[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-train-nvpanoptix3d":3,"mdc--lq0pi9-key":31,"related-org-nvidia-tao-train-nvpanoptix3d":2763,"related-repo-nvidia-tao-train-nvpanoptix3d":2921},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":20,"repoUrl":21,"updatedAt":22,"license":23,"forks":24,"topics":25,"repo":26,"sourceUrl":29,"mdContent":30},"tao-train-nvpanoptix3d","reconstruct 3D scenes with NVPanoptix3D","NVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on a VGGT backbone with a Mask2Former-style head and 3D frustum reconstruction. Use when training, evaluating, exporting, or running inference for a TAO NVPanoptix3D model. Trigger phrases include \"train NVPanoptix3D\", \"panoptic 3D reconstruction\", \"3D scene segmentation\", \"occupancy completion\".",{"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],{"name":13,"slug":14,"type":15},"Machine Learning","machine-learning","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"Computer Vision","computer-vision",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:26:47.714787","Apache-2.0",281,[],{"repoUrl":21,"stars":20,"forks":24,"topics":27,"description":28},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftao-train-nvpanoptix3d","---\nname: tao-train-nvpanoptix3d\ndescription: NVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation\n  (semantic, instance, and panoptic masks) with occupancy completion. Built on a VGGT backbone with a Mask2Former-style head\n  and 3D frustum reconstruction. Use when training, evaluating, exporting, or running inference for a TAO NVPanoptix3D model.\n  Trigger phrases include \"train NVPanoptix3D\", \"panoptic 3D reconstruction\", \"3D scene segmentation\", \"occupancy completion\".\nlicense: Apache-2.0\ncompatibility: Requires docker + nvidia-container-toolkit.\nmetadata:\n  version: \"0.1.0\"\n  author: NVIDIA Corporation\nallowed-tools: Read Bash\ntags:\n- panoptic\n- 3d\n- reconstruction\n---\n\n# NVPanoptix3D\n\nNVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on VGGT backbone with Mask2Former-style head and 3D frustum reconstruction.\n\nUses 2D and 3D stage checkpoints. Set train.checkpoint_2d and train.checkpoint_3d for staged initialization.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nFor AutoML, use `train_loss` as the optimization metric with\n`direction=minimize`, and set `train.optim.monitor_name: train_loss` in\n`spec_overrides`. NVPanoptix3D train jobs emit `PRQ`, `RSQ`, and `RRQ` in\n`status.json`, and the training progress log emits `train_loss`; short or\nminimal jobs may not emit `val_loss`, including full-trial smoke runs.\nMulti-fidelity AutoML algorithms such as Hyperband, ASHA, and BOHB may promote\na checkpoint to a resume job that completes without emitting a fresh\n`train_loss` line. In that case, the AutoML metric is the carried-forward\nmetric from the source rung job that emitted `train_loss`; still verify the\npromoted job resumed from the explicit epoch\u002Fstep checkpoint, produced a real\ncheckpoint, and is usable for evaluate\u002Finference.\nNon-train actions such as `evaluate`, `inference`, `export`, and deploy flows stay in this model skill. The per-run `automl_policy` override does not change model metadata.\n\n## Training Requirements\n\n- **Dataset type:** nvpanoptix3d\n- **Formats:** front3d, matterport\n- **Monitoring metric:** train_loss\n- **AutoML direction:** minimize\n- For AutoML train jobs, use `train_loss`. For multi-fidelity resume jobs that do not emit a fresh `train_loss`,\n  compare AutoML's carried metric to the source rung job that emitted it.\n  Validation status KPIs are `PRQ`, `RSQ`, and `RRQ`; do not use `val_loss`\n  unless a specific run is known to emit it.\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.frustum_mask_path | eval_dataset | meta\u002Ffrustum_mask.npz | No |\n| evaluate | dataset.label_map | eval_dataset | meta\u002Fcolormap.json | No |\n| evaluate | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| evaluate | dataset.val.base_dir | eval_dataset |  | No |\n| evaluate | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| evaluate | dataset.test.base_dir | inference_dataset |  | No |\n| inference | dataset.frustum_mask_path | inference_dataset | meta\u002Ffrustum_mask.npz | No |\n| inference | dataset.label_map | inference_dataset | meta\u002Fcolormap.json | No |\n| inference | inference.images_dir | inference_dataset | flat folder of `.jpg`\u002F`.png` RGB images | No |\n| train | dataset.frustum_mask_path | train_datasets | meta\u002Ffrustum_mask.npz | No |\n| train | dataset.label_map | train_datasets | meta\u002Fcolormap.json | No |\n| train | dataset.train.json_path | train_datasets | meta\u002Ftrain.json | No |\n| train | dataset.train.base_dir | train_datasets |  | No |\n| train | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| train | dataset.val.base_dir | eval_dataset |  | No |\n| train | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| train | dataset.test.base_dir | inference_dataset |  | No |\n\n### Typical Spec Overrides\n\nData source overrides are **mandatory for every action** — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in `spec_overrides`.\nFor packaged S3 folders that store scene data as `data\u002Fimages.tar.gz`, the\nskill metadata requests extraction into the parent `data\u002F` directory because\nthe TAO loader expects `base_dir\u002Fdata\u002F\u003Cscene_id>\u002F...`.\n\n```python\nS3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n```\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"model.sem_seg_head.num_classes\": 13,\n    \"dataset.frustum_mask_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.train.json_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ftrain.json\",\n    \"dataset.train.base_dir\": f\"{S3_TRAIN}\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"inference.images_dir\": \"\u002Fpath\u002Fto\u002Fflat_rgb_images\",\n}\n```\n## Eval Dataset\n\nOptional. Val\u002Ftest splits configured via dataset.val and dataset.test paths.\n\n## Important Parameters\n\n- **model.sem_seg_head.num_classes**: Number of semantic classes. Default 13.\n- **model.mode**: Prediction mode. Options: panoptic, instance, semantic. Default panoptic.\n- **model.backbone_type**: Backbone. Default vggt (only option in schema).\n- **model.mask_former.num_object_queries**: Object queries. Default 100.\n- **model.mask_former.dec_layers**: Decoder layers. Default 10.\n- **model.frustum3d.truncation**: 3D frustum truncation. Default 3.\n- **model.frustum3d.panoptic_weight**: Panoptic loss weight. Default 25.\n- **model.frustum3d.completion_weights**: Completion loss weights. Default [50, 25, 10].\n- **dataset.name**: Dataset name. Options: front3d, matterport, synthetic_hospital, synthetic_warehouse.\n- **dataset.contiguous_id**: Set `True` when the label-map JSON already\n  supplies `trainId` values for its category IDs; leaving the default `False`\n  can synthesize placeholder categories without `trainId` and fail during\n  metadata construction.\n- **dataset.downsample_factor**: Image downsample factor. Default 1 (Front3D), 2 (Matterport).\n- **dataset.target_size**: Target image size. Default [320, 240].\n- **dataset.depth_min**: Min depth. Default 0.4 meters.\n- **dataset.depth_max**: Max depth. Default 6.0 meters.\n- **train.lr**: Learning rate. Default 2e-4. backbone_multiplier=0.1.\n- **train.lr_scheduler**: Options: MultiStep, Warmuppoly. Milestones [88, 96].\n- **train.precision**: Only `fp32` is supported by the current train code.\n- **train.distributed_strategy**: Options: ddp, fsdp. activation_checkpoint=True by default.\n- **train.clip_grad_norm**: Gradient clipping norm. Default 0.1.\n- **export.onnx_file_2d**: ONNX path for 2D model component.\n- **export.max_voxels**: Max voxels for engine input. Default 700000.\n- **inference.mode**: Options: semantic, instance, panoptic.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** Lightning-managed (single `python` process, Lightning spawns workers).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.num_nodes` | Number of nodes | 1 |\n| `train.distributed_strategy` | `ddp` only | `ddp` |\n\n- **`fsdp` is NOT supported** for NVPanoptix3D (code only handles `ddp`)\n- `ddp` with activation checkpointing (enabled by default): `find_unused_parameters=False`\n- `ddp` without: `find_unused_parameters=True`\n- FAN backbones with 3D enabled auto-enable `sync_batchnorm`\n\n**Multi-node env vars** (set by orchestrator): `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT`, `NUM_GPU_PER_NODE`.\n\n## Export \u002F TRT Defaults\n\n- Exports the 2D ONNX model to `export.onnx_file_2d`. The current export\n  entrypoint calls `export_2d_model`; `export.onnx_file_3d` is present in the\n  schema but not produced by this toolkit image.\n- TRT data types: FP32, FP16 only\n- max_voxels: 700000 (engine input tensor limit)\n\n## Hardware\n\nMinimum 2 GPU(s), recommended 4 GPU(s). 40GB+ (A100 recommended) VRAM per GPU. 3D reconstruction is very memory intensive. Use `train.precision: fp32`; the current training entrypoint rejects fp16. activation_checkpoint enabled by default. FSDP for multi-node. AutoML is enabled at the model layer; preserve this GPU\u002FVRAM guidance when routing train through AutoML.\n\n## Error Patterns\n\n**`nvpanoptix3d: not found` in the PyTorch image**: Use the packaged module\nentrypoint command:\n`python -m nvidia_tao_pytorch.cv.nvpanoptix3d.entrypoint.nvpanoptix3d \u003Caction> -e \u003Cspec>`.\nThe 7.0 PyTorch image contains the NVPanoptix3D package but does not expose a\n`nvpanoptix3d` console script.\n\n**Missing frustum mask**: Ensure meta\u002Ffrustum_mask.npz is present in the dataset directory.\n\n**Downsample factor mismatch**: Use downsample_factor=2 for Matterport3D, 1 for Front3D \u002F synthetic datasets.\n\n**3D occupancy OOM**: Reduce frustum_dims or grid_dimensions if running out of GPU memory during 3D reconstruction.\n\n**fp16 precision rejected**: The schema advertises `fp16`, but the current\ntraining entrypoint raises `ValueError: Only fp32 precision is supported.` Use\n`train.precision: fp32` for train and resume\u002Fretrain.\n\n**Inference dataloader length is zero**: `inference.images_dir` is scanned only\nfor top-level `.jpg` and `.png` files. If the S3 test archive extracts to\nscene subdirectories, create or point to a flat folder of real RGB images before\nrunning inference.\n\n**3D ONNX missing after export**: The current export entrypoint only calls the\n2D ONNX exporter and writes `export.onnx_file_2d`. Do not require\n`export.onnx_file_3d` unless the toolkit image adds a 3D exporter.\n\n**Resume stops at the epoch boundary**: A one-epoch smoke run writes an\nend-of-epoch checkpoint such as `model_epoch_000_step_00020.pth`. Resuming with\n`train.num_epochs` set only one epoch beyond the original run can restore the\ncheckpoint and stop without producing a new epoch checkpoint. When validating\nactual retraining from an epoch-boundary checkpoint, set `train.num_epochs` at\nleast two epochs beyond the source smoke run and raise `train.optim.max_steps`\naccordingly. For example, resuming from `model_epoch_000_step_00020.pth` needs\n`train.num_epochs: 3` and enough max steps to produce a new exact epoch\u002Fstep\ncheckpoint such as `model_epoch_001_step_00040.pth` before handing the model to\nevaluate, inference, or export.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nModel-specific handoff mappings:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| evaluate | `encryption_key` | `key` | encryption key |\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `encryption_key` | `key` | encryption key |\n| export | `export.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| export | `export.onnx_file_2d` | `create_onnx_file_2d` | output 2D ONNX path |\n| export | `results_dir` | `output_dir` | current job results directory |\n| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.checkpoint_2d` | `parent_model_or_ptm` | parent model if available, otherwise PTM |\n| train | `train.checkpoint_3d` | `ptm` | pretrained model |\n| train | `train.resume_training_checkpoint_path` | `resume_model` | model file inferred from the current job results folder |\n\nFor `parent_model` or `parent_model_folder`, pass the upstream train\u002Fexport\u002FAutoML child job id as `parent_job_id`. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to `config.json` and do not patch generated runner scripts to guess checkpoint paths.\n",{"data":32,"body":42},{"name":4,"description":6,"license":23,"compatibility":33,"metadata":34,"allowed-tools":37,"tags":38},"Requires docker + nvidia-container-toolkit.",{"version":35,"author":36},"0.1.0","NVIDIA Corporation","Read Bash",[39,40,41],"panoptic","3d","reconstruction",{"type":43,"children":44},"root",[45,54,60,65,72,166,172,281,406,412,503,510,975,981,1024,1055,1063,1221,1229,1316,1324,1382,1388,1393,1399,1679,1685,1702,1826,1895,1940,1946,1987,1993,2006,2012,2043,2053,2063,2073,2106,2136,2160,2224,2230,2259,2264,2722,2757],{"type":46,"tag":47,"props":48,"children":50},"element","h1",{"id":49},"nvpanoptix3d",[51],{"type":52,"value":53},"text","NVPanoptix3D",{"type":46,"tag":55,"props":56,"children":57},"p",{},[58],{"type":52,"value":59},"NVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on VGGT backbone with Mask2Former-style head and 3D frustum reconstruction.",{"type":46,"tag":55,"props":61,"children":62},{},[63],{"type":52,"value":64},"Uses 2D and 3D stage checkpoints. Set train.checkpoint_2d and train.checkpoint_3d for staged initialization.",{"type":46,"tag":66,"props":67,"children":69},"h2",{"id":68},"dataclass-schemas",[70],{"type":52,"value":71},"Dataclass Schemas",{"type":46,"tag":55,"props":73,"children":74},{},[75,77,84,86,92,94,100,102,108,110,116,118,124,126,132,134,140,142,148,150,156,158,164],{"type":52,"value":76},"Generated TAO Core schemas are packaged in ",{"type":46,"tag":78,"props":79,"children":81},"code",{"className":80},[],[82],{"type":52,"value":83},"schemas\u002F\u003Caction>.schema.json",{"type":52,"value":85},", with ",{"type":46,"tag":78,"props":87,"children":89},{"className":88},[],[90],{"type":52,"value":91},"schemas\u002Fmanifest.json",{"type":52,"value":93}," listing available actions. Each generated schema also emits ",{"type":46,"tag":78,"props":95,"children":97},{"className":96},[],[98],{"type":52,"value":99},"references\u002Fspec_template_\u003Caction>.yaml",{"type":52,"value":101}," from the schema top-level ",{"type":46,"tag":78,"props":103,"children":105},{"className":104},[],[106],{"type":52,"value":107},"default",{"type":52,"value":109}," field. AutoML enablement is declared at the model layer in ",{"type":46,"tag":78,"props":111,"children":113},{"className":112},[],[114],{"type":52,"value":115},"references\u002Fskill_info.yaml",{"type":52,"value":117}," via ",{"type":46,"tag":78,"props":119,"children":121},{"className":120},[],[122],{"type":52,"value":123},"automl_enabled",{"type":52,"value":125},". Runnable AutoML still requires ",{"type":46,"tag":78,"props":127,"children":129},{"className":128},[],[130],{"type":52,"value":131},"schemas\u002Ftrain.schema.json",{"type":52,"value":133}," and ",{"type":46,"tag":78,"props":135,"children":137},{"className":136},[],[138],{"type":52,"value":139},"references\u002Fspec_template_train.yaml",{"type":52,"value":141}," to exist and parse. Use the packaged train schema for ",{"type":46,"tag":78,"props":143,"children":145},{"className":144},[],[146],{"type":52,"value":147},"automl_default_parameters",{"type":52,"value":149},", ",{"type":46,"tag":78,"props":151,"children":153},{"className":152},[],[154],{"type":52,"value":155},"automl_disabled_parameters",{"type":52,"value":157},", defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect ",{"type":46,"tag":78,"props":159,"children":161},{"className":160},[],[162],{"type":52,"value":163},"~\u002Ftao-core",{"type":52,"value":165}," at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.",{"type":46,"tag":66,"props":167,"children":169},{"id":168},"train-action-policy",[170],{"type":52,"value":171},"Train Action Policy",{"type":46,"tag":55,"props":173,"children":174},{},[175,177,182,184,190,192,198,200,206,208,214,216,222,224,229,230,236,238,243,244,249,251,257,259,265,267,272,274,279],{"type":52,"value":176},"This model is AutoML-enabled at the model layer. Before handling any train-stage request, read ",{"type":46,"tag":78,"props":178,"children":180},{"className":179},[],[181],{"type":52,"value":115},{"type":52,"value":183}," and resolve the run override from either an explicit ",{"type":46,"tag":78,"props":185,"children":187},{"className":186},[],[188],{"type":52,"value":189},"automl_policy",{"type":52,"value":191}," value or the user's workflow request. Use ",{"type":46,"tag":78,"props":193,"children":195},{"className":194},[],[196],{"type":52,"value":197},"automl_policy: on",{"type":52,"value":199}," by default and only expose ",{"type":46,"tag":78,"props":201,"children":203},{"className":202},[],[204],{"type":52,"value":205},"on",{"type":52,"value":207}," \u002F ",{"type":46,"tag":78,"props":209,"children":211},{"className":210},[],[212],{"type":52,"value":213},"off",{"type":52,"value":215}," in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as ",{"type":46,"tag":78,"props":217,"children":219},{"className":218},[],[220],{"type":52,"value":221},"automl_policy: off",{"type":52,"value":223}," for this run only. When ",{"type":46,"tag":78,"props":225,"children":227},{"className":226},[],[228],{"type":52,"value":197},{"type":52,"value":149},{"type":46,"tag":78,"props":231,"children":233},{"className":232},[],[234],{"type":52,"value":235},"automl_enabled: true",{"type":52,"value":237},", and both ",{"type":46,"tag":78,"props":239,"children":241},{"className":240},[],[242],{"type":52,"value":131},{"type":52,"value":133},{"type":46,"tag":78,"props":245,"children":247},{"className":246},[],[248],{"type":52,"value":139},{"type":52,"value":250}," are packaged, route the train action through ",{"type":46,"tag":78,"props":252,"children":254},{"className":253},[],[255],{"type":52,"value":256},"tao-skill-bank:tao-run-automl",{"type":52,"value":258}," by default with this model's ",{"type":46,"tag":78,"props":260,"children":262},{"className":261},[],[263],{"type":52,"value":264},"skill_dir",{"type":52,"value":266},". Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and ",{"type":46,"tag":78,"props":268,"children":270},{"className":269},[],[271],{"type":52,"value":189},{"type":52,"value":273},". Use direct model training only when ",{"type":46,"tag":78,"props":275,"children":277},{"className":276},[],[278],{"type":52,"value":221},{"type":52,"value":280}," or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.",{"type":46,"tag":55,"props":282,"children":283},{},[284,286,292,294,300,302,308,310,316,318,324,325,331,333,339,340,346,348,353,355,361,363,368,370,375,377,383,384,390,391,397,399,404],{"type":52,"value":285},"For AutoML, use ",{"type":46,"tag":78,"props":287,"children":289},{"className":288},[],[290],{"type":52,"value":291},"train_loss",{"type":52,"value":293}," as the optimization metric with\n",{"type":46,"tag":78,"props":295,"children":297},{"className":296},[],[298],{"type":52,"value":299},"direction=minimize",{"type":52,"value":301},", and set ",{"type":46,"tag":78,"props":303,"children":305},{"className":304},[],[306],{"type":52,"value":307},"train.optim.monitor_name: train_loss",{"type":52,"value":309}," in\n",{"type":46,"tag":78,"props":311,"children":313},{"className":312},[],[314],{"type":52,"value":315},"spec_overrides",{"type":52,"value":317},". NVPanoptix3D train jobs emit ",{"type":46,"tag":78,"props":319,"children":321},{"className":320},[],[322],{"type":52,"value":323},"PRQ",{"type":52,"value":149},{"type":46,"tag":78,"props":326,"children":328},{"className":327},[],[329],{"type":52,"value":330},"RSQ",{"type":52,"value":332},", and ",{"type":46,"tag":78,"props":334,"children":336},{"className":335},[],[337],{"type":52,"value":338},"RRQ",{"type":52,"value":309},{"type":46,"tag":78,"props":341,"children":343},{"className":342},[],[344],{"type":52,"value":345},"status.json",{"type":52,"value":347},", and the training progress log emits ",{"type":46,"tag":78,"props":349,"children":351},{"className":350},[],[352],{"type":52,"value":291},{"type":52,"value":354},"; short or\nminimal jobs may not emit ",{"type":46,"tag":78,"props":356,"children":358},{"className":357},[],[359],{"type":52,"value":360},"val_loss",{"type":52,"value":362},", including full-trial smoke runs.\nMulti-fidelity AutoML algorithms such as Hyperband, ASHA, and BOHB may promote\na checkpoint to a resume job that completes without emitting a fresh\n",{"type":46,"tag":78,"props":364,"children":366},{"className":365},[],[367],{"type":52,"value":291},{"type":52,"value":369}," line. In that case, the AutoML metric is the carried-forward\nmetric from the source rung job that emitted ",{"type":46,"tag":78,"props":371,"children":373},{"className":372},[],[374],{"type":52,"value":291},{"type":52,"value":376},"; still verify the\npromoted job resumed from the explicit epoch\u002Fstep checkpoint, produced a real\ncheckpoint, and is usable for evaluate\u002Finference.\nNon-train actions such as ",{"type":46,"tag":78,"props":378,"children":380},{"className":379},[],[381],{"type":52,"value":382},"evaluate",{"type":52,"value":149},{"type":46,"tag":78,"props":385,"children":387},{"className":386},[],[388],{"type":52,"value":389},"inference",{"type":52,"value":149},{"type":46,"tag":78,"props":392,"children":394},{"className":393},[],[395],{"type":52,"value":396},"export",{"type":52,"value":398},", and deploy flows stay in this model skill. The per-run ",{"type":46,"tag":78,"props":400,"children":402},{"className":401},[],[403],{"type":52,"value":189},{"type":52,"value":405}," override does not change model metadata.",{"type":46,"tag":66,"props":407,"children":409},{"id":408},"training-requirements",[410],{"type":52,"value":411},"Training Requirements",{"type":46,"tag":413,"props":414,"children":415},"ul",{},[416,428,438,448,458],{"type":46,"tag":417,"props":418,"children":419},"li",{},[420,426],{"type":46,"tag":421,"props":422,"children":423},"strong",{},[424],{"type":52,"value":425},"Dataset type:",{"type":52,"value":427}," nvpanoptix3d",{"type":46,"tag":417,"props":429,"children":430},{},[431,436],{"type":46,"tag":421,"props":432,"children":433},{},[434],{"type":52,"value":435},"Formats:",{"type":52,"value":437}," front3d, matterport",{"type":46,"tag":417,"props":439,"children":440},{},[441,446],{"type":46,"tag":421,"props":442,"children":443},{},[444],{"type":52,"value":445},"Monitoring metric:",{"type":52,"value":447}," train_loss",{"type":46,"tag":417,"props":449,"children":450},{},[451,456],{"type":46,"tag":421,"props":452,"children":453},{},[454],{"type":52,"value":455},"AutoML direction:",{"type":52,"value":457}," minimize",{"type":46,"tag":417,"props":459,"children":460},{},[461,463,468,470,475,477,482,483,488,489,494,496,501],{"type":52,"value":462},"For AutoML train jobs, use ",{"type":46,"tag":78,"props":464,"children":466},{"className":465},[],[467],{"type":52,"value":291},{"type":52,"value":469},". For multi-fidelity resume jobs that do not emit a fresh ",{"type":46,"tag":78,"props":471,"children":473},{"className":472},[],[474],{"type":52,"value":291},{"type":52,"value":476},",\ncompare AutoML's carried metric to the source rung job that emitted it.\nValidation status KPIs are ",{"type":46,"tag":78,"props":478,"children":480},{"className":479},[],[481],{"type":52,"value":323},{"type":52,"value":149},{"type":46,"tag":78,"props":484,"children":486},{"className":485},[],[487],{"type":52,"value":330},{"type":52,"value":332},{"type":46,"tag":78,"props":490,"children":492},{"className":491},[],[493],{"type":52,"value":338},{"type":52,"value":495},"; do not use ",{"type":46,"tag":78,"props":497,"children":499},{"className":498},[],[500],{"type":52,"value":360},{"type":52,"value":502},"\nunless a specific run is known to emit it.",{"type":46,"tag":504,"props":505,"children":507},"h3",{"id":506},"per-action-dataset-requirements",[508],{"type":52,"value":509},"Per-Action Dataset Requirements",{"type":46,"tag":511,"props":512,"children":513},"table",{},[514,548],{"type":46,"tag":515,"props":516,"children":517},"thead",{},[518],{"type":46,"tag":519,"props":520,"children":521},"tr",{},[522,528,533,538,543],{"type":46,"tag":523,"props":524,"children":525},"th",{},[526],{"type":52,"value":527},"Action",{"type":46,"tag":523,"props":529,"children":530},{},[531],{"type":52,"value":532},"Spec Key",{"type":46,"tag":523,"props":534,"children":535},{},[536],{"type":52,"value":537},"Source",{"type":46,"tag":523,"props":539,"children":540},{},[541],{"type":52,"value":542},"Files",{"type":46,"tag":523,"props":544,"children":545},{},[546],{"type":52,"value":547},"List?",{"type":46,"tag":549,"props":550,"children":551},"tbody",{},[552,580,605,630,653,679,702,725,748,789,814,837,862,885,908,930,953],{"type":46,"tag":519,"props":553,"children":554},{},[555,560,565,570,575],{"type":46,"tag":556,"props":557,"children":558},"td",{},[559],{"type":52,"value":382},{"type":46,"tag":556,"props":561,"children":562},{},[563],{"type":52,"value":564},"dataset.frustum_mask_path",{"type":46,"tag":556,"props":566,"children":567},{},[568],{"type":52,"value":569},"eval_dataset",{"type":46,"tag":556,"props":571,"children":572},{},[573],{"type":52,"value":574},"meta\u002Ffrustum_mask.npz",{"type":46,"tag":556,"props":576,"children":577},{},[578],{"type":52,"value":579},"No",{"type":46,"tag":519,"props":581,"children":582},{},[583,587,592,596,601],{"type":46,"tag":556,"props":584,"children":585},{},[586],{"type":52,"value":382},{"type":46,"tag":556,"props":588,"children":589},{},[590],{"type":52,"value":591},"dataset.label_map",{"type":46,"tag":556,"props":593,"children":594},{},[595],{"type":52,"value":569},{"type":46,"tag":556,"props":597,"children":598},{},[599],{"type":52,"value":600},"meta\u002Fcolormap.json",{"type":46,"tag":556,"props":602,"children":603},{},[604],{"type":52,"value":579},{"type":46,"tag":519,"props":606,"children":607},{},[608,612,617,621,626],{"type":46,"tag":556,"props":609,"children":610},{},[611],{"type":52,"value":382},{"type":46,"tag":556,"props":613,"children":614},{},[615],{"type":52,"value":616},"dataset.val.json_path",{"type":46,"tag":556,"props":618,"children":619},{},[620],{"type":52,"value":569},{"type":46,"tag":556,"props":622,"children":623},{},[624],{"type":52,"value":625},"meta\u002Fval.json",{"type":46,"tag":556,"props":627,"children":628},{},[629],{"type":52,"value":579},{"type":46,"tag":519,"props":631,"children":632},{},[633,637,642,646,649],{"type":46,"tag":556,"props":634,"children":635},{},[636],{"type":52,"value":382},{"type":46,"tag":556,"props":638,"children":639},{},[640],{"type":52,"value":641},"dataset.val.base_dir",{"type":46,"tag":556,"props":643,"children":644},{},[645],{"type":52,"value":569},{"type":46,"tag":556,"props":647,"children":648},{},[],{"type":46,"tag":556,"props":650,"children":651},{},[652],{"type":52,"value":579},{"type":46,"tag":519,"props":654,"children":655},{},[656,660,665,670,675],{"type":46,"tag":556,"props":657,"children":658},{},[659],{"type":52,"value":382},{"type":46,"tag":556,"props":661,"children":662},{},[663],{"type":52,"value":664},"dataset.test.json_path",{"type":46,"tag":556,"props":666,"children":667},{},[668],{"type":52,"value":669},"inference_dataset",{"type":46,"tag":556,"props":671,"children":672},{},[673],{"type":52,"value":674},"meta\u002Ftest.json",{"type":46,"tag":556,"props":676,"children":677},{},[678],{"type":52,"value":579},{"type":46,"tag":519,"props":680,"children":681},{},[682,686,691,695,698],{"type":46,"tag":556,"props":683,"children":684},{},[685],{"type":52,"value":382},{"type":46,"tag":556,"props":687,"children":688},{},[689],{"type":52,"value":690},"dataset.test.base_dir",{"type":46,"tag":556,"props":692,"children":693},{},[694],{"type":52,"value":669},{"type":46,"tag":556,"props":696,"children":697},{},[],{"type":46,"tag":556,"props":699,"children":700},{},[701],{"type":52,"value":579},{"type":46,"tag":519,"props":703,"children":704},{},[705,709,713,717,721],{"type":46,"tag":556,"props":706,"children":707},{},[708],{"type":52,"value":389},{"type":46,"tag":556,"props":710,"children":711},{},[712],{"type":52,"value":564},{"type":46,"tag":556,"props":714,"children":715},{},[716],{"type":52,"value":669},{"type":46,"tag":556,"props":718,"children":719},{},[720],{"type":52,"value":574},{"type":46,"tag":556,"props":722,"children":723},{},[724],{"type":52,"value":579},{"type":46,"tag":519,"props":726,"children":727},{},[728,732,736,740,744],{"type":46,"tag":556,"props":729,"children":730},{},[731],{"type":52,"value":389},{"type":46,"tag":556,"props":733,"children":734},{},[735],{"type":52,"value":591},{"type":46,"tag":556,"props":737,"children":738},{},[739],{"type":52,"value":669},{"type":46,"tag":556,"props":741,"children":742},{},[743],{"type":52,"value":600},{"type":46,"tag":556,"props":745,"children":746},{},[747],{"type":52,"value":579},{"type":46,"tag":519,"props":749,"children":750},{},[751,755,760,764,785],{"type":46,"tag":556,"props":752,"children":753},{},[754],{"type":52,"value":389},{"type":46,"tag":556,"props":756,"children":757},{},[758],{"type":52,"value":759},"inference.images_dir",{"type":46,"tag":556,"props":761,"children":762},{},[763],{"type":52,"value":669},{"type":46,"tag":556,"props":765,"children":766},{},[767,769,775,777,783],{"type":52,"value":768},"flat folder of ",{"type":46,"tag":78,"props":770,"children":772},{"className":771},[],[773],{"type":52,"value":774},".jpg",{"type":52,"value":776},"\u002F",{"type":46,"tag":78,"props":778,"children":780},{"className":779},[],[781],{"type":52,"value":782},".png",{"type":52,"value":784}," RGB images",{"type":46,"tag":556,"props":786,"children":787},{},[788],{"type":52,"value":579},{"type":46,"tag":519,"props":790,"children":791},{},[792,797,801,806,810],{"type":46,"tag":556,"props":793,"children":794},{},[795],{"type":52,"value":796},"train",{"type":46,"tag":556,"props":798,"children":799},{},[800],{"type":52,"value":564},{"type":46,"tag":556,"props":802,"children":803},{},[804],{"type":52,"value":805},"train_datasets",{"type":46,"tag":556,"props":807,"children":808},{},[809],{"type":52,"value":574},{"type":46,"tag":556,"props":811,"children":812},{},[813],{"type":52,"value":579},{"type":46,"tag":519,"props":815,"children":816},{},[817,821,825,829,833],{"type":46,"tag":556,"props":818,"children":819},{},[820],{"type":52,"value":796},{"type":46,"tag":556,"props":822,"children":823},{},[824],{"type":52,"value":591},{"type":46,"tag":556,"props":826,"children":827},{},[828],{"type":52,"value":805},{"type":46,"tag":556,"props":830,"children":831},{},[832],{"type":52,"value":600},{"type":46,"tag":556,"props":834,"children":835},{},[836],{"type":52,"value":579},{"type":46,"tag":519,"props":838,"children":839},{},[840,844,849,853,858],{"type":46,"tag":556,"props":841,"children":842},{},[843],{"type":52,"value":796},{"type":46,"tag":556,"props":845,"children":846},{},[847],{"type":52,"value":848},"dataset.train.json_path",{"type":46,"tag":556,"props":850,"children":851},{},[852],{"type":52,"value":805},{"type":46,"tag":556,"props":854,"children":855},{},[856],{"type":52,"value":857},"meta\u002Ftrain.json",{"type":46,"tag":556,"props":859,"children":860},{},[861],{"type":52,"value":579},{"type":46,"tag":519,"props":863,"children":864},{},[865,869,874,878,881],{"type":46,"tag":556,"props":866,"children":867},{},[868],{"type":52,"value":796},{"type":46,"tag":556,"props":870,"children":871},{},[872],{"type":52,"value":873},"dataset.train.base_dir",{"type":46,"tag":556,"props":875,"children":876},{},[877],{"type":52,"value":805},{"type":46,"tag":556,"props":879,"children":880},{},[],{"type":46,"tag":556,"props":882,"children":883},{},[884],{"type":52,"value":579},{"type":46,"tag":519,"props":886,"children":887},{},[888,892,896,900,904],{"type":46,"tag":556,"props":889,"children":890},{},[891],{"type":52,"value":796},{"type":46,"tag":556,"props":893,"children":894},{},[895],{"type":52,"value":616},{"type":46,"tag":556,"props":897,"children":898},{},[899],{"type":52,"value":569},{"type":46,"tag":556,"props":901,"children":902},{},[903],{"type":52,"value":625},{"type":46,"tag":556,"props":905,"children":906},{},[907],{"type":52,"value":579},{"type":46,"tag":519,"props":909,"children":910},{},[911,915,919,923,926],{"type":46,"tag":556,"props":912,"children":913},{},[914],{"type":52,"value":796},{"type":46,"tag":556,"props":916,"children":917},{},[918],{"type":52,"value":641},{"type":46,"tag":556,"props":920,"children":921},{},[922],{"type":52,"value":569},{"type":46,"tag":556,"props":924,"children":925},{},[],{"type":46,"tag":556,"props":927,"children":928},{},[929],{"type":52,"value":579},{"type":46,"tag":519,"props":931,"children":932},{},[933,937,941,945,949],{"type":46,"tag":556,"props":934,"children":935},{},[936],{"type":52,"value":796},{"type":46,"tag":556,"props":938,"children":939},{},[940],{"type":52,"value":664},{"type":46,"tag":556,"props":942,"children":943},{},[944],{"type":52,"value":669},{"type":46,"tag":556,"props":946,"children":947},{},[948],{"type":52,"value":674},{"type":46,"tag":556,"props":950,"children":951},{},[952],{"type":52,"value":579},{"type":46,"tag":519,"props":954,"children":955},{},[956,960,964,968,971],{"type":46,"tag":556,"props":957,"children":958},{},[959],{"type":52,"value":796},{"type":46,"tag":556,"props":961,"children":962},{},[963],{"type":52,"value":690},{"type":46,"tag":556,"props":965,"children":966},{},[967],{"type":52,"value":669},{"type":46,"tag":556,"props":969,"children":970},{},[],{"type":46,"tag":556,"props":972,"children":973},{},[974],{"type":52,"value":579},{"type":46,"tag":504,"props":976,"children":978},{"id":977},"typical-spec-overrides",[979],{"type":52,"value":980},"Typical Spec Overrides",{"type":46,"tag":55,"props":982,"children":983},{},[984,986,991,993,998,1000,1006,1008,1014,1016,1022],{"type":52,"value":985},"Data source overrides are ",{"type":46,"tag":421,"props":987,"children":988},{},[989],{"type":52,"value":990},"mandatory for every action",{"type":52,"value":992}," — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in ",{"type":46,"tag":78,"props":994,"children":996},{"className":995},[],[997],{"type":52,"value":315},{"type":52,"value":999},".\nFor packaged S3 folders that store scene data as ",{"type":46,"tag":78,"props":1001,"children":1003},{"className":1002},[],[1004],{"type":52,"value":1005},"data\u002Fimages.tar.gz",{"type":52,"value":1007},", the\nskill metadata requests extraction into the parent ",{"type":46,"tag":78,"props":1009,"children":1011},{"className":1010},[],[1012],{"type":52,"value":1013},"data\u002F",{"type":52,"value":1015}," directory because\nthe TAO loader expects ",{"type":46,"tag":78,"props":1017,"children":1019},{"className":1018},[],[1020],{"type":52,"value":1021},"base_dir\u002Fdata\u002F\u003Cscene_id>\u002F...",{"type":52,"value":1023},".",{"type":46,"tag":1025,"props":1026,"children":1031},"pre",{"className":1027,"code":1028,"language":1029,"meta":1030,"style":1030},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","S3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n","python","",[1032],{"type":46,"tag":78,"props":1033,"children":1034},{"__ignoreMap":1030},[1035,1046],{"type":46,"tag":1036,"props":1037,"children":1040},"span",{"class":1038,"line":1039},"line",1,[1041],{"type":46,"tag":1036,"props":1042,"children":1043},{},[1044],{"type":52,"value":1045},"S3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\n",{"type":46,"tag":1036,"props":1047,"children":1049},{"class":1038,"line":1048},2,[1050],{"type":46,"tag":1036,"props":1051,"children":1052},{},[1053],{"type":52,"value":1054},"S3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n",{"type":46,"tag":55,"props":1056,"children":1057},{},[1058],{"type":46,"tag":421,"props":1059,"children":1060},{},[1061],{"type":52,"value":1062},"train (mandatory data sources):",{"type":46,"tag":1025,"props":1064,"children":1066},{"className":1027,"code":1065,"language":1029,"meta":1030,"style":1030},"{\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"model.sem_seg_head.num_classes\": 13,\n    \"dataset.frustum_mask_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.train.json_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ftrain.json\",\n    \"dataset.train.base_dir\": f\"{S3_TRAIN}\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n",[1067],{"type":46,"tag":78,"props":1068,"children":1069},{"__ignoreMap":1030},[1070,1078,1086,1095,1104,1113,1122,1131,1140,1149,1158,1167,1176,1185,1194,1203,1212],{"type":46,"tag":1036,"props":1071,"children":1072},{"class":1038,"line":1039},[1073],{"type":46,"tag":1036,"props":1074,"children":1075},{},[1076],{"type":52,"value":1077},"{\n",{"type":46,"tag":1036,"props":1079,"children":1080},{"class":1038,"line":1048},[1081],{"type":46,"tag":1036,"props":1082,"children":1083},{},[1084],{"type":52,"value":1085},"    \"train.num_epochs\": 10,\n",{"type":46,"tag":1036,"props":1087,"children":1089},{"class":1038,"line":1088},3,[1090],{"type":46,"tag":1036,"props":1091,"children":1092},{},[1093],{"type":52,"value":1094},"    \"train.checkpoint_interval\": 10,\n",{"type":46,"tag":1036,"props":1096,"children":1098},{"class":1038,"line":1097},4,[1099],{"type":46,"tag":1036,"props":1100,"children":1101},{},[1102],{"type":52,"value":1103},"    \"train.validation_interval\": 10,\n",{"type":46,"tag":1036,"props":1105,"children":1107},{"class":1038,"line":1106},5,[1108],{"type":46,"tag":1036,"props":1109,"children":1110},{},[1111],{"type":52,"value":1112},"    \"train.num_gpus\": 1,\n",{"type":46,"tag":1036,"props":1114,"children":1116},{"class":1038,"line":1115},6,[1117],{"type":46,"tag":1036,"props":1118,"children":1119},{},[1120],{"type":52,"value":1121},"    \"dataset.enable_3d\": True,\n",{"type":46,"tag":1036,"props":1123,"children":1125},{"class":1038,"line":1124},7,[1126],{"type":46,"tag":1036,"props":1127,"children":1128},{},[1129],{"type":52,"value":1130},"    \"dataset.contiguous_id\": True,\n",{"type":46,"tag":1036,"props":1132,"children":1134},{"class":1038,"line":1133},8,[1135],{"type":46,"tag":1036,"props":1136,"children":1137},{},[1138],{"type":52,"value":1139},"    \"model.sem_seg_head.num_classes\": 13,\n",{"type":46,"tag":1036,"props":1141,"children":1143},{"class":1038,"line":1142},9,[1144],{"type":46,"tag":1036,"props":1145,"children":1146},{},[1147],{"type":52,"value":1148},"    \"dataset.frustum_mask_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ffrustum_mask.npz\",\n",{"type":46,"tag":1036,"props":1150,"children":1152},{"class":1038,"line":1151},10,[1153],{"type":46,"tag":1036,"props":1154,"children":1155},{},[1156],{"type":52,"value":1157},"    \"dataset.label_map\": f\"{S3_TRAIN}\u002Fmeta\u002Fcolormap.json\",\n",{"type":46,"tag":1036,"props":1159,"children":1161},{"class":1038,"line":1160},11,[1162],{"type":46,"tag":1036,"props":1163,"children":1164},{},[1165],{"type":52,"value":1166},"    \"dataset.train.json_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ftrain.json\",\n",{"type":46,"tag":1036,"props":1168,"children":1170},{"class":1038,"line":1169},12,[1171],{"type":46,"tag":1036,"props":1172,"children":1173},{},[1174],{"type":52,"value":1175},"    \"dataset.train.base_dir\": f\"{S3_TRAIN}\",\n",{"type":46,"tag":1036,"props":1177,"children":1179},{"class":1038,"line":1178},13,[1180],{"type":46,"tag":1036,"props":1181,"children":1182},{},[1183],{"type":52,"value":1184},"    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n",{"type":46,"tag":1036,"props":1186,"children":1188},{"class":1038,"line":1187},14,[1189],{"type":46,"tag":1036,"props":1190,"children":1191},{},[1192],{"type":52,"value":1193},"    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n",{"type":46,"tag":1036,"props":1195,"children":1197},{"class":1038,"line":1196},15,[1198],{"type":46,"tag":1036,"props":1199,"children":1200},{},[1201],{"type":52,"value":1202},"    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n",{"type":46,"tag":1036,"props":1204,"children":1206},{"class":1038,"line":1205},16,[1207],{"type":46,"tag":1036,"props":1208,"children":1209},{},[1210],{"type":52,"value":1211},"    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n",{"type":46,"tag":1036,"props":1213,"children":1215},{"class":1038,"line":1214},17,[1216],{"type":46,"tag":1036,"props":1217,"children":1218},{},[1219],{"type":52,"value":1220},"}\n",{"type":46,"tag":55,"props":1222,"children":1223},{},[1224],{"type":46,"tag":421,"props":1225,"children":1226},{},[1227],{"type":52,"value":1228},"evaluate (mandatory data sources):",{"type":46,"tag":1025,"props":1230,"children":1232},{"className":1027,"code":1231,"language":1029,"meta":1030,"style":1030},"{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n",[1233],{"type":46,"tag":78,"props":1234,"children":1235},{"__ignoreMap":1030},[1236,1243,1251,1258,1265,1273,1281,1288,1295,1302,1309],{"type":46,"tag":1036,"props":1237,"children":1238},{"class":1038,"line":1039},[1239],{"type":46,"tag":1036,"props":1240,"children":1241},{},[1242],{"type":52,"value":1077},{"type":46,"tag":1036,"props":1244,"children":1245},{"class":1038,"line":1048},[1246],{"type":46,"tag":1036,"props":1247,"children":1248},{},[1249],{"type":52,"value":1250},"    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":46,"tag":1036,"props":1252,"children":1253},{"class":1038,"line":1088},[1254],{"type":46,"tag":1036,"props":1255,"children":1256},{},[1257],{"type":52,"value":1121},{"type":46,"tag":1036,"props":1259,"children":1260},{"class":1038,"line":1097},[1261],{"type":46,"tag":1036,"props":1262,"children":1263},{},[1264],{"type":52,"value":1130},{"type":46,"tag":1036,"props":1266,"children":1267},{"class":1038,"line":1106},[1268],{"type":46,"tag":1036,"props":1269,"children":1270},{},[1271],{"type":52,"value":1272},"    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n",{"type":46,"tag":1036,"props":1274,"children":1275},{"class":1038,"line":1115},[1276],{"type":46,"tag":1036,"props":1277,"children":1278},{},[1279],{"type":52,"value":1280},"    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n",{"type":46,"tag":1036,"props":1282,"children":1283},{"class":1038,"line":1124},[1284],{"type":46,"tag":1036,"props":1285,"children":1286},{},[1287],{"type":52,"value":1184},{"type":46,"tag":1036,"props":1289,"children":1290},{"class":1038,"line":1133},[1291],{"type":46,"tag":1036,"props":1292,"children":1293},{},[1294],{"type":52,"value":1193},{"type":46,"tag":1036,"props":1296,"children":1297},{"class":1038,"line":1142},[1298],{"type":46,"tag":1036,"props":1299,"children":1300},{},[1301],{"type":52,"value":1202},{"type":46,"tag":1036,"props":1303,"children":1304},{"class":1038,"line":1151},[1305],{"type":46,"tag":1036,"props":1306,"children":1307},{},[1308],{"type":52,"value":1211},{"type":46,"tag":1036,"props":1310,"children":1311},{"class":1038,"line":1160},[1312],{"type":46,"tag":1036,"props":1313,"children":1314},{},[1315],{"type":52,"value":1220},{"type":46,"tag":55,"props":1317,"children":1318},{},[1319],{"type":46,"tag":421,"props":1320,"children":1321},{},[1322],{"type":52,"value":1323},"inference (mandatory data sources):",{"type":46,"tag":1025,"props":1325,"children":1327},{"className":1027,"code":1326,"language":1029,"meta":1030,"style":1030},"{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"inference.images_dir\": \"\u002Fpath\u002Fto\u002Fflat_rgb_images\",\n}\n",[1328],{"type":46,"tag":78,"props":1329,"children":1330},{"__ignoreMap":1030},[1331,1338,1346,1353,1360,1367,1375],{"type":46,"tag":1036,"props":1332,"children":1333},{"class":1038,"line":1039},[1334],{"type":46,"tag":1036,"props":1335,"children":1336},{},[1337],{"type":52,"value":1077},{"type":46,"tag":1036,"props":1339,"children":1340},{"class":1038,"line":1048},[1341],{"type":46,"tag":1036,"props":1342,"children":1343},{},[1344],{"type":52,"value":1345},"    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":46,"tag":1036,"props":1347,"children":1348},{"class":1038,"line":1088},[1349],{"type":46,"tag":1036,"props":1350,"children":1351},{},[1352],{"type":52,"value":1121},{"type":46,"tag":1036,"props":1354,"children":1355},{"class":1038,"line":1097},[1356],{"type":46,"tag":1036,"props":1357,"children":1358},{},[1359],{"type":52,"value":1272},{"type":46,"tag":1036,"props":1361,"children":1362},{"class":1038,"line":1106},[1363],{"type":46,"tag":1036,"props":1364,"children":1365},{},[1366],{"type":52,"value":1280},{"type":46,"tag":1036,"props":1368,"children":1369},{"class":1038,"line":1115},[1370],{"type":46,"tag":1036,"props":1371,"children":1372},{},[1373],{"type":52,"value":1374},"    \"inference.images_dir\": \"\u002Fpath\u002Fto\u002Fflat_rgb_images\",\n",{"type":46,"tag":1036,"props":1376,"children":1377},{"class":1038,"line":1124},[1378],{"type":46,"tag":1036,"props":1379,"children":1380},{},[1381],{"type":52,"value":1220},{"type":46,"tag":66,"props":1383,"children":1385},{"id":1384},"eval-dataset",[1386],{"type":52,"value":1387},"Eval Dataset",{"type":46,"tag":55,"props":1389,"children":1390},{},[1391],{"type":52,"value":1392},"Optional. Val\u002Ftest splits configured via dataset.val and dataset.test paths.",{"type":46,"tag":66,"props":1394,"children":1396},{"id":1395},"important-parameters",[1397],{"type":52,"value":1398},"Important Parameters",{"type":46,"tag":413,"props":1400,"children":1401},{},[1402,1412,1422,1432,1442,1452,1462,1472,1488,1498,1539,1549,1565,1575,1585,1595,1611,1629,1639,1649,1659,1669],{"type":46,"tag":417,"props":1403,"children":1404},{},[1405,1410],{"type":46,"tag":421,"props":1406,"children":1407},{},[1408],{"type":52,"value":1409},"model.sem_seg_head.num_classes",{"type":52,"value":1411},": Number of semantic classes. Default 13.",{"type":46,"tag":417,"props":1413,"children":1414},{},[1415,1420],{"type":46,"tag":421,"props":1416,"children":1417},{},[1418],{"type":52,"value":1419},"model.mode",{"type":52,"value":1421},": Prediction mode. Options: panoptic, instance, semantic. Default panoptic.",{"type":46,"tag":417,"props":1423,"children":1424},{},[1425,1430],{"type":46,"tag":421,"props":1426,"children":1427},{},[1428],{"type":52,"value":1429},"model.backbone_type",{"type":52,"value":1431},": Backbone. Default vggt (only option in schema).",{"type":46,"tag":417,"props":1433,"children":1434},{},[1435,1440],{"type":46,"tag":421,"props":1436,"children":1437},{},[1438],{"type":52,"value":1439},"model.mask_former.num_object_queries",{"type":52,"value":1441},": Object queries. Default 100.",{"type":46,"tag":417,"props":1443,"children":1444},{},[1445,1450],{"type":46,"tag":421,"props":1446,"children":1447},{},[1448],{"type":52,"value":1449},"model.mask_former.dec_layers",{"type":52,"value":1451},": Decoder layers. Default 10.",{"type":46,"tag":417,"props":1453,"children":1454},{},[1455,1460],{"type":46,"tag":421,"props":1456,"children":1457},{},[1458],{"type":52,"value":1459},"model.frustum3d.truncation",{"type":52,"value":1461},": 3D frustum truncation. Default 3.",{"type":46,"tag":417,"props":1463,"children":1464},{},[1465,1470],{"type":46,"tag":421,"props":1466,"children":1467},{},[1468],{"type":52,"value":1469},"model.frustum3d.panoptic_weight",{"type":52,"value":1471},": Panoptic loss weight. Default 25.",{"type":46,"tag":417,"props":1473,"children":1474},{},[1475,1480,1482,1487],{"type":46,"tag":421,"props":1476,"children":1477},{},[1478],{"type":52,"value":1479},"model.frustum3d.completion_weights",{"type":52,"value":1481},": Completion loss weights. Default ",{"type":46,"tag":1036,"props":1483,"children":1484},{},[1485],{"type":52,"value":1486},"50, 25, 10",{"type":52,"value":1023},{"type":46,"tag":417,"props":1489,"children":1490},{},[1491,1496],{"type":46,"tag":421,"props":1492,"children":1493},{},[1494],{"type":52,"value":1495},"dataset.name",{"type":52,"value":1497},": Dataset name. Options: front3d, matterport, synthetic_hospital, synthetic_warehouse.",{"type":46,"tag":417,"props":1499,"children":1500},{},[1501,1506,1508,1514,1516,1522,1524,1530,1532,1537],{"type":46,"tag":421,"props":1502,"children":1503},{},[1504],{"type":52,"value":1505},"dataset.contiguous_id",{"type":52,"value":1507},": Set ",{"type":46,"tag":78,"props":1509,"children":1511},{"className":1510},[],[1512],{"type":52,"value":1513},"True",{"type":52,"value":1515}," when the label-map JSON already\nsupplies ",{"type":46,"tag":78,"props":1517,"children":1519},{"className":1518},[],[1520],{"type":52,"value":1521},"trainId",{"type":52,"value":1523}," values for its category IDs; leaving the default ",{"type":46,"tag":78,"props":1525,"children":1527},{"className":1526},[],[1528],{"type":52,"value":1529},"False",{"type":52,"value":1531},"\ncan synthesize placeholder categories without ",{"type":46,"tag":78,"props":1533,"children":1535},{"className":1534},[],[1536],{"type":52,"value":1521},{"type":52,"value":1538}," and fail during\nmetadata construction.",{"type":46,"tag":417,"props":1540,"children":1541},{},[1542,1547],{"type":46,"tag":421,"props":1543,"children":1544},{},[1545],{"type":52,"value":1546},"dataset.downsample_factor",{"type":52,"value":1548},": Image downsample factor. Default 1 (Front3D), 2 (Matterport).",{"type":46,"tag":417,"props":1550,"children":1551},{},[1552,1557,1559,1564],{"type":46,"tag":421,"props":1553,"children":1554},{},[1555],{"type":52,"value":1556},"dataset.target_size",{"type":52,"value":1558},": Target image size. Default ",{"type":46,"tag":1036,"props":1560,"children":1561},{},[1562],{"type":52,"value":1563},"320, 240",{"type":52,"value":1023},{"type":46,"tag":417,"props":1566,"children":1567},{},[1568,1573],{"type":46,"tag":421,"props":1569,"children":1570},{},[1571],{"type":52,"value":1572},"dataset.depth_min",{"type":52,"value":1574},": Min depth. Default 0.4 meters.",{"type":46,"tag":417,"props":1576,"children":1577},{},[1578,1583],{"type":46,"tag":421,"props":1579,"children":1580},{},[1581],{"type":52,"value":1582},"dataset.depth_max",{"type":52,"value":1584},": Max depth. Default 6.0 meters.",{"type":46,"tag":417,"props":1586,"children":1587},{},[1588,1593],{"type":46,"tag":421,"props":1589,"children":1590},{},[1591],{"type":52,"value":1592},"train.lr",{"type":52,"value":1594},": Learning rate. Default 2e-4. backbone_multiplier=0.1.",{"type":46,"tag":417,"props":1596,"children":1597},{},[1598,1603,1605,1610],{"type":46,"tag":421,"props":1599,"children":1600},{},[1601],{"type":52,"value":1602},"train.lr_scheduler",{"type":52,"value":1604},": Options: MultiStep, Warmuppoly. Milestones ",{"type":46,"tag":1036,"props":1606,"children":1607},{},[1608],{"type":52,"value":1609},"88, 96",{"type":52,"value":1023},{"type":46,"tag":417,"props":1612,"children":1613},{},[1614,1619,1621,1627],{"type":46,"tag":421,"props":1615,"children":1616},{},[1617],{"type":52,"value":1618},"train.precision",{"type":52,"value":1620},": Only ",{"type":46,"tag":78,"props":1622,"children":1624},{"className":1623},[],[1625],{"type":52,"value":1626},"fp32",{"type":52,"value":1628}," is supported by the current train code.",{"type":46,"tag":417,"props":1630,"children":1631},{},[1632,1637],{"type":46,"tag":421,"props":1633,"children":1634},{},[1635],{"type":52,"value":1636},"train.distributed_strategy",{"type":52,"value":1638},": Options: ddp, fsdp. activation_checkpoint=True by default.",{"type":46,"tag":417,"props":1640,"children":1641},{},[1642,1647],{"type":46,"tag":421,"props":1643,"children":1644},{},[1645],{"type":52,"value":1646},"train.clip_grad_norm",{"type":52,"value":1648},": Gradient clipping norm. Default 0.1.",{"type":46,"tag":417,"props":1650,"children":1651},{},[1652,1657],{"type":46,"tag":421,"props":1653,"children":1654},{},[1655],{"type":52,"value":1656},"export.onnx_file_2d",{"type":52,"value":1658},": ONNX path for 2D model component.",{"type":46,"tag":417,"props":1660,"children":1661},{},[1662,1667],{"type":46,"tag":421,"props":1663,"children":1664},{},[1665],{"type":52,"value":1666},"export.max_voxels",{"type":52,"value":1668},": Max voxels for engine input. Default 700000.",{"type":46,"tag":417,"props":1670,"children":1671},{},[1672,1677],{"type":46,"tag":421,"props":1673,"children":1674},{},[1675],{"type":52,"value":1676},"inference.mode",{"type":52,"value":1678},": Options: semantic, instance, panoptic.",{"type":46,"tag":66,"props":1680,"children":1682},{"id":1681},"multi-gpu-multi-node",[1683],{"type":52,"value":1684},"Multi-GPU \u002F Multi-Node",{"type":46,"tag":55,"props":1686,"children":1687},{},[1688,1693,1695,1700],{"type":46,"tag":421,"props":1689,"children":1690},{},[1691],{"type":52,"value":1692},"Launch method:",{"type":52,"value":1694}," Lightning-managed (single ",{"type":46,"tag":78,"props":1696,"children":1698},{"className":1697},[],[1699],{"type":52,"value":1029},{"type":52,"value":1701}," process, Lightning spawns workers).",{"type":46,"tag":511,"props":1703,"children":1704},{},[1705,1725],{"type":46,"tag":515,"props":1706,"children":1707},{},[1708],{"type":46,"tag":519,"props":1709,"children":1710},{},[1711,1715,1720],{"type":46,"tag":523,"props":1712,"children":1713},{},[1714],{"type":52,"value":532},{"type":46,"tag":523,"props":1716,"children":1717},{},[1718],{"type":52,"value":1719},"Description",{"type":46,"tag":523,"props":1721,"children":1722},{},[1723],{"type":52,"value":1724},"Default",{"type":46,"tag":549,"props":1726,"children":1727},{},[1728,1750,1775,1796],{"type":46,"tag":519,"props":1729,"children":1730},{},[1731,1740,1745],{"type":46,"tag":556,"props":1732,"children":1733},{},[1734],{"type":46,"tag":78,"props":1735,"children":1737},{"className":1736},[],[1738],{"type":52,"value":1739},"train.num_gpus",{"type":46,"tag":556,"props":1741,"children":1742},{},[1743],{"type":52,"value":1744},"Number of GPUs",{"type":46,"tag":556,"props":1746,"children":1747},{},[1748],{"type":52,"value":1749},"1",{"type":46,"tag":519,"props":1751,"children":1752},{},[1753,1762,1767],{"type":46,"tag":556,"props":1754,"children":1755},{},[1756],{"type":46,"tag":78,"props":1757,"children":1759},{"className":1758},[],[1760],{"type":52,"value":1761},"train.gpu_ids",{"type":46,"tag":556,"props":1763,"children":1764},{},[1765],{"type":52,"value":1766},"GPU device indices",{"type":46,"tag":556,"props":1768,"children":1769},{},[1770],{"type":46,"tag":1036,"props":1771,"children":1772},{},[1773],{"type":52,"value":1774},"0",{"type":46,"tag":519,"props":1776,"children":1777},{},[1778,1787,1792],{"type":46,"tag":556,"props":1779,"children":1780},{},[1781],{"type":46,"tag":78,"props":1782,"children":1784},{"className":1783},[],[1785],{"type":52,"value":1786},"train.num_nodes",{"type":46,"tag":556,"props":1788,"children":1789},{},[1790],{"type":52,"value":1791},"Number of nodes",{"type":46,"tag":556,"props":1793,"children":1794},{},[1795],{"type":52,"value":1749},{"type":46,"tag":519,"props":1797,"children":1798},{},[1799,1807,1818],{"type":46,"tag":556,"props":1800,"children":1801},{},[1802],{"type":46,"tag":78,"props":1803,"children":1805},{"className":1804},[],[1806],{"type":52,"value":1636},{"type":46,"tag":556,"props":1808,"children":1809},{},[1810,1816],{"type":46,"tag":78,"props":1811,"children":1813},{"className":1812},[],[1814],{"type":52,"value":1815},"ddp",{"type":52,"value":1817}," only",{"type":46,"tag":556,"props":1819,"children":1820},{},[1821],{"type":46,"tag":78,"props":1822,"children":1824},{"className":1823},[],[1825],{"type":52,"value":1815},{"type":46,"tag":413,"props":1827,"children":1828},{},[1829,1852,1868,1884],{"type":46,"tag":417,"props":1830,"children":1831},{},[1832,1843,1845,1850],{"type":46,"tag":421,"props":1833,"children":1834},{},[1835,1841],{"type":46,"tag":78,"props":1836,"children":1838},{"className":1837},[],[1839],{"type":52,"value":1840},"fsdp",{"type":52,"value":1842}," is NOT supported",{"type":52,"value":1844}," for NVPanoptix3D (code only handles ",{"type":46,"tag":78,"props":1846,"children":1848},{"className":1847},[],[1849],{"type":52,"value":1815},{"type":52,"value":1851},")",{"type":46,"tag":417,"props":1853,"children":1854},{},[1855,1860,1862],{"type":46,"tag":78,"props":1856,"children":1858},{"className":1857},[],[1859],{"type":52,"value":1815},{"type":52,"value":1861}," with activation checkpointing (enabled by default): ",{"type":46,"tag":78,"props":1863,"children":1865},{"className":1864},[],[1866],{"type":52,"value":1867},"find_unused_parameters=False",{"type":46,"tag":417,"props":1869,"children":1870},{},[1871,1876,1878],{"type":46,"tag":78,"props":1872,"children":1874},{"className":1873},[],[1875],{"type":52,"value":1815},{"type":52,"value":1877}," without: ",{"type":46,"tag":78,"props":1879,"children":1881},{"className":1880},[],[1882],{"type":52,"value":1883},"find_unused_parameters=True",{"type":46,"tag":417,"props":1885,"children":1886},{},[1887,1889],{"type":52,"value":1888},"FAN backbones with 3D enabled auto-enable ",{"type":46,"tag":78,"props":1890,"children":1892},{"className":1891},[],[1893],{"type":52,"value":1894},"sync_batchnorm",{"type":46,"tag":55,"props":1896,"children":1897},{},[1898,1903,1905,1911,1912,1918,1919,1925,1926,1932,1933,1939],{"type":46,"tag":421,"props":1899,"children":1900},{},[1901],{"type":52,"value":1902},"Multi-node env vars",{"type":52,"value":1904}," (set by orchestrator): ",{"type":46,"tag":78,"props":1906,"children":1908},{"className":1907},[],[1909],{"type":52,"value":1910},"WORLD_SIZE",{"type":52,"value":149},{"type":46,"tag":78,"props":1913,"children":1915},{"className":1914},[],[1916],{"type":52,"value":1917},"NODE_RANK",{"type":52,"value":149},{"type":46,"tag":78,"props":1920,"children":1922},{"className":1921},[],[1923],{"type":52,"value":1924},"MASTER_ADDR",{"type":52,"value":149},{"type":46,"tag":78,"props":1927,"children":1929},{"className":1928},[],[1930],{"type":52,"value":1931},"MASTER_PORT",{"type":52,"value":149},{"type":46,"tag":78,"props":1934,"children":1936},{"className":1935},[],[1937],{"type":52,"value":1938},"NUM_GPU_PER_NODE",{"type":52,"value":1023},{"type":46,"tag":66,"props":1941,"children":1943},{"id":1942},"export-trt-defaults",[1944],{"type":52,"value":1945},"Export \u002F TRT Defaults",{"type":46,"tag":413,"props":1947,"children":1948},{},[1949,1977,1982],{"type":46,"tag":417,"props":1950,"children":1951},{},[1952,1954,1959,1961,1967,1969,1975],{"type":52,"value":1953},"Exports the 2D ONNX model to ",{"type":46,"tag":78,"props":1955,"children":1957},{"className":1956},[],[1958],{"type":52,"value":1656},{"type":52,"value":1960},". The current export\nentrypoint calls ",{"type":46,"tag":78,"props":1962,"children":1964},{"className":1963},[],[1965],{"type":52,"value":1966},"export_2d_model",{"type":52,"value":1968},"; ",{"type":46,"tag":78,"props":1970,"children":1972},{"className":1971},[],[1973],{"type":52,"value":1974},"export.onnx_file_3d",{"type":52,"value":1976}," is present in the\nschema but not produced by this toolkit image.",{"type":46,"tag":417,"props":1978,"children":1979},{},[1980],{"type":52,"value":1981},"TRT data types: FP32, FP16 only",{"type":46,"tag":417,"props":1983,"children":1984},{},[1985],{"type":52,"value":1986},"max_voxels: 700000 (engine input tensor limit)",{"type":46,"tag":66,"props":1988,"children":1990},{"id":1989},"hardware",[1991],{"type":52,"value":1992},"Hardware",{"type":46,"tag":55,"props":1994,"children":1995},{},[1996,1998,2004],{"type":52,"value":1997},"Minimum 2 GPU(s), recommended 4 GPU(s). 40GB+ (A100 recommended) VRAM per GPU. 3D reconstruction is very memory intensive. Use ",{"type":46,"tag":78,"props":1999,"children":2001},{"className":2000},[],[2002],{"type":52,"value":2003},"train.precision: fp32",{"type":52,"value":2005},"; the current training entrypoint rejects fp16. activation_checkpoint enabled by default. FSDP for multi-node. AutoML is enabled at the model layer; preserve this GPU\u002FVRAM guidance when routing train through AutoML.",{"type":46,"tag":66,"props":2007,"children":2009},{"id":2008},"error-patterns",[2010],{"type":52,"value":2011},"Error Patterns",{"type":46,"tag":55,"props":2013,"children":2014},{},[2015,2026,2028,2034,2036,2041],{"type":46,"tag":421,"props":2016,"children":2017},{},[2018,2024],{"type":46,"tag":78,"props":2019,"children":2021},{"className":2020},[],[2022],{"type":52,"value":2023},"nvpanoptix3d: not found",{"type":52,"value":2025}," in the PyTorch image",{"type":52,"value":2027},": Use the packaged module\nentrypoint command:\n",{"type":46,"tag":78,"props":2029,"children":2031},{"className":2030},[],[2032],{"type":52,"value":2033},"python -m nvidia_tao_pytorch.cv.nvpanoptix3d.entrypoint.nvpanoptix3d \u003Caction> -e \u003Cspec>",{"type":52,"value":2035},".\nThe 7.0 PyTorch image contains the NVPanoptix3D package but does not expose a\n",{"type":46,"tag":78,"props":2037,"children":2039},{"className":2038},[],[2040],{"type":52,"value":49},{"type":52,"value":2042}," console script.",{"type":46,"tag":55,"props":2044,"children":2045},{},[2046,2051],{"type":46,"tag":421,"props":2047,"children":2048},{},[2049],{"type":52,"value":2050},"Missing frustum mask",{"type":52,"value":2052},": Ensure meta\u002Ffrustum_mask.npz is present in the dataset directory.",{"type":46,"tag":55,"props":2054,"children":2055},{},[2056,2061],{"type":46,"tag":421,"props":2057,"children":2058},{},[2059],{"type":52,"value":2060},"Downsample factor mismatch",{"type":52,"value":2062},": Use downsample_factor=2 for Matterport3D, 1 for Front3D \u002F synthetic datasets.",{"type":46,"tag":55,"props":2064,"children":2065},{},[2066,2071],{"type":46,"tag":421,"props":2067,"children":2068},{},[2069],{"type":52,"value":2070},"3D occupancy OOM",{"type":52,"value":2072},": Reduce frustum_dims or grid_dimensions if running out of GPU memory during 3D reconstruction.",{"type":46,"tag":55,"props":2074,"children":2075},{},[2076,2081,2083,2089,2091,2097,2099,2104],{"type":46,"tag":421,"props":2077,"children":2078},{},[2079],{"type":52,"value":2080},"fp16 precision rejected",{"type":52,"value":2082},": The schema advertises ",{"type":46,"tag":78,"props":2084,"children":2086},{"className":2085},[],[2087],{"type":52,"value":2088},"fp16",{"type":52,"value":2090},", but the current\ntraining entrypoint raises ",{"type":46,"tag":78,"props":2092,"children":2094},{"className":2093},[],[2095],{"type":52,"value":2096},"ValueError: Only fp32 precision is supported.",{"type":52,"value":2098}," Use\n",{"type":46,"tag":78,"props":2100,"children":2102},{"className":2101},[],[2103],{"type":52,"value":2003},{"type":52,"value":2105}," for train and resume\u002Fretrain.",{"type":46,"tag":55,"props":2107,"children":2108},{},[2109,2114,2116,2121,2123,2128,2129,2134],{"type":46,"tag":421,"props":2110,"children":2111},{},[2112],{"type":52,"value":2113},"Inference dataloader length is zero",{"type":52,"value":2115},": ",{"type":46,"tag":78,"props":2117,"children":2119},{"className":2118},[],[2120],{"type":52,"value":759},{"type":52,"value":2122}," is scanned only\nfor top-level ",{"type":46,"tag":78,"props":2124,"children":2126},{"className":2125},[],[2127],{"type":52,"value":774},{"type":52,"value":133},{"type":46,"tag":78,"props":2130,"children":2132},{"className":2131},[],[2133],{"type":52,"value":782},{"type":52,"value":2135}," files. If the S3 test archive extracts to\nscene subdirectories, create or point to a flat folder of real RGB images before\nrunning inference.",{"type":46,"tag":55,"props":2137,"children":2138},{},[2139,2144,2146,2151,2153,2158],{"type":46,"tag":421,"props":2140,"children":2141},{},[2142],{"type":52,"value":2143},"3D ONNX missing after export",{"type":52,"value":2145},": The current export entrypoint only calls the\n2D ONNX exporter and writes ",{"type":46,"tag":78,"props":2147,"children":2149},{"className":2148},[],[2150],{"type":52,"value":1656},{"type":52,"value":2152},". Do not require\n",{"type":46,"tag":78,"props":2154,"children":2156},{"className":2155},[],[2157],{"type":52,"value":1974},{"type":52,"value":2159}," unless the toolkit image adds a 3D exporter.",{"type":46,"tag":55,"props":2161,"children":2162},{},[2163,2168,2170,2176,2178,2184,2186,2191,2193,2199,2201,2206,2208,2214,2216,2222],{"type":46,"tag":421,"props":2164,"children":2165},{},[2166],{"type":52,"value":2167},"Resume stops at the epoch boundary",{"type":52,"value":2169},": A one-epoch smoke run writes an\nend-of-epoch checkpoint such as ",{"type":46,"tag":78,"props":2171,"children":2173},{"className":2172},[],[2174],{"type":52,"value":2175},"model_epoch_000_step_00020.pth",{"type":52,"value":2177},". Resuming with\n",{"type":46,"tag":78,"props":2179,"children":2181},{"className":2180},[],[2182],{"type":52,"value":2183},"train.num_epochs",{"type":52,"value":2185}," set only one epoch beyond the original run can restore the\ncheckpoint and stop without producing a new epoch checkpoint. When validating\nactual retraining from an epoch-boundary checkpoint, set ",{"type":46,"tag":78,"props":2187,"children":2189},{"className":2188},[],[2190],{"type":52,"value":2183},{"type":52,"value":2192}," at\nleast two epochs beyond the source smoke run and raise ",{"type":46,"tag":78,"props":2194,"children":2196},{"className":2195},[],[2197],{"type":52,"value":2198},"train.optim.max_steps",{"type":52,"value":2200},"\naccordingly. For example, resuming from ",{"type":46,"tag":78,"props":2202,"children":2204},{"className":2203},[],[2205],{"type":52,"value":2175},{"type":52,"value":2207}," needs\n",{"type":46,"tag":78,"props":2209,"children":2211},{"className":2210},[],[2212],{"type":52,"value":2213},"train.num_epochs: 3",{"type":52,"value":2215}," and enough max steps to produce a new exact epoch\u002Fstep\ncheckpoint such as ",{"type":46,"tag":78,"props":2217,"children":2219},{"className":2218},[],[2220],{"type":52,"value":2221},"model_epoch_001_step_00040.pth",{"type":52,"value":2223}," before handing the model to\nevaluate, inference, or export.",{"type":46,"tag":66,"props":2225,"children":2227},{"id":2226},"spec-param-parent-model-inference",[2228],{"type":52,"value":2229},"Spec Param \u002F Parent Model Inference",{"type":46,"tag":55,"props":2231,"children":2232},{},[2233,2235,2241,2243,2249,2251,2257],{"type":52,"value":2234},"Model-specific inference mappings belong in this MD file, not in ",{"type":46,"tag":78,"props":2236,"children":2238},{"className":2237},[],[2239],{"type":52,"value":2240},"config.json",{"type":52,"value":2242},". Generated runners should read this section and apply the mappings with SDK helpers before ",{"type":46,"tag":78,"props":2244,"children":2246},{"className":2245},[],[2247],{"type":52,"value":2248},"create_job()",{"type":52,"value":2250},". This mirrors the old microservices ",{"type":46,"tag":78,"props":2252,"children":2254},{"className":2253},[],[2255],{"type":52,"value":2256},"infer_params.py",{"type":52,"value":2258}," flow.",{"type":46,"tag":55,"props":2260,"children":2261},{},[2262],{"type":52,"value":2263},"Model-specific handoff mappings:",{"type":46,"tag":511,"props":2265,"children":2266},{},[2267,2292],{"type":46,"tag":515,"props":2268,"children":2269},{},[2270],{"type":46,"tag":519,"props":2271,"children":2272},{},[2273,2277,2282,2287],{"type":46,"tag":523,"props":2274,"children":2275},{},[2276],{"type":52,"value":527},{"type":46,"tag":523,"props":2278,"children":2279},{},[2280],{"type":52,"value":2281},"Spec Field",{"type":46,"tag":523,"props":2283,"children":2284},{},[2285],{"type":52,"value":2286},"Inference Function",{"type":46,"tag":523,"props":2288,"children":2289},{},[2290],{"type":52,"value":2291},"Meaning",{"type":46,"tag":549,"props":2293,"children":2294},{},[2295,2325,2355,2385,2412,2440,2469,2496,2523,2551,2578,2605,2632,2662,2692],{"type":46,"tag":519,"props":2296,"children":2297},{},[2298,2302,2311,2320],{"type":46,"tag":556,"props":2299,"children":2300},{},[2301],{"type":52,"value":382},{"type":46,"tag":556,"props":2303,"children":2304},{},[2305],{"type":46,"tag":78,"props":2306,"children":2308},{"className":2307},[],[2309],{"type":52,"value":2310},"encryption_key",{"type":46,"tag":556,"props":2312,"children":2313},{},[2314],{"type":46,"tag":78,"props":2315,"children":2317},{"className":2316},[],[2318],{"type":52,"value":2319},"key",{"type":46,"tag":556,"props":2321,"children":2322},{},[2323],{"type":52,"value":2324},"encryption key",{"type":46,"tag":519,"props":2326,"children":2327},{},[2328,2332,2341,2350],{"type":46,"tag":556,"props":2329,"children":2330},{},[2331],{"type":52,"value":382},{"type":46,"tag":556,"props":2333,"children":2334},{},[2335],{"type":46,"tag":78,"props":2336,"children":2338},{"className":2337},[],[2339],{"type":52,"value":2340},"evaluate.checkpoint",{"type":46,"tag":556,"props":2342,"children":2343},{},[2344],{"type":46,"tag":78,"props":2345,"children":2347},{"className":2346},[],[2348],{"type":52,"value":2349},"parent_model",{"type":46,"tag":556,"props":2351,"children":2352},{},[2353],{"type":52,"value":2354},"model file inferred from the parent job results folder",{"type":46,"tag":519,"props":2356,"children":2357},{},[2358,2362,2371,2380],{"type":46,"tag":556,"props":2359,"children":2360},{},[2361],{"type":52,"value":382},{"type":46,"tag":556,"props":2363,"children":2364},{},[2365],{"type":46,"tag":78,"props":2366,"children":2368},{"className":2367},[],[2369],{"type":52,"value":2370},"results_dir",{"type":46,"tag":556,"props":2372,"children":2373},{},[2374],{"type":46,"tag":78,"props":2375,"children":2377},{"className":2376},[],[2378],{"type":52,"value":2379},"output_dir",{"type":46,"tag":556,"props":2381,"children":2382},{},[2383],{"type":52,"value":2384},"current job results directory",{"type":46,"tag":519,"props":2386,"children":2387},{},[2388,2392,2400,2408],{"type":46,"tag":556,"props":2389,"children":2390},{},[2391],{"type":52,"value":396},{"type":46,"tag":556,"props":2393,"children":2394},{},[2395],{"type":46,"tag":78,"props":2396,"children":2398},{"className":2397},[],[2399],{"type":52,"value":2310},{"type":46,"tag":556,"props":2401,"children":2402},{},[2403],{"type":46,"tag":78,"props":2404,"children":2406},{"className":2405},[],[2407],{"type":52,"value":2319},{"type":46,"tag":556,"props":2409,"children":2410},{},[2411],{"type":52,"value":2324},{"type":46,"tag":519,"props":2413,"children":2414},{},[2415,2419,2428,2436],{"type":46,"tag":556,"props":2416,"children":2417},{},[2418],{"type":52,"value":396},{"type":46,"tag":556,"props":2420,"children":2421},{},[2422],{"type":46,"tag":78,"props":2423,"children":2425},{"className":2424},[],[2426],{"type":52,"value":2427},"export.checkpoint",{"type":46,"tag":556,"props":2429,"children":2430},{},[2431],{"type":46,"tag":78,"props":2432,"children":2434},{"className":2433},[],[2435],{"type":52,"value":2349},{"type":46,"tag":556,"props":2437,"children":2438},{},[2439],{"type":52,"value":2354},{"type":46,"tag":519,"props":2441,"children":2442},{},[2443,2447,2455,2464],{"type":46,"tag":556,"props":2444,"children":2445},{},[2446],{"type":52,"value":396},{"type":46,"tag":556,"props":2448,"children":2449},{},[2450],{"type":46,"tag":78,"props":2451,"children":2453},{"className":2452},[],[2454],{"type":52,"value":1656},{"type":46,"tag":556,"props":2456,"children":2457},{},[2458],{"type":46,"tag":78,"props":2459,"children":2461},{"className":2460},[],[2462],{"type":52,"value":2463},"create_onnx_file_2d",{"type":46,"tag":556,"props":2465,"children":2466},{},[2467],{"type":52,"value":2468},"output 2D ONNX path",{"type":46,"tag":519,"props":2470,"children":2471},{},[2472,2476,2484,2492],{"type":46,"tag":556,"props":2473,"children":2474},{},[2475],{"type":52,"value":396},{"type":46,"tag":556,"props":2477,"children":2478},{},[2479],{"type":46,"tag":78,"props":2480,"children":2482},{"className":2481},[],[2483],{"type":52,"value":2370},{"type":46,"tag":556,"props":2485,"children":2486},{},[2487],{"type":46,"tag":78,"props":2488,"children":2490},{"className":2489},[],[2491],{"type":52,"value":2379},{"type":46,"tag":556,"props":2493,"children":2494},{},[2495],{"type":52,"value":2384},{"type":46,"tag":519,"props":2497,"children":2498},{},[2499,2503,2511,2519],{"type":46,"tag":556,"props":2500,"children":2501},{},[2502],{"type":52,"value":389},{"type":46,"tag":556,"props":2504,"children":2505},{},[2506],{"type":46,"tag":78,"props":2507,"children":2509},{"className":2508},[],[2510],{"type":52,"value":2310},{"type":46,"tag":556,"props":2512,"children":2513},{},[2514],{"type":46,"tag":78,"props":2515,"children":2517},{"className":2516},[],[2518],{"type":52,"value":2319},{"type":46,"tag":556,"props":2520,"children":2521},{},[2522],{"type":52,"value":2324},{"type":46,"tag":519,"props":2524,"children":2525},{},[2526,2530,2539,2547],{"type":46,"tag":556,"props":2527,"children":2528},{},[2529],{"type":52,"value":389},{"type":46,"tag":556,"props":2531,"children":2532},{},[2533],{"type":46,"tag":78,"props":2534,"children":2536},{"className":2535},[],[2537],{"type":52,"value":2538},"inference.checkpoint",{"type":46,"tag":556,"props":2540,"children":2541},{},[2542],{"type":46,"tag":78,"props":2543,"children":2545},{"className":2544},[],[2546],{"type":52,"value":2349},{"type":46,"tag":556,"props":2548,"children":2549},{},[2550],{"type":52,"value":2354},{"type":46,"tag":519,"props":2552,"children":2553},{},[2554,2558,2566,2574],{"type":46,"tag":556,"props":2555,"children":2556},{},[2557],{"type":52,"value":389},{"type":46,"tag":556,"props":2559,"children":2560},{},[2561],{"type":46,"tag":78,"props":2562,"children":2564},{"className":2563},[],[2565],{"type":52,"value":2370},{"type":46,"tag":556,"props":2567,"children":2568},{},[2569],{"type":46,"tag":78,"props":2570,"children":2572},{"className":2571},[],[2573],{"type":52,"value":2379},{"type":46,"tag":556,"props":2575,"children":2576},{},[2577],{"type":52,"value":2384},{"type":46,"tag":519,"props":2579,"children":2580},{},[2581,2585,2593,2601],{"type":46,"tag":556,"props":2582,"children":2583},{},[2584],{"type":52,"value":796},{"type":46,"tag":556,"props":2586,"children":2587},{},[2588],{"type":46,"tag":78,"props":2589,"children":2591},{"className":2590},[],[2592],{"type":52,"value":2310},{"type":46,"tag":556,"props":2594,"children":2595},{},[2596],{"type":46,"tag":78,"props":2597,"children":2599},{"className":2598},[],[2600],{"type":52,"value":2319},{"type":46,"tag":556,"props":2602,"children":2603},{},[2604],{"type":52,"value":2324},{"type":46,"tag":519,"props":2606,"children":2607},{},[2608,2612,2620,2628],{"type":46,"tag":556,"props":2609,"children":2610},{},[2611],{"type":52,"value":796},{"type":46,"tag":556,"props":2613,"children":2614},{},[2615],{"type":46,"tag":78,"props":2616,"children":2618},{"className":2617},[],[2619],{"type":52,"value":2370},{"type":46,"tag":556,"props":2621,"children":2622},{},[2623],{"type":46,"tag":78,"props":2624,"children":2626},{"className":2625},[],[2627],{"type":52,"value":2379},{"type":46,"tag":556,"props":2629,"children":2630},{},[2631],{"type":52,"value":2384},{"type":46,"tag":519,"props":2633,"children":2634},{},[2635,2639,2648,2657],{"type":46,"tag":556,"props":2636,"children":2637},{},[2638],{"type":52,"value":796},{"type":46,"tag":556,"props":2640,"children":2641},{},[2642],{"type":46,"tag":78,"props":2643,"children":2645},{"className":2644},[],[2646],{"type":52,"value":2647},"train.checkpoint_2d",{"type":46,"tag":556,"props":2649,"children":2650},{},[2651],{"type":46,"tag":78,"props":2652,"children":2654},{"className":2653},[],[2655],{"type":52,"value":2656},"parent_model_or_ptm",{"type":46,"tag":556,"props":2658,"children":2659},{},[2660],{"type":52,"value":2661},"parent model if available, otherwise PTM",{"type":46,"tag":519,"props":2663,"children":2664},{},[2665,2669,2678,2687],{"type":46,"tag":556,"props":2666,"children":2667},{},[2668],{"type":52,"value":796},{"type":46,"tag":556,"props":2670,"children":2671},{},[2672],{"type":46,"tag":78,"props":2673,"children":2675},{"className":2674},[],[2676],{"type":52,"value":2677},"train.checkpoint_3d",{"type":46,"tag":556,"props":2679,"children":2680},{},[2681],{"type":46,"tag":78,"props":2682,"children":2684},{"className":2683},[],[2685],{"type":52,"value":2686},"ptm",{"type":46,"tag":556,"props":2688,"children":2689},{},[2690],{"type":52,"value":2691},"pretrained model",{"type":46,"tag":519,"props":2693,"children":2694},{},[2695,2699,2708,2717],{"type":46,"tag":556,"props":2696,"children":2697},{},[2698],{"type":52,"value":796},{"type":46,"tag":556,"props":2700,"children":2701},{},[2702],{"type":46,"tag":78,"props":2703,"children":2705},{"className":2704},[],[2706],{"type":52,"value":2707},"train.resume_training_checkpoint_path",{"type":46,"tag":556,"props":2709,"children":2710},{},[2711],{"type":46,"tag":78,"props":2712,"children":2714},{"className":2713},[],[2715],{"type":52,"value":2716},"resume_model",{"type":46,"tag":556,"props":2718,"children":2719},{},[2720],{"type":52,"value":2721},"model file inferred from the current job results folder",{"type":46,"tag":55,"props":2723,"children":2724},{},[2725,2727,2732,2734,2740,2742,2748,2750,2755],{"type":52,"value":2726},"For ",{"type":46,"tag":78,"props":2728,"children":2730},{"className":2729},[],[2731],{"type":52,"value":2349},{"type":52,"value":2733}," or ",{"type":46,"tag":78,"props":2735,"children":2737},{"className":2736},[],[2738],{"type":52,"value":2739},"parent_model_folder",{"type":52,"value":2741},", pass the upstream train\u002Fexport\u002FAutoML child job id as ",{"type":46,"tag":78,"props":2743,"children":2745},{"className":2744},[],[2746],{"type":52,"value":2747},"parent_job_id",{"type":52,"value":2749},". The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to ",{"type":46,"tag":78,"props":2751,"children":2753},{"className":2752},[],[2754],{"type":52,"value":2240},{"type":52,"value":2756}," and do not patch generated runner scripts to guess checkpoint paths.",{"type":46,"tag":2758,"props":2759,"children":2760},"style",{},[2761],{"type":52,"value":2762},"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":2764,"total":2920},[2765,2783,2800,2811,2823,2837,2850,2862,2875,2886,2900,2909],{"slug":2766,"name":2766,"fn":2767,"description":2768,"org":2769,"tags":2770,"stars":2780,"repoUrl":2781,"updatedAt":2782},"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},[2771,2774,2777],{"name":2772,"slug":2773,"type":15},"Documentation","documentation",{"name":2775,"slug":2776,"type":15},"MCP","mcp",{"name":2778,"slug":2779,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":2784,"name":2784,"fn":2785,"description":2786,"org":2787,"tags":2788,"stars":2797,"repoUrl":2798,"updatedAt":2799},"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},[2789,2792,2795],{"name":2790,"slug":2791,"type":15},"Containers","containers",{"name":2793,"slug":2794,"type":15},"Deployment","deployment",{"name":2796,"slug":1029,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":2801,"name":2801,"fn":2802,"description":2803,"org":2804,"tags":2805,"stars":2797,"repoUrl":2798,"updatedAt":2810},"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},[2806,2809],{"name":2807,"slug":2808,"type":15},"CI\u002FCD","ci-cd",{"name":2793,"slug":2794,"type":15},"2026-07-14T05:25:59.97109",{"slug":2812,"name":2812,"fn":2813,"description":2814,"org":2815,"tags":2816,"stars":2797,"repoUrl":2798,"updatedAt":2822},"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},[2817,2818,2819],{"name":2807,"slug":2808,"type":15},{"name":2793,"slug":2794,"type":15},{"name":2820,"slug":2821,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":2824,"name":2824,"fn":2825,"description":2826,"org":2827,"tags":2828,"stars":2797,"repoUrl":2798,"updatedAt":2836},"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},[2829,2832,2833],{"name":2830,"slug":2831,"type":15},"Debugging","debugging",{"name":2820,"slug":2821,"type":15},{"name":2834,"slug":2835,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":2838,"name":2838,"fn":2839,"description":2840,"org":2841,"tags":2842,"stars":2797,"repoUrl":2798,"updatedAt":2849},"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},[2843,2846],{"name":2844,"slug":2845,"type":15},"Best Practices","best-practices",{"name":2847,"slug":2848,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":2851,"name":2851,"fn":2852,"description":2853,"org":2854,"tags":2855,"stars":2797,"repoUrl":2798,"updatedAt":2861},"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},[2856,2857,2860],{"name":13,"slug":14,"type":15},{"name":2858,"slug":2859,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":2863,"name":2863,"fn":2864,"description":2865,"org":2866,"tags":2867,"stars":2797,"repoUrl":2798,"updatedAt":2874},"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},[2868,2871],{"name":2869,"slug":2870,"type":15},"QA","qa",{"name":2872,"slug":2873,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":2876,"name":2876,"fn":2877,"description":2878,"org":2879,"tags":2880,"stars":2797,"repoUrl":2798,"updatedAt":2885},"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},[2881,2882],{"name":2793,"slug":2794,"type":15},{"name":2883,"slug":2884,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":2887,"name":2887,"fn":2888,"description":2889,"org":2890,"tags":2891,"stars":2797,"repoUrl":2798,"updatedAt":2899},"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},[2892,2895,2896],{"name":2893,"slug":2894,"type":15},"Code Review","code-review",{"name":2820,"slug":2821,"type":15},{"name":2897,"slug":2898,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":2901,"name":2901,"fn":2902,"description":2903,"org":2904,"tags":2905,"stars":2797,"repoUrl":2798,"updatedAt":2908},"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},[2906,2907],{"name":2869,"slug":2870,"type":15},{"name":2872,"slug":2873,"type":15},"2026-07-14T05:25:54.928983",{"slug":2910,"name":2910,"fn":2911,"description":2912,"org":2913,"tags":2914,"stars":2797,"repoUrl":2798,"updatedAt":2919},"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},[2915,2918],{"name":2916,"slug":2917,"type":15},"Automation","automation",{"name":2807,"slug":2808,"type":15},"2026-07-30T05:29:03.275638",496,{"items":2922,"total":3018},[2923,2940,2950,2964,2974,2989,3004],{"slug":2924,"name":2924,"fn":2925,"description":2926,"org":2927,"tags":2928,"stars":20,"repoUrl":21,"updatedAt":2939},"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},[2929,2932,2935,2936],{"name":2930,"slug":2931,"type":15},"Data Analysis","data-analysis",{"name":2933,"slug":2934,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":2937,"slug":2938,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":2941,"name":2941,"fn":2942,"description":2943,"org":2944,"tags":2945,"stars":20,"repoUrl":21,"updatedAt":2949},"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},[2946,2947,2948],{"name":2793,"slug":2794,"type":15},{"name":2883,"slug":2884,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":2951,"name":2951,"fn":2952,"description":2953,"org":2954,"tags":2955,"stars":20,"repoUrl":21,"updatedAt":2963},"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},[2956,2959,2960],{"name":2957,"slug":2958,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":2961,"slug":2962,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":2965,"name":2965,"fn":2966,"description":2967,"org":2968,"tags":2969,"stars":20,"repoUrl":21,"updatedAt":2973},"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},[2970,2971,2972],{"name":2930,"slug":2931,"type":15},{"name":9,"slug":8,"type":15},{"name":2872,"slug":2873,"type":15},"2026-07-17T05:29:03.913266",{"slug":2975,"name":2975,"fn":2976,"description":2977,"org":2978,"tags":2979,"stars":20,"repoUrl":21,"updatedAt":2988},"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},[2980,2981,2984,2985],{"name":2916,"slug":2917,"type":15},{"name":2982,"slug":2983,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":2986,"slug":2987,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":2990,"name":2990,"fn":2991,"description":2992,"org":2993,"tags":2994,"stars":20,"repoUrl":21,"updatedAt":3003},"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},[2995,2996,2999,3000],{"name":2793,"slug":2794,"type":15},{"name":2997,"slug":2998,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":3001,"slug":3002,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":3005,"name":3005,"fn":3006,"description":3007,"org":3008,"tags":3009,"stars":20,"repoUrl":21,"updatedAt":3017},"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},[3010,3011,3014],{"name":9,"slug":8,"type":15},{"name":3012,"slug":3013,"type":15},"Quantum Computing","quantum-computing",{"name":3015,"slug":3016,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305]