[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-train-pointpillars":3,"mdc--ozmxpd-key":34,"related-repo-nvidia-tao-train-pointpillars":2923,"related-org-nvidia-tao-train-pointpillars":3027},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"tao-train-pointpillars","detect 3D objects with PointPillars","PointPillars for 3D object detection from LiDAR point clouds. Encodes point clouds into a pseudo-image via a pillar-based representation, then applies 2D detection — used in autonomous driving and robotics. Use when training, evaluating, exporting, pruning, retraining, or running inference for a TAO PointPillars model. Trigger phrases include \"train PointPillars\", \"LiDAR 3D detection\", \"point-cloud object detection\", \"pillar-based 3D detector\".",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,20],{"name":13,"slug":14,"type":15},"Deep Learning","deep-learning","tag",{"name":17,"slug":18,"type":15},"Machine Learning","machine-learning",{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},"Computer Vision","computer-vision",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:27:29.322221","Apache-2.0",281,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftao-train-pointpillars","---\nname: tao-train-pointpillars\ndescription: PointPillars for 3D object detection from LiDAR point clouds. Encodes point clouds into a pseudo-image via a\n  pillar-based representation, then applies 2D detection — used in autonomous driving and robotics. Use when training,\n  evaluating, exporting, pruning, retraining, or running inference for a TAO PointPillars model. Trigger phrases include\n  \"train PointPillars\", \"LiDAR 3D detection\", \"point-cloud object detection\", \"pillar-based 3D detector\".\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- point\n- cloud\n- 3d\n- detection\n---\n\n# PointPillars\n\nPointPillars for 3D object detection from LiDAR point clouds. Encodes point clouds into a pseudo-image via pillar-based representation, then applies 2D detection. Used in autonomous driving \u002F robotics.\n\nTypically trained from scratch. Provide train.resume_training_checkpoint_path to resume.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-pointpillars.md` first. Deploy spec templates live in this skill's `references\u002F` folder with the `spec_template_deploy_*.yaml` prefix.\n\nThe packaged PyTorch PointPillars CLI supports `dataset_convert`, `train`, `evaluate`, `inference`, `export`, and `prune`. It does not expose a parent-model `gen_trt_engine` action; TensorRT engine generation is deploy-only. It also does not expose a separate `retrain` subcommand. Retraining from a pruned model uses `pointpillars train -e ...` with `train.pruned_model_path` populated.\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\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:** pointpillars\n- **Formats:** default\n- **Monitoring metric:** loss\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| dataset_convert | dataset.data_path | id |  | No |\n| evaluate | dataset.data_path | train_datasets |  | No |\n| evaluate | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | No |\n| export | dataset.data_path | train_datasets |  | No |\n| export | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | No |\n| inference | dataset.data_path | train_datasets |  | No |\n| inference | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | No |\n| prune | dataset.data_path | train_datasets |  | No |\n| prune | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | No |\n| retrain | dataset.data_path | train_datasets |  | No |\n| retrain | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | No |\n| train | dataset.data_path | train_datasets |  | No |\n| train | dataset.data_info_path | train_datasets | \u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F | 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`.\n\n```python\nDATA_ROOT = \"s3:\u002F\u002Fbucket\u002Fdata\u002Fpointpillars\"\nDATA_INFO = \"\u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\"\nCHECKPOINT = \"\u002Fresults\u002F{train_job_id}\u002Fresults_dir\u002Fcheckpoint_epoch_1.pth\"\nPRUNED_MODEL = \"\u002Fresults\u002F{prune_job_id}\u002Fresults_dir\u002Fpruned_0.1.tlt\"\n```\n\nThe raw PointPillars data root must be an extracted folder containing matching `train\u002Flidar`, `train\u002Flabel`, `val\u002Flidar`, and `val\u002Flabel` subfolders before `dataset_convert` runs. If the source dataset is packaged as separate train\u002Fval archives, extract both under the same mounted data root and point `dataset.data_path` at that root.\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_epochs\": 30,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n}\n```\n\n**resume train (mandatory checkpoint):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"train.resume_training_checkpoint_path\": CHECKPOINT,\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"evaluate.checkpoint\": CHECKPOINT,\n}\n```\n\n**export (mandatory data sources):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"export.checkpoint\": CHECKPOINT,\n    \"export.onnx_file\": \"\u002Fresults\u002F{export_job_id}\u002Fresults_dir\u002Fpointpillars.onnx\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"inference.checkpoint\": CHECKPOINT,\n}\n```\n\n**prune (mandatory data sources):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"prune.model\": CHECKPOINT,\n}\n```\n\n**retrain (mandatory data sources):**\n```python\n{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"train.pruned_model_path\": PRUNED_MODEL,\n}\n```\n\nFor local Docker, `DATA_INFO` must be visible inside every train\u002Fevaluate\u002Fexport\u002Fprune\u002Fretrain container. Use the dataset_convert job from the same results root, or mount\u002Fcopy the converted `results_dir\u002Fdata_info` folder into the current run and set `dataset.data_info_path` to that mounted container path. If the host scratch root is mounted at `\u002Fresults` and the conversion artifacts live under host `scratch\u002Fresults\u002F\u003Cjob_id>\u002Fresults_dir\u002Fdata_info`, the direct-job container path is `\u002Fresults\u002Fresults\u002F\u003Cjob_id>\u002Fresults_dir\u002Fdata_info`. Do not reuse a `\u002Fresults\u002F\u003Cjob_id>\u002F...` path from another run root unless that folder is mounted into the current job.\n\nFor AutoML train workflows, perform this as a launch preflight before calling `AutoMLRunner.run`: create or materialize the `dataset_convert` output under the current run's `RESULTS_ROOT`, set `dataset.data_info_path` to that current-run container path, and verify `dbinfos_train.pkl`, `infos_train.pkl`, and `infos_val.pkl` are present from the train container's point of view. If a runner is cloned or adapted from a prior AutoML algorithm, update the conversion artifact in the new run root; a stale `CONVERT_JOB_ID` from another results mount is not valid.\n## Eval Dataset\n\nOptional. Validation data (val.tar.gz) is separate from training. Used for mAP evaluation.\n\n## Important Parameters\n\n- **train.num_epochs**: Default 80 (much higher than other TAO models). PointPillars needs more epochs for convergence on 3D detection.\n- **train.lr**: Learning rate. Default 0.003 (adam_onecycle scheduler).\n- **dataset.class_names**: List of 3D object classes. Default 7 classes (KITTI-style). Modify to match your dataset.\n- **dataset.data_path**: Path to point cloud data directory.\n- **dataset.data_info_path**: Path to data info files from dataset_convert step.\n- **dataset.point_cloud_range**: Spatial extent of the point cloud to consider. Must match your sensor configuration.\n- **model.dense_head.anchor_generator_config**: Anchor configurations per class. Must be tuned for your object sizes and the point cloud range.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** `torchrun` (LIGHTNING_EXCLUDED_NETWORK). Uses PyTorch native `DistributedDataParallel` (NOT Lightning Trainer).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs per node | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.num_nodes` | Number of nodes | 1 |\n\n- `CUDA_VISIBLE_DEVICES` is explicitly set from `TAO_VISIBLE_DEVICES`\n- Uses `nn.parallel.DistributedDataParallel` directly (not Lightning strategy)\n- `NODE_RANK` is copied to `RANK` if `RANK` is unset\n\n**Multi-node env vars** (set by orchestrator):\n\n| Variable | Purpose |\n|----------|---------|\n| `WORLD_SIZE` | Number of nodes |\n| `NODE_RANK` | This node's rank |\n| `MASTER_ADDR` | Rank-0 node IP |\n| `MASTER_PORT` | Rank-0 port (default 29500) |\n| `NUM_GPU_PER_NODE` | GPUs per node |\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 4 GPU(s). 16GB+ (V100 or A100) VRAM per GPU. PointPillars is relatively efficient for 3D detection. The main bottleneck is data I\u002FO for large point cloud datasets.\n\n## Error Patterns\n\n**dataset_convert required**: Training will fail if `dataset.data_info_path` is not populated from a prior `dataset_convert` job. Always run convert first, and verify the train container can see `dbinfos_train.pkl` and `infos_train.pkl` under `dataset.data_info_path`. A common local-Docker failure is a stale `\u002Fresults\u002F\u003Cold_job_id>\u002F...` path from a different results root.\n\n**Point cloud range mismatch**: If point_cloud_range does not match the actual sensor data extent, detections will be poor or empty.\n\n**Epoch numbering**: PointPillars checkpoint epoch numbers may be offset by 1 from status.json reported epochs.\n\n**Checkpoint selection**: PointPillars training emits checkpoints named like `checkpoint_epoch_1.pth`. For evaluation, inference, export, prune, and resume, select the intended checkpoint through the model\u002Fjob checkpoint resolver and pass that exact file to `evaluate.checkpoint`, `inference.checkpoint`, `export.checkpoint`, `prune.model`, or `train.resume_training_checkpoint_path`. Do not guess by taking the newest `model.pth`; this model does not use that filename.\n\n**Prune\u002Fretrain key**: PointPillars prune writes an encrypted `.tlt` artifact. Keep a non-empty `key` in the prune and retrain specs; the packaged templates use the TAO default `tlt_encode`. If `key` is omitted or `null`, the toolkit can still exit with a container success code while logging a passphrase error and creating an empty `pruned_0.1.tlt`. Always verify the pruned model is nonzero before using it for retrain.\n\n**Status files matter**: Some PointPillars failures can be followed by `Execution status: PASS` in the entrypoint footer and a Docker exit code of 0. Check `results_dir\u002Fstatus.json` and the expected artifact before marking an action as passed.\n\n**Local results_dir wiring**: For direct local-Docker specs, set the top-level `results_dir` as well as any action-specific `*.results_dir` field. If only `evaluate.results_dir` is set and the top-level field is left blank, evaluate can try to write under `\u002Fopt\u002Fnvidia\u002Feval` and then still print the generic PASS footer. Treat that as a failed action unless the expected result directory and status\u002Fartifact files exist.\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\nInference mappings from TAO Core `pointpillars.config.json`:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| dataset_convert | `results_dir` | `output_dir` | current job results directory |\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `key` | `key` | encryption key |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `export.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| export | `export.onnx_file` | `create_onnx_file` | output ONNX path |\n| export | `export.save_engine` | `create_engine_file` | output TensorRT engine path |\n| export | `key` | `key` | encryption key |\n| export | `results_dir` | `output_dir` | current job results directory |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `inference.trt_engine` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `key` | `key` | encryption key |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| prune | `key` | `key` | encryption key |\n| prune | `prune.model` | `parent_model` | model file inferred from the parent job results folder |\n| prune | `results_dir` | `output_dir` | current job results directory |\n| retrain | `key` | `key` | encryption key |\n| retrain | `results_dir` | `output_dir` | current job results directory |\n| retrain | `train.pruned_model_path` | `parent_model` | model file inferred from the parent job results folder |\n| train | `key` | `key` | encryption key |\n| train | `model.pretrained_model_path` | `ptm_if_no_resume_model` | PTM when no resume checkpoint exists |\n| train | `results_dir` | `output_dir` | current job results directory |\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\n## Deployment\n\n- [tao-deploy-pointpillars](references\u002Ftao-deploy-pointpillars.md)\n",{"data":35,"body":46},{"name":4,"description":6,"license":26,"compatibility":36,"metadata":37,"allowed-tools":40,"tags":41},"Requires docker + nvidia-container-toolkit.",{"version":38,"author":39},"0.1.0","NVIDIA Corporation","Read Bash",[42,43,44,45],"point","cloud","3d","detection",{"type":47,"children":48},"root",[49,58,64,69,123,202,209,301,307,416,447,453,489,496,837,843,863,912,960,968,1043,1051,1094,1102,1145,1153,1204,1212,1255,1263,1306,1314,1357,1417,1482,1488,1493,1499,1570,1576,1602,1696,1755,1765,1870,1876,1881,1887,1939,1949,1959,2022,2079,2105,2147,2153,2182,2195,2864,2899,2905,2917],{"type":50,"tag":51,"props":52,"children":54},"element","h1",{"id":53},"pointpillars",[55],{"type":56,"value":57},"text","PointPillars",{"type":50,"tag":59,"props":60,"children":61},"p",{},[62],{"type":56,"value":63},"PointPillars for 3D object detection from LiDAR point clouds. Encodes point clouds into a pseudo-image via pillar-based representation, then applies 2D detection. Used in autonomous driving \u002F robotics.",{"type":50,"tag":59,"props":65,"children":66},{},[67],{"type":56,"value":68},"Typically trained from scratch. Provide train.resume_training_checkpoint_path to resume.",{"type":50,"tag":59,"props":70,"children":71},{},[72,74,81,83,89,91,97,99,105,107,113,115,121],{"type":56,"value":73},"For TAO Deploy TensorRT actions (",{"type":50,"tag":75,"props":76,"children":78},"code",{"className":77},[],[79],{"type":56,"value":80},"gen_trt_engine",{"type":56,"value":82},", TensorRT ",{"type":50,"tag":75,"props":84,"children":86},{"className":85},[],[87],{"type":56,"value":88},"evaluate",{"type":56,"value":90},", and TensorRT ",{"type":50,"tag":75,"props":92,"children":94},{"className":93},[],[95],{"type":56,"value":96},"inference",{"type":56,"value":98},"), read ",{"type":50,"tag":75,"props":100,"children":102},{"className":101},[],[103],{"type":56,"value":104},"references\u002Ftao-deploy-pointpillars.md",{"type":56,"value":106}," first. Deploy spec templates live in this skill's ",{"type":50,"tag":75,"props":108,"children":110},{"className":109},[],[111],{"type":56,"value":112},"references\u002F",{"type":56,"value":114}," folder with the ",{"type":50,"tag":75,"props":116,"children":118},{"className":117},[],[119],{"type":56,"value":120},"spec_template_deploy_*.yaml",{"type":56,"value":122}," prefix.",{"type":50,"tag":59,"props":124,"children":125},{},[126,128,134,136,142,143,148,149,154,155,161,163,169,171,176,178,184,186,192,194,200],{"type":56,"value":127},"The packaged PyTorch PointPillars CLI supports ",{"type":50,"tag":75,"props":129,"children":131},{"className":130},[],[132],{"type":56,"value":133},"dataset_convert",{"type":56,"value":135},", ",{"type":50,"tag":75,"props":137,"children":139},{"className":138},[],[140],{"type":56,"value":141},"train",{"type":56,"value":135},{"type":50,"tag":75,"props":144,"children":146},{"className":145},[],[147],{"type":56,"value":88},{"type":56,"value":135},{"type":50,"tag":75,"props":150,"children":152},{"className":151},[],[153],{"type":56,"value":96},{"type":56,"value":135},{"type":50,"tag":75,"props":156,"children":158},{"className":157},[],[159],{"type":56,"value":160},"export",{"type":56,"value":162},", and ",{"type":50,"tag":75,"props":164,"children":166},{"className":165},[],[167],{"type":56,"value":168},"prune",{"type":56,"value":170},". It does not expose a parent-model ",{"type":50,"tag":75,"props":172,"children":174},{"className":173},[],[175],{"type":56,"value":80},{"type":56,"value":177}," action; TensorRT engine generation is deploy-only. It also does not expose a separate ",{"type":50,"tag":75,"props":179,"children":181},{"className":180},[],[182],{"type":56,"value":183},"retrain",{"type":56,"value":185}," subcommand. Retraining from a pruned model uses ",{"type":50,"tag":75,"props":187,"children":189},{"className":188},[],[190],{"type":56,"value":191},"pointpillars train -e ...",{"type":56,"value":193}," with ",{"type":50,"tag":75,"props":195,"children":197},{"className":196},[],[198],{"type":56,"value":199},"train.pruned_model_path",{"type":56,"value":201}," populated.",{"type":50,"tag":203,"props":204,"children":206},"h2",{"id":205},"dataclass-schemas",[207],{"type":56,"value":208},"Dataclass Schemas",{"type":50,"tag":59,"props":210,"children":211},{},[212,214,220,222,228,230,236,238,244,246,252,254,260,262,268,270,276,278,284,285,291,293,299],{"type":56,"value":213},"Generated TAO Core schemas are packaged in ",{"type":50,"tag":75,"props":215,"children":217},{"className":216},[],[218],{"type":56,"value":219},"schemas\u002F\u003Caction>.schema.json",{"type":56,"value":221},", with ",{"type":50,"tag":75,"props":223,"children":225},{"className":224},[],[226],{"type":56,"value":227},"schemas\u002Fmanifest.json",{"type":56,"value":229}," listing available actions. Each generated schema also emits ",{"type":50,"tag":75,"props":231,"children":233},{"className":232},[],[234],{"type":56,"value":235},"references\u002Fspec_template_\u003Caction>.yaml",{"type":56,"value":237}," from the schema top-level ",{"type":50,"tag":75,"props":239,"children":241},{"className":240},[],[242],{"type":56,"value":243},"default",{"type":56,"value":245}," field. AutoML enablement is declared at the model layer in ",{"type":50,"tag":75,"props":247,"children":249},{"className":248},[],[250],{"type":56,"value":251},"references\u002Fskill_info.yaml",{"type":56,"value":253}," via ",{"type":50,"tag":75,"props":255,"children":257},{"className":256},[],[258],{"type":56,"value":259},"automl_enabled",{"type":56,"value":261},". Runnable AutoML still requires ",{"type":50,"tag":75,"props":263,"children":265},{"className":264},[],[266],{"type":56,"value":267},"schemas\u002Ftrain.schema.json",{"type":56,"value":269}," and ",{"type":50,"tag":75,"props":271,"children":273},{"className":272},[],[274],{"type":56,"value":275},"references\u002Fspec_template_train.yaml",{"type":56,"value":277}," to exist and parse. Use the packaged train schema for ",{"type":50,"tag":75,"props":279,"children":281},{"className":280},[],[282],{"type":56,"value":283},"automl_default_parameters",{"type":56,"value":135},{"type":50,"tag":75,"props":286,"children":288},{"className":287},[],[289],{"type":56,"value":290},"automl_disabled_parameters",{"type":56,"value":292},", defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect ",{"type":50,"tag":75,"props":294,"children":296},{"className":295},[],[297],{"type":56,"value":298},"~\u002Ftao-core",{"type":56,"value":300}," at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.",{"type":50,"tag":203,"props":302,"children":304},{"id":303},"train-action-policy",[305],{"type":56,"value":306},"Train Action Policy",{"type":50,"tag":59,"props":308,"children":309},{},[310,312,317,319,325,327,333,335,341,343,349,351,357,359,364,365,371,373,378,379,384,386,392,394,400,402,407,409,414],{"type":56,"value":311},"This model is AutoML-enabled at the model layer. Before handling any train-stage request, read ",{"type":50,"tag":75,"props":313,"children":315},{"className":314},[],[316],{"type":56,"value":251},{"type":56,"value":318}," and resolve the run override from either an explicit ",{"type":50,"tag":75,"props":320,"children":322},{"className":321},[],[323],{"type":56,"value":324},"automl_policy",{"type":56,"value":326}," value or the user's workflow request. Use ",{"type":50,"tag":75,"props":328,"children":330},{"className":329},[],[331],{"type":56,"value":332},"automl_policy: on",{"type":56,"value":334}," by default and only expose ",{"type":50,"tag":75,"props":336,"children":338},{"className":337},[],[339],{"type":56,"value":340},"on",{"type":56,"value":342}," \u002F ",{"type":50,"tag":75,"props":344,"children":346},{"className":345},[],[347],{"type":56,"value":348},"off",{"type":56,"value":350}," in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as ",{"type":50,"tag":75,"props":352,"children":354},{"className":353},[],[355],{"type":56,"value":356},"automl_policy: off",{"type":56,"value":358}," for this run only. When ",{"type":50,"tag":75,"props":360,"children":362},{"className":361},[],[363],{"type":56,"value":332},{"type":56,"value":135},{"type":50,"tag":75,"props":366,"children":368},{"className":367},[],[369],{"type":56,"value":370},"automl_enabled: true",{"type":56,"value":372},", and both ",{"type":50,"tag":75,"props":374,"children":376},{"className":375},[],[377],{"type":56,"value":267},{"type":56,"value":269},{"type":50,"tag":75,"props":380,"children":382},{"className":381},[],[383],{"type":56,"value":275},{"type":56,"value":385}," are packaged, route the train action through ",{"type":50,"tag":75,"props":387,"children":389},{"className":388},[],[390],{"type":56,"value":391},"tao-skill-bank:tao-run-automl",{"type":56,"value":393}," by default with this model's ",{"type":50,"tag":75,"props":395,"children":397},{"className":396},[],[398],{"type":56,"value":399},"skill_dir",{"type":56,"value":401},". Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and ",{"type":50,"tag":75,"props":403,"children":405},{"className":404},[],[406],{"type":56,"value":324},{"type":56,"value":408},". Use direct model training only when ",{"type":50,"tag":75,"props":410,"children":412},{"className":411},[],[413],{"type":56,"value":356},{"type":56,"value":415}," 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":50,"tag":59,"props":417,"children":418},{},[419,421,426,427,432,433,438,440,445],{"type":56,"value":420},"Non-train actions such as ",{"type":50,"tag":75,"props":422,"children":424},{"className":423},[],[425],{"type":56,"value":88},{"type":56,"value":135},{"type":50,"tag":75,"props":428,"children":430},{"className":429},[],[431],{"type":56,"value":96},{"type":56,"value":135},{"type":50,"tag":75,"props":434,"children":436},{"className":435},[],[437],{"type":56,"value":160},{"type":56,"value":439},", and deploy flows stay in this model skill. The per-run ",{"type":50,"tag":75,"props":441,"children":443},{"className":442},[],[444],{"type":56,"value":324},{"type":56,"value":446}," override does not change model metadata.",{"type":50,"tag":203,"props":448,"children":450},{"id":449},"training-requirements",[451],{"type":56,"value":452},"Training Requirements",{"type":50,"tag":454,"props":455,"children":456},"ul",{},[457,469,479],{"type":50,"tag":458,"props":459,"children":460},"li",{},[461,467],{"type":50,"tag":462,"props":463,"children":464},"strong",{},[465],{"type":56,"value":466},"Dataset type:",{"type":56,"value":468}," pointpillars",{"type":50,"tag":458,"props":470,"children":471},{},[472,477],{"type":50,"tag":462,"props":473,"children":474},{},[475],{"type":56,"value":476},"Formats:",{"type":56,"value":478}," default",{"type":50,"tag":458,"props":480,"children":481},{},[482,487],{"type":50,"tag":462,"props":483,"children":484},{},[485],{"type":56,"value":486},"Monitoring metric:",{"type":56,"value":488}," loss",{"type":50,"tag":490,"props":491,"children":493},"h3",{"id":492},"per-action-dataset-requirements",[494],{"type":56,"value":495},"Per-Action Dataset Requirements",{"type":50,"tag":497,"props":498,"children":499},"table",{},[500,534],{"type":50,"tag":501,"props":502,"children":503},"thead",{},[504],{"type":50,"tag":505,"props":506,"children":507},"tr",{},[508,514,519,524,529],{"type":50,"tag":509,"props":510,"children":511},"th",{},[512],{"type":56,"value":513},"Action",{"type":50,"tag":509,"props":515,"children":516},{},[517],{"type":56,"value":518},"Spec Key",{"type":50,"tag":509,"props":520,"children":521},{},[522],{"type":56,"value":523},"Source",{"type":50,"tag":509,"props":525,"children":526},{},[527],{"type":56,"value":528},"Files",{"type":50,"tag":509,"props":530,"children":531},{},[532],{"type":56,"value":533},"List?",{"type":50,"tag":535,"props":536,"children":537},"tbody",{},[538,564,587,612,634,657,679,702,724,747,769,792,814],{"type":50,"tag":505,"props":539,"children":540},{},[541,546,551,556,559],{"type":50,"tag":542,"props":543,"children":544},"td",{},[545],{"type":56,"value":133},{"type":50,"tag":542,"props":547,"children":548},{},[549],{"type":56,"value":550},"dataset.data_path",{"type":50,"tag":542,"props":552,"children":553},{},[554],{"type":56,"value":555},"id",{"type":50,"tag":542,"props":557,"children":558},{},[],{"type":50,"tag":542,"props":560,"children":561},{},[562],{"type":56,"value":563},"No",{"type":50,"tag":505,"props":565,"children":566},{},[567,571,575,580,583],{"type":50,"tag":542,"props":568,"children":569},{},[570],{"type":56,"value":88},{"type":50,"tag":542,"props":572,"children":573},{},[574],{"type":56,"value":550},{"type":50,"tag":542,"props":576,"children":577},{},[578],{"type":56,"value":579},"train_datasets",{"type":50,"tag":542,"props":581,"children":582},{},[],{"type":50,"tag":542,"props":584,"children":585},{},[586],{"type":56,"value":563},{"type":50,"tag":505,"props":588,"children":589},{},[590,594,599,603,608],{"type":50,"tag":542,"props":591,"children":592},{},[593],{"type":56,"value":88},{"type":50,"tag":542,"props":595,"children":596},{},[597],{"type":56,"value":598},"dataset.data_info_path",{"type":50,"tag":542,"props":600,"children":601},{},[602],{"type":56,"value":579},{"type":50,"tag":542,"props":604,"children":605},{},[606],{"type":56,"value":607},"\u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\u002F",{"type":50,"tag":542,"props":609,"children":610},{},[611],{"type":56,"value":563},{"type":50,"tag":505,"props":613,"children":614},{},[615,619,623,627,630],{"type":50,"tag":542,"props":616,"children":617},{},[618],{"type":56,"value":160},{"type":50,"tag":542,"props":620,"children":621},{},[622],{"type":56,"value":550},{"type":50,"tag":542,"props":624,"children":625},{},[626],{"type":56,"value":579},{"type":50,"tag":542,"props":628,"children":629},{},[],{"type":50,"tag":542,"props":631,"children":632},{},[633],{"type":56,"value":563},{"type":50,"tag":505,"props":635,"children":636},{},[637,641,645,649,653],{"type":50,"tag":542,"props":638,"children":639},{},[640],{"type":56,"value":160},{"type":50,"tag":542,"props":642,"children":643},{},[644],{"type":56,"value":598},{"type":50,"tag":542,"props":646,"children":647},{},[648],{"type":56,"value":579},{"type":50,"tag":542,"props":650,"children":651},{},[652],{"type":56,"value":607},{"type":50,"tag":542,"props":654,"children":655},{},[656],{"type":56,"value":563},{"type":50,"tag":505,"props":658,"children":659},{},[660,664,668,672,675],{"type":50,"tag":542,"props":661,"children":662},{},[663],{"type":56,"value":96},{"type":50,"tag":542,"props":665,"children":666},{},[667],{"type":56,"value":550},{"type":50,"tag":542,"props":669,"children":670},{},[671],{"type":56,"value":579},{"type":50,"tag":542,"props":673,"children":674},{},[],{"type":50,"tag":542,"props":676,"children":677},{},[678],{"type":56,"value":563},{"type":50,"tag":505,"props":680,"children":681},{},[682,686,690,694,698],{"type":50,"tag":542,"props":683,"children":684},{},[685],{"type":56,"value":96},{"type":50,"tag":542,"props":687,"children":688},{},[689],{"type":56,"value":598},{"type":50,"tag":542,"props":691,"children":692},{},[693],{"type":56,"value":579},{"type":50,"tag":542,"props":695,"children":696},{},[697],{"type":56,"value":607},{"type":50,"tag":542,"props":699,"children":700},{},[701],{"type":56,"value":563},{"type":50,"tag":505,"props":703,"children":704},{},[705,709,713,717,720],{"type":50,"tag":542,"props":706,"children":707},{},[708],{"type":56,"value":168},{"type":50,"tag":542,"props":710,"children":711},{},[712],{"type":56,"value":550},{"type":50,"tag":542,"props":714,"children":715},{},[716],{"type":56,"value":579},{"type":50,"tag":542,"props":718,"children":719},{},[],{"type":50,"tag":542,"props":721,"children":722},{},[723],{"type":56,"value":563},{"type":50,"tag":505,"props":725,"children":726},{},[727,731,735,739,743],{"type":50,"tag":542,"props":728,"children":729},{},[730],{"type":56,"value":168},{"type":50,"tag":542,"props":732,"children":733},{},[734],{"type":56,"value":598},{"type":50,"tag":542,"props":736,"children":737},{},[738],{"type":56,"value":579},{"type":50,"tag":542,"props":740,"children":741},{},[742],{"type":56,"value":607},{"type":50,"tag":542,"props":744,"children":745},{},[746],{"type":56,"value":563},{"type":50,"tag":505,"props":748,"children":749},{},[750,754,758,762,765],{"type":50,"tag":542,"props":751,"children":752},{},[753],{"type":56,"value":183},{"type":50,"tag":542,"props":755,"children":756},{},[757],{"type":56,"value":550},{"type":50,"tag":542,"props":759,"children":760},{},[761],{"type":56,"value":579},{"type":50,"tag":542,"props":763,"children":764},{},[],{"type":50,"tag":542,"props":766,"children":767},{},[768],{"type":56,"value":563},{"type":50,"tag":505,"props":770,"children":771},{},[772,776,780,784,788],{"type":50,"tag":542,"props":773,"children":774},{},[775],{"type":56,"value":183},{"type":50,"tag":542,"props":777,"children":778},{},[779],{"type":56,"value":598},{"type":50,"tag":542,"props":781,"children":782},{},[783],{"type":56,"value":579},{"type":50,"tag":542,"props":785,"children":786},{},[787],{"type":56,"value":607},{"type":50,"tag":542,"props":789,"children":790},{},[791],{"type":56,"value":563},{"type":50,"tag":505,"props":793,"children":794},{},[795,799,803,807,810],{"type":50,"tag":542,"props":796,"children":797},{},[798],{"type":56,"value":141},{"type":50,"tag":542,"props":800,"children":801},{},[802],{"type":56,"value":550},{"type":50,"tag":542,"props":804,"children":805},{},[806],{"type":56,"value":579},{"type":50,"tag":542,"props":808,"children":809},{},[],{"type":50,"tag":542,"props":811,"children":812},{},[813],{"type":56,"value":563},{"type":50,"tag":505,"props":815,"children":816},{},[817,821,825,829,833],{"type":50,"tag":542,"props":818,"children":819},{},[820],{"type":56,"value":141},{"type":50,"tag":542,"props":822,"children":823},{},[824],{"type":56,"value":598},{"type":50,"tag":542,"props":826,"children":827},{},[828],{"type":56,"value":579},{"type":50,"tag":542,"props":830,"children":831},{},[832],{"type":56,"value":607},{"type":50,"tag":542,"props":834,"children":835},{},[836],{"type":56,"value":563},{"type":50,"tag":490,"props":838,"children":840},{"id":839},"typical-spec-overrides",[841],{"type":56,"value":842},"Typical Spec Overrides",{"type":50,"tag":59,"props":844,"children":845},{},[846,848,853,855,861],{"type":56,"value":847},"Data source overrides are ",{"type":50,"tag":462,"props":849,"children":850},{},[851],{"type":56,"value":852},"mandatory for every action",{"type":56,"value":854}," — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in ",{"type":50,"tag":75,"props":856,"children":858},{"className":857},[],[859],{"type":56,"value":860},"spec_overrides",{"type":56,"value":862},".",{"type":50,"tag":864,"props":865,"children":870},"pre",{"className":866,"code":867,"language":868,"meta":869,"style":869},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","DATA_ROOT = \"s3:\u002F\u002Fbucket\u002Fdata\u002Fpointpillars\"\nDATA_INFO = \"\u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\"\nCHECKPOINT = \"\u002Fresults\u002F{train_job_id}\u002Fresults_dir\u002Fcheckpoint_epoch_1.pth\"\nPRUNED_MODEL = \"\u002Fresults\u002F{prune_job_id}\u002Fresults_dir\u002Fpruned_0.1.tlt\"\n","python","",[871],{"type":50,"tag":75,"props":872,"children":873},{"__ignoreMap":869},[874,885,894,903],{"type":50,"tag":875,"props":876,"children":879},"span",{"class":877,"line":878},"line",1,[880],{"type":50,"tag":875,"props":881,"children":882},{},[883],{"type":56,"value":884},"DATA_ROOT = \"s3:\u002F\u002Fbucket\u002Fdata\u002Fpointpillars\"\n",{"type":50,"tag":875,"props":886,"children":888},{"class":877,"line":887},2,[889],{"type":50,"tag":875,"props":890,"children":891},{},[892],{"type":56,"value":893},"DATA_INFO = \"\u002Fresults\u002F{dataset_convert_job_id}\u002Fresults_dir\u002Fdata_info\"\n",{"type":50,"tag":875,"props":895,"children":897},{"class":877,"line":896},3,[898],{"type":50,"tag":875,"props":899,"children":900},{},[901],{"type":56,"value":902},"CHECKPOINT = \"\u002Fresults\u002F{train_job_id}\u002Fresults_dir\u002Fcheckpoint_epoch_1.pth\"\n",{"type":50,"tag":875,"props":904,"children":906},{"class":877,"line":905},4,[907],{"type":50,"tag":875,"props":908,"children":909},{},[910],{"type":56,"value":911},"PRUNED_MODEL = \"\u002Fresults\u002F{prune_job_id}\u002Fresults_dir\u002Fpruned_0.1.tlt\"\n",{"type":50,"tag":59,"props":913,"children":914},{},[915,917,923,924,930,931,937,938,944,946,951,953,958],{"type":56,"value":916},"The raw PointPillars data root must be an extracted folder containing matching ",{"type":50,"tag":75,"props":918,"children":920},{"className":919},[],[921],{"type":56,"value":922},"train\u002Flidar",{"type":56,"value":135},{"type":50,"tag":75,"props":925,"children":927},{"className":926},[],[928],{"type":56,"value":929},"train\u002Flabel",{"type":56,"value":135},{"type":50,"tag":75,"props":932,"children":934},{"className":933},[],[935],{"type":56,"value":936},"val\u002Flidar",{"type":56,"value":162},{"type":50,"tag":75,"props":939,"children":941},{"className":940},[],[942],{"type":56,"value":943},"val\u002Flabel",{"type":56,"value":945}," subfolders before ",{"type":50,"tag":75,"props":947,"children":949},{"className":948},[],[950],{"type":56,"value":133},{"type":56,"value":952}," runs. If the source dataset is packaged as separate train\u002Fval archives, extract both under the same mounted data root and point ",{"type":50,"tag":75,"props":954,"children":956},{"className":955},[],[957],{"type":56,"value":550},{"type":56,"value":959}," at that root.",{"type":50,"tag":59,"props":961,"children":962},{},[963],{"type":50,"tag":462,"props":964,"children":965},{},[966],{"type":56,"value":967},"train (mandatory data sources):",{"type":50,"tag":864,"props":969,"children":971},{"className":866,"code":970,"language":868,"meta":869,"style":869},"{\n    \"train.num_epochs\": 30,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n}\n",[972],{"type":50,"tag":75,"props":973,"children":974},{"__ignoreMap":869},[975,983,991,999,1007,1016,1025,1034],{"type":50,"tag":875,"props":976,"children":977},{"class":877,"line":878},[978],{"type":50,"tag":875,"props":979,"children":980},{},[981],{"type":56,"value":982},"{\n",{"type":50,"tag":875,"props":984,"children":985},{"class":877,"line":887},[986],{"type":50,"tag":875,"props":987,"children":988},{},[989],{"type":56,"value":990},"    \"train.num_epochs\": 30,\n",{"type":50,"tag":875,"props":992,"children":993},{"class":877,"line":896},[994],{"type":50,"tag":875,"props":995,"children":996},{},[997],{"type":56,"value":998},"    \"train.checkpoint_interval\": 10,\n",{"type":50,"tag":875,"props":1000,"children":1001},{"class":877,"line":905},[1002],{"type":50,"tag":875,"props":1003,"children":1004},{},[1005],{"type":56,"value":1006},"    \"train.validation_interval\": 10,\n",{"type":50,"tag":875,"props":1008,"children":1010},{"class":877,"line":1009},5,[1011],{"type":50,"tag":875,"props":1012,"children":1013},{},[1014],{"type":56,"value":1015},"    \"train.num_gpus\": 1,\n",{"type":50,"tag":875,"props":1017,"children":1019},{"class":877,"line":1018},6,[1020],{"type":50,"tag":875,"props":1021,"children":1022},{},[1023],{"type":56,"value":1024},"    \"dataset.data_path\": DATA_ROOT,\n",{"type":50,"tag":875,"props":1026,"children":1028},{"class":877,"line":1027},7,[1029],{"type":50,"tag":875,"props":1030,"children":1031},{},[1032],{"type":56,"value":1033},"    \"dataset.data_info_path\": DATA_INFO,\n",{"type":50,"tag":875,"props":1035,"children":1037},{"class":877,"line":1036},8,[1038],{"type":50,"tag":875,"props":1039,"children":1040},{},[1041],{"type":56,"value":1042},"}\n",{"type":50,"tag":59,"props":1044,"children":1045},{},[1046],{"type":50,"tag":462,"props":1047,"children":1048},{},[1049],{"type":56,"value":1050},"resume train (mandatory checkpoint):",{"type":50,"tag":864,"props":1052,"children":1054},{"className":866,"code":1053,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"train.resume_training_checkpoint_path\": CHECKPOINT,\n}\n",[1055],{"type":50,"tag":75,"props":1056,"children":1057},{"__ignoreMap":869},[1058,1065,1072,1079,1087],{"type":50,"tag":875,"props":1059,"children":1060},{"class":877,"line":878},[1061],{"type":50,"tag":875,"props":1062,"children":1063},{},[1064],{"type":56,"value":982},{"type":50,"tag":875,"props":1066,"children":1067},{"class":877,"line":887},[1068],{"type":50,"tag":875,"props":1069,"children":1070},{},[1071],{"type":56,"value":1024},{"type":50,"tag":875,"props":1073,"children":1074},{"class":877,"line":896},[1075],{"type":50,"tag":875,"props":1076,"children":1077},{},[1078],{"type":56,"value":1033},{"type":50,"tag":875,"props":1080,"children":1081},{"class":877,"line":905},[1082],{"type":50,"tag":875,"props":1083,"children":1084},{},[1085],{"type":56,"value":1086},"    \"train.resume_training_checkpoint_path\": CHECKPOINT,\n",{"type":50,"tag":875,"props":1088,"children":1089},{"class":877,"line":1009},[1090],{"type":50,"tag":875,"props":1091,"children":1092},{},[1093],{"type":56,"value":1042},{"type":50,"tag":59,"props":1095,"children":1096},{},[1097],{"type":50,"tag":462,"props":1098,"children":1099},{},[1100],{"type":56,"value":1101},"evaluate (mandatory data sources):",{"type":50,"tag":864,"props":1103,"children":1105},{"className":866,"code":1104,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"evaluate.checkpoint\": CHECKPOINT,\n}\n",[1106],{"type":50,"tag":75,"props":1107,"children":1108},{"__ignoreMap":869},[1109,1116,1123,1130,1138],{"type":50,"tag":875,"props":1110,"children":1111},{"class":877,"line":878},[1112],{"type":50,"tag":875,"props":1113,"children":1114},{},[1115],{"type":56,"value":982},{"type":50,"tag":875,"props":1117,"children":1118},{"class":877,"line":887},[1119],{"type":50,"tag":875,"props":1120,"children":1121},{},[1122],{"type":56,"value":1024},{"type":50,"tag":875,"props":1124,"children":1125},{"class":877,"line":896},[1126],{"type":50,"tag":875,"props":1127,"children":1128},{},[1129],{"type":56,"value":1033},{"type":50,"tag":875,"props":1131,"children":1132},{"class":877,"line":905},[1133],{"type":50,"tag":875,"props":1134,"children":1135},{},[1136],{"type":56,"value":1137},"    \"evaluate.checkpoint\": CHECKPOINT,\n",{"type":50,"tag":875,"props":1139,"children":1140},{"class":877,"line":1009},[1141],{"type":50,"tag":875,"props":1142,"children":1143},{},[1144],{"type":56,"value":1042},{"type":50,"tag":59,"props":1146,"children":1147},{},[1148],{"type":50,"tag":462,"props":1149,"children":1150},{},[1151],{"type":56,"value":1152},"export (mandatory data sources):",{"type":50,"tag":864,"props":1154,"children":1156},{"className":866,"code":1155,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"export.checkpoint\": CHECKPOINT,\n    \"export.onnx_file\": \"\u002Fresults\u002F{export_job_id}\u002Fresults_dir\u002Fpointpillars.onnx\",\n}\n",[1157],{"type":50,"tag":75,"props":1158,"children":1159},{"__ignoreMap":869},[1160,1167,1174,1181,1189,1197],{"type":50,"tag":875,"props":1161,"children":1162},{"class":877,"line":878},[1163],{"type":50,"tag":875,"props":1164,"children":1165},{},[1166],{"type":56,"value":982},{"type":50,"tag":875,"props":1168,"children":1169},{"class":877,"line":887},[1170],{"type":50,"tag":875,"props":1171,"children":1172},{},[1173],{"type":56,"value":1024},{"type":50,"tag":875,"props":1175,"children":1176},{"class":877,"line":896},[1177],{"type":50,"tag":875,"props":1178,"children":1179},{},[1180],{"type":56,"value":1033},{"type":50,"tag":875,"props":1182,"children":1183},{"class":877,"line":905},[1184],{"type":50,"tag":875,"props":1185,"children":1186},{},[1187],{"type":56,"value":1188},"    \"export.checkpoint\": CHECKPOINT,\n",{"type":50,"tag":875,"props":1190,"children":1191},{"class":877,"line":1009},[1192],{"type":50,"tag":875,"props":1193,"children":1194},{},[1195],{"type":56,"value":1196},"    \"export.onnx_file\": \"\u002Fresults\u002F{export_job_id}\u002Fresults_dir\u002Fpointpillars.onnx\",\n",{"type":50,"tag":875,"props":1198,"children":1199},{"class":877,"line":1018},[1200],{"type":50,"tag":875,"props":1201,"children":1202},{},[1203],{"type":56,"value":1042},{"type":50,"tag":59,"props":1205,"children":1206},{},[1207],{"type":50,"tag":462,"props":1208,"children":1209},{},[1210],{"type":56,"value":1211},"inference (mandatory data sources):",{"type":50,"tag":864,"props":1213,"children":1215},{"className":866,"code":1214,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"inference.checkpoint\": CHECKPOINT,\n}\n",[1216],{"type":50,"tag":75,"props":1217,"children":1218},{"__ignoreMap":869},[1219,1226,1233,1240,1248],{"type":50,"tag":875,"props":1220,"children":1221},{"class":877,"line":878},[1222],{"type":50,"tag":875,"props":1223,"children":1224},{},[1225],{"type":56,"value":982},{"type":50,"tag":875,"props":1227,"children":1228},{"class":877,"line":887},[1229],{"type":50,"tag":875,"props":1230,"children":1231},{},[1232],{"type":56,"value":1024},{"type":50,"tag":875,"props":1234,"children":1235},{"class":877,"line":896},[1236],{"type":50,"tag":875,"props":1237,"children":1238},{},[1239],{"type":56,"value":1033},{"type":50,"tag":875,"props":1241,"children":1242},{"class":877,"line":905},[1243],{"type":50,"tag":875,"props":1244,"children":1245},{},[1246],{"type":56,"value":1247},"    \"inference.checkpoint\": CHECKPOINT,\n",{"type":50,"tag":875,"props":1249,"children":1250},{"class":877,"line":1009},[1251],{"type":50,"tag":875,"props":1252,"children":1253},{},[1254],{"type":56,"value":1042},{"type":50,"tag":59,"props":1256,"children":1257},{},[1258],{"type":50,"tag":462,"props":1259,"children":1260},{},[1261],{"type":56,"value":1262},"prune (mandatory data sources):",{"type":50,"tag":864,"props":1264,"children":1266},{"className":866,"code":1265,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"prune.model\": CHECKPOINT,\n}\n",[1267],{"type":50,"tag":75,"props":1268,"children":1269},{"__ignoreMap":869},[1270,1277,1284,1291,1299],{"type":50,"tag":875,"props":1271,"children":1272},{"class":877,"line":878},[1273],{"type":50,"tag":875,"props":1274,"children":1275},{},[1276],{"type":56,"value":982},{"type":50,"tag":875,"props":1278,"children":1279},{"class":877,"line":887},[1280],{"type":50,"tag":875,"props":1281,"children":1282},{},[1283],{"type":56,"value":1024},{"type":50,"tag":875,"props":1285,"children":1286},{"class":877,"line":896},[1287],{"type":50,"tag":875,"props":1288,"children":1289},{},[1290],{"type":56,"value":1033},{"type":50,"tag":875,"props":1292,"children":1293},{"class":877,"line":905},[1294],{"type":50,"tag":875,"props":1295,"children":1296},{},[1297],{"type":56,"value":1298},"    \"prune.model\": CHECKPOINT,\n",{"type":50,"tag":875,"props":1300,"children":1301},{"class":877,"line":1009},[1302],{"type":50,"tag":875,"props":1303,"children":1304},{},[1305],{"type":56,"value":1042},{"type":50,"tag":59,"props":1307,"children":1308},{},[1309],{"type":50,"tag":462,"props":1310,"children":1311},{},[1312],{"type":56,"value":1313},"retrain (mandatory data sources):",{"type":50,"tag":864,"props":1315,"children":1317},{"className":866,"code":1316,"language":868,"meta":869,"style":869},"{\n    \"dataset.data_path\": DATA_ROOT,\n    \"dataset.data_info_path\": DATA_INFO,\n    \"train.pruned_model_path\": PRUNED_MODEL,\n}\n",[1318],{"type":50,"tag":75,"props":1319,"children":1320},{"__ignoreMap":869},[1321,1328,1335,1342,1350],{"type":50,"tag":875,"props":1322,"children":1323},{"class":877,"line":878},[1324],{"type":50,"tag":875,"props":1325,"children":1326},{},[1327],{"type":56,"value":982},{"type":50,"tag":875,"props":1329,"children":1330},{"class":877,"line":887},[1331],{"type":50,"tag":875,"props":1332,"children":1333},{},[1334],{"type":56,"value":1024},{"type":50,"tag":875,"props":1336,"children":1337},{"class":877,"line":896},[1338],{"type":50,"tag":875,"props":1339,"children":1340},{},[1341],{"type":56,"value":1033},{"type":50,"tag":875,"props":1343,"children":1344},{"class":877,"line":905},[1345],{"type":50,"tag":875,"props":1346,"children":1347},{},[1348],{"type":56,"value":1349},"    \"train.pruned_model_path\": PRUNED_MODEL,\n",{"type":50,"tag":875,"props":1351,"children":1352},{"class":877,"line":1009},[1353],{"type":50,"tag":875,"props":1354,"children":1355},{},[1356],{"type":56,"value":1042},{"type":50,"tag":59,"props":1358,"children":1359},{},[1360,1362,1368,1370,1376,1378,1383,1385,1391,1393,1399,1401,1407,1409,1415],{"type":56,"value":1361},"For local Docker, ",{"type":50,"tag":75,"props":1363,"children":1365},{"className":1364},[],[1366],{"type":56,"value":1367},"DATA_INFO",{"type":56,"value":1369}," must be visible inside every train\u002Fevaluate\u002Fexport\u002Fprune\u002Fretrain container. Use the dataset_convert job from the same results root, or mount\u002Fcopy the converted ",{"type":50,"tag":75,"props":1371,"children":1373},{"className":1372},[],[1374],{"type":56,"value":1375},"results_dir\u002Fdata_info",{"type":56,"value":1377}," folder into the current run and set ",{"type":50,"tag":75,"props":1379,"children":1381},{"className":1380},[],[1382],{"type":56,"value":598},{"type":56,"value":1384}," to that mounted container path. If the host scratch root is mounted at ",{"type":50,"tag":75,"props":1386,"children":1388},{"className":1387},[],[1389],{"type":56,"value":1390},"\u002Fresults",{"type":56,"value":1392}," and the conversion artifacts live under host ",{"type":50,"tag":75,"props":1394,"children":1396},{"className":1395},[],[1397],{"type":56,"value":1398},"scratch\u002Fresults\u002F\u003Cjob_id>\u002Fresults_dir\u002Fdata_info",{"type":56,"value":1400},", the direct-job container path is ",{"type":50,"tag":75,"props":1402,"children":1404},{"className":1403},[],[1405],{"type":56,"value":1406},"\u002Fresults\u002Fresults\u002F\u003Cjob_id>\u002Fresults_dir\u002Fdata_info",{"type":56,"value":1408},". Do not reuse a ",{"type":50,"tag":75,"props":1410,"children":1412},{"className":1411},[],[1413],{"type":56,"value":1414},"\u002Fresults\u002F\u003Cjob_id>\u002F...",{"type":56,"value":1416}," path from another run root unless that folder is mounted into the current job.",{"type":50,"tag":59,"props":1418,"children":1419},{},[1420,1422,1428,1430,1435,1437,1443,1445,1450,1452,1458,1459,1465,1466,1472,1474,1480],{"type":56,"value":1421},"For AutoML train workflows, perform this as a launch preflight before calling ",{"type":50,"tag":75,"props":1423,"children":1425},{"className":1424},[],[1426],{"type":56,"value":1427},"AutoMLRunner.run",{"type":56,"value":1429},": create or materialize the ",{"type":50,"tag":75,"props":1431,"children":1433},{"className":1432},[],[1434],{"type":56,"value":133},{"type":56,"value":1436}," output under the current run's ",{"type":50,"tag":75,"props":1438,"children":1440},{"className":1439},[],[1441],{"type":56,"value":1442},"RESULTS_ROOT",{"type":56,"value":1444},", set ",{"type":50,"tag":75,"props":1446,"children":1448},{"className":1447},[],[1449],{"type":56,"value":598},{"type":56,"value":1451}," to that current-run container path, and verify ",{"type":50,"tag":75,"props":1453,"children":1455},{"className":1454},[],[1456],{"type":56,"value":1457},"dbinfos_train.pkl",{"type":56,"value":135},{"type":50,"tag":75,"props":1460,"children":1462},{"className":1461},[],[1463],{"type":56,"value":1464},"infos_train.pkl",{"type":56,"value":162},{"type":50,"tag":75,"props":1467,"children":1469},{"className":1468},[],[1470],{"type":56,"value":1471},"infos_val.pkl",{"type":56,"value":1473}," are present from the train container's point of view. If a runner is cloned or adapted from a prior AutoML algorithm, update the conversion artifact in the new run root; a stale ",{"type":50,"tag":75,"props":1475,"children":1477},{"className":1476},[],[1478],{"type":56,"value":1479},"CONVERT_JOB_ID",{"type":56,"value":1481}," from another results mount is not valid.",{"type":50,"tag":203,"props":1483,"children":1485},{"id":1484},"eval-dataset",[1486],{"type":56,"value":1487},"Eval Dataset",{"type":50,"tag":59,"props":1489,"children":1490},{},[1491],{"type":56,"value":1492},"Optional. Validation data (val.tar.gz) is separate from training. Used for mAP evaluation.",{"type":50,"tag":203,"props":1494,"children":1496},{"id":1495},"important-parameters",[1497],{"type":56,"value":1498},"Important Parameters",{"type":50,"tag":454,"props":1500,"children":1501},{},[1502,1512,1522,1532,1541,1550,1560],{"type":50,"tag":458,"props":1503,"children":1504},{},[1505,1510],{"type":50,"tag":462,"props":1506,"children":1507},{},[1508],{"type":56,"value":1509},"train.num_epochs",{"type":56,"value":1511},": Default 80 (much higher than other TAO models). PointPillars needs more epochs for convergence on 3D detection.",{"type":50,"tag":458,"props":1513,"children":1514},{},[1515,1520],{"type":50,"tag":462,"props":1516,"children":1517},{},[1518],{"type":56,"value":1519},"train.lr",{"type":56,"value":1521},": Learning rate. Default 0.003 (adam_onecycle scheduler).",{"type":50,"tag":458,"props":1523,"children":1524},{},[1525,1530],{"type":50,"tag":462,"props":1526,"children":1527},{},[1528],{"type":56,"value":1529},"dataset.class_names",{"type":56,"value":1531},": List of 3D object classes. Default 7 classes (KITTI-style). Modify to match your dataset.",{"type":50,"tag":458,"props":1533,"children":1534},{},[1535,1539],{"type":50,"tag":462,"props":1536,"children":1537},{},[1538],{"type":56,"value":550},{"type":56,"value":1540},": Path to point cloud data directory.",{"type":50,"tag":458,"props":1542,"children":1543},{},[1544,1548],{"type":50,"tag":462,"props":1545,"children":1546},{},[1547],{"type":56,"value":598},{"type":56,"value":1549},": Path to data info files from dataset_convert step.",{"type":50,"tag":458,"props":1551,"children":1552},{},[1553,1558],{"type":50,"tag":462,"props":1554,"children":1555},{},[1556],{"type":56,"value":1557},"dataset.point_cloud_range",{"type":56,"value":1559},": Spatial extent of the point cloud to consider. Must match your sensor configuration.",{"type":50,"tag":458,"props":1561,"children":1562},{},[1563,1568],{"type":50,"tag":462,"props":1564,"children":1565},{},[1566],{"type":56,"value":1567},"model.dense_head.anchor_generator_config",{"type":56,"value":1569},": Anchor configurations per class. Must be tuned for your object sizes and the point cloud range.",{"type":50,"tag":203,"props":1571,"children":1573},{"id":1572},"multi-gpu-multi-node",[1574],{"type":56,"value":1575},"Multi-GPU \u002F Multi-Node",{"type":50,"tag":59,"props":1577,"children":1578},{},[1579,1584,1586,1592,1594,1600],{"type":50,"tag":462,"props":1580,"children":1581},{},[1582],{"type":56,"value":1583},"Launch method:",{"type":56,"value":1585}," ",{"type":50,"tag":75,"props":1587,"children":1589},{"className":1588},[],[1590],{"type":56,"value":1591},"torchrun",{"type":56,"value":1593}," (LIGHTNING_EXCLUDED_NETWORK). Uses PyTorch native ",{"type":50,"tag":75,"props":1595,"children":1597},{"className":1596},[],[1598],{"type":56,"value":1599},"DistributedDataParallel",{"type":56,"value":1601}," (NOT Lightning Trainer).",{"type":50,"tag":497,"props":1603,"children":1604},{},[1605,1625],{"type":50,"tag":501,"props":1606,"children":1607},{},[1608],{"type":50,"tag":505,"props":1609,"children":1610},{},[1611,1615,1620],{"type":50,"tag":509,"props":1612,"children":1613},{},[1614],{"type":56,"value":518},{"type":50,"tag":509,"props":1616,"children":1617},{},[1618],{"type":56,"value":1619},"Description",{"type":50,"tag":509,"props":1621,"children":1622},{},[1623],{"type":56,"value":1624},"Default",{"type":50,"tag":535,"props":1626,"children":1627},{},[1628,1650,1675],{"type":50,"tag":505,"props":1629,"children":1630},{},[1631,1640,1645],{"type":50,"tag":542,"props":1632,"children":1633},{},[1634],{"type":50,"tag":75,"props":1635,"children":1637},{"className":1636},[],[1638],{"type":56,"value":1639},"train.num_gpus",{"type":50,"tag":542,"props":1641,"children":1642},{},[1643],{"type":56,"value":1644},"Number of GPUs per node",{"type":50,"tag":542,"props":1646,"children":1647},{},[1648],{"type":56,"value":1649},"1",{"type":50,"tag":505,"props":1651,"children":1652},{},[1653,1662,1667],{"type":50,"tag":542,"props":1654,"children":1655},{},[1656],{"type":50,"tag":75,"props":1657,"children":1659},{"className":1658},[],[1660],{"type":56,"value":1661},"train.gpu_ids",{"type":50,"tag":542,"props":1663,"children":1664},{},[1665],{"type":56,"value":1666},"GPU device indices",{"type":50,"tag":542,"props":1668,"children":1669},{},[1670],{"type":50,"tag":875,"props":1671,"children":1672},{},[1673],{"type":56,"value":1674},"0",{"type":50,"tag":505,"props":1676,"children":1677},{},[1678,1687,1692],{"type":50,"tag":542,"props":1679,"children":1680},{},[1681],{"type":50,"tag":75,"props":1682,"children":1684},{"className":1683},[],[1685],{"type":56,"value":1686},"train.num_nodes",{"type":50,"tag":542,"props":1688,"children":1689},{},[1690],{"type":56,"value":1691},"Number of nodes",{"type":50,"tag":542,"props":1693,"children":1694},{},[1695],{"type":56,"value":1649},{"type":50,"tag":454,"props":1697,"children":1698},{},[1699,1716,1729],{"type":50,"tag":458,"props":1700,"children":1701},{},[1702,1708,1710],{"type":50,"tag":75,"props":1703,"children":1705},{"className":1704},[],[1706],{"type":56,"value":1707},"CUDA_VISIBLE_DEVICES",{"type":56,"value":1709}," is explicitly set from ",{"type":50,"tag":75,"props":1711,"children":1713},{"className":1712},[],[1714],{"type":56,"value":1715},"TAO_VISIBLE_DEVICES",{"type":50,"tag":458,"props":1717,"children":1718},{},[1719,1721,1727],{"type":56,"value":1720},"Uses ",{"type":50,"tag":75,"props":1722,"children":1724},{"className":1723},[],[1725],{"type":56,"value":1726},"nn.parallel.DistributedDataParallel",{"type":56,"value":1728}," directly (not Lightning strategy)",{"type":50,"tag":458,"props":1730,"children":1731},{},[1732,1738,1740,1746,1748,1753],{"type":50,"tag":75,"props":1733,"children":1735},{"className":1734},[],[1736],{"type":56,"value":1737},"NODE_RANK",{"type":56,"value":1739}," is copied to ",{"type":50,"tag":75,"props":1741,"children":1743},{"className":1742},[],[1744],{"type":56,"value":1745},"RANK",{"type":56,"value":1747}," if ",{"type":50,"tag":75,"props":1749,"children":1751},{"className":1750},[],[1752],{"type":56,"value":1745},{"type":56,"value":1754}," is unset",{"type":50,"tag":59,"props":1756,"children":1757},{},[1758,1763],{"type":50,"tag":462,"props":1759,"children":1760},{},[1761],{"type":56,"value":1762},"Multi-node env vars",{"type":56,"value":1764}," (set by orchestrator):",{"type":50,"tag":497,"props":1766,"children":1767},{},[1768,1784],{"type":50,"tag":501,"props":1769,"children":1770},{},[1771],{"type":50,"tag":505,"props":1772,"children":1773},{},[1774,1779],{"type":50,"tag":509,"props":1775,"children":1776},{},[1777],{"type":56,"value":1778},"Variable",{"type":50,"tag":509,"props":1780,"children":1781},{},[1782],{"type":56,"value":1783},"Purpose",{"type":50,"tag":535,"props":1785,"children":1786},{},[1787,1803,1819,1836,1853],{"type":50,"tag":505,"props":1788,"children":1789},{},[1790,1799],{"type":50,"tag":542,"props":1791,"children":1792},{},[1793],{"type":50,"tag":75,"props":1794,"children":1796},{"className":1795},[],[1797],{"type":56,"value":1798},"WORLD_SIZE",{"type":50,"tag":542,"props":1800,"children":1801},{},[1802],{"type":56,"value":1691},{"type":50,"tag":505,"props":1804,"children":1805},{},[1806,1814],{"type":50,"tag":542,"props":1807,"children":1808},{},[1809],{"type":50,"tag":75,"props":1810,"children":1812},{"className":1811},[],[1813],{"type":56,"value":1737},{"type":50,"tag":542,"props":1815,"children":1816},{},[1817],{"type":56,"value":1818},"This node's rank",{"type":50,"tag":505,"props":1820,"children":1821},{},[1822,1831],{"type":50,"tag":542,"props":1823,"children":1824},{},[1825],{"type":50,"tag":75,"props":1826,"children":1828},{"className":1827},[],[1829],{"type":56,"value":1830},"MASTER_ADDR",{"type":50,"tag":542,"props":1832,"children":1833},{},[1834],{"type":56,"value":1835},"Rank-0 node IP",{"type":50,"tag":505,"props":1837,"children":1838},{},[1839,1848],{"type":50,"tag":542,"props":1840,"children":1841},{},[1842],{"type":50,"tag":75,"props":1843,"children":1845},{"className":1844},[],[1846],{"type":56,"value":1847},"MASTER_PORT",{"type":50,"tag":542,"props":1849,"children":1850},{},[1851],{"type":56,"value":1852},"Rank-0 port (default 29500)",{"type":50,"tag":505,"props":1854,"children":1855},{},[1856,1865],{"type":50,"tag":542,"props":1857,"children":1858},{},[1859],{"type":50,"tag":75,"props":1860,"children":1862},{"className":1861},[],[1863],{"type":56,"value":1864},"NUM_GPU_PER_NODE",{"type":50,"tag":542,"props":1866,"children":1867},{},[1868],{"type":56,"value":1869},"GPUs per node",{"type":50,"tag":203,"props":1871,"children":1873},{"id":1872},"hardware",[1874],{"type":56,"value":1875},"Hardware",{"type":50,"tag":59,"props":1877,"children":1878},{},[1879],{"type":56,"value":1880},"Minimum 1 GPU(s), recommended 4 GPU(s). 16GB+ (V100 or A100) VRAM per GPU. PointPillars is relatively efficient for 3D detection. The main bottleneck is data I\u002FO for large point cloud datasets.",{"type":50,"tag":203,"props":1882,"children":1884},{"id":1883},"error-patterns",[1885],{"type":56,"value":1886},"Error Patterns",{"type":50,"tag":59,"props":1888,"children":1889},{},[1890,1895,1897,1902,1904,1909,1911,1916,1917,1922,1924,1929,1931,1937],{"type":50,"tag":462,"props":1891,"children":1892},{},[1893],{"type":56,"value":1894},"dataset_convert required",{"type":56,"value":1896},": Training will fail if ",{"type":50,"tag":75,"props":1898,"children":1900},{"className":1899},[],[1901],{"type":56,"value":598},{"type":56,"value":1903}," is not populated from a prior ",{"type":50,"tag":75,"props":1905,"children":1907},{"className":1906},[],[1908],{"type":56,"value":133},{"type":56,"value":1910}," job. Always run convert first, and verify the train container can see ",{"type":50,"tag":75,"props":1912,"children":1914},{"className":1913},[],[1915],{"type":56,"value":1457},{"type":56,"value":269},{"type":50,"tag":75,"props":1918,"children":1920},{"className":1919},[],[1921],{"type":56,"value":1464},{"type":56,"value":1923}," under ",{"type":50,"tag":75,"props":1925,"children":1927},{"className":1926},[],[1928],{"type":56,"value":598},{"type":56,"value":1930},". A common local-Docker failure is a stale ",{"type":50,"tag":75,"props":1932,"children":1934},{"className":1933},[],[1935],{"type":56,"value":1936},"\u002Fresults\u002F\u003Cold_job_id>\u002F...",{"type":56,"value":1938}," path from a different results root.",{"type":50,"tag":59,"props":1940,"children":1941},{},[1942,1947],{"type":50,"tag":462,"props":1943,"children":1944},{},[1945],{"type":56,"value":1946},"Point cloud range mismatch",{"type":56,"value":1948},": If point_cloud_range does not match the actual sensor data extent, detections will be poor or empty.",{"type":50,"tag":59,"props":1950,"children":1951},{},[1952,1957],{"type":50,"tag":462,"props":1953,"children":1954},{},[1955],{"type":56,"value":1956},"Epoch numbering",{"type":56,"value":1958},": PointPillars checkpoint epoch numbers may be offset by 1 from status.json reported epochs.",{"type":50,"tag":59,"props":1960,"children":1961},{},[1962,1967,1969,1975,1977,1983,1984,1990,1991,1997,1998,2004,2006,2012,2014,2020],{"type":50,"tag":462,"props":1963,"children":1964},{},[1965],{"type":56,"value":1966},"Checkpoint selection",{"type":56,"value":1968},": PointPillars training emits checkpoints named like ",{"type":50,"tag":75,"props":1970,"children":1972},{"className":1971},[],[1973],{"type":56,"value":1974},"checkpoint_epoch_1.pth",{"type":56,"value":1976},". For evaluation, inference, export, prune, and resume, select the intended checkpoint through the model\u002Fjob checkpoint resolver and pass that exact file to ",{"type":50,"tag":75,"props":1978,"children":1980},{"className":1979},[],[1981],{"type":56,"value":1982},"evaluate.checkpoint",{"type":56,"value":135},{"type":50,"tag":75,"props":1985,"children":1987},{"className":1986},[],[1988],{"type":56,"value":1989},"inference.checkpoint",{"type":56,"value":135},{"type":50,"tag":75,"props":1992,"children":1994},{"className":1993},[],[1995],{"type":56,"value":1996},"export.checkpoint",{"type":56,"value":135},{"type":50,"tag":75,"props":1999,"children":2001},{"className":2000},[],[2002],{"type":56,"value":2003},"prune.model",{"type":56,"value":2005},", or ",{"type":50,"tag":75,"props":2007,"children":2009},{"className":2008},[],[2010],{"type":56,"value":2011},"train.resume_training_checkpoint_path",{"type":56,"value":2013},". Do not guess by taking the newest ",{"type":50,"tag":75,"props":2015,"children":2017},{"className":2016},[],[2018],{"type":56,"value":2019},"model.pth",{"type":56,"value":2021},"; this model does not use that filename.",{"type":50,"tag":59,"props":2023,"children":2024},{},[2025,2030,2032,2038,2040,2046,2048,2054,2056,2061,2063,2069,2071,2077],{"type":50,"tag":462,"props":2026,"children":2027},{},[2028],{"type":56,"value":2029},"Prune\u002Fretrain key",{"type":56,"value":2031},": PointPillars prune writes an encrypted ",{"type":50,"tag":75,"props":2033,"children":2035},{"className":2034},[],[2036],{"type":56,"value":2037},".tlt",{"type":56,"value":2039}," artifact. Keep a non-empty ",{"type":50,"tag":75,"props":2041,"children":2043},{"className":2042},[],[2044],{"type":56,"value":2045},"key",{"type":56,"value":2047}," in the prune and retrain specs; the packaged templates use the TAO default ",{"type":50,"tag":75,"props":2049,"children":2051},{"className":2050},[],[2052],{"type":56,"value":2053},"tlt_encode",{"type":56,"value":2055},". If ",{"type":50,"tag":75,"props":2057,"children":2059},{"className":2058},[],[2060],{"type":56,"value":2045},{"type":56,"value":2062}," is omitted or ",{"type":50,"tag":75,"props":2064,"children":2066},{"className":2065},[],[2067],{"type":56,"value":2068},"null",{"type":56,"value":2070},", the toolkit can still exit with a container success code while logging a passphrase error and creating an empty ",{"type":50,"tag":75,"props":2072,"children":2074},{"className":2073},[],[2075],{"type":56,"value":2076},"pruned_0.1.tlt",{"type":56,"value":2078},". Always verify the pruned model is nonzero before using it for retrain.",{"type":50,"tag":59,"props":2080,"children":2081},{},[2082,2087,2089,2095,2097,2103],{"type":50,"tag":462,"props":2083,"children":2084},{},[2085],{"type":56,"value":2086},"Status files matter",{"type":56,"value":2088},": Some PointPillars failures can be followed by ",{"type":50,"tag":75,"props":2090,"children":2092},{"className":2091},[],[2093],{"type":56,"value":2094},"Execution status: PASS",{"type":56,"value":2096}," in the entrypoint footer and a Docker exit code of 0. Check ",{"type":50,"tag":75,"props":2098,"children":2100},{"className":2099},[],[2101],{"type":56,"value":2102},"results_dir\u002Fstatus.json",{"type":56,"value":2104}," and the expected artifact before marking an action as passed.",{"type":50,"tag":59,"props":2106,"children":2107},{},[2108,2113,2115,2121,2123,2129,2131,2137,2139,2145],{"type":50,"tag":462,"props":2109,"children":2110},{},[2111],{"type":56,"value":2112},"Local results_dir wiring",{"type":56,"value":2114},": For direct local-Docker specs, set the top-level ",{"type":50,"tag":75,"props":2116,"children":2118},{"className":2117},[],[2119],{"type":56,"value":2120},"results_dir",{"type":56,"value":2122}," as well as any action-specific ",{"type":50,"tag":75,"props":2124,"children":2126},{"className":2125},[],[2127],{"type":56,"value":2128},"*.results_dir",{"type":56,"value":2130}," field. If only ",{"type":50,"tag":75,"props":2132,"children":2134},{"className":2133},[],[2135],{"type":56,"value":2136},"evaluate.results_dir",{"type":56,"value":2138}," is set and the top-level field is left blank, evaluate can try to write under ",{"type":50,"tag":75,"props":2140,"children":2142},{"className":2141},[],[2143],{"type":56,"value":2144},"\u002Fopt\u002Fnvidia\u002Feval",{"type":56,"value":2146}," and then still print the generic PASS footer. Treat that as a failed action unless the expected result directory and status\u002Fartifact files exist.",{"type":50,"tag":203,"props":2148,"children":2150},{"id":2149},"spec-param-parent-model-inference",[2151],{"type":56,"value":2152},"Spec Param \u002F Parent Model Inference",{"type":50,"tag":59,"props":2154,"children":2155},{},[2156,2158,2164,2166,2172,2174,2180],{"type":56,"value":2157},"Model-specific inference mappings belong in this MD file, not in ",{"type":50,"tag":75,"props":2159,"children":2161},{"className":2160},[],[2162],{"type":56,"value":2163},"config.json",{"type":56,"value":2165},". Generated runners should read this section and apply the mappings with SDK helpers before ",{"type":50,"tag":75,"props":2167,"children":2169},{"className":2168},[],[2170],{"type":56,"value":2171},"create_job()",{"type":56,"value":2173},". This mirrors the old microservices ",{"type":50,"tag":75,"props":2175,"children":2177},{"className":2176},[],[2178],{"type":56,"value":2179},"infer_params.py",{"type":56,"value":2181}," flow.",{"type":50,"tag":59,"props":2183,"children":2184},{},[2185,2187,2193],{"type":56,"value":2186},"Inference mappings from TAO Core ",{"type":50,"tag":75,"props":2188,"children":2190},{"className":2189},[],[2191],{"type":56,"value":2192},"pointpillars.config.json",{"type":56,"value":2194},":",{"type":50,"tag":497,"props":2196,"children":2197},{},[2198,2223],{"type":50,"tag":501,"props":2199,"children":2200},{},[2201],{"type":50,"tag":505,"props":2202,"children":2203},{},[2204,2208,2213,2218],{"type":50,"tag":509,"props":2205,"children":2206},{},[2207],{"type":56,"value":513},{"type":50,"tag":509,"props":2209,"children":2210},{},[2211],{"type":56,"value":2212},"Spec Field",{"type":50,"tag":509,"props":2214,"children":2215},{},[2216],{"type":56,"value":2217},"Inference Function",{"type":50,"tag":509,"props":2219,"children":2220},{},[2221],{"type":56,"value":2222},"Meaning",{"type":50,"tag":535,"props":2224,"children":2225},{},[2226,2255,2284,2312,2339,2366,2396,2426,2453,2480,2507,2535,2562,2589,2616,2643,2670,2697,2724,2751,2778,2808,2835],{"type":50,"tag":505,"props":2227,"children":2228},{},[2229,2233,2241,2250],{"type":50,"tag":542,"props":2230,"children":2231},{},[2232],{"type":56,"value":133},{"type":50,"tag":542,"props":2234,"children":2235},{},[2236],{"type":50,"tag":75,"props":2237,"children":2239},{"className":2238},[],[2240],{"type":56,"value":2120},{"type":50,"tag":542,"props":2242,"children":2243},{},[2244],{"type":50,"tag":75,"props":2245,"children":2247},{"className":2246},[],[2248],{"type":56,"value":2249},"output_dir",{"type":50,"tag":542,"props":2251,"children":2252},{},[2253],{"type":56,"value":2254},"current job results directory",{"type":50,"tag":505,"props":2256,"children":2257},{},[2258,2262,2270,2279],{"type":50,"tag":542,"props":2259,"children":2260},{},[2261],{"type":56,"value":88},{"type":50,"tag":542,"props":2263,"children":2264},{},[2265],{"type":50,"tag":75,"props":2266,"children":2268},{"className":2267},[],[2269],{"type":56,"value":1982},{"type":50,"tag":542,"props":2271,"children":2272},{},[2273],{"type":50,"tag":75,"props":2274,"children":2276},{"className":2275},[],[2277],{"type":56,"value":2278},"parent_model",{"type":50,"tag":542,"props":2280,"children":2281},{},[2282],{"type":56,"value":2283},"model file inferred from the parent job results folder",{"type":50,"tag":505,"props":2285,"children":2286},{},[2287,2291,2299,2307],{"type":50,"tag":542,"props":2288,"children":2289},{},[2290],{"type":56,"value":88},{"type":50,"tag":542,"props":2292,"children":2293},{},[2294],{"type":50,"tag":75,"props":2295,"children":2297},{"className":2296},[],[2298],{"type":56,"value":2045},{"type":50,"tag":542,"props":2300,"children":2301},{},[2302],{"type":50,"tag":75,"props":2303,"children":2305},{"className":2304},[],[2306],{"type":56,"value":2045},{"type":50,"tag":542,"props":2308,"children":2309},{},[2310],{"type":56,"value":2311},"encryption key",{"type":50,"tag":505,"props":2313,"children":2314},{},[2315,2319,2327,2335],{"type":50,"tag":542,"props":2316,"children":2317},{},[2318],{"type":56,"value":88},{"type":50,"tag":542,"props":2320,"children":2321},{},[2322],{"type":50,"tag":75,"props":2323,"children":2325},{"className":2324},[],[2326],{"type":56,"value":2120},{"type":50,"tag":542,"props":2328,"children":2329},{},[2330],{"type":50,"tag":75,"props":2331,"children":2333},{"className":2332},[],[2334],{"type":56,"value":2249},{"type":50,"tag":542,"props":2336,"children":2337},{},[2338],{"type":56,"value":2254},{"type":50,"tag":505,"props":2340,"children":2341},{},[2342,2346,2354,2362],{"type":50,"tag":542,"props":2343,"children":2344},{},[2345],{"type":56,"value":160},{"type":50,"tag":542,"props":2347,"children":2348},{},[2349],{"type":50,"tag":75,"props":2350,"children":2352},{"className":2351},[],[2353],{"type":56,"value":1996},{"type":50,"tag":542,"props":2355,"children":2356},{},[2357],{"type":50,"tag":75,"props":2358,"children":2360},{"className":2359},[],[2361],{"type":56,"value":2278},{"type":50,"tag":542,"props":2363,"children":2364},{},[2365],{"type":56,"value":2283},{"type":50,"tag":505,"props":2367,"children":2368},{},[2369,2373,2382,2391],{"type":50,"tag":542,"props":2370,"children":2371},{},[2372],{"type":56,"value":160},{"type":50,"tag":542,"props":2374,"children":2375},{},[2376],{"type":50,"tag":75,"props":2377,"children":2379},{"className":2378},[],[2380],{"type":56,"value":2381},"export.onnx_file",{"type":50,"tag":542,"props":2383,"children":2384},{},[2385],{"type":50,"tag":75,"props":2386,"children":2388},{"className":2387},[],[2389],{"type":56,"value":2390},"create_onnx_file",{"type":50,"tag":542,"props":2392,"children":2393},{},[2394],{"type":56,"value":2395},"output ONNX path",{"type":50,"tag":505,"props":2397,"children":2398},{},[2399,2403,2412,2421],{"type":50,"tag":542,"props":2400,"children":2401},{},[2402],{"type":56,"value":160},{"type":50,"tag":542,"props":2404,"children":2405},{},[2406],{"type":50,"tag":75,"props":2407,"children":2409},{"className":2408},[],[2410],{"type":56,"value":2411},"export.save_engine",{"type":50,"tag":542,"props":2413,"children":2414},{},[2415],{"type":50,"tag":75,"props":2416,"children":2418},{"className":2417},[],[2419],{"type":56,"value":2420},"create_engine_file",{"type":50,"tag":542,"props":2422,"children":2423},{},[2424],{"type":56,"value":2425},"output TensorRT engine path",{"type":50,"tag":505,"props":2427,"children":2428},{},[2429,2433,2441,2449],{"type":50,"tag":542,"props":2430,"children":2431},{},[2432],{"type":56,"value":160},{"type":50,"tag":542,"props":2434,"children":2435},{},[2436],{"type":50,"tag":75,"props":2437,"children":2439},{"className":2438},[],[2440],{"type":56,"value":2045},{"type":50,"tag":542,"props":2442,"children":2443},{},[2444],{"type":50,"tag":75,"props":2445,"children":2447},{"className":2446},[],[2448],{"type":56,"value":2045},{"type":50,"tag":542,"props":2450,"children":2451},{},[2452],{"type":56,"value":2311},{"type":50,"tag":505,"props":2454,"children":2455},{},[2456,2460,2468,2476],{"type":50,"tag":542,"props":2457,"children":2458},{},[2459],{"type":56,"value":160},{"type":50,"tag":542,"props":2461,"children":2462},{},[2463],{"type":50,"tag":75,"props":2464,"children":2466},{"className":2465},[],[2467],{"type":56,"value":2120},{"type":50,"tag":542,"props":2469,"children":2470},{},[2471],{"type":50,"tag":75,"props":2472,"children":2474},{"className":2473},[],[2475],{"type":56,"value":2249},{"type":50,"tag":542,"props":2477,"children":2478},{},[2479],{"type":56,"value":2254},{"type":50,"tag":505,"props":2481,"children":2482},{},[2483,2487,2495,2503],{"type":50,"tag":542,"props":2484,"children":2485},{},[2486],{"type":56,"value":96},{"type":50,"tag":542,"props":2488,"children":2489},{},[2490],{"type":50,"tag":75,"props":2491,"children":2493},{"className":2492},[],[2494],{"type":56,"value":1989},{"type":50,"tag":542,"props":2496,"children":2497},{},[2498],{"type":50,"tag":75,"props":2499,"children":2501},{"className":2500},[],[2502],{"type":56,"value":2278},{"type":50,"tag":542,"props":2504,"children":2505},{},[2506],{"type":56,"value":2283},{"type":50,"tag":505,"props":2508,"children":2509},{},[2510,2514,2523,2531],{"type":50,"tag":542,"props":2511,"children":2512},{},[2513],{"type":56,"value":96},{"type":50,"tag":542,"props":2515,"children":2516},{},[2517],{"type":50,"tag":75,"props":2518,"children":2520},{"className":2519},[],[2521],{"type":56,"value":2522},"inference.trt_engine",{"type":50,"tag":542,"props":2524,"children":2525},{},[2526],{"type":50,"tag":75,"props":2527,"children":2529},{"className":2528},[],[2530],{"type":56,"value":2278},{"type":50,"tag":542,"props":2532,"children":2533},{},[2534],{"type":56,"value":2283},{"type":50,"tag":505,"props":2536,"children":2537},{},[2538,2542,2550,2558],{"type":50,"tag":542,"props":2539,"children":2540},{},[2541],{"type":56,"value":96},{"type":50,"tag":542,"props":2543,"children":2544},{},[2545],{"type":50,"tag":75,"props":2546,"children":2548},{"className":2547},[],[2549],{"type":56,"value":2045},{"type":50,"tag":542,"props":2551,"children":2552},{},[2553],{"type":50,"tag":75,"props":2554,"children":2556},{"className":2555},[],[2557],{"type":56,"value":2045},{"type":50,"tag":542,"props":2559,"children":2560},{},[2561],{"type":56,"value":2311},{"type":50,"tag":505,"props":2563,"children":2564},{},[2565,2569,2577,2585],{"type":50,"tag":542,"props":2566,"children":2567},{},[2568],{"type":56,"value":96},{"type":50,"tag":542,"props":2570,"children":2571},{},[2572],{"type":50,"tag":75,"props":2573,"children":2575},{"className":2574},[],[2576],{"type":56,"value":2120},{"type":50,"tag":542,"props":2578,"children":2579},{},[2580],{"type":50,"tag":75,"props":2581,"children":2583},{"className":2582},[],[2584],{"type":56,"value":2249},{"type":50,"tag":542,"props":2586,"children":2587},{},[2588],{"type":56,"value":2254},{"type":50,"tag":505,"props":2590,"children":2591},{},[2592,2596,2604,2612],{"type":50,"tag":542,"props":2593,"children":2594},{},[2595],{"type":56,"value":168},{"type":50,"tag":542,"props":2597,"children":2598},{},[2599],{"type":50,"tag":75,"props":2600,"children":2602},{"className":2601},[],[2603],{"type":56,"value":2045},{"type":50,"tag":542,"props":2605,"children":2606},{},[2607],{"type":50,"tag":75,"props":2608,"children":2610},{"className":2609},[],[2611],{"type":56,"value":2045},{"type":50,"tag":542,"props":2613,"children":2614},{},[2615],{"type":56,"value":2311},{"type":50,"tag":505,"props":2617,"children":2618},{},[2619,2623,2631,2639],{"type":50,"tag":542,"props":2620,"children":2621},{},[2622],{"type":56,"value":168},{"type":50,"tag":542,"props":2624,"children":2625},{},[2626],{"type":50,"tag":75,"props":2627,"children":2629},{"className":2628},[],[2630],{"type":56,"value":2003},{"type":50,"tag":542,"props":2632,"children":2633},{},[2634],{"type":50,"tag":75,"props":2635,"children":2637},{"className":2636},[],[2638],{"type":56,"value":2278},{"type":50,"tag":542,"props":2640,"children":2641},{},[2642],{"type":56,"value":2283},{"type":50,"tag":505,"props":2644,"children":2645},{},[2646,2650,2658,2666],{"type":50,"tag":542,"props":2647,"children":2648},{},[2649],{"type":56,"value":168},{"type":50,"tag":542,"props":2651,"children":2652},{},[2653],{"type":50,"tag":75,"props":2654,"children":2656},{"className":2655},[],[2657],{"type":56,"value":2120},{"type":50,"tag":542,"props":2659,"children":2660},{},[2661],{"type":50,"tag":75,"props":2662,"children":2664},{"className":2663},[],[2665],{"type":56,"value":2249},{"type":50,"tag":542,"props":2667,"children":2668},{},[2669],{"type":56,"value":2254},{"type":50,"tag":505,"props":2671,"children":2672},{},[2673,2677,2685,2693],{"type":50,"tag":542,"props":2674,"children":2675},{},[2676],{"type":56,"value":183},{"type":50,"tag":542,"props":2678,"children":2679},{},[2680],{"type":50,"tag":75,"props":2681,"children":2683},{"className":2682},[],[2684],{"type":56,"value":2045},{"type":50,"tag":542,"props":2686,"children":2687},{},[2688],{"type":50,"tag":75,"props":2689,"children":2691},{"className":2690},[],[2692],{"type":56,"value":2045},{"type":50,"tag":542,"props":2694,"children":2695},{},[2696],{"type":56,"value":2311},{"type":50,"tag":505,"props":2698,"children":2699},{},[2700,2704,2712,2720],{"type":50,"tag":542,"props":2701,"children":2702},{},[2703],{"type":56,"value":183},{"type":50,"tag":542,"props":2705,"children":2706},{},[2707],{"type":50,"tag":75,"props":2708,"children":2710},{"className":2709},[],[2711],{"type":56,"value":2120},{"type":50,"tag":542,"props":2713,"children":2714},{},[2715],{"type":50,"tag":75,"props":2716,"children":2718},{"className":2717},[],[2719],{"type":56,"value":2249},{"type":50,"tag":542,"props":2721,"children":2722},{},[2723],{"type":56,"value":2254},{"type":50,"tag":505,"props":2725,"children":2726},{},[2727,2731,2739,2747],{"type":50,"tag":542,"props":2728,"children":2729},{},[2730],{"type":56,"value":183},{"type":50,"tag":542,"props":2732,"children":2733},{},[2734],{"type":50,"tag":75,"props":2735,"children":2737},{"className":2736},[],[2738],{"type":56,"value":199},{"type":50,"tag":542,"props":2740,"children":2741},{},[2742],{"type":50,"tag":75,"props":2743,"children":2745},{"className":2744},[],[2746],{"type":56,"value":2278},{"type":50,"tag":542,"props":2748,"children":2749},{},[2750],{"type":56,"value":2283},{"type":50,"tag":505,"props":2752,"children":2753},{},[2754,2758,2766,2774],{"type":50,"tag":542,"props":2755,"children":2756},{},[2757],{"type":56,"value":141},{"type":50,"tag":542,"props":2759,"children":2760},{},[2761],{"type":50,"tag":75,"props":2762,"children":2764},{"className":2763},[],[2765],{"type":56,"value":2045},{"type":50,"tag":542,"props":2767,"children":2768},{},[2769],{"type":50,"tag":75,"props":2770,"children":2772},{"className":2771},[],[2773],{"type":56,"value":2045},{"type":50,"tag":542,"props":2775,"children":2776},{},[2777],{"type":56,"value":2311},{"type":50,"tag":505,"props":2779,"children":2780},{},[2781,2785,2794,2803],{"type":50,"tag":542,"props":2782,"children":2783},{},[2784],{"type":56,"value":141},{"type":50,"tag":542,"props":2786,"children":2787},{},[2788],{"type":50,"tag":75,"props":2789,"children":2791},{"className":2790},[],[2792],{"type":56,"value":2793},"model.pretrained_model_path",{"type":50,"tag":542,"props":2795,"children":2796},{},[2797],{"type":50,"tag":75,"props":2798,"children":2800},{"className":2799},[],[2801],{"type":56,"value":2802},"ptm_if_no_resume_model",{"type":50,"tag":542,"props":2804,"children":2805},{},[2806],{"type":56,"value":2807},"PTM when no resume checkpoint exists",{"type":50,"tag":505,"props":2809,"children":2810},{},[2811,2815,2823,2831],{"type":50,"tag":542,"props":2812,"children":2813},{},[2814],{"type":56,"value":141},{"type":50,"tag":542,"props":2816,"children":2817},{},[2818],{"type":50,"tag":75,"props":2819,"children":2821},{"className":2820},[],[2822],{"type":56,"value":2120},{"type":50,"tag":542,"props":2824,"children":2825},{},[2826],{"type":50,"tag":75,"props":2827,"children":2829},{"className":2828},[],[2830],{"type":56,"value":2249},{"type":50,"tag":542,"props":2832,"children":2833},{},[2834],{"type":56,"value":2254},{"type":50,"tag":505,"props":2836,"children":2837},{},[2838,2842,2850,2859],{"type":50,"tag":542,"props":2839,"children":2840},{},[2841],{"type":56,"value":141},{"type":50,"tag":542,"props":2843,"children":2844},{},[2845],{"type":50,"tag":75,"props":2846,"children":2848},{"className":2847},[],[2849],{"type":56,"value":2011},{"type":50,"tag":542,"props":2851,"children":2852},{},[2853],{"type":50,"tag":75,"props":2854,"children":2856},{"className":2855},[],[2857],{"type":56,"value":2858},"resume_model",{"type":50,"tag":542,"props":2860,"children":2861},{},[2862],{"type":56,"value":2863},"model file inferred from the current job results folder",{"type":50,"tag":59,"props":2865,"children":2866},{},[2867,2869,2874,2876,2882,2884,2890,2892,2897],{"type":56,"value":2868},"For ",{"type":50,"tag":75,"props":2870,"children":2872},{"className":2871},[],[2873],{"type":56,"value":2278},{"type":56,"value":2875}," or ",{"type":50,"tag":75,"props":2877,"children":2879},{"className":2878},[],[2880],{"type":56,"value":2881},"parent_model_folder",{"type":56,"value":2883},", pass the upstream train\u002Fexport\u002FAutoML child job id as ",{"type":50,"tag":75,"props":2885,"children":2887},{"className":2886},[],[2888],{"type":56,"value":2889},"parent_job_id",{"type":56,"value":2891},". 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":50,"tag":75,"props":2893,"children":2895},{"className":2894},[],[2896],{"type":56,"value":2163},{"type":56,"value":2898}," and do not patch generated runner scripts to guess checkpoint paths.",{"type":50,"tag":203,"props":2900,"children":2902},{"id":2901},"deployment",[2903],{"type":56,"value":2904},"Deployment",{"type":50,"tag":454,"props":2906,"children":2907},{},[2908],{"type":50,"tag":458,"props":2909,"children":2910},{},[2911],{"type":50,"tag":2912,"props":2913,"children":2914},"a",{"href":104},[2915],{"type":56,"value":2916},"tao-deploy-pointpillars",{"type":50,"tag":2918,"props":2919,"children":2920},"style",{},[2921],{"type":56,"value":2922},"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":2924,"total":3026},[2925,2942,2954,2968,2980,2997,3012],{"slug":2926,"name":2926,"fn":2927,"description":2928,"org":2929,"tags":2930,"stars":23,"repoUrl":24,"updatedAt":2941},"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},[2931,2934,2937,2938],{"name":2932,"slug":2933,"type":15},"Data Analysis","data-analysis",{"name":2935,"slug":2936,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":2939,"slug":2940,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":2943,"name":2943,"fn":2944,"description":2945,"org":2946,"tags":2947,"stars":23,"repoUrl":24,"updatedAt":2953},"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},[2948,2949,2952],{"name":2904,"slug":2901,"type":15},{"name":2950,"slug":2951,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":2955,"name":2955,"fn":2956,"description":2957,"org":2958,"tags":2959,"stars":23,"repoUrl":24,"updatedAt":2967},"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},[2960,2963,2964],{"name":2961,"slug":2962,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":2965,"slug":2966,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":2969,"name":2969,"fn":2970,"description":2971,"org":2972,"tags":2973,"stars":23,"repoUrl":24,"updatedAt":2979},"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},[2974,2975,2976],{"name":2932,"slug":2933,"type":15},{"name":9,"slug":8,"type":15},{"name":2977,"slug":2978,"type":15},"Testing","testing","2026-07-17T05:29:03.913266",{"slug":2981,"name":2981,"fn":2982,"description":2983,"org":2984,"tags":2985,"stars":23,"repoUrl":24,"updatedAt":2996},"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},[2986,2989,2992,2993],{"name":2987,"slug":2988,"type":15},"Automation","automation",{"name":2990,"slug":2991,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":2994,"slug":2995,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":2998,"name":2998,"fn":2999,"description":3000,"org":3001,"tags":3002,"stars":23,"repoUrl":24,"updatedAt":3011},"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},[3003,3004,3007,3008],{"name":2904,"slug":2901,"type":15},{"name":3005,"slug":3006,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":3009,"slug":3010,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":3013,"name":3013,"fn":3014,"description":3015,"org":3016,"tags":3017,"stars":23,"repoUrl":24,"updatedAt":3025},"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},[3018,3019,3022],{"name":9,"slug":8,"type":15},{"name":3020,"slug":3021,"type":15},"Quantum Computing","quantum-computing",{"name":3023,"slug":3024,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305,{"items":3028,"total":3176},[3029,3047,3062,3073,3085,3099,3112,3124,3135,3144,3158,3167],{"slug":3030,"name":3030,"fn":3031,"description":3032,"org":3033,"tags":3034,"stars":3044,"repoUrl":3045,"updatedAt":3046},"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},[3035,3038,3041],{"name":3036,"slug":3037,"type":15},"Documentation","documentation",{"name":3039,"slug":3040,"type":15},"MCP","mcp",{"name":3042,"slug":3043,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":3048,"name":3048,"fn":3049,"description":3050,"org":3051,"tags":3052,"stars":3059,"repoUrl":3060,"updatedAt":3061},"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},[3053,3056,3057],{"name":3054,"slug":3055,"type":15},"Containers","containers",{"name":2904,"slug":2901,"type":15},{"name":3058,"slug":868,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":3063,"name":3063,"fn":3064,"description":3065,"org":3066,"tags":3067,"stars":3059,"repoUrl":3060,"updatedAt":3072},"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},[3068,3071],{"name":3069,"slug":3070,"type":15},"CI\u002FCD","ci-cd",{"name":2904,"slug":2901,"type":15},"2026-07-14T05:25:59.97109",{"slug":3074,"name":3074,"fn":3075,"description":3076,"org":3077,"tags":3078,"stars":3059,"repoUrl":3060,"updatedAt":3084},"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},[3079,3080,3081],{"name":3069,"slug":3070,"type":15},{"name":2904,"slug":2901,"type":15},{"name":3082,"slug":3083,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":3086,"name":3086,"fn":3087,"description":3088,"org":3089,"tags":3090,"stars":3059,"repoUrl":3060,"updatedAt":3098},"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},[3091,3094,3095],{"name":3092,"slug":3093,"type":15},"Debugging","debugging",{"name":3082,"slug":3083,"type":15},{"name":3096,"slug":3097,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":3100,"name":3100,"fn":3101,"description":3102,"org":3103,"tags":3104,"stars":3059,"repoUrl":3060,"updatedAt":3111},"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},[3105,3108],{"name":3106,"slug":3107,"type":15},"Best Practices","best-practices",{"name":3109,"slug":3110,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":3113,"name":3113,"fn":3114,"description":3115,"org":3116,"tags":3117,"stars":3059,"repoUrl":3060,"updatedAt":3123},"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},[3118,3119,3122],{"name":17,"slug":18,"type":15},{"name":3120,"slug":3121,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":3125,"name":3125,"fn":3126,"description":3127,"org":3128,"tags":3129,"stars":3059,"repoUrl":3060,"updatedAt":3134},"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},[3130,3133],{"name":3131,"slug":3132,"type":15},"QA","qa",{"name":2977,"slug":2978,"type":15},"2026-07-14T05:25:53.673039",{"slug":3136,"name":3136,"fn":3137,"description":3138,"org":3139,"tags":3140,"stars":3059,"repoUrl":3060,"updatedAt":3143},"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},[3141,3142],{"name":2904,"slug":2901,"type":15},{"name":2950,"slug":2951,"type":15},"2026-07-14T05:25:49.362534",{"slug":3145,"name":3145,"fn":3146,"description":3147,"org":3148,"tags":3149,"stars":3059,"repoUrl":3060,"updatedAt":3157},"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},[3150,3153,3154],{"name":3151,"slug":3152,"type":15},"Code Review","code-review",{"name":3082,"slug":3083,"type":15},{"name":3155,"slug":3156,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":3159,"name":3159,"fn":3160,"description":3161,"org":3162,"tags":3163,"stars":3059,"repoUrl":3060,"updatedAt":3166},"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},[3164,3165],{"name":3131,"slug":3132,"type":15},{"name":2977,"slug":2978,"type":15},"2026-07-14T05:25:54.928983",{"slug":3168,"name":3168,"fn":3169,"description":3170,"org":3171,"tags":3172,"stars":3059,"repoUrl":3060,"updatedAt":3175},"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},[3173,3174],{"name":2987,"slug":2988,"type":15},{"name":3069,"slug":3070,"type":15},"2026-07-30T05:29:03.275638",496]