Tao Train Mask Auto Encoder

NVIDIAOfficialCreated by NVIDIA Corporation

Skill
Tool integration
Data and AI
Apache-2.0
Medium risk

From the author

Masked Auto-Encoder (MAE) for self-supervised pretraining and fine-tuning. Masks random patches and reconstructs them to learn visual representations; supports pretrain and finetune stages. Use when training, evaluating, exporting, or running inference for a TAO MAE backbone. Trigger phrases include "pretrain MAE", "self-supervised vision pretraining", "Masked Autoencoder", "Mask Auto-Encoder", "MAE fine-tune".

— the description by NVIDIA Corporation

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The tao-train-mask-auto-encoder skill provides a framework for Masked Auto-Encoder (MAE) operations. It supports self-supervised pretraining by masking random image patches and reconstructing them to learn visual representations, as well as fine-tuning for specific classification tasks. The skill integrates with the TAO Toolkit to manage training, evaluation, inference, and export workflows, including AutoML capabilities for fine-tuning stages.

What it does

Performs self-supervised pretraining by masking and reconstructing image patches.
Supports fine-tuning of MAE backbones for image classification tasks.
Enables evaluation, inference, and model export workflows.
Integrates with TAO AutoML for automated hyperparameter optimization during training.
Provides spec templates for configuring training, evaluation, and export actions.

When to use it

When you need to perform self-supervised vision pretraining on image datasets.
When fine-tuning a pretrained MAE model for a specific classification task.
When you need to export a trained MAE model for deployment.
When you require automated hyperparameter optimization for MAE training runs.

Advantages

Enables learning visual representations without requiring labeled data during pretraining.
Supports flexible training stages including pretraining and fine-tuning.
Automates hyperparameter tuning through integrated AutoML policies.
Provides structured spec templates to ensure consistent configuration across actions.

Usage examples

“Pretrain an MAE model using my training dataset.”: The agent initiates the MAE pretraining stage using the provided dataset path and default training parameters.
“Fine-tune the MAE model for a 10-class classification task.”: The agent configures the fine-tuning stage with the specified class count and loads the pretrained weights.
“Export my trained MAE model to ONNX format.”: The agent executes the export action, generating an ONNX artifact based on the provided checkpoint and configuration.

Install

Download the skill as a ZIP:
https://mp-staging.autogineer.com/agent-skills/nvidia/tao-train-mask-auto-encoder/download

Unzip it into the cross-client skills folder of your project:

mkdir -p .agents/skills && unzip tao-train-mask-auto-encoder.zip -d .agents/skills

Documentation by NVIDIA Corporation

MAE

Standalone install? If this session was not initialized by the TAO skill bank plugin, run the tao-setup skill first (host preflight, credentials, cross-skill discovery).

MAE (Masked Autoencoder) for self-supervised pretraining and fine-tuning. Masks random patches and reconstructs them to learn visual representations. Supports pretrain and finetune stages.

Set train.pretrained_model_path for pretrained MAE weights when fine-tuning.

For TAO Deploy TensorRT actions (gen_trt_engine), read references/tao-deploy-mask-auto-encoder.md first. Deploy spec templates live in this skill's references/ folder with the spec_template_deploy_*.yaml prefix.

The parent PyTorch mae CLI supports train, evaluate, inference, and export. Build TensorRT engines through the deploy workflow, not the model skill.

Dataclass Schemas

Generated TAO Core schemas are packaged in schemas/<action>.schema.json, with schemas/manifest.json listing available actions. Each generated schema also emits references/spec_template_<action>.yaml from the schema top-level default field. AutoML enablement is declared at the model layer in references/skill_info.yaml via automl_enabled. Runnable AutoML for an action requires schemas/<action>.schema.json and references/spec_template_<action>.yaml to exist and parse. Use the packaged selected-action schema for automl_default_parameters, automl_disabled_parameters, defaults, min/max bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect ~/tao-core at runtime; maintainers regenerate schemas/templates before packaging the skill bank.

Train Action Policy

This model is AutoML-enabled at the model layer. Before handling any train-stage request, read references/skill_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 / 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/train.schema.json and references/spec_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/application overrides for datasets, specs, output directories, GPU/platform settings, parent checkpoints, and automl_policy. Use direct model training only when automl_policy: off or the packaged train schema/template is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.

Non-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.

Training Requirements

Dataset type: image_classification
Formats: ssl
Accepted dataset intents: training, evaluation, testing
Pretraining monitoring metric: train_loss, minimized.
AutoML metric contract: for fine-tuning, use ACC_all with maximize direction. Fine-tune validation and the packaged evaluate action emit ACC_all (plus val_loss); they do not emit train_loss. Use ACC_all for trial and final checkpoint selection when train.stage: finetune.

Per-Action Dataset Requirements

Action	Spec Key	Source	Files	List?
train	dataset.train_data_sources	train_datasets	images_train.tar.gz	No
train	dataset.val_data_sources	eval_dataset	images_val.tar.gz	No
evaluate	dataset.val_data_sources	eval_dataset	images_val.tar.gz	No
inference	dataset.test_data_sources	inference_dataset	images_test.tar.gz	No

For SDK/app job inputs, the images_*.tar.gz archives are uploaded as the action inputs. For direct local Docker runs against host-mounted data, extract the archives first and point dataset.train_data_sources, dataset.val_data_sources, and dataset.test_data_sources at the extracted images_train, images_val, and images_test folders. Passing a local tar path directly to the MAE CLI can produce a zero-sample dataloader because the local dataloader does not unpack that archive path.

Typical Spec Overrides

Data 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.

S3_TRAIN = "s3://bucket/data/train"
S3_EVAL = "s3://bucket/data/eval"

train (mandatory data sources):

{
    "dataset.train_data_sources": f"{S3_TRAIN}/images_train.tar.gz",
    "dataset.val_data_sources": f"{S3_EVAL}/images_val.tar.gz",
    "train.num_epochs": 10,
    "train.optim.lr": 2e-4,
}

evaluate (mandatory data sources):

{
    "dataset.val_data_sources": f"{S3_EVAL}/images_val.tar.gz",
    "evaluate.checkpoint": "<selected train/AutoML checkpoint>",
    "train.stage": "finetune",
}

inference (mandatory data sources):

{
    "dataset.test_data_sources": f"{S3_EVAL}/images_test.tar.gz",
    "inference.checkpoint": "<selected train/AutoML checkpoint>",
    "train.stage": "finetune",
}

Eval Dataset

Optional. Pretraining does not need eval data. Fine-tuning optionally uses val set.

Important Parameters

train.stage: Training stage. Options: pretrain, finetune. Pretrain learns representations via masking. Finetune adds a classification head.
model.arch: Architecture. Default convnextv2_base. For local smoke AutoML, use convnextv2_atto rather than unsupported names such as vit_tiny_patch16. Supported families include vit_base_patch16 and larger ViTs, ConvNeXtV2 atto/femto/pico/nano/tiny/base/large/huge, and Hiera tiny/small/base/large/huge.
model.num_classes: Number of classes for fine-tuning. Default 1000 (ImageNet). Only relevant in finetune stage.
model.mask_ratio: Fraction of patches to mask during pretraining. Typically 0.75.
model.norm_pix_loss: Whether to normalize pixel values in reconstruction loss.
dataset.augmentation.input_size: Keep the local smoke profile at 224 for ConvNeXtV2 MAE. Reducing to 112 can make the MAE mask grid incompatible with feature-map dimensions.
MAE does not expose a dataset.workers spec field. Do not add it to smoke-test overrides; Hydra rejects unknown dataset keys before training.
train.optim.lr: Learning rate. Default 2e-4.
dataset.augmentation: Augmentation settings including mixup, cutmix for fine-tuning.

Multi-GPU / Multi-Node

Launch method: Lightning-managed (single python process, Lightning spawns workers).

Spec Key	Description	Default
`train.num_gpus`	Number of GPUs	1
`train.gpu_ids`	GPU device indices	[0]
`train.num_nodes`	Number of nodes	1
`train.distributed_strategy`	`ddp` or `fsdp`	`ddp`

ddp uses find_unused_parameters=True
fsdp forces FP16
Multi-GPU strongly recommended for pretraining (large batch sizes needed)

Multi-node env vars (set by orchestrator): WORLD_SIZE, NODE_RANK, MASTER_ADDR, MASTER_PORT, NUM_GPU_PER_NODE.

Hardware

Minimum 2 GPU(s), recommended 8 GPU(s). 24GB+ (A100 recommended) VRAM per GPU. MAE pretraining benefits from large batch sizes across many GPUs. Fine-tuning is more modest in resource requirements.

Error Patterns

Stage mismatch: Ensure train.stage matches your intent (pretrain vs finetune). Fine-tuning without a pretrained_model_path trains from scratch.

Inference with pretrain checkpoints: The MAE predict dataloader raises NotImplementedError for train.stage: pretrain. Use a finetune checkpoint for inference and classification-style evaluation, or restrict a pretrain-only run to train/evaluate/export.

num_classes mismatch (finetune only): Ensure model.num_classes matches your dataset class count when fine-tuning.

Spec Param / Parent Model Inference

Model-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.

Inference mappings from TAO Core mae.config.json:

Action	Spec Field	Inference Function	Meaning
evaluate	`encryption_key`	`key`	encryption key
evaluate	`evaluate.checkpoint`	`parent_model`	model file inferred from the parent job results folder
evaluate	`evaluate.trt_engine`	`parent_model`	model file inferred from the parent job results folder
evaluate	`results_dir`	`output_dir`	current job results directory
export	`encryption_key`	`key`	encryption key
export	`export.checkpoint`	`parent_model`	model file inferred from the parent job results folder
export	`export.onnx_file`	`create_onnx_file`	output ONNX path
export	`results_dir`	`output_dir`	current job results directory
inference	`encryption_key`	`key`	encryption key
inference	`inference.checkpoint`	`parent_model`	model file inferred from the parent job results folder
inference	`inference.trt_engine`	`parent_model`	model file inferred from the parent job results folder
inference	`results_dir`	`output_dir`	current job results directory
train	`encryption_key`	`key`	encryption key
train	`results_dir`	`output_dir`	current job results directory
train	`train.pretrained_model_path`	`ptm_if_no_resume_model`	PTM when no resume checkpoint exists
train	`train.resume_training_checkpoint_path`	`resume_model`	model file inferred from the current job results folder

For parent_model or parent_model_folder, pass the upstream train/export/AutoML 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.

When resolving checkpoints outside the SDK resolver, select the intended epoch/step artifact exactly, for example model_epoch_000_step_00099.pth. Use the convnextv2_atto_latest.pth or other latest symlink only when latest is explicitly requested. Carry train.stage, model.arch, model.num_classes, and export input size forward into evaluate, inference, export, and deploy specs so the checkpoint and ONNX/engine shapes match.

Deployment

tao-deploy-mask-auto-encoder

Files

ATTRIBUTION.mdAdded by AutoGineer951 B
BENCHMARK.md7.0 KiB
LICENSELicence file added by AutoGineer11.2 KiB
SKILL.md11.2 KiB
config/skillspector-baseline.yaml834 B
evals/evals.json846 B
references/skill_info.yaml3.3 KiB
references/spec_template_deploy_gen_trt_engine.yaml750 B
references/spec_template_evaluate.yaml1.7 KiB
references/spec_template_export.yaml1.8 KiB
references/spec_template_gen_trt_engine.yaml1.9 KiB
references/spec_template_inference.yaml1.7 KiB
references/spec_template_train.yaml1.6 KiB
references/tao-deploy-mask-auto-encoder.md3.4 KiB
references/tao-deploy-mask-auto-encoder.skill_info.yaml1.1 KiB
schemas/evaluate.schema.json29.1 KiB
schemas/export.schema.json30.0 KiB
schemas/gen_trt_engine.schema.json31.8 KiB
schemas/inference.schema.json29.1 KiB
schemas/manifest.json7.7 KiB
schemas/train.schema.json26.6 KiB
skill-card.md4.5 KiB
skill.oms.sig8.4 KiB

schemas/train.schema.json

{
  "automl_default_parameters": [
    "train.optim.weight_decay",
    "train.optim.backbone_multiplier",
    "train.optim.momentum",
    "train.optim.lr",
    "train.optim.warmup_epochs"
  ],
  "automl_disabled_parameters": [
    "train.cudnn",
    "train.gpu_ids",
    "wandb.tags",
    "train.optim.milestones",
    "evaluate",
    "inference",
    "train",
    "dataset.augmentation",
    "gen_trt_engine",
    "gen_trt_engine.tensorrt.layers_precision",
    "dataset",
    "gen_trt_engine.tensorrt",
    "model",
    "train.freeze",
    "evaluate.gpu_ids",
    "train.optim",
    "dataset.augmentation.std",
    "export",
    "wandb",
    "inference.gpu_ids",
    "dataset.augmentation.mean"
  ],
  "default": {
    "dataset": {
      "augmentation": {
        "auto_aug": "rand-m9-mstd0.5-inc1",
        "color_jitter": 0.0,
        "cutmix": 1.0,
        "hflip": 0.5,
        "input_size": 224,
        "interpolation": "bilinear",
        "max_ratio": 1.33,
        "max_scale": 2.0,
        "mean": [
          0.485,
          0.456,
          0.406
        ],
        "min_ratio": 0.75,
        "min_scale": 0.1,
        "mixup": 0.8,
        "mixup_mode": "batch",
        "mixup_prob": 1.0,
        "mixup_switch_prob": 0.5,
        "re_prob": 0.0,
        "smoothing": 0.1,
        "std": [
          0.229,
          0.224,
          0.225
        ]
      },
      "batch_size": 1,
      "num_workers_per_gpu": 2,
      "train_data_sources": "",
      "val_data_sources": ""
    },
    "encryption_key": "",
    "model": {
      "arch": "convnextv2_base",
      "decoder_depth": 1,
      "decoder_embed_dim": 512,
      "drop_path_rate": 0.1,
      "global_pool": true,
      "num_classes": 1
    },
    "model_name": "",
    "results_dir": "",
    "train": {
      "accum_grad_batches": 1,
      "checkpoint_interval": 1,
      "checkpoint_interval_unit": "epoch",
      "cudnn": {
        "benchmark": false,
        "deterministic": true
      },
      "distributed_strategy": "ddp",
      "freeze": [],
      "gpu_ids": [
        0
      ],
      "mask_ratio": 0.75,
      "norm_pix_loss": true,
      "num_epochs": 10,
      "num_gpus": 1,
      "num_nodes": 1,
      "optim": {
        "backbone_multiplier": 0.1,
        "gamma": 0.1,
        "layer_decay": 0.75,
        "lr": 0.0002,
        "lr_scheduler": "MultiStep",
        "milestones": [
          88,
          96
        ],
        "momentum": 0.9,
        "monitor_name": "train_loss",
        "type": "AdamW",
        "warmup_epochs": 1,
        "weight_decay": 0.05
      },
      "precision": "fp32",
      "pretrained_model_path": "",
      "results_dir": "",
      "resume_training_checkpoint_path": "",
      "seed": 1234,
      "stage": "pretrain",
      "validation_interval": 1
    },
    "wandb": {
      "enable": true,
      "entity": "",
      "group": "",
      "name": "TAO Toolkit Training",
      "project": "TAO Toolkit",
      "reinit": false,
      "run_id": "",
      "save_code": false,
      "sync_tensorboard": false,
      "tags": [
        "tao-toolkit"
      ]
    }
  },
  "popular": {
    "evaluate": {
      "gpu_ids": [
        0
      ],
      "num_gpus": 1,
      "num_nodes": 1
    },
    "gen_trt_engine": {
      "batch_size": -1,
      "gpu_id": 0,
      "tensorrt": {
        "max_batch_size": 1,
        "min_batch_size": 1,
        "opt_batch_size": 1
      }
    },
    "inference": {
      "gpu_ids": [
        0
      ],
      "num_gpus": 1,
      "num_nodes": 1
    },
    "train": {
      "checkpoint_interval": 1,
      "gpu_ids": [
        0
      ],
      "num_epochs": 10,
      "num_gpus": 1,
      "num_nodes": 1,
      "optim": {
        "backbone_multiplier": 0.1,
        "layer_decay": 0.75,
        "momentum": 0.9,
        "weight_decay": 0.05
      },
      "validation_interval": 1
    }
  },
  "properties": {
    "automl_disabled_parameters": [
      "wandb",
      "dataset",
      "train",
      "model",
      "inference",
      "evaluate",
      "gen_trt_engine",
      "export"
    ],
    "dataset": {
      "automl_disabled_parameters": [
        "dataset.augmentation"
      ],
      "automl_enabled": false,
      "default": {
        "augmentation": {
          "auto_aug": "rand-m9-mstd0.5-inc1",
          "color_jitter": 0.0,
          "cutmix": 1.0,
          "hflip": 0.5,
          "input_size": 224,
          "interpolation": "bilinear",
          "max_ratio": 1.33,
          "max_scale": 2.0,
          "mean": [
            0.485,
            0.456,
            0.406
          ],
          "min_ratio": 0.75,
          "min_scale": 0.1,
          "mixup": 0.8,
          "mixup_mode": "batch",
          "mixup_prob": 1.0,
          "mixup_switch_prob": 0.5,
          "re_prob": 0.0,
          "smoothing": 0.1,
          "std": [
            0.229,
            0.224,
            0.225
          ]
        },
        "batch_size": 1,
        "num_workers_per_gpu": 2,
        "train_data_sources": "",
        "val_data_sources": ""
      },
      "properties": {
        "augmentation": {
          "automl_disabled_parameters": [
            "dataset.augmentation.mean",
            "dataset.augmentation.std"
          ],
          "automl_enabled": false,
          "default": {
            "auto_aug": "rand-m9-mstd0.5-inc1",
            "color_jitter": 0.0,
            "cutmix": 1.0,
            "hflip": 0.5,
            "input_size": 224,
            "interpolation": "bilinear",
            "max_ratio": 1.33,
            "max_scale": 2.0,
            "mean": [
              0.485,
              0.456,
              0.406
            ],
            "min_ratio": 0.75,
            "min_scale": 0.1,
            "mixup": 0.8,
            "mixup_mode": "batch",
            "mixup_prob": 1.0,
            "mixup_switch_prob": 0.5,
            "re_prob": 0.0,
            "smoothing": 0.1,
            "std": [
              0.229,
              0.224,
              0.225
            ]
          },
          "description": "Configuration parameters for data augmentation",
          "properties": {
            "auto_aug": {
              "default": "rand-m9-mstd0.5-inc1",
              "description": "Auto augmentation settings",
              "title": "Auto augmentation settings.",
              "type": "string"
            },
            "color_jitter": {
              "default": 0.0,
              "description": "Color jittering",
              "title": "Color jittering.",
              "type": "float"
            },
            "cutmix": {
              "default": 1.0,
              "description": "Cutmix augmentation",
              "title": "Cutmix augmentation.",
              "type": "float"
            },
            "cutmix_minmax": {
              "description": "Cutmix minmax augmentation",
              "title": "Cutmix minmax augmentation.",
              "type": "float"
            },
            "hflip": {
              "default": 0.5,
              "description": "Horizontal flip probability",
              "title": "Horizontal flip probability.",
              "type": "float"
            },
            "input_size": {
              "default": 224,
              "description": "Input size.",
              "title": "Input size",
              "type": "int"
            },
            "interpolation": {
              "default": "bilinear",
              "description": "Interpolation mode during training",
              "enum": [
                "bilinear",
                "bicubic",
                "random"
              ],
              "title": "Interpolation mode.",
              "type": "categorical"
            },
            "max_ratio": {
              "default": 1.33,
              "description": "Max ratio for resizing augmentation",
              "title": "Max ratio.",
              "type": "float"
            },
            "max_scale": {
              "default": 2.0,
              "description": "Max scale for resizing augmentation",
              "title": "Max scale.",
              "type": "float"
            },
            "mean": {
              "automl_enabled": false,
              "default": [
                0.485,
                0.456,
                0.406
              ],
              "description": "Image mean.",
              "title": "Mean for the image normalization",
              "type": "list"
            },
            "min_ratio": {
              "default": 0.75,
              "description": "Min ratio for resizing augmentation",
              "title": "Min ratio.",
              "type": "float"
            },
            "min_scale": {
              "default": 0.1,
              "description": "Min scale for resizing augmentation",
              "title": "Min scale.",
              "type": "float"
            },
            "mixup": {
              "default": 0.8,
              "description": "Mixup augmentation",
              "title": "Mixup augmentation.",
              "type": "float"
            },
            "mixup_mode": {
              "default": "batch",
              "description": "Mixup mode",
              "enum": [
                "batch",
                "pair",
                "elem"
              ],
              "title": "Mixup mode.",
              "type": "categorical"
            },
            "mixup_prob": {
              "default": 1.0,
              "description": "Mixup probability",
              "title": "Mixup probability.",
              "type": "float"
            },
            "mixup_switch_prob": {
              "default": 0.5,
              "description": "Mixup switch probability",
              "title": "Mixup switch probability.",
              "type": "float"
            },
            "re_prob": {
              "default": 0.0,
              "description": "Random erasing probability",
              "title": "Random erasing probability.",
              "type": "float"
            },
            "smoothing": {
              "default": 0.1,
              "description": "Label smoothing",
              "title": "Label smoothing.",
              "type": "float"
            },
            "std": {
              "automl_enabled": false,
              "default": [
                0.229,
                0.224,
                0.225
              ],
              "description": "Standard deviation for the image normalization",
              "title": "Image standard deviation",
              "type": "list"
            }
          },
          "title": "augmentation",
          "type": "collection"
        },
        "batch_size": {
          "default": 1,
          "description": "Batch size.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "Batch size",
          "type": "int"
        },
        "num_workers_per_gpu": {
          "default": 2,
          "description": "Number of workers per GPU",
          "title": "Number of workers per GPU.",
          "type": "int"
        },
        "test_data_sources": {
          "title": "Image directory of the test set",
          "type": "string"
        },
        "train_data_sources": {
          "default": "",
          "title": "Image directory of the training set",
          "type": "string"
        },
        "val_data_sources": {
          "default": "",
          "title": "Image directory of the validation set",
          "type": "string"
        }
      },
      "type": "collection"
    },
    "encryption_key": {
      "default": "",
      "description": "Key for encrypting model checkpoints",
      "title": "Encryption key",
      "type": "string"
    },
    "model": {
      "automl_enabled": false,
      "default": {
        "arch": "convnextv2_base",
        "decoder_depth": 1,
        "decoder_embed_dim": 512,
        "drop_path_rate": 0.1,
        "global_pool": true,
        "num_classes": 1
      },
      "properties": {
        "arch": {
          "default": "convnextv2_base",
          "description": "Model architecture.",
          "enum": [
            "convnextv2_atto",
            "convnextv2_femto",
            "convnextv2_pico",
            "convnextv2_nano",
            "convnextv2_tiny",
            "convnextv2_base",
            "convnextv2_large",
            "convnextv2_huge",
            "hiera_tiny_224",
            "hiera_small_224",
            "hiera_base_224",
            "hiera_large_224",
            "hiera_huge_224",
            "vit_base_patch16",
            "vit_large_patch16",
            "vit_huge_patch14"
          ],
          "title": "Model arch",
          "type": "categorical"
        },
        "decoder_depth": {
          "default": 1,
          "description": "Decoder depth of MAE models.",
          "title": "Decoder depth",
          "type": "int"
        },
        "decoder_embed_dim": {
          "default": 512,
          "type": "int"
        },
        "drop_path_rate": {
          "default": 0.1,
          "description": "Drop path rate.",
          "title": "Drop path rate",
          "type": "float"
        },
        "global_pool": {
          "default": true,
          "description": "Whether to use global pooling in ViT or Hiera models.",
          "title": "Global pooling",
          "type": "bool"
        },
        "num_classes": {
          "default": 1,
          "description": "Number of classes.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "Number of classes",
          "type": "int"
        }
      },
      "type": "collection"
    },
    "model_name": {
      "default": "",
      "description": "Name of model if invoking task via :code:`model_agnostic`",
      "title": "Model name",
      "type": "string"
    },
    "results_dir": {
      "default": "",
      "description": "\n        Path to where all the assets generated from a task are stored.\n        ",
      "title": "Results directory",
      "type": "string"
    },
    "train": {
      "automl_disabled_parameters": [
        "train.gpu_ids",
        "train.cudnn",
        "train.optim",
        "train.freeze"
      ],
      "automl_enabled": false,
      "default": {
        "accum_grad_batches": 1,
        "checkpoint_interval": 1,
        "checkpoint_interval_unit": "epoch",
        "cudnn": {
          "benchmark": false,
          "deterministic": true
        },
        "distributed_strategy": "ddp",
        "freeze": [],
        "gpu_ids": [
          0
        ],
        "mask_ratio": 0.75,
        "norm_pix_loss": true,
        "num_epochs": 10,
        "num_gpus": 1,
        "num_nodes": 1,
        "optim": {
          "backbone_multiplier": 0.1,
          "gamma": 0.1,
          "layer_decay": 0.75,
          "lr": 0.0002,
          "lr_scheduler": "MultiStep",
          "milestones": [
            88,
            96
          ],
          "momentum": 0.9,
          "monitor_name": "train_loss",
          "type": "AdamW",
          "warmup_epochs": 1,
          "weight_decay": 0.05
        },
        "precision": "fp32",
        "pretrained_model_path": "",
        "results_dir": "",
        "resume_training_checkpoint_path": "",
        "seed": 1234,
        "stage": "pretrain",
        "validation_interval": 1
      },
      "popular": [
        "optim",
        "validation_interval",
        "checkpoint_interval",
        "num_gpus",
        "num_nodes",
        "num_epochs",
        "gpu_ids"
      ],
      "properties": {
        "accum_grad_batches": {
          "default": 1,
          "description": "Number of accumulated gradient batches",
          "maximum": Infinity,
          "minimum": 1,
          "title": "Accum gradient batches",
          "type": "int"
        },
        "checkpoint_interval": {
          "default": 1,
          "description": "The interval (in epochs) at which a checkpoint will be saved. Helps resume training.",
          "minimum": 1,
          "popular": true,
          "title": "Checkpoint interval",
          "type": "int"
        },
        "checkpoint_interval_unit": {
          "default": "epoch",
          "description": "The unit of the checkpoint interval.",
          "enum": [
            "epoch",
            "step"
          ],
          "title": "Checkpoint interval unit",
          "type": "categorical"
        },
        "cudnn": {
          "automl_enabled": false,
          "default": {
            "benchmark": false,
            "deterministic": true
          },
          "properties": {
            "benchmark": {
              "default": false,
              "type": "bool"
            },
            "deterministic": {
              "default": true,
              "type": "bool"
            }
          },
          "type": "collection"
        },
        "distributed_strategy": {
          "default": "ddp",
          "description": "\n        The multi-GPU training strategy.\n        DDP (Distributed Data Parallel) and Fully Sharded DDP are supported.",
          "enum": [
            "ddp",
            "fsdp"
          ],
          "title": "distributed_strategy",
          "type": "categorical"
        },
        "freeze": {
          "automl_enabled": false,
          "default": [],
          "description": "List of layer names to freeze.",
          "title": "freeze",
          "type": "list"
        },
        "gpu_ids": {
          "automl_enabled": false,
          "default": [
            0
          ],
          "description": "\n        List of GPU IDs to run the training on. The length of this list\n        must be equal to the number of gpus in train.num_gpus.",
          "popular": true,
          "title": "GPU IDs",
          "type": "list"
        },
        "mask_ratio": {
          "default": 0.75,
          "description": "Mask ratio",
          "title": "Mask ratio.",
          "type": "float"
        },
        "norm_pix_loss": {
          "default": true,
          "description": "Whether to normalize pixel loss",
          "title": "Normalize pixel loss",
          "type": "bool"
        },
        "num_epochs": {
          "default": 10,
          "description": "Number of epochs to run the training.",
          "maximum": Infinity,
          "minimum": 1,
          "popular": true,
          "title": "Number of epochs",
          "type": "int"
        },
        "num_gpus": {
          "default": 1,
          "description": "The number of GPUs to run the train job.",
          "minimum": 1,
          "popular": true,
          "title": "Number of GPUs",
          "type": "int"
        },
        "num_nodes": {
          "default": 1,
          "description": "Number of nodes to run the training on. If > 1, then multi-node is enabled.",
          "minimum": 1,
          "popular": true,
          "title": "Number of nodes",
          "type": "int"
        },
        "optim": {
          "automl_default_parameters": [
            "train.optim.lr",
            "train.optim.backbone_multiplier",
            "train.optim.momentum",
            "train.optim.weight_decay",
            "train.optim.warmup_epochs"
          ],
          "automl_disabled_parameters": [
            "train.optim.milestones"
          ],
          "automl_enabled": false,
          "default": {
            "backbone_multiplier": 0.1,
            "gamma": 0.1,
            "layer_decay": 0.75,
            "lr": 0.0002,
            "lr_scheduler": "MultiStep",
            "milestones": [
              88,
              96
            ],
            "momentum": 0.9,
            "monitor_name": "train_loss",
            "type": "AdamW",
            "warmup_epochs": 1,
            "weight_decay": 0.05
          },
          "description": "Hyper parameters to configure the optimizer.",
          "popular": [
            "momentum",
            "weight_decay",
            "backbone_multiplier",
            "layer_decay"
          ],
          "properties": {
            "backbone_multiplier": {
              "automl_enabled": true,
              "default": 0.1,
              "description": "A multiplier for backbone learning rate.",
              "math_cond": "> 0.0",
              "maximum": 1.0,
              "minimum": 0.0,
              "popular": true,
              "title": "backbone learning rate multiplier",
              "type": "float"
            },
            "gamma": {
              "default": 0.1,
              "description": "Multiplicative factor of learning rate decay.",
              "math_cond": "> 0.0",
              "title": "gamma",
              "type": "float"
            },
            "layer_decay": {
              "default": 0.75,
              "description": "The layer decay coefficient.",
              "math_cond": "> 0.0",
              "popular": true,
              "title": "layer decay",
              "type": "float"
            },
            "lr": {
              "automl_enabled": true,
              "default": 0.0002,
              "description": "The initial learning rate for training the model.",
              "math_cond": "> 0.0",
              "maximum": 1.0,
              "minimum": 0.0,
              "title": "learning rate",
              "type": "float"
            },
            "lr_scheduler": {
              "default": "MultiStep",
              "description": "The learning scheduler:\n                    * MultiStep : Decrease the lr by lr_decay from lr_steps\n                    * cosine : Poly learning rate schedule.",
              "enum": [
                "MultiStep",
                "cosine"
              ],
              "title": "learning rate scheduler",
              "type": "categorical"
            },
            "milestones": {
              "automl_enabled": false,
              "default": [
                88,
                96
              ],
              "description": "learning rate decay epochs.",
              "title": "learning rate decay epochs.",
              "type": "list"
            },
            "momentum": {
              "automl_enabled": true,
              "default": 0.9,
              "description": "The momentum for the AdamW optimizer.",
              "math_cond": "> 0.0",
              "maximum": 1.0,
              "minimum": 0.0,
              "popular": true,
              "title": "momentum - AdamW",
              "type": "float"
            },
            "monitor_name": {
              "default": "train_loss",
              "description": "The metric value to be monitored for the :code:`AutoReduce` Scheduler.",
              "enum": [
                "val_loss",
                "train_loss"
              ],
              "title": "monitor_name",
              "type": "categorical"
            },
            "type": {
              "default": "AdamW",
              "description": "Type of optimizer used to train the network.",
              "enum": [
                "AdamW"
              ],
              "type": "categorical"
            },
            "warmup_epochs": {
              "automl_enabled": true,
              "default": 1,
              "description": "Warmup epochs.",
              "math_cond": ">= 0",
              "maximum": 100,
              "minimum": 0,
              "title": "Warmup epochs",
              "type": "int"
            },
            "weight_decay": {
              "automl_enabled": true,
              "default": 0.05,
              "description": "The weight decay coefficient.",
              "math_cond": "> 0.0",
              "maximum": 1.0,
              "minimum": 0.0,
              "popular": true,
              "title": "weight decay",
              "type": "float"
            }
          },
          "title": "optimizer",
          "type": "collection"
        },
        "precision": {
          "default": "fp32",
          "description": "Precision to run the training on.",
          "enum": [
            "fp16",
            "bf16",
            "fp32"
          ],
          "title": "precision",
          "type": "categorical"
        },
        "pretrained_model_path": {
          "default": "",
          "description": "Path to the pretrained model",
          "title": "Pretrained model",
          "type": "string"
        },
        "results_dir": {
          "default": "",
          "description": "\n        Path to where all the assets generated from a task are stored.\n        ",
          "title": "Results directory",
          "type": "string"
        },
        "resume_training_checkpoint_path": {
          "default": "",
          "description": "Path to the checkpoint to resume training from.",
          "title": "Resume checkpoint path",
          "type": "string"
        },
        "seed": {
          "default": 1234,
          "description": "The seed for the initializer in PyTorch. If < 0, disable fixed seed.",
          "maximum": Infinity,
          "minimum": -1,
          "title": "Seed for randomization",
          "type": "int"
        },
        "stage": {
          "default": "pretrain",
          "description": "Training stage.",
          "enum": [
            "pretrain",
            "finetune"
          ],
          "title": "Stage",
          "type": "categorical"
        },
        "validation_interval": {
          "default": 1,
          "description": "\n        The interval (in epochs) at which a evaluation\n        will be triggered on the validation dataset.",
          "minimum": 1,
          "popular": true,
          "title": "Validation interval",
          "type": "int"
        }
      },
      "type": "collection"
    },
    "wandb": {
      "automl_disabled_parameters": [
        "wandb.tags"
      ],
      "automl_enabled": false,
      "default": {
        "enable": true,
        "entity": "",
        "group": "",
        "name": "TAO Toolkit Training",
        "project": "TAO Toolkit",
        "reinit": false,
        "run_id": "",
        "save_code": false,
        "sync_tensorboard": false,
        "tags": [
          "tao-toolkit"
        ]
      },
      "properties": {
        "enable": {
          "default": true,
          "type": "bool"
        },
        "entity": {
          "default": "",
          "type": "string"
        },
        "group": {
          "default": "",
          "type": "string"
        },
        "name": {
          "default": "TAO Toolkit Training",
          "type": "string"
        },
        "project": {
          "default": "TAO Toolkit",
          "type": "string"
        },
        "reinit": {
          "default": false,
          "type": "bool"
        },
        "run_id": {
          "default": "",
          "type": "string"
        },
        "save_code": {
          "default": false,
          "type": "bool"
        },
        "sync_tensorboard": {
          "default": false,
          "type": "bool"
        },
        "tags": {
          "automl_enabled": false,
          "default": [
            "tao-toolkit"
          ],
          "type": "list"
        }
      },
      "type": "collection"
    }
  },
  "type": "object",
  "x_tao_schema": {
    "action": "train",
    "core_module": "mae",
    "model": "mae",
    "network_arch": "mae",
    "schema_action": "train",
    "schema_version": 1,
    "source": "tao-core dataclass config"
  }
}

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Declares allowed tools, Needs Python, Needs Node.js

The author says: Requires docker + nvidia-container-toolkit.

Security

What our scanner noticed· Medium risk

Large base64-encoded blobs

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by

NVIDIA Corporation

Publisher

NVIDIA (Official)

Source

NVIDIA/skills · skills/tao-train-mask-auto-encoder · commit fce3cca

Licence

Apache License 2.0 (redistribution allowed)

LICENSE (added by AutoGineer from LICENSE-APACHE)
ATTRIBUTION.md (written by AutoGineer)

Synced

2026-11-08

AutoGineer is not affiliated with or endorsed by NVIDIA Corporation or NVIDIA.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download tao-train-mask-auto-encoder.zip