[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78AXP2RCZ9E4SYCHSD9KYYV":3,"skill-same:01M78AXP2RCZ9E4SYCHSD9KYYV":127},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":32,"commitSha":33,"skillDirectory":34,"forksCount":14,"repositoryHtmlUrl":35,"repositoryHomepageUrl":36,"ownerHtmlUrl":37,"ownerBlogUrl":38,"ownerDescription":39,"updatedAt":40,"install":41,"resources":54},"01M78AXP2RCZ9E4SYCHSD9KYYV","tao-train-ocdnet",null,"OCDNet for scene text detection. Detects arbitrary-oriented text regions in natural images using a differentiable binarization approach. Use when training, evaluating, exporting, pruning, quantizing, retraining, or running inference for a TAO OCDNet model. Trigger phrases include \"train OCDNet\", \"scene text detection\", \"arbitrary-oriented text boxes\", \"differentiable binarization detector\".","Solizardking\u002Fskillhub-main","Solizardking","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F108096810?v=4","01M786KDB10AYV042N0F9BD5B5","Clawd","01M786NYYXSZK1D05WM8JQPC48",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-train-ocdnet\ndescription: OCDNet for scene text detection. Detects arbitrary-oriented text regions in natural images using a\n  differentiable binarization approach. Use when training, evaluating, exporting, pruning, quantizing, retraining, or running\n  inference for a TAO OCDNet model. Trigger phrases include \"train OCDNet\", \"scene text detection\", \"arbitrary-oriented text\n  boxes\", \"differentiable binarization 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- text\n- detection\n---\n\n# OCDNet\n\nOCDNet for scene text detection. Detects arbitrary-oriented text regions in natural images using a differentiable binarization approach.\n\nSet `model.pretrained_model_path` for pretrained weights.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-ocdnet.md` first. Deploy spec templates live in this skill's `references\u002F` folder with the `spec_template_deploy_*.yaml` prefix.\n\nThe PyT OCDNet CLI supports `train`, `evaluate`, `export`, `inference`, `prune`, `quantize`, and `default_specs`. It does not expose PyT-side `retrain` or `gen_trt_engine` subcommands. The model skill exposes `retrain` by running `ocdnet train` with `model.load_pruned_graph: true` and `model.pruned_graph_path`. Resume from an epoch checkpoint uses `ocdnet train` plus `train.resume_training_checkpoint_path`. TensorRT engine generation is owned by the deploy workflow.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nFor AutoML train, use `train_loss_epoch` or `train_loss` as the optimization\nmetric with `direction=minimize`. The Lightning progress log emits\n`train_loss_epoch`, and TAO `status.json` records the same final value under\n`train_loss`. For one-epoch local AutoML smoke runs, set\n`train.lr_scheduler.args.warmup_epoch: 0`; leaving warmup equal to the epoch\nbudget causes the trainer to fail before a recommendation can report a metric.\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:** ocdnet\n- **Formats:** default\n- **Monitoring metric:** hmean\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Runtime value | List? |\n|---|---|---|---|---|\n| evaluate | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| inference | inference.input_folder | inference_dataset or eval_dataset | extracted image folder | No |\n| prune | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.quant_calibration_dataset.images_dir | train_datasets or calibration_dataset | extracted calibration image folder | No |\n| train | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| train | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| retrain | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| retrain | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\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`. OCDNet does not unpack dataset archives at runtime. If the source is `train.tar.gz`, `test.tar.gz`, or `img.tar.gz`, extract it first and pass the split folder or image folder into the spec. The split folder must contain `img\u002F` and `gt\u002F`; alternatively, pass a UTF-8 datalist text file whose lines map image paths to label paths.\n\n```python\nTRAIN_ROOT = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftrain\"\nEVAL_ROOT = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftest\"\nINFER_IMG_DIR = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftest\u002Fimg\"\nCALIB_IMG_DIR = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftrain\u002Fimg\"\n```\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.train_dataset.loader.batch_size\": 16,\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"inference.input_folder\": INFER_IMG_DIR,\n}\n```\n\n**prune (mandatory data sources):**\n```python\n{\n    \"prune.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train checkpoint or exported ONNX>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n    \"dataset.quant_calibration_dataset.images_dir\": CALIB_IMG_DIR,\n}\n```\n\n**resume training (mandatory data sources):**\n```python\n{\n    \"train.resume_training_checkpoint_path\": \"\u003Cexact model_epoch checkpoint>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**retrain from prune output (mandatory data sources):**\n```python\n{\n    \"model.load_pruned_graph\": True,\n    \"model.pruned_graph_path\": \"\u003Cselected prune output>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**default_specs:**\n```python\n{\n    \"results_dir\": \"\u003Cwritable output directory>\",\n}\n```\n## Eval Dataset\n\nOptional. Test dataset provided as separate tarball.\n\n## Important Parameters\n\n- **model.backbone**: Default deformable_resnet18. Deformable convolutions improve text region detection for irregular text.\n- **train.optimizer.args.lr**: Learning rate. Default 0.001 (Adam).\n- **postprocess.thresh**: Binarization threshold for text region extraction.\n- **postprocess.box_thresh**: Box confidence threshold for filtering detections.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** Lightning-managed (single `python` process, Lightning spawns workers).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.distributed_strategy` | `ddp`, `fsdp`, or `deepspeed_stage_3_offload` | `ddp` |\n\n- `ddp` with activation checkpointing: `find_unused_parameters=False`\n- `ddp` without: `find_unused_parameters=True`\n- `fsdp` forces FP16\n- **`deepspeed_stage_3_offload`** is uniquely supported for OCDNet (forces FP16)\n- FAN backbones auto-enable `sync_batchnorm`\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCDNet is lightweight. Single GPU is sufficient for most datasets.\n\n## Error Patterns\n\n**Low detection rate**: Tune postprocess.thresh and box_thresh. Default thresholds may be too aggressive for some datasets.\n\n**One-epoch smoke train with default scheduler**: `train.num_epochs` must not equal `train.lr_scheduler.args.warmup_epoch`. For one-epoch validation, set `warmup_epoch: 0`; for normal starter runs, keep `num_epochs > warmup_epoch`.\n\n**Archive passed as dataset path**: `dataset.*.data_path` is not an archive path for OCDNet. Passing `train.tar.gz` or `test.tar.gz` directly causes the dataloader to open the gzip as a UTF-8 datalist. Extract the archive and pass the split folder containing `img\u002F` and `gt\u002F`, or pass a real UTF-8 datalist file.\n\n**Quantize checkpoint type**: Do not pass `model_best.pth` to the PyTorch quantize path. Some older PyT runtimes wrote `model_best.pth` without full Lightning checkpoint metadata. The default `torchao` quantize path should use the intended full `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` checkpoint and write `quantized_model_torchao.pth`.\n\n**Default specs output directory**: `ocdnet default_specs` requires a writable `results_dir` override, for example `results_dir=\u002Fworkspace\u002Frun\u002Fresults\u002Fdefault_specs`.\n\n## Checkpoint Handoff\n\nOCDNet train writes `model_best.pth` plus full Lightning epoch checkpoints such as `model_epoch_001_step_00046.pth`; it may also write `ocd_model_latest.pth` as a latest symlink. Use `model_best.pth` for `evaluate.checkpoint`, `inference.checkpoint`, `export.checkpoint`, and `prune.checkpoint` when the user asks for the best checkpoint. Use a specific `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` for `train.resume_training_checkpoint_path` and for any action that explicitly needs a full Lightning checkpoint. Prune writes artifacts such as `pruned_\u003Cch_sparsity>.pth`; use the exact pruned `.pth` artifact for `model.pruned_graph_path` when retraining from a pruned graph. Use a latest checkpoint only when the user explicitly asks for latest.\n\nIf quantize is retried with a PyTorch backend, resolve the full `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` that corresponds to the intended best epoch or requested epoch; do not pass `model_best.pth` to the PyTorch quantize path. If quantize is retried with `modelopt.onnx`, pass the exported ONNX as `quantize.model_path` and verify that the runtime image actually contains `modelopt.onnx.quantization`.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nModel handoff mappings:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `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 | `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 | `results_dir` | `output_dir` | current job results directory |\n| prune | `prune.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| prune | `results_dir` | `output_dir` | current job results directory |\n| quantize | `quantize.model_path` | `parent_model` | model file inferred from the parent job results folder |\n| quantize | `results_dir` | `output_dir` | current job results directory |\n| retrain from prune | `model.pruned_graph_path` | `parent_model` | exact pruned model file inferred from the parent prune results folder |\n| retrain from prune | `results_dir` | `output_dir` | current job results directory |\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-ocdnet](references\u002Ftao-deploy-ocdnet.md)\n","# OCDNet\n\nOCDNet for scene text detection. Detects arbitrary-oriented text regions in natural images using a differentiable binarization approach.\n\nSet `model.pretrained_model_path` for pretrained weights.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-ocdnet.md` first. Deploy spec templates live in this skill's `references\u002F` folder with the `spec_template_deploy_*.yaml` prefix.\n\nThe PyT OCDNet CLI supports `train`, `evaluate`, `export`, `inference`, `prune`, `quantize`, and `default_specs`. It does not expose PyT-side `retrain` or `gen_trt_engine` subcommands. The model skill exposes `retrain` by running `ocdnet train` with `model.load_pruned_graph: true` and `model.pruned_graph_path`. Resume from an epoch checkpoint uses `ocdnet train` plus `train.resume_training_checkpoint_path`. TensorRT engine generation is owned by the deploy workflow.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nFor AutoML train, use `train_loss_epoch` or `train_loss` as the optimization\nmetric with `direction=minimize`. The Lightning progress log emits\n`train_loss_epoch`, and TAO `status.json` records the same final value under\n`train_loss`. For one-epoch local AutoML smoke runs, set\n`train.lr_scheduler.args.warmup_epoch: 0`; leaving warmup equal to the epoch\nbudget causes the trainer to fail before a recommendation can report a metric.\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:** ocdnet\n- **Formats:** default\n- **Monitoring metric:** hmean\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Runtime value | List? |\n|---|---|---|---|---|\n| evaluate | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| inference | inference.input_folder | inference_dataset or eval_dataset | extracted image folder | No |\n| prune | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| quantize | dataset.quant_calibration_dataset.images_dir | train_datasets or calibration_dataset | extracted calibration image folder | No |\n| train | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| train | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\n| retrain | dataset.train_dataset.data_path | train_datasets | extracted train split folder with `img\u002F` and `gt\u002F` | Yes |\n| retrain | dataset.validate_dataset.data_path | eval_dataset | extracted validation split folder with `img\u002F` and `gt\u002F` | Yes |\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`. OCDNet does not unpack dataset archives at runtime. If the source is `train.tar.gz`, `test.tar.gz`, or `img.tar.gz`, extract it first and pass the split folder or image folder into the spec. The split folder must contain `img\u002F` and `gt\u002F`; alternatively, pass a UTF-8 datalist text file whose lines map image paths to label paths.\n\n```python\nTRAIN_ROOT = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftrain\"\nEVAL_ROOT = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftest\"\nINFER_IMG_DIR = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftest\u002Fimg\"\nCALIB_IMG_DIR = \"\u002Fpath\u002Fto\u002Fextracted\u002Ftrain\u002Fimg\"\n```\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.train_dataset.loader.batch_size\": 16,\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"inference.input_folder\": INFER_IMG_DIR,\n}\n```\n\n**prune (mandatory data sources):**\n```python\n{\n    \"prune.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train checkpoint or exported ONNX>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n    \"dataset.quant_calibration_dataset.images_dir\": CALIB_IMG_DIR,\n}\n```\n\n**resume training (mandatory data sources):**\n```python\n{\n    \"train.resume_training_checkpoint_path\": \"\u003Cexact model_epoch checkpoint>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**retrain from prune output (mandatory data sources):**\n```python\n{\n    \"model.load_pruned_graph\": True,\n    \"model.pruned_graph_path\": \"\u003Cselected prune output>\",\n    \"dataset.train_dataset.data_path\": [TRAIN_ROOT],\n    \"dataset.validate_dataset.data_path\": [EVAL_ROOT],\n}\n```\n\n**default_specs:**\n```python\n{\n    \"results_dir\": \"\u003Cwritable output directory>\",\n}\n```\n## Eval Dataset\n\nOptional. Test dataset provided as separate tarball.\n\n## Important Parameters\n\n- **model.backbone**: Default deformable_resnet18. Deformable convolutions improve text region detection for irregular text.\n- **train.optimizer.args.lr**: Learning rate. Default 0.001 (Adam).\n- **postprocess.thresh**: Binarization threshold for text region extraction.\n- **postprocess.box_thresh**: Box confidence threshold for filtering detections.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** Lightning-managed (single `python` process, Lightning spawns workers).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.distributed_strategy` | `ddp`, `fsdp`, or `deepspeed_stage_3_offload` | `ddp` |\n\n- `ddp` with activation checkpointing: `find_unused_parameters=False`\n- `ddp` without: `find_unused_parameters=True`\n- `fsdp` forces FP16\n- **`deepspeed_stage_3_offload`** is uniquely supported for OCDNet (forces FP16)\n- FAN backbones auto-enable `sync_batchnorm`\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCDNet is lightweight. Single GPU is sufficient for most datasets.\n\n## Error Patterns\n\n**Low detection rate**: Tune postprocess.thresh and box_thresh. Default thresholds may be too aggressive for some datasets.\n\n**One-epoch smoke train with default scheduler**: `train.num_epochs` must not equal `train.lr_scheduler.args.warmup_epoch`. For one-epoch validation, set `warmup_epoch: 0`; for normal starter runs, keep `num_epochs > warmup_epoch`.\n\n**Archive passed as dataset path**: `dataset.*.data_path` is not an archive path for OCDNet. Passing `train.tar.gz` or `test.tar.gz` directly causes the dataloader to open the gzip as a UTF-8 datalist. Extract the archive and pass the split folder containing `img\u002F` and `gt\u002F`, or pass a real UTF-8 datalist file.\n\n**Quantize checkpoint type**: Do not pass `model_best.pth` to the PyTorch quantize path. Some older PyT runtimes wrote `model_best.pth` without full Lightning checkpoint metadata. The default `torchao` quantize path should use the intended full `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` checkpoint and write `quantized_model_torchao.pth`.\n\n**Default specs output directory**: `ocdnet default_specs` requires a writable `results_dir` override, for example `results_dir=\u002Fworkspace\u002Frun\u002Fresults\u002Fdefault_specs`.\n\n## Checkpoint Handoff\n\nOCDNet train writes `model_best.pth` plus full Lightning epoch checkpoints such as `model_epoch_001_step_00046.pth`; it may also write `ocd_model_latest.pth` as a latest symlink. Use `model_best.pth` for `evaluate.checkpoint`, `inference.checkpoint`, `export.checkpoint`, and `prune.checkpoint` when the user asks for the best checkpoint. Use a specific `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` for `train.resume_training_checkpoint_path` and for any action that explicitly needs a full Lightning checkpoint. Prune writes artifacts such as `pruned_\u003Cch_sparsity>.pth`; use the exact pruned `.pth` artifact for `model.pruned_graph_path` when retraining from a pruned graph. Use a latest checkpoint only when the user explicitly asks for latest.\n\nIf quantize is retried with a PyTorch backend, resolve the full `model_epoch_\u003Cepoch>_step_\u003Cstep>.pth` that corresponds to the intended best epoch or requested epoch; do not pass `model_best.pth` to the PyTorch quantize path. If quantize is retried with `modelopt.onnx`, pass the exported ONNX as `quantize.model_path` and verify that the runtime image actually contains `modelopt.onnx.quantization`.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nModel handoff mappings:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `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 | `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 | `results_dir` | `output_dir` | current job results directory |\n| prune | `prune.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| prune | `results_dir` | `output_dir` | current job results directory |\n| quantize | `quantize.model_path` | `parent_model` | model file inferred from the parent job results folder |\n| quantize | `results_dir` | `output_dir` | current job results directory |\n| retrain from prune | `model.pruned_graph_path` | `parent_model` | exact pruned model file inferred from the parent prune results folder |\n| retrain from prune | `results_dir` | `output_dir` | current job results directory |\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-ocdnet](references\u002Ftao-deploy-ocdnet.md)\n",{"name":5,"tags":24,"license":15,"metadata":27,"description":7,"allowed-tools":30,"compatibility":31},[25,26],"text","detection",{"author":28,"version":29},"NVIDIA Corporation","0.1.0","Read Bash","Requires docker + nvidia-container-toolkit.","main","be3de0244dc1a57e323d8160787ebb40fb64930e","public\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main","https:\u002F\u002Fskills.x402.wtf","https:\u002F\u002Fgithub.com\u002FSolizardking","https:\u002F\u002Fmusebook.trade","Got a mile high and never came down","2026-10-08 23:20:13",{"source":8,"skillName":5,"commands":42,"prompt":52,"zipAvailable":53},[43,46,49],{"tool":44,"command":45},"npx","npx skills add Solizardking\u002Fskillhub-main --skill tao-train-ocdnet",{"tool":47,"command":48},"bunx","bunx skills add Solizardking\u002Fskillhub-main --skill tao-train-ocdnet",{"tool":50,"command":51},"pnpm","pnpm dlx skills add Solizardking\u002Fskillhub-main --skill tao-train-ocdnet","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002FSKILL.md",true,[55,61,65,69,73,77,81,85,89,93,97,101,105,108,111,115,119,123],{"path":56,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":59,"sourceHtmlUrl":60},"BENCHMARK.md","REFERENCE","参考文档",4048,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002FBENCHMARK.md",{"path":62,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":63,"sourceHtmlUrl":64},"metadata.json",599,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Fmetadata.json",{"path":66,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fskill_info.yaml",6137,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fskill_info.yaml",{"path":70,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fspec_template_deploy_evaluate.yaml",814,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_deploy_evaluate.yaml",{"path":74,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fspec_template_deploy_gen_trt_engine.yaml",445,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_deploy_gen_trt_engine.yaml",{"path":78,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fspec_template_deploy_inference.yaml",413,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_deploy_inference.yaml",{"path":82,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":83,"sourceHtmlUrl":84},"references\u002Fspec_template_evaluate.yaml",3357,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_evaluate.yaml",{"path":86,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":87,"sourceHtmlUrl":88},"references\u002Fspec_template_export.yaml",3143,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_export.yaml",{"path":90,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":91,"sourceHtmlUrl":92},"references\u002Fspec_template_gen_trt_engine.yaml",3445,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_gen_trt_engine.yaml",{"path":94,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":95,"sourceHtmlUrl":96},"references\u002Fspec_template_inference.yaml",3382,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_inference.yaml",{"path":98,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":99,"sourceHtmlUrl":100},"references\u002Fspec_template_prune.yaml",3119,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_prune.yaml",{"path":102,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":103,"sourceHtmlUrl":104},"references\u002Fspec_template_quantize.yaml",3006,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_quantize.yaml",{"path":106,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":103,"sourceHtmlUrl":107},"references\u002Fspec_template_retrain.yaml","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_retrain.yaml",{"path":109,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":103,"sourceHtmlUrl":110},"references\u002Fspec_template_train.yaml","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Fspec_template_train.yaml",{"path":112,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":113,"sourceHtmlUrl":114},"references\u002Ftao-deploy-ocdnet.md",6233,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Ftao-deploy-ocdnet.md",{"path":116,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":117,"sourceHtmlUrl":118},"references\u002Ftao-deploy-ocdnet.skill_info.yaml",3378,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Freferences\u002Ftao-deploy-ocdnet.skill_info.yaml",{"path":120,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":121,"sourceHtmlUrl":122},"skill-card.md",3720,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Fskill-card.md",{"path":124,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":125,"sourceHtmlUrl":126},"verification.json",4114,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-ocdnet\u002Fverification.json",[128,136,143,150,157,165],{"skillNo":129,"name":130,"nameZh":6,"description":131,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":135,"publishedAt":19},"01M78E0TGCAMNPS207BA3G07S7","zkrouter","Self-hosted OpenAI-compatible LLM router — smart tier-based routing, ZK-stamped routing decisions, mode overrides, free OpenRouter at install via the Birth bot. Save 60-80% on inference costs by routing to the cheapest capable model across 12+ OpenRouter models.","MIT","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzkrouter\u002FSKILL.md",[],[],{"skillNo":137,"name":138,"nameZh":6,"description":139,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":6,"sourceHtmlUrl":140,"categories":141,"keywords":142,"publishedAt":19},"01M78E0R1Z8BDT6EDTHY17MTNB","zk","For custom ZK Solana programs and privacy-preserving applications to prevent double spending. Guide to integrate rent-free nullifier PDAs for double-spend prevention.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzk\u002FSKILL.md",[],[],{"skillNo":144,"name":145,"nameZh":6,"description":146,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":6,"sourceHtmlUrl":147,"categories":148,"keywords":149,"publishedAt":19},"01M78E0P3ZTZ1YWF8MD5MM23QE","zk-omni-messaging","Zero-knowledge omnichain messaging between Robinhood Chain and Solana using CheshireZkOmniMessenger (msgType 4) with nullifier anti-replay and the zk-omni-relayer service.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzk-omni-messaging\u002FSKILL.md",[],[],{"skillNo":151,"name":152,"nameZh":6,"description":153,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":6,"sourceHtmlUrl":154,"categories":155,"keywords":156,"publishedAt":19},"01M78E0JXZ2PNB1896VXAGDBWD","zerion-wallet-data","Use when the user asks to check a wallet's portfolio, token balances, DeFi positions, PnL, transaction history, or NFTs. Provides interpreted on-chain data across 40+ EVM chains and Solana via x402 pay-per-request or API key — no raw RPC decoding needed.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzerion-wallet-data\u002FSKILL.md",[],[],{"skillNo":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":161,"sourceHtmlUrl":162,"categories":163,"keywords":164,"publishedAt":19},"01M78E0G14FH9GRRTQYMKD5E59","youtube-clipper","YouTube 视频智能剪辑工具。下载视频和字幕，AI 分析生成精细章节（几分钟级别）， 用户选择片段后自动剪辑、翻译字幕为中英双语、烧录字幕到视频，并生成总结文案。 使用场景：当用户需要剪辑 YouTube 视频、生成短视频片段、制作双语字幕版本时。 关键词：视频剪辑、YouTube、字幕翻译、双语字幕、视频下载、clip video","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FSKILL.md",[],[],{"skillNo":166,"name":167,"nameZh":6,"description":168,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":132,"licenseFileUrl":6,"sourceHtmlUrl":169,"categories":170,"keywords":171,"publishedAt":19},"01M78E0C2ZYA99CFFDNB7ZMK5S","yield-optimization","Multi-chain yield optimization via yield.xyz (StakeKit). Discover 2,988+ yield opportunities across 75+ blockchains, build deposit\u002Fwithdrawal transactions with shell scripts, and sign them with a MoonPay wallet. Use when the user wants to earn yield, find the best lending or staking rate, enter\u002Fexit a position, or check portfolio balances.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyield-optimization\u002FSKILL.md",[],[]]