895},{"type":96,"value":1108},"), gRPC services (",{"type":90,"tag":172,"props":1110,"children":1112},{"className":1111},[],[1113],{"type":96,"value":1114},"nvidia.riva.*",{"type":96,"value":1116},"), and NVIDIA documentation URLs still use ",{"type":90,"tag":107,"props":1118,"children":1119},{},[1120],{"type":96,"value":132},{"type":96,"value":1122}," — follow official docs and catalogs for naming, do not rename these in commands or code",{"type":90,"tag":160,"props":1124,"children":1126},{"id":1125},"next-steps",[1127],{"type":96,"value":1128},"Next Steps",{"type":90,"tag":216,"props":1130,"children":1131},{},[1132,1145,1158,1171],{"type":90,"tag":220,"props":1133,"children":1134},{},[1135,1137],{"type":96,"value":1136},"Verify hardware compatibility: ",{"type":90,"tag":233,"props":1138,"children":1139},{"href":409},[1140],{"type":90,"tag":172,"props":1141,"children":1143},{"className":1142},[],[1144],{"type":96,"value":409},{"type":90,"tag":220,"props":1146,"children":1147},{},[1148,1150],{"type":96,"value":1149},"Set up the environment: ",{"type":90,"tag":233,"props":1151,"children":1152},{"href":235},[1153],{"type":90,"tag":172,"props":1154,"children":1156},{"className":1155},[],[1157],{"type":96,"value":235},{"type":90,"tag":220,"props":1159,"children":1160},{},[1161,1163],{"type":96,"value":1162},"Pick a model: ",{"type":90,"tag":233,"props":1164,"children":1165},{"href":424},[1166],{"type":90,"tag":172,"props":1167,"children":1169},{"className":1168},[],[1170],{"type":96,"value":424},{"type":90,"tag":220,"props":1172,"children":1173},{},[1174,1176,1184,1185,1193,1195],{"type":96,"value":1175},"Deploy: ",{"type":90,"tag":233,"props":1177,"children":1178},{"href":439},[1179],{"type":90,"tag":172,"props":1180,"children":1182},{"className":1181},[],[1183],{"type":96,"value":439},{"type":96,"value":875},{"type":90,"tag":233,"props":1186,"children":1187},{"href":492},[1188],{"type":90,"tag":172,"props":1189,"children":1191},{"className":1190},[],[1192],{"type":96,"value":492},{"type":96,"value":1194},", or ",{"type":90,"tag":233,"props":1196,"children":1197},{"href":507},[1198],{"type":90,"tag":172,"props":1199,"children":1201},{"className":1200},[],[1202],{"type":96,"value":507},{"items":1204,"total":1306},[1205,1222,1234,1248,1260,1277,1292],{"slug":1206,"name":1206,"fn":1207,"description":1208,"org":1209,"tags":1210,"stars":23,"repoUrl":24,"updatedAt":1221},"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},[1211,1214,1217,1218],{"name":1212,"slug":1213,"type":15},"Data Analysis","data-analysis",{"name":1215,"slug":1216,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":1219,"slug":1220,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":1223,"name":1223,"fn":1224,"description":1225,"org":1226,"tags":1227,"stars":23,"repoUrl":24,"updatedAt":1233},"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},[1228,1229,1232],{"name":13,"slug":14,"type":15},{"name":1230,"slug":1231,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":1235,"name":1235,"fn":1236,"description":1237,"org":1238,"tags":1239,"stars":23,"repoUrl":24,"updatedAt":1247},"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},[1240,1243,1244],{"name":1241,"slug":1242,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":1245,"slug":1246,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":1249,"name":1249,"fn":1250,"description":1251,"org":1252,"tags":1253,"stars":23,"repoUrl":24,"updatedAt":1259},"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},[1254,1255,1256],{"name":1212,"slug":1213,"type":15},{"name":9,"slug":8,"type":15},{"name":1257,"slug":1258,"type":15},"Testing","testing","2026-07-17T05:29:03.913266",{"slug":1261,"name":1261,"fn":1262,"description":1263,"org":1264,"tags":1265,"stars":23,"repoUrl":24,"updatedAt":1276},"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},[1266,1269,1272,1273],{"name":1267,"slug":1268,"type":15},"Automation","automation",{"name":1270,"slug":1271,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":1274,"slug":1275,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":1278,"name":1278,"fn":1279,"description":1280,"org":1281,"tags":1282,"stars":23,"repoUrl":24,"updatedAt":1291},"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},[1283,1284,1287,1288],{"name":13,"slug":14,"type":15},{"name":1285,"slug":1286,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":1289,"slug":1290,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":1293,"name":1293,"fn":1294,"description":1295,"org":1296,"tags":1297,"stars":23,"repoUrl":24,"updatedAt":1305},"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},[1298,1299,1302],{"name":9,"slug":8,"type":15},{"name":1300,"slug":1301,"type":15},"Quantum Computing","quantum-computing",{"name":1303,"slug":1304,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305,{"items":1308,"total":1459},[1309,1327,1343,1354,1366,1380,1393,1407,1418,1427,1441,1450],{"slug":1310,"name":1310,"fn":1311,"description":1312,"org":1313,"tags":1314,"stars":1324,"repoUrl":1325,"updatedAt":1326},"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},[1315,1318,1321],{"name":1316,"slug":1317,"type":15},"Documentation","documentation",{"name":1319,"slug":1320,"type":15},"MCP","mcp",{"name":1322,"slug":1323,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1328,"name":1328,"fn":1329,"description":1330,"org":1331,"tags":1332,"stars":1340,"repoUrl":1341,"updatedAt":1342},"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},[1333,1336,1337],{"name":1334,"slug":1335,"type":15},"Containers","containers",{"name":13,"slug":14,"type":15},{"name":1338,"slug":1339,"type":15},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1344,"name":1344,"fn":1345,"description":1346,"org":1347,"tags":1348,"stars":1340,"repoUrl":1341,"updatedAt":1353},"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},[1349,1352],{"name":1350,"slug":1351,"type":15},"CI\u002FCD","ci-cd",{"name":13,"slug":14,"type":15},"2026-07-14T05:25:59.97109",{"slug":1355,"name":1355,"fn":1356,"description":1357,"org":1358,"tags":1359,"stars":1340,"repoUrl":1341,"updatedAt":1365},"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},[1360,1361,1362],{"name":1350,"slug":1351,"type":15},{"name":13,"slug":14,"type":15},{"name":1363,"slug":1364,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1367,"name":1367,"fn":1368,"description":1369,"org":1370,"tags":1371,"stars":1340,"repoUrl":1341,"updatedAt":1379},"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},[1372,1375,1376],{"name":1373,"slug":1374,"type":15},"Debugging","debugging",{"name":1363,"slug":1364,"type":15},{"name":1377,"slug":1378,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1381,"name":1381,"fn":1382,"description":1383,"org":1384,"tags":1385,"stars":1340,"repoUrl":1341,"updatedAt":1392},"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},[1386,1389],{"name":1387,"slug":1388,"type":15},"Best Practices","best-practices",{"name":1390,"slug":1391,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1394,"name":1394,"fn":1395,"description":1396,"org":1397,"tags":1398,"stars":1340,"repoUrl":1341,"updatedAt":1406},"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},[1399,1402,1405],{"name":1400,"slug":1401,"type":15},"Machine Learning","machine-learning",{"name":1403,"slug":1404,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1408,"name":1408,"fn":1409,"description":1410,"org":1411,"tags":1412,"stars":1340,"repoUrl":1341,"updatedAt":1417},"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},[1413,1416],{"name":1414,"slug":1415,"type":15},"QA","qa",{"name":1257,"slug":1258,"type":15},"2026-07-14T05:25:53.673039",{"slug":1419,"name":1419,"fn":1420,"description":1421,"org":1422,"tags":1423,"stars":1340,"repoUrl":1341,"updatedAt":1426},"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},[1424,1425],{"name":13,"slug":14,"type":15},{"name":1230,"slug":1231,"type":15},"2026-07-14T05:25:49.362534",{"slug":1428,"name":1428,"fn":1429,"description":1430,"org":1431,"tags":1432,"stars":1340,"repoUrl":1341,"updatedAt":1440},"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},[1433,1436,1437],{"name":1434,"slug":1435,"type":15},"Code Review","code-review",{"name":1363,"slug":1364,"type":15},{"name":1438,"slug":1439,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1442,"name":1442,"fn":1443,"description":1444,"org":1445,"tags":1446,"stars":1340,"repoUrl":1341,"updatedAt":1449},"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},[1447,1448],{"name":1414,"slug":1415,"type":15},{"name":1257,"slug":1258,"type":15},"2026-07-14T05:25:54.928983",{"slug":1451,"name":1451,"fn":1452,"description":1453,"org":1454,"tags":1455,"stars":1340,"repoUrl":1341,"updatedAt":1458},"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},[1456,1457],{"name":1267,"slug":1268,"type":15},{"name":1350,"slug":1351,"type":15},"2026-07-30T05:29:03.275638",496]