[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-nv-segment-ct-finetune":3,"mdc--33x6px-key":34,"related-repo-nvidia-nv-segment-ct-finetune":2050,"related-org-nvidia-nv-segment-ct-finetune":2156},{"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},"nv-segment-ct-finetune","finetune CT segmentation models","Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,20],{"name":13,"slug":14,"type":15},"Healthcare","healthcare","tag",{"name":17,"slug":18,"type":15},"Deep Learning","deep-learning",{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},"Computer Vision","computer-vision",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:27:31.785224","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\u002Fnv-segment-ct-finetune","---\nname: nv-segment-ct-finetune\ndescription: Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.\nlicense: Apache-2.0\nallowed-tools: Bash\nmetadata:\n  author: NVIDIA MedTech Team\n  tags:\n    - MedTech\n    - CT\n    - finetuning\n    - segmentation\n---\n\n# NV-Segment-CT Finetune\n\n## Purpose\n\n- Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.\n- Wraps the upstream MONAI bundle entrypoint; do not replace it with handwritten training or inference code.\n- Manifest inputs are `dataset_dir`, `datalist`, `target_anatomy`, `label_mapping`, `smoke`, `sanity`, `auto_seg`, and `skip_formal_eval`.\n- Manifest outputs are `finetuned_ckpt` and schema-checked `result_json`.\n\n## Instructions\n\n- Run `scripts\u002Frun_finetune.py`; do not patch files under `bundle\u002F` or upstream checkouts during normal skill use.\n- For standalone Bash, include the fresh-environment setup line before the wrapper; benchmark venvs start empty.\n- Run the committed script in place from the repo root. Do not copy this skill to a runtime directory, and do not use `rm` or cleanup commands in generated invocations.\n- If a host exposes `run_script`, use `run_script(\"scripts\u002Frun_finetune.py\", args=[...])`; otherwise run from the repo root.\n- For the shortest workflow check, use `--smoke`; for MSD Task06 Lung Tumor reproduction, use `--sanity`.\n- Read `references\u002Ftask06-and-results.md` only when you need Task06 reference details, output-field definitions, or manual bundle setup notes.\n\n## Available Scripts\n\n| Script | Purpose | Arguments |\n|---|---|---|\n| `scripts\u002Frun_finetune.py` | Primary entrypoint declared by `skill_manifest.yaml`; stages configs, runs MONAI, and writes `output.json`. | `[FIXTURE_OR_DATASET] --output-dir OUT_DIR [--smoke] [--sanity] [--auto-seg] [--dataset-dir DIR] [--datalist JSON] [--target-anatomy TEXT] [--label-mapping JSON] [--patch-size JSON]` |\n\n## Prerequisites\n\n- Python 3.10+ with CUDA-capable Torch for GPU runs.\n- Runtime packages from `skill_manifest.yaml`, especially `monai==1.4.0`, `numpy\u003C2`, `nibabel`, `scipy`, `typer`, `PyYAML`, `fire`, `pytorch-ignite`, `einops`, and `huggingface_hub`.\n- Optional environment variables: `CUDA_VISIBLE_DEVICES` restricts visible GPUs; `NPROC_PER_NODE` overrides GPU count and values `>=2` select multi-GPU mode for non-sanity runs.\n- Side effects: writes generated bundle configs under `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002F`, including `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fauto_override.json`, `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Ftrain_continual_task06_lung.json`, and `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fdfw_no_logging.json`; writes checkpoints\u002Fevidence under `--output-dir`, may cache model assets under `~\u002F.cache\u002Fhuggingface\u002F`, and may contact `https:\u002F\u002Fhuggingface.co` or `https:\u002F\u002Fraw.githubusercontent.com`.\n\nFresh environment setup:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub\n```\n\nKnown upstream compatibility constraints:\n\n- DFW Task06 reference: Python `3.10.16`, MONAI `1.4.0`, Torch `2.7.0+cu126`.\n- Use exact `monai==1.4.0` for smoke, sanity, and evidence runs; MONAI 1.5.x can crash the upstream finetune loss on boolean labels.\n- Do not float the dependency as `monai>=1.4,\u003C1.6` in generated commands.\n\n## Usage\n\nSmoke-scale workflow check:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub && \\\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  PATH_TO_DATASET \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nUse the staged dataset as `PATH_TO_DATASET`. For the micro fixture, use `skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro`. Smoke mode proves wiring, config generation, checkpoint loading, and runtime compatibility; it is not a quality bar.\n\nMSD Task06 Lung Tumor sanity reproduction:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  \u002Fpath\u002Fto\u002FTask06 \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\nThe sanity preset follows the single-GPU DFW recipe: fold-0 validation, label mapping `[[1, 23]]` for `lung tumor`, automatic class-prompt segmentation, patch `[128,128,128]`, 5 epochs, and original-spacing `configs\u002Fevaluate.json` scoring before and after training. Expected reference range is pretrained Dice about `0.6697`, training-best Dice about `0.6905`, and fine-tuned formal Dice about `0.6836`.\n\nUser-data finetune:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  --dataset-dir \u002Fpath\u002Fto\u002Fdataset \\\n  --datalist \u002Fpath\u002Fto\u002Fdatalist.json \\\n  --target-anatomy \"lung tumor\" \\\n  --auto-seg \\\n  --epochs 5 \\\n  --patch-size '[128,128,128]' \\\n  --output-dir runs\u002Fnvseg_user_finetune\n```\n\nUse `--label-mapping '[[1, 23]]'` when local label values are custom or the anatomy name is ambiguous.\n\n## Examples\n\nSmoke run on a staged tiny dataset:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  runs\u002Fwith_vs_without_nv\u002F_inputs\u002Fnv_segment_ct_finetune\u002Finput_dataset \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nTask06 sanity run on a local MSD cache:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  .workbench_data\u002Fdatasets\u002FTask06_Lung \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\n## Data Contract\n\n- Preferred layout: `dataset\u002FimagesTr\u002F*.nii.gz` and `dataset\u002FlabelsTr\u002F*.nii.gz`.\n- Labels must align one-to-one with images by basename.\n- The target label value must be present in the training labels.\n- Use a datalist when patient-level splitting matters. The bundle default `fold` is `0`, so `fold: 0` entries are validation and all other folds are training.\n- Every trained foreground label must map to an existing VISTA3D global class id from `bundle\u002Flabel_dict.json`; this skill cannot invent a new class.\n\n## Results\n\nCheck `output.json` in the run directory first:\n\n- `formal_pretrained_val_dice` and `formal_finetuned_val_dice`: original-spacing pre\u002Fpost scores when formal eval is enabled.\n- `training_start_val_dice`, `val_dice_per_epoch`, and `training_best_val_dice`: training-time validation trace.\n- `finetuned_ckpt_matches_pretrained_weights`: detects the epoch-0 checkpoint trap when `val_at_start=true`.\n- `recommended_ckpt`: checkpoint to keep. Do not blindly use the last epoch or `model_finetune.pt`.\n- `runtime.oom`, `runtime.peak_gpu_mb`, and phase logs: distinguish OOM, slow validation, and process failure.\n\nDecision rule: prefer formal original-spacing pre\u002Fpost scores when present; reject tensor-identical \"fine-tuned\" checkpoints for sanity recovery; treat `improved: false` as valid evidence rather than a wrapper failure.\n\n## Limitations\n\n- Thin wrapper. Training, validation, transforms, and checkpointing are delegated to the upstream bundle in `bundle\u002F`.\n- Reproduction record only: the successful five-epoch Task06 run used Python\n  `3.12.3`, PyTorch `2.12.0+cu130` with CUDA `13.0`, MONAI `1.4.0`, NumPy\n  `1.26.4`, PyTorch-Ignite `0.5.4`, NiBabel `5.4.2`, SciPy `1.16.0`, einops\n  `0.8.2`, Fire `0.7.1`, Hugging Face Hub `0.36.2`, Transformers `4.57.6`,\n  Typer `0.25.1`, PyYAML `6.0.3`, and MLflow `3.14.0` on one NVIDIA RTX 6000\n  Ada 48 GB GPU. These versions document the evidence environment; they are\n  not additional package constraints or a claim that other versions cannot\n  work.\n- The auto-derived plan is heuristic; caller-provided `--patch-size`, `--cache-rate`, `--epochs`, and `--learning-rate` win.\n- The Task06 sanity recipe intentionally forces single-GPU execution to match the DFW reference. Multi-GPU mode for other datasets requires host `torchrun` support.\n- The paired verifier is CPU-only and audits the evidence pack; it does not re-run GPU segmentation.\n- Not for clinical deployment, clinical interpretation, autonomous diagnosis, or regulatory submission.\n\n## Troubleshooting\n\n| Error | Cause | Fix |\n|---|---|---|\n| Missing dependency or import error | Runtime drift from `skill_manifest.yaml`. | Install the packages above or use the documented environment. |\n| Low Task06 pretrained Dice | Wrong config, wrong checkpoint, data split drift, or dependency drift. | Compare environment fields and staged configs before changing training logic. |\n| `model_finetune.pt` matches pretrained | `val_at_start=true` selected epoch 0 as best. | Use `recommended_ckpt`; treat sanity recovery as failed unless a changed checkpoint improves formal Dice. |\n| Missing formal Dice fields | Formal eval failed or was skipped. | Inspect `eval_pretrained.log`, `eval_finetuned.log`, and `metrics.csv`. |\n| GPU out of memory | Patch\u002Fcache settings too large. | Reduce `--patch-size`, lower `--cache-rate`, or reduce workers. |\n| No validation cases | Datalist lacks `fold: 0`. | Provide at least one validation entry. |\n\n## Verification\n\nRun the implemented verifier when quality gates matter:\n\n```bash\npython -m eval_engine.run_trusted skills\u002Fnv-segment-ct-finetune \\\n  --fixture skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro \\\n  --out runs\u002Fnvseg_trusted\n```\n",{"data":35,"body":44},{"name":4,"description":6,"license":26,"allowed-tools":36,"metadata":37},"Bash",{"author":38,"tags":39},"NVIDIA MedTech Team",[40,41,42,43],"MedTech","CT","finetuning","segmentation",{"type":45,"children":46},"root",[47,55,62,161,167,263,269,341,347,532,538,644,649,705,711,716,895,916,921,979,1039,1044,1191,1204,1210,1215,1294,1299,1355,1361,1436,1442,1454,1554,1567,1573,1767,1773,1971,1977,1982,2044],{"type":48,"tag":49,"props":50,"children":51},"element","h1",{"id":4},[52],{"type":53,"value":54},"text","NV-Segment-CT Finetune",{"type":48,"tag":56,"props":57,"children":59},"h2",{"id":58},"purpose",[60],{"type":53,"value":61},"Purpose",{"type":48,"tag":63,"props":64,"children":65},"ul",{},[66,71,76,141],{"type":48,"tag":67,"props":68,"children":69},"li",{},[70],{"type":53,"value":6},{"type":48,"tag":67,"props":72,"children":73},{},[74],{"type":53,"value":75},"Wraps the upstream MONAI bundle entrypoint; do not replace it with handwritten training or inference code.",{"type":48,"tag":67,"props":77,"children":78},{},[79,81,88,90,96,97,103,104,110,111,117,118,124,125,131,133,139],{"type":53,"value":80},"Manifest inputs are ",{"type":48,"tag":82,"props":83,"children":85},"code",{"className":84},[],[86],{"type":53,"value":87},"dataset_dir",{"type":53,"value":89},", ",{"type":48,"tag":82,"props":91,"children":93},{"className":92},[],[94],{"type":53,"value":95},"datalist",{"type":53,"value":89},{"type":48,"tag":82,"props":98,"children":100},{"className":99},[],[101],{"type":53,"value":102},"target_anatomy",{"type":53,"value":89},{"type":48,"tag":82,"props":105,"children":107},{"className":106},[],[108],{"type":53,"value":109},"label_mapping",{"type":53,"value":89},{"type":48,"tag":82,"props":112,"children":114},{"className":113},[],[115],{"type":53,"value":116},"smoke",{"type":53,"value":89},{"type":48,"tag":82,"props":119,"children":121},{"className":120},[],[122],{"type":53,"value":123},"sanity",{"type":53,"value":89},{"type":48,"tag":82,"props":126,"children":128},{"className":127},[],[129],{"type":53,"value":130},"auto_seg",{"type":53,"value":132},", and ",{"type":48,"tag":82,"props":134,"children":136},{"className":135},[],[137],{"type":53,"value":138},"skip_formal_eval",{"type":53,"value":140},".",{"type":48,"tag":67,"props":142,"children":143},{},[144,146,152,154,160],{"type":53,"value":145},"Manifest outputs are ",{"type":48,"tag":82,"props":147,"children":149},{"className":148},[],[150],{"type":53,"value":151},"finetuned_ckpt",{"type":53,"value":153}," and schema-checked ",{"type":48,"tag":82,"props":155,"children":157},{"className":156},[],[158],{"type":53,"value":159},"result_json",{"type":53,"value":140},{"type":48,"tag":56,"props":162,"children":164},{"id":163},"instructions",[165],{"type":53,"value":166},"Instructions",{"type":48,"tag":63,"props":168,"children":169},{},[170,191,196,209,230,250],{"type":48,"tag":67,"props":171,"children":172},{},[173,175,181,183,189],{"type":53,"value":174},"Run ",{"type":48,"tag":82,"props":176,"children":178},{"className":177},[],[179],{"type":53,"value":180},"scripts\u002Frun_finetune.py",{"type":53,"value":182},"; do not patch files under ",{"type":48,"tag":82,"props":184,"children":186},{"className":185},[],[187],{"type":53,"value":188},"bundle\u002F",{"type":53,"value":190}," or upstream checkouts during normal skill use.",{"type":48,"tag":67,"props":192,"children":193},{},[194],{"type":53,"value":195},"For standalone Bash, include the fresh-environment setup line before the wrapper; benchmark venvs start empty.",{"type":48,"tag":67,"props":197,"children":198},{},[199,201,207],{"type":53,"value":200},"Run the committed script in place from the repo root. Do not copy this skill to a runtime directory, and do not use ",{"type":48,"tag":82,"props":202,"children":204},{"className":203},[],[205],{"type":53,"value":206},"rm",{"type":53,"value":208}," or cleanup commands in generated invocations.",{"type":48,"tag":67,"props":210,"children":211},{},[212,214,220,222,228],{"type":53,"value":213},"If a host exposes ",{"type":48,"tag":82,"props":215,"children":217},{"className":216},[],[218],{"type":53,"value":219},"run_script",{"type":53,"value":221},", use ",{"type":48,"tag":82,"props":223,"children":225},{"className":224},[],[226],{"type":53,"value":227},"run_script(\"scripts\u002Frun_finetune.py\", args=[...])",{"type":53,"value":229},"; otherwise run from the repo root.",{"type":48,"tag":67,"props":231,"children":232},{},[233,235,241,243,249],{"type":53,"value":234},"For the shortest workflow check, use ",{"type":48,"tag":82,"props":236,"children":238},{"className":237},[],[239],{"type":53,"value":240},"--smoke",{"type":53,"value":242},"; for MSD Task06 Lung Tumor reproduction, use ",{"type":48,"tag":82,"props":244,"children":246},{"className":245},[],[247],{"type":53,"value":248},"--sanity",{"type":53,"value":140},{"type":48,"tag":67,"props":251,"children":252},{},[253,255,261],{"type":53,"value":254},"Read ",{"type":48,"tag":82,"props":256,"children":258},{"className":257},[],[259],{"type":53,"value":260},"references\u002Ftask06-and-results.md",{"type":53,"value":262}," only when you need Task06 reference details, output-field definitions, or manual bundle setup notes.",{"type":48,"tag":56,"props":264,"children":266},{"id":265},"available-scripts",[267],{"type":53,"value":268},"Available Scripts",{"type":48,"tag":270,"props":271,"children":272},"table",{},[273,296],{"type":48,"tag":274,"props":275,"children":276},"thead",{},[277],{"type":48,"tag":278,"props":279,"children":280},"tr",{},[281,287,291],{"type":48,"tag":282,"props":283,"children":284},"th",{},[285],{"type":53,"value":286},"Script",{"type":48,"tag":282,"props":288,"children":289},{},[290],{"type":53,"value":61},{"type":48,"tag":282,"props":292,"children":293},{},[294],{"type":53,"value":295},"Arguments",{"type":48,"tag":297,"props":298,"children":299},"tbody",{},[300],{"type":48,"tag":278,"props":301,"children":302},{},[303,312,332],{"type":48,"tag":304,"props":305,"children":306},"td",{},[307],{"type":48,"tag":82,"props":308,"children":310},{"className":309},[],[311],{"type":53,"value":180},{"type":48,"tag":304,"props":313,"children":314},{},[315,317,323,325,331],{"type":53,"value":316},"Primary entrypoint declared by ",{"type":48,"tag":82,"props":318,"children":320},{"className":319},[],[321],{"type":53,"value":322},"skill_manifest.yaml",{"type":53,"value":324},"; stages configs, runs MONAI, and writes ",{"type":48,"tag":82,"props":326,"children":328},{"className":327},[],[329],{"type":53,"value":330},"output.json",{"type":53,"value":140},{"type":48,"tag":304,"props":333,"children":334},{},[335],{"type":48,"tag":82,"props":336,"children":338},{"className":337},[],[339],{"type":53,"value":340},"[FIXTURE_OR_DATASET] --output-dir OUT_DIR [--smoke] [--sanity] [--auto-seg] [--dataset-dir DIR] [--datalist JSON] [--target-anatomy TEXT] [--label-mapping JSON] [--patch-size JSON]",{"type":48,"tag":56,"props":342,"children":344},{"id":343},"prerequisites",[345],{"type":53,"value":346},"Prerequisites",{"type":48,"tag":63,"props":348,"children":349},{},[350,355,437,466],{"type":48,"tag":67,"props":351,"children":352},{},[353],{"type":53,"value":354},"Python 3.10+ with CUDA-capable Torch for GPU runs.",{"type":48,"tag":67,"props":356,"children":357},{},[358,360,365,367,373,374,380,381,387,388,394,395,401,402,408,409,415,416,422,423,429,430,436],{"type":53,"value":359},"Runtime packages from ",{"type":48,"tag":82,"props":361,"children":363},{"className":362},[],[364],{"type":53,"value":322},{"type":53,"value":366},", especially ",{"type":48,"tag":82,"props":368,"children":370},{"className":369},[],[371],{"type":53,"value":372},"monai==1.4.0",{"type":53,"value":89},{"type":48,"tag":82,"props":375,"children":377},{"className":376},[],[378],{"type":53,"value":379},"numpy\u003C2",{"type":53,"value":89},{"type":48,"tag":82,"props":382,"children":384},{"className":383},[],[385],{"type":53,"value":386},"nibabel",{"type":53,"value":89},{"type":48,"tag":82,"props":389,"children":391},{"className":390},[],[392],{"type":53,"value":393},"scipy",{"type":53,"value":89},{"type":48,"tag":82,"props":396,"children":398},{"className":397},[],[399],{"type":53,"value":400},"typer",{"type":53,"value":89},{"type":48,"tag":82,"props":403,"children":405},{"className":404},[],[406],{"type":53,"value":407},"PyYAML",{"type":53,"value":89},{"type":48,"tag":82,"props":410,"children":412},{"className":411},[],[413],{"type":53,"value":414},"fire",{"type":53,"value":89},{"type":48,"tag":82,"props":417,"children":419},{"className":418},[],[420],{"type":53,"value":421},"pytorch-ignite",{"type":53,"value":89},{"type":48,"tag":82,"props":424,"children":426},{"className":425},[],[427],{"type":53,"value":428},"einops",{"type":53,"value":132},{"type":48,"tag":82,"props":431,"children":433},{"className":432},[],[434],{"type":53,"value":435},"huggingface_hub",{"type":53,"value":140},{"type":48,"tag":67,"props":438,"children":439},{},[440,442,448,450,456,458,464],{"type":53,"value":441},"Optional environment variables: ",{"type":48,"tag":82,"props":443,"children":445},{"className":444},[],[446],{"type":53,"value":447},"CUDA_VISIBLE_DEVICES",{"type":53,"value":449}," restricts visible GPUs; ",{"type":48,"tag":82,"props":451,"children":453},{"className":452},[],[454],{"type":53,"value":455},"NPROC_PER_NODE",{"type":53,"value":457}," overrides GPU count and values ",{"type":48,"tag":82,"props":459,"children":461},{"className":460},[],[462],{"type":53,"value":463},">=2",{"type":53,"value":465}," select multi-GPU mode for non-sanity runs.",{"type":48,"tag":67,"props":467,"children":468},{},[469,471,477,479,485,486,492,493,499,501,507,509,515,517,523,525,531],{"type":53,"value":470},"Side effects: writes generated bundle configs under ",{"type":48,"tag":82,"props":472,"children":474},{"className":473},[],[475],{"type":53,"value":476},"skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002F",{"type":53,"value":478},", including ",{"type":48,"tag":82,"props":480,"children":482},{"className":481},[],[483],{"type":53,"value":484},"skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fauto_override.json",{"type":53,"value":89},{"type":48,"tag":82,"props":487,"children":489},{"className":488},[],[490],{"type":53,"value":491},"skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Ftrain_continual_task06_lung.json",{"type":53,"value":132},{"type":48,"tag":82,"props":494,"children":496},{"className":495},[],[497],{"type":53,"value":498},"skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fdfw_no_logging.json",{"type":53,"value":500},"; writes checkpoints\u002Fevidence under ",{"type":48,"tag":82,"props":502,"children":504},{"className":503},[],[505],{"type":53,"value":506},"--output-dir",{"type":53,"value":508},", may cache model assets under ",{"type":48,"tag":82,"props":510,"children":512},{"className":511},[],[513],{"type":53,"value":514},"~\u002F.cache\u002Fhuggingface\u002F",{"type":53,"value":516},", and may contact ",{"type":48,"tag":82,"props":518,"children":520},{"className":519},[],[521],{"type":53,"value":522},"https:\u002F\u002Fhuggingface.co",{"type":53,"value":524}," or ",{"type":48,"tag":82,"props":526,"children":528},{"className":527},[],[529],{"type":53,"value":530},"https:\u002F\u002Fraw.githubusercontent.com",{"type":53,"value":140},{"type":48,"tag":533,"props":534,"children":535},"p",{},[536],{"type":53,"value":537},"Fresh environment setup:",{"type":48,"tag":539,"props":540,"children":545},"pre",{"className":541,"code":542,"language":543,"meta":544,"style":544},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","python -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub\n","bash","",[546],{"type":48,"tag":82,"props":547,"children":548},{"__ignoreMap":544},[549],{"type":48,"tag":550,"props":551,"children":554},"span",{"class":552,"line":553},"line",1,[555,561,567,572,577,583,587,592,596,600,604,609,614,619,624,629,634,639],{"type":48,"tag":550,"props":556,"children":558},{"style":557},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[559],{"type":53,"value":560},"python",{"type":48,"tag":550,"props":562,"children":564},{"style":563},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[565],{"type":53,"value":566}," -m",{"type":48,"tag":550,"props":568,"children":569},{"style":563},[570],{"type":53,"value":571}," pip",{"type":48,"tag":550,"props":573,"children":574},{"style":563},[575],{"type":53,"value":576}," install",{"type":48,"tag":550,"props":578,"children":580},{"style":579},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[581],{"type":53,"value":582}," \"",{"type":48,"tag":550,"props":584,"children":585},{"style":563},[586],{"type":53,"value":372},{"type":48,"tag":550,"props":588,"children":589},{"style":579},[590],{"type":53,"value":591},"\"",{"type":48,"tag":550,"props":593,"children":594},{"style":579},[595],{"type":53,"value":582},{"type":48,"tag":550,"props":597,"children":598},{"style":563},[599],{"type":53,"value":379},{"type":48,"tag":550,"props":601,"children":602},{"style":579},[603],{"type":53,"value":591},{"type":48,"tag":550,"props":605,"children":606},{"style":563},[607],{"type":53,"value":608}," pytorch-ignite",{"type":48,"tag":550,"props":610,"children":611},{"style":563},[612],{"type":53,"value":613}," einops",{"type":48,"tag":550,"props":615,"children":616},{"style":563},[617],{"type":53,"value":618}," nibabel",{"type":48,"tag":550,"props":620,"children":621},{"style":563},[622],{"type":53,"value":623}," scipy",{"type":48,"tag":550,"props":625,"children":626},{"style":563},[627],{"type":53,"value":628}," typer",{"type":48,"tag":550,"props":630,"children":631},{"style":563},[632],{"type":53,"value":633}," PyYAML",{"type":48,"tag":550,"props":635,"children":636},{"style":563},[637],{"type":53,"value":638}," fire",{"type":48,"tag":550,"props":640,"children":641},{"style":563},[642],{"type":53,"value":643}," huggingface_hub\n",{"type":48,"tag":533,"props":645,"children":646},{},[647],{"type":53,"value":648},"Known upstream compatibility constraints:",{"type":48,"tag":63,"props":650,"children":651},{},[652,680,692],{"type":48,"tag":67,"props":653,"children":654},{},[655,657,663,665,671,673,679],{"type":53,"value":656},"DFW Task06 reference: Python ",{"type":48,"tag":82,"props":658,"children":660},{"className":659},[],[661],{"type":53,"value":662},"3.10.16",{"type":53,"value":664},", MONAI ",{"type":48,"tag":82,"props":666,"children":668},{"className":667},[],[669],{"type":53,"value":670},"1.4.0",{"type":53,"value":672},", Torch ",{"type":48,"tag":82,"props":674,"children":676},{"className":675},[],[677],{"type":53,"value":678},"2.7.0+cu126",{"type":53,"value":140},{"type":48,"tag":67,"props":681,"children":682},{},[683,685,690],{"type":53,"value":684},"Use exact ",{"type":48,"tag":82,"props":686,"children":688},{"className":687},[],[689],{"type":53,"value":372},{"type":53,"value":691}," for smoke, sanity, and evidence runs; MONAI 1.5.x can crash the upstream finetune loss on boolean labels.",{"type":48,"tag":67,"props":693,"children":694},{},[695,697,703],{"type":53,"value":696},"Do not float the dependency as ",{"type":48,"tag":82,"props":698,"children":700},{"className":699},[],[701],{"type":53,"value":702},"monai>=1.4,\u003C1.6",{"type":53,"value":704}," in generated commands.",{"type":48,"tag":56,"props":706,"children":708},{"id":707},"usage",[709],{"type":53,"value":710},"Usage",{"type":48,"tag":533,"props":712,"children":713},{},[714],{"type":53,"value":715},"Smoke-scale workflow check:",{"type":48,"tag":539,"props":717,"children":719},{"className":541,"code":718,"language":543,"meta":544,"style":544},"python -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub && \\\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  PATH_TO_DATASET \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n",[720],{"type":48,"tag":82,"props":721,"children":722},{"__ignoreMap":544},[723,810,827,840,853,881],{"type":48,"tag":550,"props":724,"children":725},{"class":552,"line":553},[726,730,734,738,742,746,750,754,758,762,766,770,774,778,782,786,790,794,799,804],{"type":48,"tag":550,"props":727,"children":728},{"style":557},[729],{"type":53,"value":560},{"type":48,"tag":550,"props":731,"children":732},{"style":563},[733],{"type":53,"value":566},{"type":48,"tag":550,"props":735,"children":736},{"style":563},[737],{"type":53,"value":571},{"type":48,"tag":550,"props":739,"children":740},{"style":563},[741],{"type":53,"value":576},{"type":48,"tag":550,"props":743,"children":744},{"style":579},[745],{"type":53,"value":582},{"type":48,"tag":550,"props":747,"children":748},{"style":563},[749],{"type":53,"value":372},{"type":48,"tag":550,"props":751,"children":752},{"style":579},[753],{"type":53,"value":591},{"type":48,"tag":550,"props":755,"children":756},{"style":579},[757],{"type":53,"value":582},{"type":48,"tag":550,"props":759,"children":760},{"style":563},[761],{"type":53,"value":379},{"type":48,"tag":550,"props":763,"children":764},{"style":579},[765],{"type":53,"value":591},{"type":48,"tag":550,"props":767,"children":768},{"style":563},[769],{"type":53,"value":608},{"type":48,"tag":550,"props":771,"children":772},{"style":563},[773],{"type":53,"value":613},{"type":48,"tag":550,"props":775,"children":776},{"style":563},[777],{"type":53,"value":618},{"type":48,"tag":550,"props":779,"children":780},{"style":563},[781],{"type":53,"value":623},{"type":48,"tag":550,"props":783,"children":784},{"style":563},[785],{"type":53,"value":628},{"type":48,"tag":550,"props":787,"children":788},{"style":563},[789],{"type":53,"value":633},{"type":48,"tag":550,"props":791,"children":792},{"style":563},[793],{"type":53,"value":638},{"type":48,"tag":550,"props":795,"children":796},{"style":563},[797],{"type":53,"value":798}," huggingface_hub",{"type":48,"tag":550,"props":800,"children":801},{"style":579},[802],{"type":53,"value":803}," &&",{"type":48,"tag":550,"props":805,"children":807},{"style":806},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[808],{"type":53,"value":809}," \\\n",{"type":48,"tag":550,"props":811,"children":813},{"class":552,"line":812},2,[814,818,823],{"type":48,"tag":550,"props":815,"children":816},{"style":557},[817],{"type":53,"value":560},{"type":48,"tag":550,"props":819,"children":820},{"style":563},[821],{"type":53,"value":822}," skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py",{"type":48,"tag":550,"props":824,"children":825},{"style":806},[826],{"type":53,"value":809},{"type":48,"tag":550,"props":828,"children":830},{"class":552,"line":829},3,[831,836],{"type":48,"tag":550,"props":832,"children":833},{"style":563},[834],{"type":53,"value":835},"  PATH_TO_DATASET",{"type":48,"tag":550,"props":837,"children":838},{"style":806},[839],{"type":53,"value":809},{"type":48,"tag":550,"props":841,"children":843},{"class":552,"line":842},4,[844,849],{"type":48,"tag":550,"props":845,"children":846},{"style":563},[847],{"type":53,"value":848},"  --smoke",{"type":48,"tag":550,"props":850,"children":851},{"style":806},[852],{"type":53,"value":809},{"type":48,"tag":550,"props":854,"children":856},{"class":552,"line":855},5,[857,862,867,872,877],{"type":48,"tag":550,"props":858,"children":859},{"style":563},[860],{"type":53,"value":861},"  --patch-size",{"type":48,"tag":550,"props":863,"children":864},{"style":579},[865],{"type":53,"value":866}," '",{"type":48,"tag":550,"props":868,"children":869},{"style":563},[870],{"type":53,"value":871},"[64,64,64]",{"type":48,"tag":550,"props":873,"children":874},{"style":579},[875],{"type":53,"value":876},"'",{"type":48,"tag":550,"props":878,"children":879},{"style":806},[880],{"type":53,"value":809},{"type":48,"tag":550,"props":882,"children":884},{"class":552,"line":883},6,[885,890],{"type":48,"tag":550,"props":886,"children":887},{"style":563},[888],{"type":53,"value":889},"  --output-dir",{"type":48,"tag":550,"props":891,"children":892},{"style":563},[893],{"type":53,"value":894}," runs\u002Fnvseg_smoke\n",{"type":48,"tag":533,"props":896,"children":897},{},[898,900,906,908,914],{"type":53,"value":899},"Use the staged dataset as ",{"type":48,"tag":82,"props":901,"children":903},{"className":902},[],[904],{"type":53,"value":905},"PATH_TO_DATASET",{"type":53,"value":907},". For the micro fixture, use ",{"type":48,"tag":82,"props":909,"children":911},{"className":910},[],[912],{"type":53,"value":913},"skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro",{"type":53,"value":915},". Smoke mode proves wiring, config generation, checkpoint loading, and runtime compatibility; it is not a quality bar.",{"type":48,"tag":533,"props":917,"children":918},{},[919],{"type":53,"value":920},"MSD Task06 Lung Tumor sanity reproduction:",{"type":48,"tag":539,"props":922,"children":924},{"className":541,"code":923,"language":543,"meta":544,"style":544},"python skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  \u002Fpath\u002Fto\u002FTask06 \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n",[925],{"type":48,"tag":82,"props":926,"children":927},{"__ignoreMap":544},[928,943,955,967],{"type":48,"tag":550,"props":929,"children":930},{"class":552,"line":553},[931,935,939],{"type":48,"tag":550,"props":932,"children":933},{"style":557},[934],{"type":53,"value":560},{"type":48,"tag":550,"props":936,"children":937},{"style":563},[938],{"type":53,"value":822},{"type":48,"tag":550,"props":940,"children":941},{"style":806},[942],{"type":53,"value":809},{"type":48,"tag":550,"props":944,"children":945},{"class":552,"line":812},[946,951],{"type":48,"tag":550,"props":947,"children":948},{"style":563},[949],{"type":53,"value":950},"  \u002Fpath\u002Fto\u002FTask06",{"type":48,"tag":550,"props":952,"children":953},{"style":806},[954],{"type":53,"value":809},{"type":48,"tag":550,"props":956,"children":957},{"class":552,"line":829},[958,963],{"type":48,"tag":550,"props":959,"children":960},{"style":563},[961],{"type":53,"value":962},"  --sanity",{"type":48,"tag":550,"props":964,"children":965},{"style":806},[966],{"type":53,"value":809},{"type":48,"tag":550,"props":968,"children":969},{"class":552,"line":842},[970,974],{"type":48,"tag":550,"props":971,"children":972},{"style":563},[973],{"type":53,"value":889},{"type":48,"tag":550,"props":975,"children":976},{"style":563},[977],{"type":53,"value":978}," runs\u002Fnvseg_task06_sanity\n",{"type":48,"tag":533,"props":980,"children":981},{},[982,984,990,992,998,1000,1006,1008,1014,1016,1022,1024,1030,1032,1038],{"type":53,"value":983},"The sanity preset follows the single-GPU DFW recipe: fold-0 validation, label mapping ",{"type":48,"tag":82,"props":985,"children":987},{"className":986},[],[988],{"type":53,"value":989},"[[1, 23]]",{"type":53,"value":991}," for ",{"type":48,"tag":82,"props":993,"children":995},{"className":994},[],[996],{"type":53,"value":997},"lung tumor",{"type":53,"value":999},", automatic class-prompt segmentation, patch ",{"type":48,"tag":82,"props":1001,"children":1003},{"className":1002},[],[1004],{"type":53,"value":1005},"[128,128,128]",{"type":53,"value":1007},", 5 epochs, and original-spacing ",{"type":48,"tag":82,"props":1009,"children":1011},{"className":1010},[],[1012],{"type":53,"value":1013},"configs\u002Fevaluate.json",{"type":53,"value":1015}," scoring before and after training. Expected reference range is pretrained Dice about ",{"type":48,"tag":82,"props":1017,"children":1019},{"className":1018},[],[1020],{"type":53,"value":1021},"0.6697",{"type":53,"value":1023},", training-best Dice about ",{"type":48,"tag":82,"props":1025,"children":1027},{"className":1026},[],[1028],{"type":53,"value":1029},"0.6905",{"type":53,"value":1031},", and fine-tuned formal Dice about ",{"type":48,"tag":82,"props":1033,"children":1035},{"className":1034},[],[1036],{"type":53,"value":1037},"0.6836",{"type":53,"value":140},{"type":48,"tag":533,"props":1040,"children":1041},{},[1042],{"type":53,"value":1043},"User-data finetune:",{"type":48,"tag":539,"props":1045,"children":1047},{"className":541,"code":1046,"language":543,"meta":544,"style":544},"python skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  --dataset-dir \u002Fpath\u002Fto\u002Fdataset \\\n  --datalist \u002Fpath\u002Fto\u002Fdatalist.json \\\n  --target-anatomy \"lung tumor\" \\\n  --auto-seg \\\n  --epochs 5 \\\n  --patch-size '[128,128,128]' \\\n  --output-dir runs\u002Fnvseg_user_finetune\n",[1048],{"type":48,"tag":82,"props":1049,"children":1050},{"__ignoreMap":544},[1051,1066,1083,1100,1124,1136,1154,1178],{"type":48,"tag":550,"props":1052,"children":1053},{"class":552,"line":553},[1054,1058,1062],{"type":48,"tag":550,"props":1055,"children":1056},{"style":557},[1057],{"type":53,"value":560},{"type":48,"tag":550,"props":1059,"children":1060},{"style":563},[1061],{"type":53,"value":822},{"type":48,"tag":550,"props":1063,"children":1064},{"style":806},[1065],{"type":53,"value":809},{"type":48,"tag":550,"props":1067,"children":1068},{"class":552,"line":812},[1069,1074,1079],{"type":48,"tag":550,"props":1070,"children":1071},{"style":563},[1072],{"type":53,"value":1073},"  --dataset-dir",{"type":48,"tag":550,"props":1075,"children":1076},{"style":563},[1077],{"type":53,"value":1078}," \u002Fpath\u002Fto\u002Fdataset",{"type":48,"tag":550,"props":1080,"children":1081},{"style":806},[1082],{"type":53,"value":809},{"type":48,"tag":550,"props":1084,"children":1085},{"class":552,"line":829},[1086,1091,1096],{"type":48,"tag":550,"props":1087,"children":1088},{"style":563},[1089],{"type":53,"value":1090},"  --datalist",{"type":48,"tag":550,"props":1092,"children":1093},{"style":563},[1094],{"type":53,"value":1095}," \u002Fpath\u002Fto\u002Fdatalist.json",{"type":48,"tag":550,"props":1097,"children":1098},{"style":806},[1099],{"type":53,"value":809},{"type":48,"tag":550,"props":1101,"children":1102},{"class":552,"line":842},[1103,1108,1112,1116,1120],{"type":48,"tag":550,"props":1104,"children":1105},{"style":563},[1106],{"type":53,"value":1107},"  --target-anatomy",{"type":48,"tag":550,"props":1109,"children":1110},{"style":579},[1111],{"type":53,"value":582},{"type":48,"tag":550,"props":1113,"children":1114},{"style":563},[1115],{"type":53,"value":997},{"type":48,"tag":550,"props":1117,"children":1118},{"style":579},[1119],{"type":53,"value":591},{"type":48,"tag":550,"props":1121,"children":1122},{"style":806},[1123],{"type":53,"value":809},{"type":48,"tag":550,"props":1125,"children":1126},{"class":552,"line":855},[1127,1132],{"type":48,"tag":550,"props":1128,"children":1129},{"style":563},[1130],{"type":53,"value":1131},"  --auto-seg",{"type":48,"tag":550,"props":1133,"children":1134},{"style":806},[1135],{"type":53,"value":809},{"type":48,"tag":550,"props":1137,"children":1138},{"class":552,"line":883},[1139,1144,1150],{"type":48,"tag":550,"props":1140,"children":1141},{"style":563},[1142],{"type":53,"value":1143},"  --epochs",{"type":48,"tag":550,"props":1145,"children":1147},{"style":1146},"--shiki-light:#F76D47;--shiki-default:#F78C6C;--shiki-dark:#F78C6C",[1148],{"type":53,"value":1149}," 5",{"type":48,"tag":550,"props":1151,"children":1152},{"style":806},[1153],{"type":53,"value":809},{"type":48,"tag":550,"props":1155,"children":1157},{"class":552,"line":1156},7,[1158,1162,1166,1170,1174],{"type":48,"tag":550,"props":1159,"children":1160},{"style":563},[1161],{"type":53,"value":861},{"type":48,"tag":550,"props":1163,"children":1164},{"style":579},[1165],{"type":53,"value":866},{"type":48,"tag":550,"props":1167,"children":1168},{"style":563},[1169],{"type":53,"value":1005},{"type":48,"tag":550,"props":1171,"children":1172},{"style":579},[1173],{"type":53,"value":876},{"type":48,"tag":550,"props":1175,"children":1176},{"style":806},[1177],{"type":53,"value":809},{"type":48,"tag":550,"props":1179,"children":1181},{"class":552,"line":1180},8,[1182,1186],{"type":48,"tag":550,"props":1183,"children":1184},{"style":563},[1185],{"type":53,"value":889},{"type":48,"tag":550,"props":1187,"children":1188},{"style":563},[1189],{"type":53,"value":1190}," runs\u002Fnvseg_user_finetune\n",{"type":48,"tag":533,"props":1192,"children":1193},{},[1194,1196,1202],{"type":53,"value":1195},"Use ",{"type":48,"tag":82,"props":1197,"children":1199},{"className":1198},[],[1200],{"type":53,"value":1201},"--label-mapping '[[1, 23]]'",{"type":53,"value":1203}," when local label values are custom or the anatomy name is ambiguous.",{"type":48,"tag":56,"props":1205,"children":1207},{"id":1206},"examples",[1208],{"type":53,"value":1209},"Examples",{"type":48,"tag":533,"props":1211,"children":1212},{},[1213],{"type":53,"value":1214},"Smoke run on a staged tiny dataset:",{"type":48,"tag":539,"props":1216,"children":1218},{"className":541,"code":1217,"language":543,"meta":544,"style":544},"python skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  runs\u002Fwith_vs_without_nv\u002F_inputs\u002Fnv_segment_ct_finetune\u002Finput_dataset \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n",[1219],{"type":48,"tag":82,"props":1220,"children":1221},{"__ignoreMap":544},[1222,1237,1249,1260,1283],{"type":48,"tag":550,"props":1223,"children":1224},{"class":552,"line":553},[1225,1229,1233],{"type":48,"tag":550,"props":1226,"children":1227},{"style":557},[1228],{"type":53,"value":560},{"type":48,"tag":550,"props":1230,"children":1231},{"style":563},[1232],{"type":53,"value":822},{"type":48,"tag":550,"props":1234,"children":1235},{"style":806},[1236],{"type":53,"value":809},{"type":48,"tag":550,"props":1238,"children":1239},{"class":552,"line":812},[1240,1245],{"type":48,"tag":550,"props":1241,"children":1242},{"style":563},[1243],{"type":53,"value":1244},"  runs\u002Fwith_vs_without_nv\u002F_inputs\u002Fnv_segment_ct_finetune\u002Finput_dataset",{"type":48,"tag":550,"props":1246,"children":1247},{"style":806},[1248],{"type":53,"value":809},{"type":48,"tag":550,"props":1250,"children":1251},{"class":552,"line":829},[1252,1256],{"type":48,"tag":550,"props":1253,"children":1254},{"style":563},[1255],{"type":53,"value":848},{"type":48,"tag":550,"props":1257,"children":1258},{"style":806},[1259],{"type":53,"value":809},{"type":48,"tag":550,"props":1261,"children":1262},{"class":552,"line":842},[1263,1267,1271,1275,1279],{"type":48,"tag":550,"props":1264,"children":1265},{"style":563},[1266],{"type":53,"value":861},{"type":48,"tag":550,"props":1268,"children":1269},{"style":579},[1270],{"type":53,"value":866},{"type":48,"tag":550,"props":1272,"children":1273},{"style":563},[1274],{"type":53,"value":871},{"type":48,"tag":550,"props":1276,"children":1277},{"style":579},[1278],{"type":53,"value":876},{"type":48,"tag":550,"props":1280,"children":1281},{"style":806},[1282],{"type":53,"value":809},{"type":48,"tag":550,"props":1284,"children":1285},{"class":552,"line":855},[1286,1290],{"type":48,"tag":550,"props":1287,"children":1288},{"style":563},[1289],{"type":53,"value":889},{"type":48,"tag":550,"props":1291,"children":1292},{"style":563},[1293],{"type":53,"value":894},{"type":48,"tag":533,"props":1295,"children":1296},{},[1297],{"type":53,"value":1298},"Task06 sanity run on a local MSD cache:",{"type":48,"tag":539,"props":1300,"children":1302},{"className":541,"code":1301,"language":543,"meta":544,"style":544},"python skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  .workbench_data\u002Fdatasets\u002FTask06_Lung \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n",[1303],{"type":48,"tag":82,"props":1304,"children":1305},{"__ignoreMap":544},[1306,1321,1333,1344],{"type":48,"tag":550,"props":1307,"children":1308},{"class":552,"line":553},[1309,1313,1317],{"type":48,"tag":550,"props":1310,"children":1311},{"style":557},[1312],{"type":53,"value":560},{"type":48,"tag":550,"props":1314,"children":1315},{"style":563},[1316],{"type":53,"value":822},{"type":48,"tag":550,"props":1318,"children":1319},{"style":806},[1320],{"type":53,"value":809},{"type":48,"tag":550,"props":1322,"children":1323},{"class":552,"line":812},[1324,1329],{"type":48,"tag":550,"props":1325,"children":1326},{"style":563},[1327],{"type":53,"value":1328},"  .workbench_data\u002Fdatasets\u002FTask06_Lung",{"type":48,"tag":550,"props":1330,"children":1331},{"style":806},[1332],{"type":53,"value":809},{"type":48,"tag":550,"props":1334,"children":1335},{"class":552,"line":829},[1336,1340],{"type":48,"tag":550,"props":1337,"children":1338},{"style":563},[1339],{"type":53,"value":962},{"type":48,"tag":550,"props":1341,"children":1342},{"style":806},[1343],{"type":53,"value":809},{"type":48,"tag":550,"props":1345,"children":1346},{"class":552,"line":842},[1347,1351],{"type":48,"tag":550,"props":1348,"children":1349},{"style":563},[1350],{"type":53,"value":889},{"type":48,"tag":550,"props":1352,"children":1353},{"style":563},[1354],{"type":53,"value":978},{"type":48,"tag":56,"props":1356,"children":1358},{"id":1357},"data-contract",[1359],{"type":53,"value":1360},"Data Contract",{"type":48,"tag":63,"props":1362,"children":1363},{},[1364,1384,1389,1394,1423],{"type":48,"tag":67,"props":1365,"children":1366},{},[1367,1369,1375,1377,1383],{"type":53,"value":1368},"Preferred layout: ",{"type":48,"tag":82,"props":1370,"children":1372},{"className":1371},[],[1373],{"type":53,"value":1374},"dataset\u002FimagesTr\u002F*.nii.gz",{"type":53,"value":1376}," and ",{"type":48,"tag":82,"props":1378,"children":1380},{"className":1379},[],[1381],{"type":53,"value":1382},"dataset\u002FlabelsTr\u002F*.nii.gz",{"type":53,"value":140},{"type":48,"tag":67,"props":1385,"children":1386},{},[1387],{"type":53,"value":1388},"Labels must align one-to-one with images by basename.",{"type":48,"tag":67,"props":1390,"children":1391},{},[1392],{"type":53,"value":1393},"The target label value must be present in the training labels.",{"type":48,"tag":67,"props":1395,"children":1396},{},[1397,1399,1405,1407,1413,1415,1421],{"type":53,"value":1398},"Use a datalist when patient-level splitting matters. The bundle default ",{"type":48,"tag":82,"props":1400,"children":1402},{"className":1401},[],[1403],{"type":53,"value":1404},"fold",{"type":53,"value":1406}," is ",{"type":48,"tag":82,"props":1408,"children":1410},{"className":1409},[],[1411],{"type":53,"value":1412},"0",{"type":53,"value":1414},", so ",{"type":48,"tag":82,"props":1416,"children":1418},{"className":1417},[],[1419],{"type":53,"value":1420},"fold: 0",{"type":53,"value":1422}," entries are validation and all other folds are training.",{"type":48,"tag":67,"props":1424,"children":1425},{},[1426,1428,1434],{"type":53,"value":1427},"Every trained foreground label must map to an existing VISTA3D global class id from ",{"type":48,"tag":82,"props":1429,"children":1431},{"className":1430},[],[1432],{"type":53,"value":1433},"bundle\u002Flabel_dict.json",{"type":53,"value":1435},"; this skill cannot invent a new class.",{"type":48,"tag":56,"props":1437,"children":1439},{"id":1438},"results",[1440],{"type":53,"value":1441},"Results",{"type":48,"tag":533,"props":1443,"children":1444},{},[1445,1447,1452],{"type":53,"value":1446},"Check ",{"type":48,"tag":82,"props":1448,"children":1450},{"className":1449},[],[1451],{"type":53,"value":330},{"type":53,"value":1453}," in the run directory first:",{"type":48,"tag":63,"props":1455,"children":1456},{},[1457,1475,1500,1518,1536],{"type":48,"tag":67,"props":1458,"children":1459},{},[1460,1466,1467,1473],{"type":48,"tag":82,"props":1461,"children":1463},{"className":1462},[],[1464],{"type":53,"value":1465},"formal_pretrained_val_dice",{"type":53,"value":1376},{"type":48,"tag":82,"props":1468,"children":1470},{"className":1469},[],[1471],{"type":53,"value":1472},"formal_finetuned_val_dice",{"type":53,"value":1474},": original-spacing pre\u002Fpost scores when formal eval is enabled.",{"type":48,"tag":67,"props":1476,"children":1477},{},[1478,1484,1485,1491,1492,1498],{"type":48,"tag":82,"props":1479,"children":1481},{"className":1480},[],[1482],{"type":53,"value":1483},"training_start_val_dice",{"type":53,"value":89},{"type":48,"tag":82,"props":1486,"children":1488},{"className":1487},[],[1489],{"type":53,"value":1490},"val_dice_per_epoch",{"type":53,"value":132},{"type":48,"tag":82,"props":1493,"children":1495},{"className":1494},[],[1496],{"type":53,"value":1497},"training_best_val_dice",{"type":53,"value":1499},": training-time validation trace.",{"type":48,"tag":67,"props":1501,"children":1502},{},[1503,1509,1511,1517],{"type":48,"tag":82,"props":1504,"children":1506},{"className":1505},[],[1507],{"type":53,"value":1508},"finetuned_ckpt_matches_pretrained_weights",{"type":53,"value":1510},": detects the epoch-0 checkpoint trap when ",{"type":48,"tag":82,"props":1512,"children":1514},{"className":1513},[],[1515],{"type":53,"value":1516},"val_at_start=true",{"type":53,"value":140},{"type":48,"tag":67,"props":1519,"children":1520},{},[1521,1527,1529,1535],{"type":48,"tag":82,"props":1522,"children":1524},{"className":1523},[],[1525],{"type":53,"value":1526},"recommended_ckpt",{"type":53,"value":1528},": checkpoint to keep. Do not blindly use the last epoch or ",{"type":48,"tag":82,"props":1530,"children":1532},{"className":1531},[],[1533],{"type":53,"value":1534},"model_finetune.pt",{"type":53,"value":140},{"type":48,"tag":67,"props":1537,"children":1538},{},[1539,1545,1546,1552],{"type":48,"tag":82,"props":1540,"children":1542},{"className":1541},[],[1543],{"type":53,"value":1544},"runtime.oom",{"type":53,"value":89},{"type":48,"tag":82,"props":1547,"children":1549},{"className":1548},[],[1550],{"type":53,"value":1551},"runtime.peak_gpu_mb",{"type":53,"value":1553},", and phase logs: distinguish OOM, slow validation, and process failure.",{"type":48,"tag":533,"props":1555,"children":1556},{},[1557,1559,1565],{"type":53,"value":1558},"Decision rule: prefer formal original-spacing pre\u002Fpost scores when present; reject tensor-identical \"fine-tuned\" checkpoints for sanity recovery; treat ",{"type":48,"tag":82,"props":1560,"children":1562},{"className":1561},[],[1563],{"type":53,"value":1564},"improved: false",{"type":53,"value":1566}," as valid evidence rather than a wrapper failure.",{"type":48,"tag":56,"props":1568,"children":1570},{"id":1569},"limitations",[1571],{"type":53,"value":1572},"Limitations",{"type":48,"tag":63,"props":1574,"children":1575},{},[1576,1587,1710,1744,1757,1762],{"type":48,"tag":67,"props":1577,"children":1578},{},[1579,1581,1586],{"type":53,"value":1580},"Thin wrapper. Training, validation, transforms, and checkpointing are delegated to the upstream bundle in ",{"type":48,"tag":82,"props":1582,"children":1584},{"className":1583},[],[1585],{"type":53,"value":188},{"type":53,"value":140},{"type":48,"tag":67,"props":1588,"children":1589},{},[1590,1592,1598,1600,1606,1608,1614,1615,1620,1622,1628,1630,1636,1638,1644,1646,1652,1654,1660,1662,1668,1670,1676,1678,1684,1686,1692,1694,1700,1702,1708],{"type":53,"value":1591},"Reproduction record only: the successful five-epoch Task06 run used Python\n",{"type":48,"tag":82,"props":1593,"children":1595},{"className":1594},[],[1596],{"type":53,"value":1597},"3.12.3",{"type":53,"value":1599},", PyTorch ",{"type":48,"tag":82,"props":1601,"children":1603},{"className":1602},[],[1604],{"type":53,"value":1605},"2.12.0+cu130",{"type":53,"value":1607}," with CUDA ",{"type":48,"tag":82,"props":1609,"children":1611},{"className":1610},[],[1612],{"type":53,"value":1613},"13.0",{"type":53,"value":664},{"type":48,"tag":82,"props":1616,"children":1618},{"className":1617},[],[1619],{"type":53,"value":670},{"type":53,"value":1621},", NumPy\n",{"type":48,"tag":82,"props":1623,"children":1625},{"className":1624},[],[1626],{"type":53,"value":1627},"1.26.4",{"type":53,"value":1629},", PyTorch-Ignite ",{"type":48,"tag":82,"props":1631,"children":1633},{"className":1632},[],[1634],{"type":53,"value":1635},"0.5.4",{"type":53,"value":1637},", NiBabel ",{"type":48,"tag":82,"props":1639,"children":1641},{"className":1640},[],[1642],{"type":53,"value":1643},"5.4.2",{"type":53,"value":1645},", SciPy ",{"type":48,"tag":82,"props":1647,"children":1649},{"className":1648},[],[1650],{"type":53,"value":1651},"1.16.0",{"type":53,"value":1653},", einops\n",{"type":48,"tag":82,"props":1655,"children":1657},{"className":1656},[],[1658],{"type":53,"value":1659},"0.8.2",{"type":53,"value":1661},", Fire ",{"type":48,"tag":82,"props":1663,"children":1665},{"className":1664},[],[1666],{"type":53,"value":1667},"0.7.1",{"type":53,"value":1669},", Hugging Face Hub ",{"type":48,"tag":82,"props":1671,"children":1673},{"className":1672},[],[1674],{"type":53,"value":1675},"0.36.2",{"type":53,"value":1677},", Transformers ",{"type":48,"tag":82,"props":1679,"children":1681},{"className":1680},[],[1682],{"type":53,"value":1683},"4.57.6",{"type":53,"value":1685},",\nTyper ",{"type":48,"tag":82,"props":1687,"children":1689},{"className":1688},[],[1690],{"type":53,"value":1691},"0.25.1",{"type":53,"value":1693},", PyYAML ",{"type":48,"tag":82,"props":1695,"children":1697},{"className":1696},[],[1698],{"type":53,"value":1699},"6.0.3",{"type":53,"value":1701},", and MLflow ",{"type":48,"tag":82,"props":1703,"children":1705},{"className":1704},[],[1706],{"type":53,"value":1707},"3.14.0",{"type":53,"value":1709}," on one NVIDIA RTX 6000\nAda 48 GB GPU. These versions document the evidence environment; they are\nnot additional package constraints or a claim that other versions cannot\nwork.",{"type":48,"tag":67,"props":1711,"children":1712},{},[1713,1715,1721,1722,1728,1729,1735,1736,1742],{"type":53,"value":1714},"The auto-derived plan is heuristic; caller-provided ",{"type":48,"tag":82,"props":1716,"children":1718},{"className":1717},[],[1719],{"type":53,"value":1720},"--patch-size",{"type":53,"value":89},{"type":48,"tag":82,"props":1723,"children":1725},{"className":1724},[],[1726],{"type":53,"value":1727},"--cache-rate",{"type":53,"value":89},{"type":48,"tag":82,"props":1730,"children":1732},{"className":1731},[],[1733],{"type":53,"value":1734},"--epochs",{"type":53,"value":132},{"type":48,"tag":82,"props":1737,"children":1739},{"className":1738},[],[1740],{"type":53,"value":1741},"--learning-rate",{"type":53,"value":1743}," win.",{"type":48,"tag":67,"props":1745,"children":1746},{},[1747,1749,1755],{"type":53,"value":1748},"The Task06 sanity recipe intentionally forces single-GPU execution to match the DFW reference. Multi-GPU mode for other datasets requires host ",{"type":48,"tag":82,"props":1750,"children":1752},{"className":1751},[],[1753],{"type":53,"value":1754},"torchrun",{"type":53,"value":1756}," support.",{"type":48,"tag":67,"props":1758,"children":1759},{},[1760],{"type":53,"value":1761},"The paired verifier is CPU-only and audits the evidence pack; it does not re-run GPU segmentation.",{"type":48,"tag":67,"props":1763,"children":1764},{},[1765],{"type":53,"value":1766},"Not for clinical deployment, clinical interpretation, autonomous diagnosis, or regulatory submission.",{"type":48,"tag":56,"props":1768,"children":1770},{"id":1769},"troubleshooting",[1771],{"type":53,"value":1772},"Troubleshooting",{"type":48,"tag":270,"props":1774,"children":1775},{},[1776,1797],{"type":48,"tag":274,"props":1777,"children":1778},{},[1779],{"type":48,"tag":278,"props":1780,"children":1781},{},[1782,1787,1792],{"type":48,"tag":282,"props":1783,"children":1784},{},[1785],{"type":53,"value":1786},"Error",{"type":48,"tag":282,"props":1788,"children":1789},{},[1790],{"type":53,"value":1791},"Cause",{"type":48,"tag":282,"props":1793,"children":1794},{},[1795],{"type":53,"value":1796},"Fix",{"type":48,"tag":297,"props":1798,"children":1799},{},[1800,1824,1842,1876,1915,1947],{"type":48,"tag":278,"props":1801,"children":1802},{},[1803,1808,1819],{"type":48,"tag":304,"props":1804,"children":1805},{},[1806],{"type":53,"value":1807},"Missing dependency or import error",{"type":48,"tag":304,"props":1809,"children":1810},{},[1811,1813,1818],{"type":53,"value":1812},"Runtime drift from ",{"type":48,"tag":82,"props":1814,"children":1816},{"className":1815},[],[1817],{"type":53,"value":322},{"type":53,"value":140},{"type":48,"tag":304,"props":1820,"children":1821},{},[1822],{"type":53,"value":1823},"Install the packages above or use the documented environment.",{"type":48,"tag":278,"props":1825,"children":1826},{},[1827,1832,1837],{"type":48,"tag":304,"props":1828,"children":1829},{},[1830],{"type":53,"value":1831},"Low Task06 pretrained Dice",{"type":48,"tag":304,"props":1833,"children":1834},{},[1835],{"type":53,"value":1836},"Wrong config, wrong checkpoint, data split drift, or dependency drift.",{"type":48,"tag":304,"props":1838,"children":1839},{},[1840],{"type":53,"value":1841},"Compare environment fields and staged configs before changing training logic.",{"type":48,"tag":278,"props":1843,"children":1844},{},[1845,1855,1865],{"type":48,"tag":304,"props":1846,"children":1847},{},[1848,1853],{"type":48,"tag":82,"props":1849,"children":1851},{"className":1850},[],[1852],{"type":53,"value":1534},{"type":53,"value":1854}," matches pretrained",{"type":48,"tag":304,"props":1856,"children":1857},{},[1858,1863],{"type":48,"tag":82,"props":1859,"children":1861},{"className":1860},[],[1862],{"type":53,"value":1516},{"type":53,"value":1864}," selected epoch 0 as best.",{"type":48,"tag":304,"props":1866,"children":1867},{},[1868,1869,1874],{"type":53,"value":1195},{"type":48,"tag":82,"props":1870,"children":1872},{"className":1871},[],[1873],{"type":53,"value":1526},{"type":53,"value":1875},"; treat sanity recovery as failed unless a changed checkpoint improves formal Dice.",{"type":48,"tag":278,"props":1877,"children":1878},{},[1879,1884,1889],{"type":48,"tag":304,"props":1880,"children":1881},{},[1882],{"type":53,"value":1883},"Missing formal Dice fields",{"type":48,"tag":304,"props":1885,"children":1886},{},[1887],{"type":53,"value":1888},"Formal eval failed or was skipped.",{"type":48,"tag":304,"props":1890,"children":1891},{},[1892,1894,1900,1901,1907,1908,1914],{"type":53,"value":1893},"Inspect ",{"type":48,"tag":82,"props":1895,"children":1897},{"className":1896},[],[1898],{"type":53,"value":1899},"eval_pretrained.log",{"type":53,"value":89},{"type":48,"tag":82,"props":1902,"children":1904},{"className":1903},[],[1905],{"type":53,"value":1906},"eval_finetuned.log",{"type":53,"value":132},{"type":48,"tag":82,"props":1909,"children":1911},{"className":1910},[],[1912],{"type":53,"value":1913},"metrics.csv",{"type":53,"value":140},{"type":48,"tag":278,"props":1916,"children":1917},{},[1918,1923,1928],{"type":48,"tag":304,"props":1919,"children":1920},{},[1921],{"type":53,"value":1922},"GPU out of memory",{"type":48,"tag":304,"props":1924,"children":1925},{},[1926],{"type":53,"value":1927},"Patch\u002Fcache settings too large.",{"type":48,"tag":304,"props":1929,"children":1930},{},[1931,1933,1938,1940,1945],{"type":53,"value":1932},"Reduce ",{"type":48,"tag":82,"props":1934,"children":1936},{"className":1935},[],[1937],{"type":53,"value":1720},{"type":53,"value":1939},", lower ",{"type":48,"tag":82,"props":1941,"children":1943},{"className":1942},[],[1944],{"type":53,"value":1727},{"type":53,"value":1946},", or reduce workers.",{"type":48,"tag":278,"props":1948,"children":1949},{},[1950,1955,1966],{"type":48,"tag":304,"props":1951,"children":1952},{},[1953],{"type":53,"value":1954},"No validation cases",{"type":48,"tag":304,"props":1956,"children":1957},{},[1958,1960,1965],{"type":53,"value":1959},"Datalist lacks ",{"type":48,"tag":82,"props":1961,"children":1963},{"className":1962},[],[1964],{"type":53,"value":1420},{"type":53,"value":140},{"type":48,"tag":304,"props":1967,"children":1968},{},[1969],{"type":53,"value":1970},"Provide at least one validation entry.",{"type":48,"tag":56,"props":1972,"children":1974},{"id":1973},"verification",[1975],{"type":53,"value":1976},"Verification",{"type":48,"tag":533,"props":1978,"children":1979},{},[1980],{"type":53,"value":1981},"Run the implemented verifier when quality gates matter:",{"type":48,"tag":539,"props":1983,"children":1985},{"className":541,"code":1984,"language":543,"meta":544,"style":544},"python -m eval_engine.run_trusted skills\u002Fnv-segment-ct-finetune \\\n  --fixture skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro \\\n  --out runs\u002Fnvseg_trusted\n",[1986],{"type":48,"tag":82,"props":1987,"children":1988},{"__ignoreMap":544},[1989,2014,2031],{"type":48,"tag":550,"props":1990,"children":1991},{"class":552,"line":553},[1992,1996,2000,2005,2010],{"type":48,"tag":550,"props":1993,"children":1994},{"style":557},[1995],{"type":53,"value":560},{"type":48,"tag":550,"props":1997,"children":1998},{"style":563},[1999],{"type":53,"value":566},{"type":48,"tag":550,"props":2001,"children":2002},{"style":563},[2003],{"type":53,"value":2004}," eval_engine.run_trusted",{"type":48,"tag":550,"props":2006,"children":2007},{"style":563},[2008],{"type":53,"value":2009}," skills\u002Fnv-segment-ct-finetune",{"type":48,"tag":550,"props":2011,"children":2012},{"style":806},[2013],{"type":53,"value":809},{"type":48,"tag":550,"props":2015,"children":2016},{"class":552,"line":812},[2017,2022,2027],{"type":48,"tag":550,"props":2018,"children":2019},{"style":563},[2020],{"type":53,"value":2021},"  --fixture",{"type":48,"tag":550,"props":2023,"children":2024},{"style":563},[2025],{"type":53,"value":2026}," skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro",{"type":48,"tag":550,"props":2028,"children":2029},{"style":806},[2030],{"type":53,"value":809},{"type":48,"tag":550,"props":2032,"children":2033},{"class":552,"line":829},[2034,2039],{"type":48,"tag":550,"props":2035,"children":2036},{"style":563},[2037],{"type":53,"value":2038},"  --out",{"type":48,"tag":550,"props":2040,"children":2041},{"style":563},[2042],{"type":53,"value":2043}," runs\u002Fnvseg_trusted\n",{"type":48,"tag":2045,"props":2046,"children":2047},"style",{},[2048],{"type":53,"value":2049},"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":2051,"total":2155},[2052,2069,2083,2097,2109,2126,2141],{"slug":2053,"name":2053,"fn":2054,"description":2055,"org":2056,"tags":2057,"stars":23,"repoUrl":24,"updatedAt":2068},"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},[2058,2061,2064,2065],{"name":2059,"slug":2060,"type":15},"Data Analysis","data-analysis",{"name":2062,"slug":2063,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":2066,"slug":2067,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":2070,"name":2070,"fn":2071,"description":2072,"org":2073,"tags":2074,"stars":23,"repoUrl":24,"updatedAt":2082},"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},[2075,2078,2081],{"name":2076,"slug":2077,"type":15},"Deployment","deployment",{"name":2079,"slug":2080,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":2084,"name":2084,"fn":2085,"description":2086,"org":2087,"tags":2088,"stars":23,"repoUrl":24,"updatedAt":2096},"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},[2089,2092,2093],{"name":2090,"slug":2091,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":2094,"slug":2095,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":2098,"name":2098,"fn":2099,"description":2100,"org":2101,"tags":2102,"stars":23,"repoUrl":24,"updatedAt":2108},"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},[2103,2104,2105],{"name":2059,"slug":2060,"type":15},{"name":9,"slug":8,"type":15},{"name":2106,"slug":2107,"type":15},"Testing","testing","2026-07-17T05:29:03.913266",{"slug":2110,"name":2110,"fn":2111,"description":2112,"org":2113,"tags":2114,"stars":23,"repoUrl":24,"updatedAt":2125},"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},[2115,2118,2121,2122],{"name":2116,"slug":2117,"type":15},"Automation","automation",{"name":2119,"slug":2120,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":2123,"slug":2124,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":2127,"name":2127,"fn":2128,"description":2129,"org":2130,"tags":2131,"stars":23,"repoUrl":24,"updatedAt":2140},"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},[2132,2133,2136,2137],{"name":2076,"slug":2077,"type":15},{"name":2134,"slug":2135,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":2138,"slug":2139,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":2142,"name":2142,"fn":2143,"description":2144,"org":2145,"tags":2146,"stars":23,"repoUrl":24,"updatedAt":2154},"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},[2147,2148,2151],{"name":9,"slug":8,"type":15},{"name":2149,"slug":2150,"type":15},"Quantum Computing","quantum-computing",{"name":2152,"slug":2153,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305,{"items":2157,"total":2307},[2158,2176,2191,2202,2214,2228,2241,2255,2266,2275,2289,2298],{"slug":2159,"name":2159,"fn":2160,"description":2161,"org":2162,"tags":2163,"stars":2173,"repoUrl":2174,"updatedAt":2175},"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},[2164,2167,2170],{"name":2165,"slug":2166,"type":15},"Documentation","documentation",{"name":2168,"slug":2169,"type":15},"MCP","mcp",{"name":2171,"slug":2172,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":2177,"name":2177,"fn":2178,"description":2179,"org":2180,"tags":2181,"stars":2188,"repoUrl":2189,"updatedAt":2190},"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},[2182,2185,2186],{"name":2183,"slug":2184,"type":15},"Containers","containers",{"name":2076,"slug":2077,"type":15},{"name":2187,"slug":560,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":2192,"name":2192,"fn":2193,"description":2194,"org":2195,"tags":2196,"stars":2188,"repoUrl":2189,"updatedAt":2201},"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},[2197,2200],{"name":2198,"slug":2199,"type":15},"CI\u002FCD","ci-cd",{"name":2076,"slug":2077,"type":15},"2026-07-14T05:25:59.97109",{"slug":2203,"name":2203,"fn":2204,"description":2205,"org":2206,"tags":2207,"stars":2188,"repoUrl":2189,"updatedAt":2213},"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},[2208,2209,2210],{"name":2198,"slug":2199,"type":15},{"name":2076,"slug":2077,"type":15},{"name":2211,"slug":2212,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":2215,"name":2215,"fn":2216,"description":2217,"org":2218,"tags":2219,"stars":2188,"repoUrl":2189,"updatedAt":2227},"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},[2220,2223,2224],{"name":2221,"slug":2222,"type":15},"Debugging","debugging",{"name":2211,"slug":2212,"type":15},{"name":2225,"slug":2226,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":2229,"name":2229,"fn":2230,"description":2231,"org":2232,"tags":2233,"stars":2188,"repoUrl":2189,"updatedAt":2240},"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},[2234,2237],{"name":2235,"slug":2236,"type":15},"Best Practices","best-practices",{"name":2238,"slug":2239,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":2242,"name":2242,"fn":2243,"description":2244,"org":2245,"tags":2246,"stars":2188,"repoUrl":2189,"updatedAt":2254},"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},[2247,2250,2253],{"name":2248,"slug":2249,"type":15},"Machine Learning","machine-learning",{"name":2251,"slug":2252,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":2256,"name":2256,"fn":2257,"description":2258,"org":2259,"tags":2260,"stars":2188,"repoUrl":2189,"updatedAt":2265},"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},[2261,2264],{"name":2262,"slug":2263,"type":15},"QA","qa",{"name":2106,"slug":2107,"type":15},"2026-07-14T05:25:53.673039",{"slug":2267,"name":2267,"fn":2268,"description":2269,"org":2270,"tags":2271,"stars":2188,"repoUrl":2189,"updatedAt":2274},"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},[2272,2273],{"name":2076,"slug":2077,"type":15},{"name":2079,"slug":2080,"type":15},"2026-07-14T05:25:49.362534",{"slug":2276,"name":2276,"fn":2277,"description":2278,"org":2279,"tags":2280,"stars":2188,"repoUrl":2189,"updatedAt":2288},"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},[2281,2284,2285],{"name":2282,"slug":2283,"type":15},"Code Review","code-review",{"name":2211,"slug":2212,"type":15},{"name":2286,"slug":2287,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":2290,"name":2290,"fn":2291,"description":2292,"org":2293,"tags":2294,"stars":2188,"repoUrl":2189,"updatedAt":2297},"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},[2295,2296],{"name":2262,"slug":2263,"type":15},{"name":2106,"slug":2107,"type":15},"2026-07-14T05:25:54.928983",{"slug":2299,"name":2299,"fn":2300,"description":2301,"org":2302,"tags":2303,"stars":2188,"repoUrl":2189,"updatedAt":2306},"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},[2304,2305],{"name":2116,"slug":2117,"type":15},{"name":2198,"slug":2199,"type":15},"2026-07-30T05:29:03.275638",496]