Tao Train Ocrnet

NVIDIAOfficialCreated by NVIDIA Corporation

Skill
Tool integration
Data and AI
Apache-2.0
Medium risk

From the author

OCRNet for scene text recognition. Recognizes text content from cropped text-region images and supports CTC and attention-based decoders. Use when training, evaluating, exporting, pruning, quantizing, retraining, or running inference for a TAO OCRNet model. Trigger phrases include "train OCRNet", "scene text recognition", "OCR cropped text", "CTC / attention text decoder".

— the description by NVIDIA Corporation

AutoGineer summary

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

The tao-train-ocrnet skill provides a comprehensive interface for training, evaluating, pruning, quantizing, and exporting OCRNet models designed for scene text recognition. It supports both CTC and attention-based decoders, allowing users to process cropped text-region images. The skill integrates with the TAO Toolkit ecosystem, supporting AutoML for training optimization and providing pathways for deployment via TensorRT engines.

What it does

Trains OCRNet models using CTC or attention-based decoders.
Converts raw image datasets into the required LMDB format.
Supports model pruning and quantization for optimized inference.
Exports trained models to ONNX format for deployment.
Facilitates evaluation and inference on test datasets.
Integrates with TAO AutoML for automated hyperparameter tuning.

When to use it

When you need to build a custom scene text recognition model.
When you have cropped text-region images and need to extract text content.
When optimizing a model for production deployment using TensorRT.
When you need to prune or quantize a model to reduce its footprint.

Advantages

Supports both CTC and attention-based decoding architectures.
Automated hyperparameter tuning via built-in AutoML capabilities.
Seamless integration with TAO Toolkit deployment workflows.
Provides clear paths for model compression and optimization.

Usage examples

“Train an OCRNet model using my dataset.”: The agent will initiate the training process, utilizing the specified dataset and configuration, with AutoML enabled by default to optimize performance.
“Convert my raw text images for OCRNet training.”: The agent will execute the dataset_convert action to transform your raw images and ground truth files into the required LMDB format.
“Prune my trained OCRNet model.”: The agent will apply pruning to the specified checkpoint to create a smaller, more efficient model graph.
“Export my OCRNet model to ONNX.”: The agent will export the trained model checkpoint into an ONNX artifact suitable for further deployment or engine generation.

Install

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

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

mkdir -p .agents/skills && unzip tao-train-ocrnet.zip -d .agents/skills

Documentation by NVIDIA Corporation

OCRNet

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).

OCRNet for scene text recognition. Recognizes text content from cropped text region images. Supports CTC and attention-based decoders.

Set train.pretrained_model_path for pretrained OCR weights.

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

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: ocrnet
Formats: default
AutoML training metric: val_acc, with direction=maximize; TAO writes this key to status.json
Lightning progress alias: val_acc_1
Standalone evaluation metric: test_acc, with direction=maximize

Per-Action Dataset Requirements

Action	Spec Key	Source	Files	List?
dataset_convert	dataset_convert.input_img_dir	train_datasets or eval_dataset	extracted folder containing cropped text images	No
dataset_convert	dataset_convert.gt_file	train_datasets or eval_dataset	train/gt_new.txt or test/gt_new.txt	No
evaluate	dataset.character_list_file	eval_dataset	character_list	No
evaluate	evaluate.test_dataset_dir	eval_dataset	extracted test image folder	No
evaluate	evaluate.test_dataset_gt_file	eval_dataset	test/gt_new.txt	No
evaluate	evaluate.checkpoint	parent train/AutoML job	best_accuracy.pth or exact requested epoch checkpoint	No
export	dataset.character_list_file	eval_dataset	character_list	No
export	export.checkpoint	parent train/AutoML job	best_accuracy.pth or exact requested epoch checkpoint	No
deploy/gen_trt_engine	gen_trt_engine.tensorrt.calibration.cal_image_dir	calibration_dataset	extracted calibration image folder for INT8 calibration	Yes
deploy/gen_trt_engine	gen_trt_engine.onnx_file	parent export job	exported .onnx artifact	No
deploy/gen_trt_engine	dataset.character_list_file	eval_dataset	character_list	No
inference	dataset.character_list_file	eval_dataset	character_list	No
inference	inference.inference_dataset_dir	inference_dataset	extracted inference image folder	No
inference	inference.checkpoint	parent train/AutoML job	best_accuracy.pth or exact requested epoch checkpoint	No
prune	dataset.character_list_file	eval_dataset	character_list	No
prune	prune.checkpoint	parent train/AutoML job	best_accuracy.pth or exact requested epoch checkpoint	No
quantize	dataset.train_dataset_dir	dataset_convert train job	LMDB folder containing data.mdb and lock.mdb	Yes
quantize	dataset.val_dataset_dir	dataset_convert eval job	LMDB folder containing data.mdb and lock.mdb	No
quantize	dataset.character_list_file	eval_dataset	character_list	No
quantize	dataset.quant_calibration_dataset.images_dir	train_datasets	extracted calibration image folder	No
quantize	quantize.model_path	parent train/AutoML job	checkpoint selected by resolver	No
retrain	dataset.train_dataset_dir	dataset_convert train job	LMDB folder containing data.mdb and lock.mdb	Yes
retrain	dataset.val_dataset_dir	dataset_convert eval job	LMDB folder containing data.mdb and lock.mdb	No
retrain	dataset.character_list_file	eval_dataset	character_list	No
retrain	model.pruned_graph_path	parent prune job	pruned .pth artifact	No
train	dataset.train_dataset_dir	dataset_convert train job	LMDB folder containing data.mdb and lock.mdb	Yes
train	dataset.train_gt_file	train_datasets	train/gt_new.txt when using raw folders instead of LMDB	No
train	dataset.val_dataset_dir	dataset_convert eval job	LMDB folder containing data.mdb and lock.mdb	No
train	dataset.val_gt_file	eval_dataset	test/gt_new.txt when using raw folders instead of LMDB	No
train	dataset.character_list_file	eval_dataset	character_list	No

Checkpoint Selection

OCRNet training writes both best_accuracy.pth and epoch-step checkpoints such as model_epoch_000_step_00003.pth. Use the SDK/model checkpoint resolver through the spec_params mappings in references/skill_info.yaml; do not guess by sorting for the newest .pth.

Use best_accuracy.pth for best-checkpoint evaluate, inference, export, and prune requests.
Use the exact requested model_epoch_*_step_*.pth for epoch/step-specific actions.
Use train.resume_training_checkpoint_path only for resume training, and use model.pruned_graph_path for retrain from a prune output. OCRNet does not expose a separate ocrnet retrain CLI subtask in the PyT image; the model-skill retrain action routes through ocrnet train -e with the pruned graph path set.
OCRNet quantize loads the model through PyTorch. For trusted checkpoints created by the same local run, set TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD=1 if PyTorch 2.6+ rejects the checkpoint as a weights-only load.

Typical Spec Overrides

Data source overrides are mandatory for every action. Run dataset_convert separately for train and validation splits, then pass the LMDB folders that directly contain data.mdb and lock.mdb into train, quantize, and retrain. Tarballs from remote storage must be extracted before they are used as image directories.

TRAIN_IMAGES = "<extracted train image folder>"
TRAIN_GT = "<train gt_new.txt>"
EVAL_IMAGES = "<extracted eval image folder>"
EVAL_GT = "<eval gt_new.txt>"
TRAIN_LMDB = "<train dataset_convert results_dir>"
EVAL_LMDB = "<eval dataset_convert results_dir>"
CHAR_LIST = "<character_list>"

dataset_convert (run once per split):

{
    "dataset_convert.input_img_dir": TRAIN_IMAGES,
    "dataset_convert.gt_file": TRAIN_GT,
}

train (mandatory data sources):

{
    "train.num_epochs": 30,
    "train.checkpoint_interval": 10,
    "train.validation_interval": 10,
    "train.num_gpus": 1,
    "dataset.batch_size": 16,
    "dataset.train_dataset_dir": [TRAIN_LMDB],
    "dataset.val_dataset_dir": EVAL_LMDB,
    "dataset.train_gt_file": "",
    "dataset.val_gt_file": "",
    "dataset.character_list_file": CHAR_LIST,
}

deploy/gen_trt_engine (mandatory data sources):

{
    "gen_trt_engine.onnx_file": "<selected export ONNX>",
    "gen_trt_engine.trt_engine": "<output engine path>",
    "gen_trt_engine.tensorrt.calibration.cal_cache_file": "<output calibration cache path>",
    "gen_trt_engine.tensorrt.data_type": "fp16",
    "gen_trt_engine.tensorrt.calibration.cal_image_dir": [TRAIN_IMAGES],
    "dataset.character_list_file": CHAR_LIST,
}

evaluate (mandatory data sources):

{
    "evaluate.checkpoint": "<selected train/AutoML checkpoint>",
    "dataset.character_list_file": CHAR_LIST,
    "evaluate.test_dataset_dir": EVAL_IMAGES,
    "evaluate.test_dataset_gt_file": EVAL_GT,
}

export (mandatory data sources):

{
    "export.checkpoint": "<selected train/AutoML checkpoint>",
    "export.onnx_file": "<output ONNX path>",
    "dataset.character_list_file": CHAR_LIST,
}

inference (mandatory data sources):

{
    "inference.checkpoint": "<selected train/AutoML checkpoint>",
    "dataset.character_list_file": CHAR_LIST,
    "inference.inference_dataset_dir": EVAL_IMAGES,
}

prune (mandatory data sources):

{
    "prune.checkpoint": "<selected train/AutoML checkpoint>",
    "prune.pruned_file": "<output pruned PTH path>",
    "dataset.character_list_file": CHAR_LIST,
}

quantize (mandatory data sources):

{
    "dataset.train_dataset_dir": [TRAIN_LMDB],
    "dataset.val_dataset_dir": EVAL_LMDB,
    "dataset.character_list_file": CHAR_LIST,
    "dataset.quant_calibration_dataset.images_dir": TRAIN_IMAGES,
    "quantize.model_path": "<selected train/AutoML checkpoint>",
}

retrain (mandatory data sources):

{
    "dataset.train_dataset_dir": [TRAIN_LMDB],
    "dataset.val_dataset_dir": EVAL_LMDB,
    "dataset.character_list_file": CHAR_LIST,
    "model.pruned_graph_path": "<selected prune output>",
}

Eval Dataset

Optional. Test data provided as separate tarball.

Important Parameters

dataset.character_list_file: Path to character list defining the supported character set. This determines the output vocabulary size.
model.backbone: Default ResNet.
model.prediction: Decoder type. CTC or Attn (attention-based).
train.optim.lr: Learning rate. Default 1.0 (Adadelta optimizer). High default is specific to Adadelta.
dataset.batch_size: Per-GPU batch size. Default 16.

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.distributed_strategy`	Strategy name	`auto`

Strategy: auto for single-GPU, reads train.distributed_strategy from config when multi-GPU
No explicit num_nodes in train script — single-node oriented
Lightweight model, single GPU typically sufficient

Hardware

Minimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCR text recognition is lightweight. Single GPU is typically sufficient.

Error Patterns

dataset_convert required: If using raw images + gt files, run dataset_convert first to produce LMDB format.

dataset_convert output folder: Direct ocrnet dataset_convert writes data.mdb and lock.mdb directly under dataset_convert.results_dir. Use that folder itself for dataset.train_dataset_dir, dataset.val_dataset_dir, quantize, and retrain inputs. SDK-backed runs may wrap the same LMDB folder inside job artifact directories; resolve the actual folder containing data.mdb and lock.mdb.

GT file BOM: Some text-recognition GT files can start with a UTF-8 BOM on the first filename. If dataset conversion logs a missing path with an invisible prefix before the first image name, strip the BOM from a local copy of the GT file before conversion or evaluation.

Character list mismatch: All characters in training data must be present in the character_list file.

Export/prune output fields required: export.onnx_file and prune.pruned_file must be writable output paths. These are declared in references/skill_info.yaml so SDK-backed model runs can create the paths automatically.

TensorRT lives in deploy: The PyT OCRNet CLI exposes dataset_convert, evaluate, export, inference, prune, quantize, and train, but not gen_trt_engine. Use references/tao-deploy-ocrnet.md and deploy/skill_info.yaml for TensorRT engine generation and TensorRT-backed evaluate/inference.

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 ocrnet.config.json:

Action	Spec Field	Inference Function	Meaning
dataset_convert	`results_dir`	`output_dir`	current job results directory
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	`model.pruned_graph_path`	`pruned_model`	parent pruned model
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
deploy/gen_trt_engine	`encryption_key`	`key`	encryption key
deploy/gen_trt_engine	`gen_trt_engine.onnx_file`	`parent_model`	ONNX file inferred from the parent export job results folder
deploy/gen_trt_engine	`gen_trt_engine.tensorrt.calibration.cal_cache_file`	`create_cal_cache`	calibration cache path
deploy/gen_trt_engine	`gen_trt_engine.trt_engine`	`create_engine_file`	output TensorRT engine path
deploy/gen_trt_engine	`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	`model.pruned_graph_path`	`pruned_model`	parent pruned model
inference	`results_dir`	`output_dir`	current job results directory
prune	`encryption_key`	`key`	encryption key
prune	`prune.checkpoint`	`parent_model`	model file inferred from the parent job results folder
prune	`prune.pruned_file`	`create_pth_file`	output PTH path
prune	`results_dir`	`output_dir`	current job results directory
quantize	`encryption_key`	`key`	encryption key
quantize	`quantize.model_path`	`parent_model`	model file inferred from the parent job results folder
quantize	`results_dir`	`output_dir`	current job results directory
retrain	`encryption_key`	`key`	encryption key
retrain	`model.pruned_graph_path`	`parent_model`	model file inferred from the parent job results folder
retrain	`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.

Deployment

tao-deploy-ocrnet

Files

ATTRIBUTION.mdAdded by AutoGineer918 B
BENCHMARK.md6.9 KiB
LICENSELicence file added by AutoGineer11.2 KiB
SKILL.md17.1 KiB
config/skillspector-baseline.yaml834 B
evals/evals.json762 B
references/skill_info.yaml8.2 KiB
references/spec_template_dataset_convert.yaml1.8 KiB
references/spec_template_deploy_experiment.yaml2.9 KiB
references/spec_template_evaluate.yaml1.9 KiB
references/spec_template_export.yaml1.8 KiB
references/spec_template_gen_trt_engine.yaml2.1 KiB
references/spec_template_inference.yaml1.9 KiB
references/spec_template_prune.yaml1.9 KiB
references/spec_template_quantize.yaml1.7 KiB
references/spec_template_retrain.yaml1.7 KiB
references/spec_template_train.yaml1.7 KiB
references/tao-deploy-ocrnet.md5.5 KiB
references/tao-deploy-ocrnet.skill_info.yaml4.0 KiB
schemas/dataset_convert.schema.json28.9 KiB
schemas/evaluate.schema.json31.3 KiB
schemas/export.schema.json29.0 KiB
schemas/gen_trt_engine.schema.json36.3 KiB
schemas/inference.schema.json31.1 KiB
schemas/manifest.json19.9 KiB
schemas/prune.schema.json30.9 KiB
schemas/quantize.schema.json28.0 KiB
schemas/retrain.schema.json28.0 KiB
schemas/train.schema.json28.0 KiB
skill-card.md4.7 KiB
skill.oms.sig10.3 KiB

schemas/quantize.schema.json

{
  "automl_default_parameters": [
    "dataset.augmentation.reverse_color_prob",
    "dataset.augmentation.aug_prob",
    "dataset.augmentation.blur_prob",
    "dataset.augmentation.rotate_prob",
    "train.optim.lr"
  ],
  "automl_disabled_parameters": [
    "quantize",
    "train.cudnn",
    "gen_trt_engine.tensorrt.calibration.cal_image_dir",
    "quantize.backend_kwargs",
    "train.gpu_ids",
    "wandb.tags",
    "quantize.skip_names",
    "evaluate",
    "inference",
    "train",
    "dataset.augmentation",
    "gen_trt_engine",
    "dataset.train_dataset_dir",
    "gen_trt_engine.tensorrt.layers_precision",
    "dataset",
    "gen_trt_engine.tensorrt",
    "quantize.layers",
    "dataset_convert",
    "dataset.quant_calibration_dataset",
    "prune.prune_setting",
    "model",
    "train.optim.lr_steps",
    "evaluate.gpu_ids",
    "train.optim",
    "gen_trt_engine.tensorrt.calibration",
    "dataset.augmentation.gaussian_radius_list",
    "export",
    "wandb",
    "inference.gpu_ids",
    "prune"
  ],
  "default": {
    "dataset": {
      "augmentation": {
        "aug_prob": 0.0,
        "blur_prob": 0.5,
        "gaussian_radius_list": [
          1,
          2,
          3,
          4
        ],
        "keep_aspect_ratio": false,
        "max_rotation_degree": 5,
        "reverse_color_prob": 0.5,
        "rotate_prob": 0.5
      },
      "batch_size": 16,
      "character_list_file": "",
      "max_label_length": 25,
      "quant_calibration_dataset": {
        "images_dir": ""
      },
      "train_dataset_dir": [],
      "train_gt_file": "",
      "val_dataset_dir": "",
      "val_gt_file": "",
      "workers": 8
    },
    "encryption_key": "",
    "model": {
      "TPS": false,
      "backbone": "ResNet",
      "feature_channel": 512,
      "hidden_size": 256,
      "input_channel": 1,
      "input_height": 32,
      "input_width": 100,
      "num_fiducial": 3,
      "prediction": "CTC",
      "pruned_graph_path": "",
      "quantize": false,
      "quantize_model_path": "",
      "sequence": "BiLSTM"
    },
    "model_name": "",
    "quantize": {
      "algorithm": "minmax",
      "backend": "torchao",
      "backend_kwargs": {},
      "device": "cuda",
      "layers": [],
      "mode": "weight_only_ptq",
      "model_path": "",
      "results_dir": "",
      "skip_names": []
    },
    "results_dir": "",
    "train": {
      "checkpoint_interval": 1,
      "checkpoint_interval_unit": "epoch",
      "clip_grad_norm": 5.0,
      "cudnn": {
        "benchmark": false,
        "deterministic": true
      },
      "distributed_strategy": "ddp",
      "gpu_ids": [
        0
      ],
      "model_ema": false,
      "num_epochs": 10,
      "num_gpus": 1,
      "num_nodes": 1,
      "optim": {
        "lr": 1.0,
        "lr_decay": 0.1,
        "lr_monitor": "val_loss",
        "lr_scheduler": "MultiStep",
        "lr_steps": [
          15,
          25
        ],
        "min_lr": 0.0001,
        "momentum": 0.9,
        "name": "adadelta",
        "patience": 1,
        "weight_decay": 0.0005
      },
      "pretrained_model_path": "",
      "results_dir": "",
      "resume_training_checkpoint_path": "",
      "seed": 1234,
      "use_distributed_sampler": false,
      "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": {
        "calibration": {
          "cal_batch_size": 1,
          "cal_batches": 1
        },
        "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,
      "validation_interval": 1
    }
  },
  "properties": {
    "automl_disabled_parameters": [
      "wandb",
      "model",
      "dataset",
      "train",
      "evaluate",
      "export",
      "inference",
      "prune",
      "dataset_convert",
      "gen_trt_engine",
      "quantize"
    ],
    "dataset": {
      "automl_disabled_parameters": [
        "dataset.quant_calibration_dataset",
        "dataset.train_dataset_dir",
        "dataset.augmentation"
      ],
      "automl_enabled": false,
      "default": {
        "augmentation": {
          "aug_prob": 0.0,
          "blur_prob": 0.5,
          "gaussian_radius_list": [
            1,
            2,
            3,
            4
          ],
          "keep_aspect_ratio": false,
          "max_rotation_degree": 5,
          "reverse_color_prob": 0.5,
          "rotate_prob": 0.5
        },
        "batch_size": 16,
        "character_list_file": "",
        "max_label_length": 25,
        "quant_calibration_dataset": {
          "images_dir": ""
        },
        "train_dataset_dir": [],
        "train_gt_file": "",
        "val_dataset_dir": "",
        "val_gt_file": "",
        "workers": 8
      },
      "description": "Configurable parameters for the dataset.",
      "properties": {
        "augmentation": {
          "automl_default_parameters": [
            "dataset.augmentation.aug_prob",
            "dataset.augmentation.reverse_color_prob",
            "dataset.augmentation.rotate_prob",
            "dataset.augmentation.blur_prob"
          ],
          "automl_disabled_parameters": [
            "dataset.augmentation.gaussian_radius_list"
          ],
          "automl_enabled": false,
          "default": {
            "aug_prob": 0.0,
            "blur_prob": 0.5,
            "gaussian_radius_list": [
              1,
              2,
              3,
              4
            ],
            "keep_aspect_ratio": false,
            "max_rotation_degree": 5,
            "reverse_color_prob": 0.5,
            "rotate_prob": 0.5
          },
          "description": "Configurable parameters for augmentation.",
          "properties": {
            "aug_prob": {
              "automl_enabled": true,
              "default": 0.0,
              "description": "The probability to augment the input images.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            },
            "blur_prob": {
              "automl_enabled": true,
              "default": 0.5,
              "description": "The probability to blur the input images.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            },
            "gaussian_radius_list": {
              "automl_enabled": false,
              "default": [
                1,
                2,
                3,
                4
              ],
              "description": "The gaussian raidus list for gaussian blur.",
              "type": "list_2"
            },
            "keep_aspect_ratio": {
              "default": false,
              "description": "The bool flag to keep aspect ratio in resizing input images.",
              "type": "bool"
            },
            "max_rotation_degree": {
              "default": 5,
              "description": "The maximum rotation degree.",
              "maximum": 360,
              "minimum": 0,
              "type": "int"
            },
            "reverse_color_prob": {
              "automl_enabled": true,
              "default": 0.5,
              "description": "The probability to reverse the color of input images.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            },
            "rotate_prob": {
              "automl_enabled": true,
              "default": 0.5,
              "description": "The probability to rotate the input images.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            }
          },
          "type": "collection"
        },
        "batch_size": {
          "default": 16,
          "description": "Batch size of model input.",
          "type": "int"
        },
        "character_list_file": {
          "default": "",
          "description": "The absolute path to the character list.",
          "type": "string"
        },
        "max_label_length": {
          "default": 25,
          "description": "The maximum length of the labels.",
          "type": "int"
        },
        "quant_calibration_dataset": {
          "automl_enabled": false,
          "default": {
            "images_dir": ""
          },
          "description": "Configurable parameters for quantization calibration dataset.",
          "properties": {
            "images_dir": {
              "default": "",
              "description": "Path to the directory containing calibration images.",
              "title": "calibration images directory",
              "type": "string"
            }
          },
          "type": "collection"
        },
        "train_dataset_dir": {
          "automl_enabled": false,
          "default": [],
          "description": "The absolute path to the train dataset directory.",
          "type": "list"
        },
        "train_gt_file": {
          "default": "",
          "description": "The absolute path to the train dataset ground truth file.",
          "type": "string"
        },
        "val_dataset_dir": {
          "default": "",
          "description": "The absolute path to the validation dataset directory.",
          "type": "string"
        },
        "val_gt_file": {
          "default": "",
          "description": "The absolute path to the validation dataset ground truth file.",
          "type": "string"
        },
        "workers": {
          "default": 8,
          "description": "The number of workers to process the dataset.",
          "type": "int"
        }
      },
      "type": "collection"
    },
    "encryption_key": {
      "default": "",
      "description": "Key for encrypting model checkpoints",
      "title": "Encryption key",
      "type": "string"
    },
    "model": {
      "automl_enabled": false,
      "default": {
        "TPS": false,
        "backbone": "ResNet",
        "feature_channel": 512,
        "hidden_size": 256,
        "input_channel": 1,
        "input_height": 32,
        "input_width": 100,
        "num_fiducial": 3,
        "prediction": "CTC",
        "pruned_graph_path": "",
        "quantize": false,
        "quantize_model_path": "",
        "sequence": "BiLSTM"
      },
      "description": "Configurable parameters for the model.",
      "properties": {
        "TPS": {
          "default": false,
          "description": "The bool flag to apply Thin-Plate-Spline interpolation to the model.",
          "title": "Thin-Plate-Spline interpolation",
          "type": "bool"
        },
        "backbone": {
          "default": "ResNet",
          "description": "Backbone of the model.",
          "enum": [
            "ResNet",
            "ResNet2X",
            "FAN_tiny_2X"
          ],
          "title": "backbone",
          "type": "categorical"
        },
        "feature_channel": {
          "default": 512,
          "description": "The number of backbone's feature output channel.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "feature channel",
          "type": "int"
        },
        "hidden_size": {
          "default": 256,
          "description": "The number of hidden uints in BiLSTM layers.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "hidden_size",
          "type": "int"
        },
        "input_channel": {
          "default": 1,
          "description": "The input channel of the model.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "input channel",
          "type": "int"
        },
        "input_height": {
          "default": 32,
          "description": "The input height of the model.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "input height",
          "type": "int"
        },
        "input_width": {
          "default": 100,
          "description": "The input width of the model.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "input width",
          "type": "int"
        },
        "num_fiducial": {
          "default": 3,
          "description": "The number of fiducial/keypoints points for TPS.",
          "maximum": Infinity,
          "minimum": 1,
          "title": "num fiducial",
          "type": "int"
        },
        "prediction": {
          "default": "CTC",
          "description": "The sequence decoding method.",
          "enum": [
            "CTC",
            "Attn"
          ],
          "title": "prediction",
          "type": "categorical"
        },
        "pruned_graph_path": {
          "default": "",
          "description": "The pruned model to be loaded.",
          "title": "pruned graph path",
          "type": "string"
        },
        "quantize": {
          "default": false,
          "description": "The bool flag to apply quantization to the model.",
          "title": "quantize",
          "type": "bool"
        },
        "quantize_model_path": {
          "default": "",
          "description": "The quantized model to be loaded.",
          "title": "quantize model path",
          "type": "string"
        },
        "sequence": {
          "default": "BiLSTM",
          "description": "The sequence fature modeling type.",
          "enum": [
            "BiLSTM"
          ],
          "title": "sequence",
          "type": "categorical"
        }
      },
      "type": "collection"
    },
    "model_name": {
      "default": "",
      "description": "Name of model if invoking task via :code:`model_agnostic`",
      "title": "Model name",
      "type": "string"
    },
    "quantize": {
      "automl_disabled_parameters": [
        "quantize.layers",
        "quantize.skip_names",
        "quantize.backend_kwargs"
      ],
      "automl_enabled": false,
      "default": {
        "algorithm": "minmax",
        "backend": "torchao",
        "backend_kwargs": {},
        "device": "cuda",
        "layers": [],
        "mode": "weight_only_ptq",
        "model_path": "",
        "results_dir": "",
        "skip_names": []
      },
      "description": "Configurable parameters for model quantization.",
      "properties": {
        "algorithm": {
          "default": "minmax",
          "description": "Calibration/optimization algorithm. Used by ModelOpt backends (modelopt.pytorch and modelopt.onnx). Ignored by torchao backend.",
          "enum": [
            "minmax",
            "max",
            "entropy",
            "awq_clip",
            "awq_lite",
            "awq_full",
            "rtn_dq"
          ],
          "title": "Calibration algorithm",
          "type": "categorical"
        },
        "backend": {
          "default": "torchao",
          "description": "The quantization backend to use",
          "enum": [
            "modelopt.pytorch",
            "torchao",
            "modelopt.onnx"
          ],
          "title": "Quantization backend",
          "type": "categorical"
        },
        "backend_kwargs": {
          "automl_enabled": false,
          "description": "Additional keyword arguments to pass to the backend",
          "title": "Backend kwargs",
          "type": "collection"
        },
        "device": {
          "default": "cuda",
          "description": "Device to use for calibration. Accepts 'cuda' (uses default GPU), 'cpu', 'trt' (TensorRT for ONNX backend), or specific GPU device like 'cuda:0', 'cuda:1', etc. If 'cuda' is specified but no GPU is available, the framework will automatically fall back to 'cpu'. Note: 'trt' is only supported by the modelopt.onnx backend for ONNX Runtime with TensorRT execution provider.",
          "pattern": "^(cuda|cpu|trt|cuda:[0-9]+)$",
          "title": "Calibration device",
          "type": "string"
        },
        "layers": {
          "automl_enabled": false,
          "default": [],
          "description": "List of per-module quantization configurations. Each entry specifies which modules to quantize and their data types. This is the primary way to configure quantization.",
          "title": "Layer quantization configs",
          "type": "list"
        },
        "mode": {
          "default": "weight_only_ptq",
          "description": "The quantization mode to use",
          "enum": [
            "static_ptq",
            "weight_only_ptq"
          ],
          "title": "Quantization mode",
          "type": "categorical"
        },
        "model_path": {
          "default": "",
          "description": "Path to the model to be quantized. For ONNX backend, path to ONNX file.",
          "title": "Model path",
          "type": "string"
        },
        "results_dir": {
          "default": "",
          "description": "Path to where all the assets generated from a task are stored.",
          "title": "Results directory",
          "type": "string"
        },
        "skip_names": {
          "automl_enabled": false,
          "default": [],
          "description": "List of module/layer names or patterns to exclude from quantization",
          "title": "Skip names",
          "type": "list"
        }
      },
      "required": [
        "model_path",
        "results_dir"
      ],
      "type": "collection"
    },
    "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"
      ],
      "automl_enabled": false,
      "default": {
        "checkpoint_interval": 1,
        "checkpoint_interval_unit": "epoch",
        "clip_grad_norm": 5.0,
        "cudnn": {
          "benchmark": false,
          "deterministic": true
        },
        "distributed_strategy": "ddp",
        "gpu_ids": [
          0
        ],
        "model_ema": false,
        "num_epochs": 10,
        "num_gpus": 1,
        "num_nodes": 1,
        "optim": {
          "lr": 1.0,
          "lr_decay": 0.1,
          "lr_monitor": "val_loss",
          "lr_scheduler": "MultiStep",
          "lr_steps": [
            15,
            25
          ],
          "min_lr": 0.0001,
          "momentum": 0.9,
          "name": "adadelta",
          "patience": 1,
          "weight_decay": 0.0005
        },
        "pretrained_model_path": "",
        "results_dir": "",
        "resume_training_checkpoint_path": "",
        "seed": 1234,
        "use_distributed_sampler": false,
        "validation_interval": 1
      },
      "description": "Configurable parameters for the training.",
      "popular": [
        "validation_interval",
        "checkpoint_interval",
        "num_gpus",
        "num_nodes",
        "num_epochs",
        "gpu_ids"
      ],
      "properties": {
        "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"
        },
        "clip_grad_norm": {
          "default": 5.0,
          "description": "The L2 magnitude of graident to be clipped in the training.",
          "type": "float"
        },
        "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": "The distributed strategy for multi-gpu training.",
          "type": "string"
        },
        "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"
        },
        "model_ema": {
          "default": false,
          "description": "The bool flag to enable model EMA.",
          "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"
          ],
          "automl_disabled_parameters": [
            "train.optim.lr_steps"
          ],
          "automl_enabled": false,
          "default": {
            "lr": 1.0,
            "lr_decay": 0.1,
            "lr_monitor": "val_loss",
            "lr_scheduler": "MultiStep",
            "lr_steps": [
              15,
              25
            ],
            "min_lr": 0.0001,
            "momentum": 0.9,
            "name": "adadelta",
            "patience": 1,
            "weight_decay": 0.0005
          },
          "description": "Configurable parameters for the optimizer.",
          "properties": {
            "lr": {
              "automl_enabled": true,
              "default": 1.0,
              "description": "Learning rate.",
              "maximum": Infinity,
              "minimum": 0.0,
              "title": "lr",
              "type": "float"
            },
            "lr_decay": {
              "default": 0.1,
              "description": "Learning rate decay factor in learning rate scheduler.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            },
            "lr_monitor": {
              "default": "val_loss",
              "description": "Learning rate monitor for AutoReduce learning rate scheduler.",
              "enum": [
                "val_loss",
                "train_loss"
              ],
              "type": "categorical"
            },
            "lr_scheduler": {
              "default": "MultiStep",
              "description": "Learning rate scheduler.",
              "enum": [
                "MultiStep",
                "AutoReduce"
              ],
              "type": "categorical"
            },
            "lr_steps": {
              "automl_enabled": false,
              "default": [
                15,
                25
              ],
              "description": "Steps to change learning rate in MultiStep scheduler.",
              "type": "list"
            },
            "min_lr": {
              "default": 0.0001,
              "description": "Minimum learning rate.",
              "maximum": Infinity,
              "minimum": 0.0,
              "type": "float"
            },
            "momentum": {
              "default": 0.9,
              "description": "Momentum coefficient.",
              "maximum": 1.0,
              "minimum": 0.0,
              "type": "float"
            },
            "name": {
              "default": "adadelta",
              "description": "Name of the optimizer.",
              "enum": [
                "adadelta",
                "adam"
              ],
              "title": "name",
              "type": "categorical"
            },
            "patience": {
              "default": 1,
              "description": "Number of epochs for AutoReduce learning rate scheduler tolerance.",
              "type": "int"
            },
            "weight_decay": {
              "default": 0.0005,
              "description": "Weight decay coefficient for trainng.",
              "maximum": Infinity,
              "minimum": 0.0,
              "type": "float"
            }
          },
          "type": "collection"
        },
        "pretrained_model_path": {
          "default": "",
          "description": "The absolute path to 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"
        },
        "use_distributed_sampler": {
          "default": false,
          "description": "Use distributed sampler for multi-GPU training",
          "type": "bool"
        },
        "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"
    }
  },
  "required": [
    "quantize"
  ],
  "type": "object",
  "x_tao_schema": {
    "action": "quantize",
    "core_module": "ocrnet",
    "model": "ocrnet",
    "network_arch": "ocrnet",
    "schema_action": "quantize",
    "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-ocrnet · 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-ocrnet.zip