[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-train-ocrnet":3,"mdc-z62n48-key":31,"related-org-nvidia-tao-train-ocrnet":3589,"related-repo-nvidia-tao-train-ocrnet":3747},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":20,"repoUrl":21,"updatedAt":22,"license":23,"forks":24,"topics":25,"repo":26,"sourceUrl":29,"mdContent":30},"tao-train-ocrnet","train OCRNet for scene text recognition","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 \u002F attention text decoder\".",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,17],{"name":13,"slug":14,"type":15},"Deep Learning","deep-learning","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"Computer Vision","computer-vision",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:30:47.797337","Apache-2.0",281,[],{"repoUrl":21,"stars":20,"forks":24,"topics":27,"description":28},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftao-train-ocrnet","---\nname: tao-train-ocrnet\ndescription: OCRNet for scene text recognition. Recognizes text content from cropped text-region images and supports CTC\n  and attention-based decoders. Use when training, evaluating, exporting, pruning, quantizing, retraining, or running\n  inference for a TAO OCRNet model. Trigger phrases include \"train OCRNet\", \"scene text recognition\", \"OCR cropped text\",\n  \"CTC \u002F attention text decoder\".\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- recognition\n---\n\n# OCRNet\n\nOCRNet for scene text recognition. Recognizes text content from cropped text region images. Supports CTC and attention-based decoders.\n\nSet train.pretrained_model_path for pretrained OCR weights.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-ocrnet.md` first. Deploy spec templates live in this skill's `references\u002F` folder with the `spec_template_deploy_*.yaml` prefix.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nNon-train actions such as `evaluate`, `inference`, `export`, and deploy flows stay in this model skill. The per-run `automl_policy` override does not change model metadata.\n\n## Training Requirements\n\n- **Dataset type:** ocrnet\n- **Formats:** default\n- **Monitoring metric:** val_acc_1\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| dataset_convert | dataset_convert.input_img_dir | train_datasets or eval_dataset | extracted folder containing cropped text images | No |\n| dataset_convert | dataset_convert.gt_file | train_datasets or eval_dataset | train\u002Fgt_new.txt or test\u002Fgt_new.txt | No |\n| evaluate | dataset.character_list_file | eval_dataset | character_list | No |\n| evaluate | evaluate.test_dataset_dir | eval_dataset | extracted test image folder | No |\n| evaluate | evaluate.test_dataset_gt_file | eval_dataset | test\u002Fgt_new.txt | No |\n| evaluate | evaluate.checkpoint | parent train\u002FAutoML job | best_accuracy.pth or exact requested epoch checkpoint | No |\n| export | dataset.character_list_file | eval_dataset | character_list | No |\n| export | export.checkpoint | parent train\u002FAutoML job | best_accuracy.pth or exact requested epoch checkpoint | No |\n| deploy\u002Fgen_trt_engine | gen_trt_engine.tensorrt.calibration.cal_image_dir | calibration_dataset | extracted calibration image folder for INT8 calibration | Yes |\n| deploy\u002Fgen_trt_engine | gen_trt_engine.onnx_file | parent export job | exported .onnx artifact | No |\n| deploy\u002Fgen_trt_engine | dataset.character_list_file | eval_dataset | character_list | No |\n| inference | dataset.character_list_file | eval_dataset | character_list | No |\n| inference | inference.inference_dataset_dir | inference_dataset | extracted inference image folder | No |\n| inference | inference.checkpoint | parent train\u002FAutoML job | best_accuracy.pth or exact requested epoch checkpoint | No |\n| prune | dataset.character_list_file | eval_dataset | character_list | No |\n| prune | prune.checkpoint | parent train\u002FAutoML job | best_accuracy.pth or exact requested epoch checkpoint | No |\n| quantize | dataset.train_dataset_dir | dataset_convert train job | LMDB folder containing data.mdb and lock.mdb | Yes |\n| quantize | dataset.val_dataset_dir | dataset_convert eval job | LMDB folder containing data.mdb and lock.mdb | No |\n| quantize | dataset.character_list_file | eval_dataset | character_list | No |\n| quantize | dataset.quant_calibration_dataset.images_dir | train_datasets | extracted calibration image folder | No |\n| quantize | quantize.model_path | parent train\u002FAutoML job | checkpoint selected by resolver | No |\n| retrain | dataset.train_dataset_dir | dataset_convert train job | LMDB folder containing data.mdb and lock.mdb | Yes |\n| retrain | dataset.val_dataset_dir | dataset_convert eval job | LMDB folder containing data.mdb and lock.mdb | No |\n| retrain | dataset.character_list_file | eval_dataset | character_list | No |\n| retrain | model.pruned_graph_path | parent prune job | pruned .pth artifact | No |\n| train | dataset.train_dataset_dir | dataset_convert train job | LMDB folder containing data.mdb and lock.mdb | Yes |\n| train | dataset.train_gt_file | train_datasets | train\u002Fgt_new.txt when using raw folders instead of LMDB | No |\n| train | dataset.val_dataset_dir | dataset_convert eval job | LMDB folder containing data.mdb and lock.mdb | No |\n| train | dataset.val_gt_file | eval_dataset | test\u002Fgt_new.txt when using raw folders instead of LMDB | No |\n| train | dataset.character_list_file | eval_dataset | character_list | No |\n\n### Checkpoint Selection\n\nOCRNet training writes both `best_accuracy.pth` and epoch-step checkpoints such as `model_epoch_000_step_00003.pth`. Use the SDK\u002Fmodel checkpoint resolver through the `spec_params` mappings in `references\u002Fskill_info.yaml`; do not guess by sorting for the newest `.pth`.\n\n- Use `best_accuracy.pth` for best-checkpoint `evaluate`, `inference`, `export`, and `prune` requests.\n- Use the exact requested `model_epoch_*_step_*.pth` for epoch\u002Fstep-specific actions.\n- 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.\n- 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.\n\n### Typical Spec Overrides\n\nData 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.\n\n```python\nTRAIN_IMAGES = \"\u003Cextracted train image folder>\"\nTRAIN_GT = \"\u003Ctrain gt_new.txt>\"\nEVAL_IMAGES = \"\u003Cextracted eval image folder>\"\nEVAL_GT = \"\u003Ceval gt_new.txt>\"\nTRAIN_LMDB = \"\u003Ctrain dataset_convert results_dir>\"\nEVAL_LMDB = \"\u003Ceval dataset_convert results_dir>\"\nCHAR_LIST = \"\u003Ccharacter_list>\"\n```\n\n**dataset_convert (run once per split):**\n```python\n{\n    \"dataset_convert.input_img_dir\": TRAIN_IMAGES,\n    \"dataset_convert.gt_file\": TRAIN_GT,\n}\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.batch_size\": 16,\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.train_gt_file\": \"\",\n    \"dataset.val_gt_file\": \"\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n```\n\n**deploy\u002Fgen_trt_engine (mandatory data sources):**\n```python\n{\n    \"gen_trt_engine.onnx_file\": \"\u003Cselected export ONNX>\",\n    \"gen_trt_engine.trt_engine\": \"\u003Coutput engine path>\",\n    \"gen_trt_engine.tensorrt.calibration.cal_cache_file\": \"\u003Coutput calibration cache path>\",\n    \"gen_trt_engine.tensorrt.data_type\": \"fp16\",\n    \"gen_trt_engine.tensorrt.calibration.cal_image_dir\": [TRAIN_IMAGES],\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"evaluate.test_dataset_dir\": EVAL_IMAGES,\n    \"evaluate.test_dataset_gt_file\": EVAL_GT,\n}\n```\n\n**export (mandatory data sources):**\n```python\n{\n    \"export.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"export.onnx_file\": \"\u003Coutput ONNX path>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"inference.inference_dataset_dir\": EVAL_IMAGES,\n}\n```\n\n**prune (mandatory data sources):**\n```python\n{\n    \"prune.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"prune.pruned_file\": \"\u003Coutput pruned PTH path>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"dataset.quant_calibration_dataset.images_dir\": TRAIN_IMAGES,\n    \"quantize.model_path\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n}\n```\n\n**retrain (mandatory data sources):**\n```python\n{\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"model.pruned_graph_path\": \"\u003Cselected prune output>\",\n}\n```\n## Eval Dataset\n\nOptional. Test data provided as separate tarball.\n\n## Important Parameters\n\n- **dataset.character_list_file**: Path to character list defining the supported character set. This determines the output vocabulary size.\n- **model.backbone**: Default ResNet.\n- **model.prediction**: Decoder type. CTC or Attn (attention-based).\n- **train.optim.lr**: Learning rate. Default 1.0 (Adadelta optimizer). High default is specific to Adadelta.\n- **dataset.batch_size**: Per-GPU batch size. Default 16.\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` | Strategy name | `auto` |\n\n- Strategy: `auto` for single-GPU, reads `train.distributed_strategy` from config when multi-GPU\n- No explicit `num_nodes` in train script — single-node oriented\n- Lightweight model, single GPU typically sufficient\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCR text recognition is lightweight. Single GPU is typically sufficient.\n\n## Error Patterns\n\n**dataset_convert required**: If using raw images + gt files, run dataset_convert first to produce LMDB format.\n\n**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`.\n\n**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.\n\n**Character list mismatch**: All characters in training data must be present in the character_list file.\n\n**Export\u002Fprune output fields required**: `export.onnx_file` and `prune.pruned_file` must be writable output paths. These are declared in `references\u002Fskill_info.yaml` so SDK-backed model runs can create the paths automatically.\n\n**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\u002Ftao-deploy-ocrnet.md` and `deploy\u002Fskill_info.yaml` for TensorRT engine generation and TensorRT-backed evaluate\u002Finference.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nInference mappings from TAO Core `ocrnet.config.json`:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| dataset_convert | `results_dir` | `output_dir` | current job results directory |\n| evaluate | `encryption_key` | `key` | encryption key |\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `evaluate.trt_engine` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `model.pruned_graph_path` | `pruned_model` | parent pruned model |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `encryption_key` | `key` | encryption key |\n| export | `export.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| export | `export.onnx_file` | `create_onnx_file` | output ONNX path |\n| export | `results_dir` | `output_dir` | current job results directory |\n| deploy\u002Fgen_trt_engine | `encryption_key` | `key` | encryption key |\n| deploy\u002Fgen_trt_engine | `gen_trt_engine.onnx_file` | `parent_model` | ONNX file inferred from the parent export job results folder |\n| deploy\u002Fgen_trt_engine | `gen_trt_engine.tensorrt.calibration.cal_cache_file` | `create_cal_cache` | calibration cache path |\n| deploy\u002Fgen_trt_engine | `gen_trt_engine.trt_engine` | `create_engine_file` | output TensorRT engine path |\n| deploy\u002Fgen_trt_engine | `results_dir` | `output_dir` | current job results directory |\n| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `inference.trt_engine` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `model.pruned_graph_path` | `pruned_model` | parent pruned model |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| prune | `encryption_key` | `key` | encryption key |\n| prune | `prune.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| prune | `prune.pruned_file` | `create_pth_file` | output PTH path |\n| prune | `results_dir` | `output_dir` | current job results directory |\n| quantize | `encryption_key` | `key` | encryption key |\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 | `encryption_key` | `key` | encryption key |\n| retrain | `model.pruned_graph_path` | `parent_model` | model file inferred from the parent job results folder |\n| retrain | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.pretrained_model_path` | `ptm_if_no_resume_model` | PTM when no resume checkpoint exists |\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-ocrnet](references\u002Ftao-deploy-ocrnet.md)\n",{"data":32,"body":41},{"name":4,"description":6,"license":23,"compatibility":33,"metadata":34,"allowed-tools":37,"tags":38},"Requires docker + nvidia-container-toolkit.",{"version":35,"author":36},"0.1.0","NVIDIA Corporation","Read Bash",[39,40],"text","recognition",{"type":42,"children":43},"root",[44,52,58,63,117,124,217,223,332,364,370,406,413,1198,1204,1248,1364,1370,1404,1480,1488,1527,1535,1641,1649,1717,1725,1777,1785,1829,1837,1881,1889,1933,1941,1999,2007,2057,2063,2068,2074,2126,2132,2149,2248,2288,2294,2299,2305,2315,2379,2389,2399,2431,2505,2511,2540,2553,3530,3565,3571,3583],{"type":45,"tag":46,"props":47,"children":49},"element","h1",{"id":48},"ocrnet",[50],{"type":39,"value":51},"OCRNet",{"type":45,"tag":53,"props":54,"children":55},"p",{},[56],{"type":39,"value":57},"OCRNet for scene text recognition. Recognizes text content from cropped text region images. Supports CTC and attention-based decoders.",{"type":45,"tag":53,"props":59,"children":60},{},[61],{"type":39,"value":62},"Set train.pretrained_model_path for pretrained OCR weights.",{"type":45,"tag":53,"props":64,"children":65},{},[66,68,75,77,83,85,91,93,99,101,107,109,115],{"type":39,"value":67},"For TAO Deploy TensorRT actions (",{"type":45,"tag":69,"props":70,"children":72},"code",{"className":71},[],[73],{"type":39,"value":74},"gen_trt_engine",{"type":39,"value":76},", TensorRT ",{"type":45,"tag":69,"props":78,"children":80},{"className":79},[],[81],{"type":39,"value":82},"evaluate",{"type":39,"value":84},", and TensorRT ",{"type":45,"tag":69,"props":86,"children":88},{"className":87},[],[89],{"type":39,"value":90},"inference",{"type":39,"value":92},"), read ",{"type":45,"tag":69,"props":94,"children":96},{"className":95},[],[97],{"type":39,"value":98},"references\u002Ftao-deploy-ocrnet.md",{"type":39,"value":100}," first. Deploy spec templates live in this skill's ",{"type":45,"tag":69,"props":102,"children":104},{"className":103},[],[105],{"type":39,"value":106},"references\u002F",{"type":39,"value":108}," folder with the ",{"type":45,"tag":69,"props":110,"children":112},{"className":111},[],[113],{"type":39,"value":114},"spec_template_deploy_*.yaml",{"type":39,"value":116}," prefix.",{"type":45,"tag":118,"props":119,"children":121},"h2",{"id":120},"dataclass-schemas",[122],{"type":39,"value":123},"Dataclass Schemas",{"type":45,"tag":53,"props":125,"children":126},{},[127,129,135,137,143,145,151,153,159,161,167,169,175,177,183,185,191,193,199,201,207,209,215],{"type":39,"value":128},"Generated TAO Core schemas are packaged in ",{"type":45,"tag":69,"props":130,"children":132},{"className":131},[],[133],{"type":39,"value":134},"schemas\u002F\u003Caction>.schema.json",{"type":39,"value":136},", with ",{"type":45,"tag":69,"props":138,"children":140},{"className":139},[],[141],{"type":39,"value":142},"schemas\u002Fmanifest.json",{"type":39,"value":144}," listing available actions. Each generated schema also emits ",{"type":45,"tag":69,"props":146,"children":148},{"className":147},[],[149],{"type":39,"value":150},"references\u002Fspec_template_\u003Caction>.yaml",{"type":39,"value":152}," from the schema top-level ",{"type":45,"tag":69,"props":154,"children":156},{"className":155},[],[157],{"type":39,"value":158},"default",{"type":39,"value":160}," field. AutoML enablement is declared at the model layer in ",{"type":45,"tag":69,"props":162,"children":164},{"className":163},[],[165],{"type":39,"value":166},"references\u002Fskill_info.yaml",{"type":39,"value":168}," via ",{"type":45,"tag":69,"props":170,"children":172},{"className":171},[],[173],{"type":39,"value":174},"automl_enabled",{"type":39,"value":176},". Runnable AutoML still requires ",{"type":45,"tag":69,"props":178,"children":180},{"className":179},[],[181],{"type":39,"value":182},"schemas\u002Ftrain.schema.json",{"type":39,"value":184}," and ",{"type":45,"tag":69,"props":186,"children":188},{"className":187},[],[189],{"type":39,"value":190},"references\u002Fspec_template_train.yaml",{"type":39,"value":192}," to exist and parse. Use the packaged train schema for ",{"type":45,"tag":69,"props":194,"children":196},{"className":195},[],[197],{"type":39,"value":198},"automl_default_parameters",{"type":39,"value":200},", ",{"type":45,"tag":69,"props":202,"children":204},{"className":203},[],[205],{"type":39,"value":206},"automl_disabled_parameters",{"type":39,"value":208},", defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect ",{"type":45,"tag":69,"props":210,"children":212},{"className":211},[],[213],{"type":39,"value":214},"~\u002Ftao-core",{"type":39,"value":216}," at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.",{"type":45,"tag":118,"props":218,"children":220},{"id":219},"train-action-policy",[221],{"type":39,"value":222},"Train Action Policy",{"type":45,"tag":53,"props":224,"children":225},{},[226,228,233,235,241,243,249,251,257,259,265,267,273,275,280,281,287,289,294,295,300,302,308,310,316,318,323,325,330],{"type":39,"value":227},"This model is AutoML-enabled at the model layer. Before handling any train-stage request, read ",{"type":45,"tag":69,"props":229,"children":231},{"className":230},[],[232],{"type":39,"value":166},{"type":39,"value":234}," and resolve the run override from either an explicit ",{"type":45,"tag":69,"props":236,"children":238},{"className":237},[],[239],{"type":39,"value":240},"automl_policy",{"type":39,"value":242}," value or the user's workflow request. Use ",{"type":45,"tag":69,"props":244,"children":246},{"className":245},[],[247],{"type":39,"value":248},"automl_policy: on",{"type":39,"value":250}," by default and only expose ",{"type":45,"tag":69,"props":252,"children":254},{"className":253},[],[255],{"type":39,"value":256},"on",{"type":39,"value":258}," \u002F ",{"type":45,"tag":69,"props":260,"children":262},{"className":261},[],[263],{"type":39,"value":264},"off",{"type":39,"value":266}," in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as ",{"type":45,"tag":69,"props":268,"children":270},{"className":269},[],[271],{"type":39,"value":272},"automl_policy: off",{"type":39,"value":274}," for this run only. When ",{"type":45,"tag":69,"props":276,"children":278},{"className":277},[],[279],{"type":39,"value":248},{"type":39,"value":200},{"type":45,"tag":69,"props":282,"children":284},{"className":283},[],[285],{"type":39,"value":286},"automl_enabled: true",{"type":39,"value":288},", and both ",{"type":45,"tag":69,"props":290,"children":292},{"className":291},[],[293],{"type":39,"value":182},{"type":39,"value":184},{"type":45,"tag":69,"props":296,"children":298},{"className":297},[],[299],{"type":39,"value":190},{"type":39,"value":301}," are packaged, route the train action through ",{"type":45,"tag":69,"props":303,"children":305},{"className":304},[],[306],{"type":39,"value":307},"tao-skill-bank:tao-run-automl",{"type":39,"value":309}," by default with this model's ",{"type":45,"tag":69,"props":311,"children":313},{"className":312},[],[314],{"type":39,"value":315},"skill_dir",{"type":39,"value":317},". Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and ",{"type":45,"tag":69,"props":319,"children":321},{"className":320},[],[322],{"type":39,"value":240},{"type":39,"value":324},". Use direct model training only when ",{"type":45,"tag":69,"props":326,"children":328},{"className":327},[],[329],{"type":39,"value":272},{"type":39,"value":331}," or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.",{"type":45,"tag":53,"props":333,"children":334},{},[335,337,342,343,348,349,355,357,362],{"type":39,"value":336},"Non-train actions such as ",{"type":45,"tag":69,"props":338,"children":340},{"className":339},[],[341],{"type":39,"value":82},{"type":39,"value":200},{"type":45,"tag":69,"props":344,"children":346},{"className":345},[],[347],{"type":39,"value":90},{"type":39,"value":200},{"type":45,"tag":69,"props":350,"children":352},{"className":351},[],[353],{"type":39,"value":354},"export",{"type":39,"value":356},", and deploy flows stay in this model skill. The per-run ",{"type":45,"tag":69,"props":358,"children":360},{"className":359},[],[361],{"type":39,"value":240},{"type":39,"value":363}," override does not change model metadata.",{"type":45,"tag":118,"props":365,"children":367},{"id":366},"training-requirements",[368],{"type":39,"value":369},"Training Requirements",{"type":45,"tag":371,"props":372,"children":373},"ul",{},[374,386,396],{"type":45,"tag":375,"props":376,"children":377},"li",{},[378,384],{"type":45,"tag":379,"props":380,"children":381},"strong",{},[382],{"type":39,"value":383},"Dataset type:",{"type":39,"value":385}," ocrnet",{"type":45,"tag":375,"props":387,"children":388},{},[389,394],{"type":45,"tag":379,"props":390,"children":391},{},[392],{"type":39,"value":393},"Formats:",{"type":39,"value":395}," default",{"type":45,"tag":375,"props":397,"children":398},{},[399,404],{"type":45,"tag":379,"props":400,"children":401},{},[402],{"type":39,"value":403},"Monitoring metric:",{"type":39,"value":405}," val_acc_1",{"type":45,"tag":407,"props":408,"children":410},"h3",{"id":409},"per-action-dataset-requirements",[411],{"type":39,"value":412},"Per-Action Dataset Requirements",{"type":45,"tag":414,"props":415,"children":416},"table",{},[417,451],{"type":45,"tag":418,"props":419,"children":420},"thead",{},[421],{"type":45,"tag":422,"props":423,"children":424},"tr",{},[425,431,436,441,446],{"type":45,"tag":426,"props":427,"children":428},"th",{},[429],{"type":39,"value":430},"Action",{"type":45,"tag":426,"props":432,"children":433},{},[434],{"type":39,"value":435},"Spec Key",{"type":45,"tag":426,"props":437,"children":438},{},[439],{"type":39,"value":440},"Source",{"type":45,"tag":426,"props":442,"children":443},{},[444],{"type":39,"value":445},"Files",{"type":45,"tag":426,"props":447,"children":448},{},[449],{"type":39,"value":450},"List?",{"type":45,"tag":452,"props":453,"children":454},"tbody",{},[455,484,509,535,560,585,611,634,658,686,712,735,758,784,808,832,856,883,908,931,957,982,1006,1029,1052,1078,1102,1127,1150,1175],{"type":45,"tag":422,"props":456,"children":457},{},[458,464,469,474,479],{"type":45,"tag":459,"props":460,"children":461},"td",{},[462],{"type":39,"value":463},"dataset_convert",{"type":45,"tag":459,"props":465,"children":466},{},[467],{"type":39,"value":468},"dataset_convert.input_img_dir",{"type":45,"tag":459,"props":470,"children":471},{},[472],{"type":39,"value":473},"train_datasets or eval_dataset",{"type":45,"tag":459,"props":475,"children":476},{},[477],{"type":39,"value":478},"extracted folder containing cropped text images",{"type":45,"tag":459,"props":480,"children":481},{},[482],{"type":39,"value":483},"No",{"type":45,"tag":422,"props":485,"children":486},{},[487,491,496,500,505],{"type":45,"tag":459,"props":488,"children":489},{},[490],{"type":39,"value":463},{"type":45,"tag":459,"props":492,"children":493},{},[494],{"type":39,"value":495},"dataset_convert.gt_file",{"type":45,"tag":459,"props":497,"children":498},{},[499],{"type":39,"value":473},{"type":45,"tag":459,"props":501,"children":502},{},[503],{"type":39,"value":504},"train\u002Fgt_new.txt or test\u002Fgt_new.txt",{"type":45,"tag":459,"props":506,"children":507},{},[508],{"type":39,"value":483},{"type":45,"tag":422,"props":510,"children":511},{},[512,516,521,526,531],{"type":45,"tag":459,"props":513,"children":514},{},[515],{"type":39,"value":82},{"type":45,"tag":459,"props":517,"children":518},{},[519],{"type":39,"value":520},"dataset.character_list_file",{"type":45,"tag":459,"props":522,"children":523},{},[524],{"type":39,"value":525},"eval_dataset",{"type":45,"tag":459,"props":527,"children":528},{},[529],{"type":39,"value":530},"character_list",{"type":45,"tag":459,"props":532,"children":533},{},[534],{"type":39,"value":483},{"type":45,"tag":422,"props":536,"children":537},{},[538,542,547,551,556],{"type":45,"tag":459,"props":539,"children":540},{},[541],{"type":39,"value":82},{"type":45,"tag":459,"props":543,"children":544},{},[545],{"type":39,"value":546},"evaluate.test_dataset_dir",{"type":45,"tag":459,"props":548,"children":549},{},[550],{"type":39,"value":525},{"type":45,"tag":459,"props":552,"children":553},{},[554],{"type":39,"value":555},"extracted test image folder",{"type":45,"tag":459,"props":557,"children":558},{},[559],{"type":39,"value":483},{"type":45,"tag":422,"props":561,"children":562},{},[563,567,572,576,581],{"type":45,"tag":459,"props":564,"children":565},{},[566],{"type":39,"value":82},{"type":45,"tag":459,"props":568,"children":569},{},[570],{"type":39,"value":571},"evaluate.test_dataset_gt_file",{"type":45,"tag":459,"props":573,"children":574},{},[575],{"type":39,"value":525},{"type":45,"tag":459,"props":577,"children":578},{},[579],{"type":39,"value":580},"test\u002Fgt_new.txt",{"type":45,"tag":459,"props":582,"children":583},{},[584],{"type":39,"value":483},{"type":45,"tag":422,"props":586,"children":587},{},[588,592,597,602,607],{"type":45,"tag":459,"props":589,"children":590},{},[591],{"type":39,"value":82},{"type":45,"tag":459,"props":593,"children":594},{},[595],{"type":39,"value":596},"evaluate.checkpoint",{"type":45,"tag":459,"props":598,"children":599},{},[600],{"type":39,"value":601},"parent train\u002FAutoML job",{"type":45,"tag":459,"props":603,"children":604},{},[605],{"type":39,"value":606},"best_accuracy.pth or exact requested epoch checkpoint",{"type":45,"tag":459,"props":608,"children":609},{},[610],{"type":39,"value":483},{"type":45,"tag":422,"props":612,"children":613},{},[614,618,622,626,630],{"type":45,"tag":459,"props":615,"children":616},{},[617],{"type":39,"value":354},{"type":45,"tag":459,"props":619,"children":620},{},[621],{"type":39,"value":520},{"type":45,"tag":459,"props":623,"children":624},{},[625],{"type":39,"value":525},{"type":45,"tag":459,"props":627,"children":628},{},[629],{"type":39,"value":530},{"type":45,"tag":459,"props":631,"children":632},{},[633],{"type":39,"value":483},{"type":45,"tag":422,"props":635,"children":636},{},[637,641,646,650,654],{"type":45,"tag":459,"props":638,"children":639},{},[640],{"type":39,"value":354},{"type":45,"tag":459,"props":642,"children":643},{},[644],{"type":39,"value":645},"export.checkpoint",{"type":45,"tag":459,"props":647,"children":648},{},[649],{"type":39,"value":601},{"type":45,"tag":459,"props":651,"children":652},{},[653],{"type":39,"value":606},{"type":45,"tag":459,"props":655,"children":656},{},[657],{"type":39,"value":483},{"type":45,"tag":422,"props":659,"children":660},{},[661,666,671,676,681],{"type":45,"tag":459,"props":662,"children":663},{},[664],{"type":39,"value":665},"deploy\u002Fgen_trt_engine",{"type":45,"tag":459,"props":667,"children":668},{},[669],{"type":39,"value":670},"gen_trt_engine.tensorrt.calibration.cal_image_dir",{"type":45,"tag":459,"props":672,"children":673},{},[674],{"type":39,"value":675},"calibration_dataset",{"type":45,"tag":459,"props":677,"children":678},{},[679],{"type":39,"value":680},"extracted calibration image folder for INT8 calibration",{"type":45,"tag":459,"props":682,"children":683},{},[684],{"type":39,"value":685},"Yes",{"type":45,"tag":422,"props":687,"children":688},{},[689,693,698,703,708],{"type":45,"tag":459,"props":690,"children":691},{},[692],{"type":39,"value":665},{"type":45,"tag":459,"props":694,"children":695},{},[696],{"type":39,"value":697},"gen_trt_engine.onnx_file",{"type":45,"tag":459,"props":699,"children":700},{},[701],{"type":39,"value":702},"parent export job",{"type":45,"tag":459,"props":704,"children":705},{},[706],{"type":39,"value":707},"exported .onnx artifact",{"type":45,"tag":459,"props":709,"children":710},{},[711],{"type":39,"value":483},{"type":45,"tag":422,"props":713,"children":714},{},[715,719,723,727,731],{"type":45,"tag":459,"props":716,"children":717},{},[718],{"type":39,"value":665},{"type":45,"tag":459,"props":720,"children":721},{},[722],{"type":39,"value":520},{"type":45,"tag":459,"props":724,"children":725},{},[726],{"type":39,"value":525},{"type":45,"tag":459,"props":728,"children":729},{},[730],{"type":39,"value":530},{"type":45,"tag":459,"props":732,"children":733},{},[734],{"type":39,"value":483},{"type":45,"tag":422,"props":736,"children":737},{},[738,742,746,750,754],{"type":45,"tag":459,"props":739,"children":740},{},[741],{"type":39,"value":90},{"type":45,"tag":459,"props":743,"children":744},{},[745],{"type":39,"value":520},{"type":45,"tag":459,"props":747,"children":748},{},[749],{"type":39,"value":525},{"type":45,"tag":459,"props":751,"children":752},{},[753],{"type":39,"value":530},{"type":45,"tag":459,"props":755,"children":756},{},[757],{"type":39,"value":483},{"type":45,"tag":422,"props":759,"children":760},{},[761,765,770,775,780],{"type":45,"tag":459,"props":762,"children":763},{},[764],{"type":39,"value":90},{"type":45,"tag":459,"props":766,"children":767},{},[768],{"type":39,"value":769},"inference.inference_dataset_dir",{"type":45,"tag":459,"props":771,"children":772},{},[773],{"type":39,"value":774},"inference_dataset",{"type":45,"tag":459,"props":776,"children":777},{},[778],{"type":39,"value":779},"extracted inference image folder",{"type":45,"tag":459,"props":781,"children":782},{},[783],{"type":39,"value":483},{"type":45,"tag":422,"props":785,"children":786},{},[787,791,796,800,804],{"type":45,"tag":459,"props":788,"children":789},{},[790],{"type":39,"value":90},{"type":45,"tag":459,"props":792,"children":793},{},[794],{"type":39,"value":795},"inference.checkpoint",{"type":45,"tag":459,"props":797,"children":798},{},[799],{"type":39,"value":601},{"type":45,"tag":459,"props":801,"children":802},{},[803],{"type":39,"value":606},{"type":45,"tag":459,"props":805,"children":806},{},[807],{"type":39,"value":483},{"type":45,"tag":422,"props":809,"children":810},{},[811,816,820,824,828],{"type":45,"tag":459,"props":812,"children":813},{},[814],{"type":39,"value":815},"prune",{"type":45,"tag":459,"props":817,"children":818},{},[819],{"type":39,"value":520},{"type":45,"tag":459,"props":821,"children":822},{},[823],{"type":39,"value":525},{"type":45,"tag":459,"props":825,"children":826},{},[827],{"type":39,"value":530},{"type":45,"tag":459,"props":829,"children":830},{},[831],{"type":39,"value":483},{"type":45,"tag":422,"props":833,"children":834},{},[835,839,844,848,852],{"type":45,"tag":459,"props":836,"children":837},{},[838],{"type":39,"value":815},{"type":45,"tag":459,"props":840,"children":841},{},[842],{"type":39,"value":843},"prune.checkpoint",{"type":45,"tag":459,"props":845,"children":846},{},[847],{"type":39,"value":601},{"type":45,"tag":459,"props":849,"children":850},{},[851],{"type":39,"value":606},{"type":45,"tag":459,"props":853,"children":854},{},[855],{"type":39,"value":483},{"type":45,"tag":422,"props":857,"children":858},{},[859,864,869,874,879],{"type":45,"tag":459,"props":860,"children":861},{},[862],{"type":39,"value":863},"quantize",{"type":45,"tag":459,"props":865,"children":866},{},[867],{"type":39,"value":868},"dataset.train_dataset_dir",{"type":45,"tag":459,"props":870,"children":871},{},[872],{"type":39,"value":873},"dataset_convert train job",{"type":45,"tag":459,"props":875,"children":876},{},[877],{"type":39,"value":878},"LMDB folder containing data.mdb and lock.mdb",{"type":45,"tag":459,"props":880,"children":881},{},[882],{"type":39,"value":685},{"type":45,"tag":422,"props":884,"children":885},{},[886,890,895,900,904],{"type":45,"tag":459,"props":887,"children":888},{},[889],{"type":39,"value":863},{"type":45,"tag":459,"props":891,"children":892},{},[893],{"type":39,"value":894},"dataset.val_dataset_dir",{"type":45,"tag":459,"props":896,"children":897},{},[898],{"type":39,"value":899},"dataset_convert eval job",{"type":45,"tag":459,"props":901,"children":902},{},[903],{"type":39,"value":878},{"type":45,"tag":459,"props":905,"children":906},{},[907],{"type":39,"value":483},{"type":45,"tag":422,"props":909,"children":910},{},[911,915,919,923,927],{"type":45,"tag":459,"props":912,"children":913},{},[914],{"type":39,"value":863},{"type":45,"tag":459,"props":916,"children":917},{},[918],{"type":39,"value":520},{"type":45,"tag":459,"props":920,"children":921},{},[922],{"type":39,"value":525},{"type":45,"tag":459,"props":924,"children":925},{},[926],{"type":39,"value":530},{"type":45,"tag":459,"props":928,"children":929},{},[930],{"type":39,"value":483},{"type":45,"tag":422,"props":932,"children":933},{},[934,938,943,948,953],{"type":45,"tag":459,"props":935,"children":936},{},[937],{"type":39,"value":863},{"type":45,"tag":459,"props":939,"children":940},{},[941],{"type":39,"value":942},"dataset.quant_calibration_dataset.images_dir",{"type":45,"tag":459,"props":944,"children":945},{},[946],{"type":39,"value":947},"train_datasets",{"type":45,"tag":459,"props":949,"children":950},{},[951],{"type":39,"value":952},"extracted calibration image folder",{"type":45,"tag":459,"props":954,"children":955},{},[956],{"type":39,"value":483},{"type":45,"tag":422,"props":958,"children":959},{},[960,964,969,973,978],{"type":45,"tag":459,"props":961,"children":962},{},[963],{"type":39,"value":863},{"type":45,"tag":459,"props":965,"children":966},{},[967],{"type":39,"value":968},"quantize.model_path",{"type":45,"tag":459,"props":970,"children":971},{},[972],{"type":39,"value":601},{"type":45,"tag":459,"props":974,"children":975},{},[976],{"type":39,"value":977},"checkpoint selected by resolver",{"type":45,"tag":459,"props":979,"children":980},{},[981],{"type":39,"value":483},{"type":45,"tag":422,"props":983,"children":984},{},[985,990,994,998,1002],{"type":45,"tag":459,"props":986,"children":987},{},[988],{"type":39,"value":989},"retrain",{"type":45,"tag":459,"props":991,"children":992},{},[993],{"type":39,"value":868},{"type":45,"tag":459,"props":995,"children":996},{},[997],{"type":39,"value":873},{"type":45,"tag":459,"props":999,"children":1000},{},[1001],{"type":39,"value":878},{"type":45,"tag":459,"props":1003,"children":1004},{},[1005],{"type":39,"value":685},{"type":45,"tag":422,"props":1007,"children":1008},{},[1009,1013,1017,1021,1025],{"type":45,"tag":459,"props":1010,"children":1011},{},[1012],{"type":39,"value":989},{"type":45,"tag":459,"props":1014,"children":1015},{},[1016],{"type":39,"value":894},{"type":45,"tag":459,"props":1018,"children":1019},{},[1020],{"type":39,"value":899},{"type":45,"tag":459,"props":1022,"children":1023},{},[1024],{"type":39,"value":878},{"type":45,"tag":459,"props":1026,"children":1027},{},[1028],{"type":39,"value":483},{"type":45,"tag":422,"props":1030,"children":1031},{},[1032,1036,1040,1044,1048],{"type":45,"tag":459,"props":1033,"children":1034},{},[1035],{"type":39,"value":989},{"type":45,"tag":459,"props":1037,"children":1038},{},[1039],{"type":39,"value":520},{"type":45,"tag":459,"props":1041,"children":1042},{},[1043],{"type":39,"value":525},{"type":45,"tag":459,"props":1045,"children":1046},{},[1047],{"type":39,"value":530},{"type":45,"tag":459,"props":1049,"children":1050},{},[1051],{"type":39,"value":483},{"type":45,"tag":422,"props":1053,"children":1054},{},[1055,1059,1064,1069,1074],{"type":45,"tag":459,"props":1056,"children":1057},{},[1058],{"type":39,"value":989},{"type":45,"tag":459,"props":1060,"children":1061},{},[1062],{"type":39,"value":1063},"model.pruned_graph_path",{"type":45,"tag":459,"props":1065,"children":1066},{},[1067],{"type":39,"value":1068},"parent prune job",{"type":45,"tag":459,"props":1070,"children":1071},{},[1072],{"type":39,"value":1073},"pruned .pth artifact",{"type":45,"tag":459,"props":1075,"children":1076},{},[1077],{"type":39,"value":483},{"type":45,"tag":422,"props":1079,"children":1080},{},[1081,1086,1090,1094,1098],{"type":45,"tag":459,"props":1082,"children":1083},{},[1084],{"type":39,"value":1085},"train",{"type":45,"tag":459,"props":1087,"children":1088},{},[1089],{"type":39,"value":868},{"type":45,"tag":459,"props":1091,"children":1092},{},[1093],{"type":39,"value":873},{"type":45,"tag":459,"props":1095,"children":1096},{},[1097],{"type":39,"value":878},{"type":45,"tag":459,"props":1099,"children":1100},{},[1101],{"type":39,"value":685},{"type":45,"tag":422,"props":1103,"children":1104},{},[1105,1109,1114,1118,1123],{"type":45,"tag":459,"props":1106,"children":1107},{},[1108],{"type":39,"value":1085},{"type":45,"tag":459,"props":1110,"children":1111},{},[1112],{"type":39,"value":1113},"dataset.train_gt_file",{"type":45,"tag":459,"props":1115,"children":1116},{},[1117],{"type":39,"value":947},{"type":45,"tag":459,"props":1119,"children":1120},{},[1121],{"type":39,"value":1122},"train\u002Fgt_new.txt when using raw folders instead of LMDB",{"type":45,"tag":459,"props":1124,"children":1125},{},[1126],{"type":39,"value":483},{"type":45,"tag":422,"props":1128,"children":1129},{},[1130,1134,1138,1142,1146],{"type":45,"tag":459,"props":1131,"children":1132},{},[1133],{"type":39,"value":1085},{"type":45,"tag":459,"props":1135,"children":1136},{},[1137],{"type":39,"value":894},{"type":45,"tag":459,"props":1139,"children":1140},{},[1141],{"type":39,"value":899},{"type":45,"tag":459,"props":1143,"children":1144},{},[1145],{"type":39,"value":878},{"type":45,"tag":459,"props":1147,"children":1148},{},[1149],{"type":39,"value":483},{"type":45,"tag":422,"props":1151,"children":1152},{},[1153,1157,1162,1166,1171],{"type":45,"tag":459,"props":1154,"children":1155},{},[1156],{"type":39,"value":1085},{"type":45,"tag":459,"props":1158,"children":1159},{},[1160],{"type":39,"value":1161},"dataset.val_gt_file",{"type":45,"tag":459,"props":1163,"children":1164},{},[1165],{"type":39,"value":525},{"type":45,"tag":459,"props":1167,"children":1168},{},[1169],{"type":39,"value":1170},"test\u002Fgt_new.txt when using raw folders instead of LMDB",{"type":45,"tag":459,"props":1172,"children":1173},{},[1174],{"type":39,"value":483},{"type":45,"tag":422,"props":1176,"children":1177},{},[1178,1182,1186,1190,1194],{"type":45,"tag":459,"props":1179,"children":1180},{},[1181],{"type":39,"value":1085},{"type":45,"tag":459,"props":1183,"children":1184},{},[1185],{"type":39,"value":520},{"type":45,"tag":459,"props":1187,"children":1188},{},[1189],{"type":39,"value":525},{"type":45,"tag":459,"props":1191,"children":1192},{},[1193],{"type":39,"value":530},{"type":45,"tag":459,"props":1195,"children":1196},{},[1197],{"type":39,"value":483},{"type":45,"tag":407,"props":1199,"children":1201},{"id":1200},"checkpoint-selection",[1202],{"type":39,"value":1203},"Checkpoint Selection",{"type":45,"tag":53,"props":1205,"children":1206},{},[1207,1209,1215,1217,1223,1225,1231,1233,1238,1240,1246],{"type":39,"value":1208},"OCRNet training writes both ",{"type":45,"tag":69,"props":1210,"children":1212},{"className":1211},[],[1213],{"type":39,"value":1214},"best_accuracy.pth",{"type":39,"value":1216}," and epoch-step checkpoints such as ",{"type":45,"tag":69,"props":1218,"children":1220},{"className":1219},[],[1221],{"type":39,"value":1222},"model_epoch_000_step_00003.pth",{"type":39,"value":1224},". Use the SDK\u002Fmodel checkpoint resolver through the ",{"type":45,"tag":69,"props":1226,"children":1228},{"className":1227},[],[1229],{"type":39,"value":1230},"spec_params",{"type":39,"value":1232}," mappings in ",{"type":45,"tag":69,"props":1234,"children":1236},{"className":1235},[],[1237],{"type":39,"value":166},{"type":39,"value":1239},"; do not guess by sorting for the newest ",{"type":45,"tag":69,"props":1241,"children":1243},{"className":1242},[],[1244],{"type":39,"value":1245},".pth",{"type":39,"value":1247},".",{"type":45,"tag":371,"props":1249,"children":1250},{},[1251,1289,1302,1344],{"type":45,"tag":375,"props":1252,"children":1253},{},[1254,1256,1261,1263,1268,1269,1274,1275,1280,1282,1287],{"type":39,"value":1255},"Use ",{"type":45,"tag":69,"props":1257,"children":1259},{"className":1258},[],[1260],{"type":39,"value":1214},{"type":39,"value":1262}," for best-checkpoint ",{"type":45,"tag":69,"props":1264,"children":1266},{"className":1265},[],[1267],{"type":39,"value":82},{"type":39,"value":200},{"type":45,"tag":69,"props":1270,"children":1272},{"className":1271},[],[1273],{"type":39,"value":90},{"type":39,"value":200},{"type":45,"tag":69,"props":1276,"children":1278},{"className":1277},[],[1279],{"type":39,"value":354},{"type":39,"value":1281},", and ",{"type":45,"tag":69,"props":1283,"children":1285},{"className":1284},[],[1286],{"type":39,"value":815},{"type":39,"value":1288}," requests.",{"type":45,"tag":375,"props":1290,"children":1291},{},[1292,1294,1300],{"type":39,"value":1293},"Use the exact requested ",{"type":45,"tag":69,"props":1295,"children":1297},{"className":1296},[],[1298],{"type":39,"value":1299},"model_epoch_*_step_*.pth",{"type":39,"value":1301}," for epoch\u002Fstep-specific actions.",{"type":45,"tag":375,"props":1303,"children":1304},{},[1305,1306,1312,1314,1319,1321,1327,1329,1334,1336,1342],{"type":39,"value":1255},{"type":45,"tag":69,"props":1307,"children":1309},{"className":1308},[],[1310],{"type":39,"value":1311},"train.resume_training_checkpoint_path",{"type":39,"value":1313}," only for resume training, and use ",{"type":45,"tag":69,"props":1315,"children":1317},{"className":1316},[],[1318],{"type":39,"value":1063},{"type":39,"value":1320}," for retrain from a prune output. OCRNet does not expose a separate ",{"type":45,"tag":69,"props":1322,"children":1324},{"className":1323},[],[1325],{"type":39,"value":1326},"ocrnet retrain",{"type":39,"value":1328}," CLI subtask in the PyT image; the model-skill ",{"type":45,"tag":69,"props":1330,"children":1332},{"className":1331},[],[1333],{"type":39,"value":989},{"type":39,"value":1335}," action routes through ",{"type":45,"tag":69,"props":1337,"children":1339},{"className":1338},[],[1340],{"type":39,"value":1341},"ocrnet train -e",{"type":39,"value":1343}," with the pruned graph path set.",{"type":45,"tag":375,"props":1345,"children":1346},{},[1347,1349,1354,1356,1362],{"type":39,"value":1348},"OCRNet ",{"type":45,"tag":69,"props":1350,"children":1352},{"className":1351},[],[1353],{"type":39,"value":863},{"type":39,"value":1355}," loads the model through PyTorch. For trusted checkpoints created by the same local run, set ",{"type":45,"tag":69,"props":1357,"children":1359},{"className":1358},[],[1360],{"type":39,"value":1361},"TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD=1",{"type":39,"value":1363}," if PyTorch 2.6+ rejects the checkpoint as a weights-only load.",{"type":45,"tag":407,"props":1365,"children":1367},{"id":1366},"typical-spec-overrides",[1368],{"type":39,"value":1369},"Typical Spec Overrides",{"type":45,"tag":53,"props":1371,"children":1372},{},[1373,1375,1380,1382,1387,1389,1395,1396,1402],{"type":39,"value":1374},"Data source overrides are ",{"type":45,"tag":379,"props":1376,"children":1377},{},[1378],{"type":39,"value":1379},"mandatory for every action",{"type":39,"value":1381},". Run ",{"type":45,"tag":69,"props":1383,"children":1385},{"className":1384},[],[1386],{"type":39,"value":463},{"type":39,"value":1388}," separately for train and validation splits, then pass the LMDB folders that directly contain ",{"type":45,"tag":69,"props":1390,"children":1392},{"className":1391},[],[1393],{"type":39,"value":1394},"data.mdb",{"type":39,"value":184},{"type":45,"tag":69,"props":1397,"children":1399},{"className":1398},[],[1400],{"type":39,"value":1401},"lock.mdb",{"type":39,"value":1403}," into train, quantize, and retrain. Tarballs from remote storage must be extracted before they are used as image directories.",{"type":45,"tag":1405,"props":1406,"children":1411},"pre",{"className":1407,"code":1408,"language":1409,"meta":1410,"style":1410},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","TRAIN_IMAGES = \"\u003Cextracted train image folder>\"\nTRAIN_GT = \"\u003Ctrain gt_new.txt>\"\nEVAL_IMAGES = \"\u003Cextracted eval image folder>\"\nEVAL_GT = \"\u003Ceval gt_new.txt>\"\nTRAIN_LMDB = \"\u003Ctrain dataset_convert results_dir>\"\nEVAL_LMDB = \"\u003Ceval dataset_convert results_dir>\"\nCHAR_LIST = \"\u003Ccharacter_list>\"\n","python","",[1412],{"type":45,"tag":69,"props":1413,"children":1414},{"__ignoreMap":1410},[1415,1426,1435,1444,1453,1462,1471],{"type":45,"tag":1416,"props":1417,"children":1420},"span",{"class":1418,"line":1419},"line",1,[1421],{"type":45,"tag":1416,"props":1422,"children":1423},{},[1424],{"type":39,"value":1425},"TRAIN_IMAGES = \"\u003Cextracted train image folder>\"\n",{"type":45,"tag":1416,"props":1427,"children":1429},{"class":1418,"line":1428},2,[1430],{"type":45,"tag":1416,"props":1431,"children":1432},{},[1433],{"type":39,"value":1434},"TRAIN_GT = \"\u003Ctrain gt_new.txt>\"\n",{"type":45,"tag":1416,"props":1436,"children":1438},{"class":1418,"line":1437},3,[1439],{"type":45,"tag":1416,"props":1440,"children":1441},{},[1442],{"type":39,"value":1443},"EVAL_IMAGES = \"\u003Cextracted eval image folder>\"\n",{"type":45,"tag":1416,"props":1445,"children":1447},{"class":1418,"line":1446},4,[1448],{"type":45,"tag":1416,"props":1449,"children":1450},{},[1451],{"type":39,"value":1452},"EVAL_GT = \"\u003Ceval gt_new.txt>\"\n",{"type":45,"tag":1416,"props":1454,"children":1456},{"class":1418,"line":1455},5,[1457],{"type":45,"tag":1416,"props":1458,"children":1459},{},[1460],{"type":39,"value":1461},"TRAIN_LMDB = \"\u003Ctrain dataset_convert results_dir>\"\n",{"type":45,"tag":1416,"props":1463,"children":1465},{"class":1418,"line":1464},6,[1466],{"type":45,"tag":1416,"props":1467,"children":1468},{},[1469],{"type":39,"value":1470},"EVAL_LMDB = \"\u003Ceval dataset_convert results_dir>\"\n",{"type":45,"tag":1416,"props":1472,"children":1474},{"class":1418,"line":1473},7,[1475],{"type":45,"tag":1416,"props":1476,"children":1477},{},[1478],{"type":39,"value":1479},"CHAR_LIST = \"\u003Ccharacter_list>\"\n",{"type":45,"tag":53,"props":1481,"children":1482},{},[1483],{"type":45,"tag":379,"props":1484,"children":1485},{},[1486],{"type":39,"value":1487},"dataset_convert (run once per split):",{"type":45,"tag":1405,"props":1489,"children":1491},{"className":1407,"code":1490,"language":1409,"meta":1410,"style":1410},"{\n    \"dataset_convert.input_img_dir\": TRAIN_IMAGES,\n    \"dataset_convert.gt_file\": TRAIN_GT,\n}\n",[1492],{"type":45,"tag":69,"props":1493,"children":1494},{"__ignoreMap":1410},[1495,1503,1511,1519],{"type":45,"tag":1416,"props":1496,"children":1497},{"class":1418,"line":1419},[1498],{"type":45,"tag":1416,"props":1499,"children":1500},{},[1501],{"type":39,"value":1502},"{\n",{"type":45,"tag":1416,"props":1504,"children":1505},{"class":1418,"line":1428},[1506],{"type":45,"tag":1416,"props":1507,"children":1508},{},[1509],{"type":39,"value":1510},"    \"dataset_convert.input_img_dir\": TRAIN_IMAGES,\n",{"type":45,"tag":1416,"props":1512,"children":1513},{"class":1418,"line":1437},[1514],{"type":45,"tag":1416,"props":1515,"children":1516},{},[1517],{"type":39,"value":1518},"    \"dataset_convert.gt_file\": TRAIN_GT,\n",{"type":45,"tag":1416,"props":1520,"children":1521},{"class":1418,"line":1446},[1522],{"type":45,"tag":1416,"props":1523,"children":1524},{},[1525],{"type":39,"value":1526},"}\n",{"type":45,"tag":53,"props":1528,"children":1529},{},[1530],{"type":45,"tag":379,"props":1531,"children":1532},{},[1533],{"type":39,"value":1534},"train (mandatory data sources):",{"type":45,"tag":1405,"props":1536,"children":1538},{"className":1407,"code":1537,"language":1409,"meta":1410,"style":1410},"{\n    \"train.num_epochs\": 30,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.batch_size\": 16,\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.train_gt_file\": \"\",\n    \"dataset.val_gt_file\": \"\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n",[1539],{"type":45,"tag":69,"props":1540,"children":1541},{"__ignoreMap":1410},[1542,1549,1557,1565,1573,1581,1589,1597,1606,1615,1624,1633],{"type":45,"tag":1416,"props":1543,"children":1544},{"class":1418,"line":1419},[1545],{"type":45,"tag":1416,"props":1546,"children":1547},{},[1548],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1550,"children":1551},{"class":1418,"line":1428},[1552],{"type":45,"tag":1416,"props":1553,"children":1554},{},[1555],{"type":39,"value":1556},"    \"train.num_epochs\": 30,\n",{"type":45,"tag":1416,"props":1558,"children":1559},{"class":1418,"line":1437},[1560],{"type":45,"tag":1416,"props":1561,"children":1562},{},[1563],{"type":39,"value":1564},"    \"train.checkpoint_interval\": 10,\n",{"type":45,"tag":1416,"props":1566,"children":1567},{"class":1418,"line":1446},[1568],{"type":45,"tag":1416,"props":1569,"children":1570},{},[1571],{"type":39,"value":1572},"    \"train.validation_interval\": 10,\n",{"type":45,"tag":1416,"props":1574,"children":1575},{"class":1418,"line":1455},[1576],{"type":45,"tag":1416,"props":1577,"children":1578},{},[1579],{"type":39,"value":1580},"    \"train.num_gpus\": 1,\n",{"type":45,"tag":1416,"props":1582,"children":1583},{"class":1418,"line":1464},[1584],{"type":45,"tag":1416,"props":1585,"children":1586},{},[1587],{"type":39,"value":1588},"    \"dataset.batch_size\": 16,\n",{"type":45,"tag":1416,"props":1590,"children":1591},{"class":1418,"line":1473},[1592],{"type":45,"tag":1416,"props":1593,"children":1594},{},[1595],{"type":39,"value":1596},"    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n",{"type":45,"tag":1416,"props":1598,"children":1600},{"class":1418,"line":1599},8,[1601],{"type":45,"tag":1416,"props":1602,"children":1603},{},[1604],{"type":39,"value":1605},"    \"dataset.val_dataset_dir\": EVAL_LMDB,\n",{"type":45,"tag":1416,"props":1607,"children":1609},{"class":1418,"line":1608},9,[1610],{"type":45,"tag":1416,"props":1611,"children":1612},{},[1613],{"type":39,"value":1614},"    \"dataset.train_gt_file\": \"\",\n",{"type":45,"tag":1416,"props":1616,"children":1618},{"class":1418,"line":1617},10,[1619],{"type":45,"tag":1416,"props":1620,"children":1621},{},[1622],{"type":39,"value":1623},"    \"dataset.val_gt_file\": \"\",\n",{"type":45,"tag":1416,"props":1625,"children":1627},{"class":1418,"line":1626},11,[1628],{"type":45,"tag":1416,"props":1629,"children":1630},{},[1631],{"type":39,"value":1632},"    \"dataset.character_list_file\": CHAR_LIST,\n",{"type":45,"tag":1416,"props":1634,"children":1636},{"class":1418,"line":1635},12,[1637],{"type":45,"tag":1416,"props":1638,"children":1639},{},[1640],{"type":39,"value":1526},{"type":45,"tag":53,"props":1642,"children":1643},{},[1644],{"type":45,"tag":379,"props":1645,"children":1646},{},[1647],{"type":39,"value":1648},"deploy\u002Fgen_trt_engine (mandatory data sources):",{"type":45,"tag":1405,"props":1650,"children":1652},{"className":1407,"code":1651,"language":1409,"meta":1410,"style":1410},"{\n    \"gen_trt_engine.onnx_file\": \"\u003Cselected export ONNX>\",\n    \"gen_trt_engine.trt_engine\": \"\u003Coutput engine path>\",\n    \"gen_trt_engine.tensorrt.calibration.cal_cache_file\": \"\u003Coutput calibration cache path>\",\n    \"gen_trt_engine.tensorrt.data_type\": \"fp16\",\n    \"gen_trt_engine.tensorrt.calibration.cal_image_dir\": [TRAIN_IMAGES],\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n",[1653],{"type":45,"tag":69,"props":1654,"children":1655},{"__ignoreMap":1410},[1656,1663,1671,1679,1687,1695,1703,1710],{"type":45,"tag":1416,"props":1657,"children":1658},{"class":1418,"line":1419},[1659],{"type":45,"tag":1416,"props":1660,"children":1661},{},[1662],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1664,"children":1665},{"class":1418,"line":1428},[1666],{"type":45,"tag":1416,"props":1667,"children":1668},{},[1669],{"type":39,"value":1670},"    \"gen_trt_engine.onnx_file\": \"\u003Cselected export ONNX>\",\n",{"type":45,"tag":1416,"props":1672,"children":1673},{"class":1418,"line":1437},[1674],{"type":45,"tag":1416,"props":1675,"children":1676},{},[1677],{"type":39,"value":1678},"    \"gen_trt_engine.trt_engine\": \"\u003Coutput engine path>\",\n",{"type":45,"tag":1416,"props":1680,"children":1681},{"class":1418,"line":1446},[1682],{"type":45,"tag":1416,"props":1683,"children":1684},{},[1685],{"type":39,"value":1686},"    \"gen_trt_engine.tensorrt.calibration.cal_cache_file\": \"\u003Coutput calibration cache path>\",\n",{"type":45,"tag":1416,"props":1688,"children":1689},{"class":1418,"line":1455},[1690],{"type":45,"tag":1416,"props":1691,"children":1692},{},[1693],{"type":39,"value":1694},"    \"gen_trt_engine.tensorrt.data_type\": \"fp16\",\n",{"type":45,"tag":1416,"props":1696,"children":1697},{"class":1418,"line":1464},[1698],{"type":45,"tag":1416,"props":1699,"children":1700},{},[1701],{"type":39,"value":1702},"    \"gen_trt_engine.tensorrt.calibration.cal_image_dir\": [TRAIN_IMAGES],\n",{"type":45,"tag":1416,"props":1704,"children":1705},{"class":1418,"line":1473},[1706],{"type":45,"tag":1416,"props":1707,"children":1708},{},[1709],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1711,"children":1712},{"class":1418,"line":1599},[1713],{"type":45,"tag":1416,"props":1714,"children":1715},{},[1716],{"type":39,"value":1526},{"type":45,"tag":53,"props":1718,"children":1719},{},[1720],{"type":45,"tag":379,"props":1721,"children":1722},{},[1723],{"type":39,"value":1724},"evaluate (mandatory data sources):",{"type":45,"tag":1405,"props":1726,"children":1728},{"className":1407,"code":1727,"language":1409,"meta":1410,"style":1410},"{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"evaluate.test_dataset_dir\": EVAL_IMAGES,\n    \"evaluate.test_dataset_gt_file\": EVAL_GT,\n}\n",[1729],{"type":45,"tag":69,"props":1730,"children":1731},{"__ignoreMap":1410},[1732,1739,1747,1754,1762,1770],{"type":45,"tag":1416,"props":1733,"children":1734},{"class":1418,"line":1419},[1735],{"type":45,"tag":1416,"props":1736,"children":1737},{},[1738],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1740,"children":1741},{"class":1418,"line":1428},[1742],{"type":45,"tag":1416,"props":1743,"children":1744},{},[1745],{"type":39,"value":1746},"    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":45,"tag":1416,"props":1748,"children":1749},{"class":1418,"line":1437},[1750],{"type":45,"tag":1416,"props":1751,"children":1752},{},[1753],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1755,"children":1756},{"class":1418,"line":1446},[1757],{"type":45,"tag":1416,"props":1758,"children":1759},{},[1760],{"type":39,"value":1761},"    \"evaluate.test_dataset_dir\": EVAL_IMAGES,\n",{"type":45,"tag":1416,"props":1763,"children":1764},{"class":1418,"line":1455},[1765],{"type":45,"tag":1416,"props":1766,"children":1767},{},[1768],{"type":39,"value":1769},"    \"evaluate.test_dataset_gt_file\": EVAL_GT,\n",{"type":45,"tag":1416,"props":1771,"children":1772},{"class":1418,"line":1464},[1773],{"type":45,"tag":1416,"props":1774,"children":1775},{},[1776],{"type":39,"value":1526},{"type":45,"tag":53,"props":1778,"children":1779},{},[1780],{"type":45,"tag":379,"props":1781,"children":1782},{},[1783],{"type":39,"value":1784},"export (mandatory data sources):",{"type":45,"tag":1405,"props":1786,"children":1788},{"className":1407,"code":1787,"language":1409,"meta":1410,"style":1410},"{\n    \"export.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"export.onnx_file\": \"\u003Coutput ONNX path>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n",[1789],{"type":45,"tag":69,"props":1790,"children":1791},{"__ignoreMap":1410},[1792,1799,1807,1815,1822],{"type":45,"tag":1416,"props":1793,"children":1794},{"class":1418,"line":1419},[1795],{"type":45,"tag":1416,"props":1796,"children":1797},{},[1798],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1800,"children":1801},{"class":1418,"line":1428},[1802],{"type":45,"tag":1416,"props":1803,"children":1804},{},[1805],{"type":39,"value":1806},"    \"export.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":45,"tag":1416,"props":1808,"children":1809},{"class":1418,"line":1437},[1810],{"type":45,"tag":1416,"props":1811,"children":1812},{},[1813],{"type":39,"value":1814},"    \"export.onnx_file\": \"\u003Coutput ONNX path>\",\n",{"type":45,"tag":1416,"props":1816,"children":1817},{"class":1418,"line":1446},[1818],{"type":45,"tag":1416,"props":1819,"children":1820},{},[1821],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1823,"children":1824},{"class":1418,"line":1455},[1825],{"type":45,"tag":1416,"props":1826,"children":1827},{},[1828],{"type":39,"value":1526},{"type":45,"tag":53,"props":1830,"children":1831},{},[1832],{"type":45,"tag":379,"props":1833,"children":1834},{},[1835],{"type":39,"value":1836},"inference (mandatory data sources):",{"type":45,"tag":1405,"props":1838,"children":1840},{"className":1407,"code":1839,"language":1409,"meta":1410,"style":1410},"{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"inference.inference_dataset_dir\": EVAL_IMAGES,\n}\n",[1841],{"type":45,"tag":69,"props":1842,"children":1843},{"__ignoreMap":1410},[1844,1851,1859,1866,1874],{"type":45,"tag":1416,"props":1845,"children":1846},{"class":1418,"line":1419},[1847],{"type":45,"tag":1416,"props":1848,"children":1849},{},[1850],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1852,"children":1853},{"class":1418,"line":1428},[1854],{"type":45,"tag":1416,"props":1855,"children":1856},{},[1857],{"type":39,"value":1858},"    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":45,"tag":1416,"props":1860,"children":1861},{"class":1418,"line":1437},[1862],{"type":45,"tag":1416,"props":1863,"children":1864},{},[1865],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1867,"children":1868},{"class":1418,"line":1446},[1869],{"type":45,"tag":1416,"props":1870,"children":1871},{},[1872],{"type":39,"value":1873},"    \"inference.inference_dataset_dir\": EVAL_IMAGES,\n",{"type":45,"tag":1416,"props":1875,"children":1876},{"class":1418,"line":1455},[1877],{"type":45,"tag":1416,"props":1878,"children":1879},{},[1880],{"type":39,"value":1526},{"type":45,"tag":53,"props":1882,"children":1883},{},[1884],{"type":45,"tag":379,"props":1885,"children":1886},{},[1887],{"type":39,"value":1888},"prune (mandatory data sources):",{"type":45,"tag":1405,"props":1890,"children":1892},{"className":1407,"code":1891,"language":1409,"meta":1410,"style":1410},"{\n    \"prune.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"prune.pruned_file\": \"\u003Coutput pruned PTH path>\",\n    \"dataset.character_list_file\": CHAR_LIST,\n}\n",[1893],{"type":45,"tag":69,"props":1894,"children":1895},{"__ignoreMap":1410},[1896,1903,1911,1919,1926],{"type":45,"tag":1416,"props":1897,"children":1898},{"class":1418,"line":1419},[1899],{"type":45,"tag":1416,"props":1900,"children":1901},{},[1902],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1904,"children":1905},{"class":1418,"line":1428},[1906],{"type":45,"tag":1416,"props":1907,"children":1908},{},[1909],{"type":39,"value":1910},"    \"prune.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":45,"tag":1416,"props":1912,"children":1913},{"class":1418,"line":1437},[1914],{"type":45,"tag":1416,"props":1915,"children":1916},{},[1917],{"type":39,"value":1918},"    \"prune.pruned_file\": \"\u003Coutput pruned PTH path>\",\n",{"type":45,"tag":1416,"props":1920,"children":1921},{"class":1418,"line":1446},[1922],{"type":45,"tag":1416,"props":1923,"children":1924},{},[1925],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1927,"children":1928},{"class":1418,"line":1455},[1929],{"type":45,"tag":1416,"props":1930,"children":1931},{},[1932],{"type":39,"value":1526},{"type":45,"tag":53,"props":1934,"children":1935},{},[1936],{"type":45,"tag":379,"props":1937,"children":1938},{},[1939],{"type":39,"value":1940},"quantize (mandatory data sources):",{"type":45,"tag":1405,"props":1942,"children":1944},{"className":1407,"code":1943,"language":1409,"meta":1410,"style":1410},"{\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"dataset.quant_calibration_dataset.images_dir\": TRAIN_IMAGES,\n    \"quantize.model_path\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n}\n",[1945],{"type":45,"tag":69,"props":1946,"children":1947},{"__ignoreMap":1410},[1948,1955,1962,1969,1976,1984,1992],{"type":45,"tag":1416,"props":1949,"children":1950},{"class":1418,"line":1419},[1951],{"type":45,"tag":1416,"props":1952,"children":1953},{},[1954],{"type":39,"value":1502},{"type":45,"tag":1416,"props":1956,"children":1957},{"class":1418,"line":1428},[1958],{"type":45,"tag":1416,"props":1959,"children":1960},{},[1961],{"type":39,"value":1596},{"type":45,"tag":1416,"props":1963,"children":1964},{"class":1418,"line":1437},[1965],{"type":45,"tag":1416,"props":1966,"children":1967},{},[1968],{"type":39,"value":1605},{"type":45,"tag":1416,"props":1970,"children":1971},{"class":1418,"line":1446},[1972],{"type":45,"tag":1416,"props":1973,"children":1974},{},[1975],{"type":39,"value":1632},{"type":45,"tag":1416,"props":1977,"children":1978},{"class":1418,"line":1455},[1979],{"type":45,"tag":1416,"props":1980,"children":1981},{},[1982],{"type":39,"value":1983},"    \"dataset.quant_calibration_dataset.images_dir\": TRAIN_IMAGES,\n",{"type":45,"tag":1416,"props":1985,"children":1986},{"class":1418,"line":1464},[1987],{"type":45,"tag":1416,"props":1988,"children":1989},{},[1990],{"type":39,"value":1991},"    \"quantize.model_path\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n",{"type":45,"tag":1416,"props":1993,"children":1994},{"class":1418,"line":1473},[1995],{"type":45,"tag":1416,"props":1996,"children":1997},{},[1998],{"type":39,"value":1526},{"type":45,"tag":53,"props":2000,"children":2001},{},[2002],{"type":45,"tag":379,"props":2003,"children":2004},{},[2005],{"type":39,"value":2006},"retrain (mandatory data sources):",{"type":45,"tag":1405,"props":2008,"children":2010},{"className":1407,"code":2009,"language":1409,"meta":1410,"style":1410},"{\n    \"dataset.train_dataset_dir\": [TRAIN_LMDB],\n    \"dataset.val_dataset_dir\": EVAL_LMDB,\n    \"dataset.character_list_file\": CHAR_LIST,\n    \"model.pruned_graph_path\": \"\u003Cselected prune output>\",\n}\n",[2011],{"type":45,"tag":69,"props":2012,"children":2013},{"__ignoreMap":1410},[2014,2021,2028,2035,2042,2050],{"type":45,"tag":1416,"props":2015,"children":2016},{"class":1418,"line":1419},[2017],{"type":45,"tag":1416,"props":2018,"children":2019},{},[2020],{"type":39,"value":1502},{"type":45,"tag":1416,"props":2022,"children":2023},{"class":1418,"line":1428},[2024],{"type":45,"tag":1416,"props":2025,"children":2026},{},[2027],{"type":39,"value":1596},{"type":45,"tag":1416,"props":2029,"children":2030},{"class":1418,"line":1437},[2031],{"type":45,"tag":1416,"props":2032,"children":2033},{},[2034],{"type":39,"value":1605},{"type":45,"tag":1416,"props":2036,"children":2037},{"class":1418,"line":1446},[2038],{"type":45,"tag":1416,"props":2039,"children":2040},{},[2041],{"type":39,"value":1632},{"type":45,"tag":1416,"props":2043,"children":2044},{"class":1418,"line":1455},[2045],{"type":45,"tag":1416,"props":2046,"children":2047},{},[2048],{"type":39,"value":2049},"    \"model.pruned_graph_path\": \"\u003Cselected prune output>\",\n",{"type":45,"tag":1416,"props":2051,"children":2052},{"class":1418,"line":1464},[2053],{"type":45,"tag":1416,"props":2054,"children":2055},{},[2056],{"type":39,"value":1526},{"type":45,"tag":118,"props":2058,"children":2060},{"id":2059},"eval-dataset",[2061],{"type":39,"value":2062},"Eval Dataset",{"type":45,"tag":53,"props":2064,"children":2065},{},[2066],{"type":39,"value":2067},"Optional. Test data provided as separate tarball.",{"type":45,"tag":118,"props":2069,"children":2071},{"id":2070},"important-parameters",[2072],{"type":39,"value":2073},"Important Parameters",{"type":45,"tag":371,"props":2075,"children":2076},{},[2077,2086,2096,2106,2116],{"type":45,"tag":375,"props":2078,"children":2079},{},[2080,2084],{"type":45,"tag":379,"props":2081,"children":2082},{},[2083],{"type":39,"value":520},{"type":39,"value":2085},": Path to character list defining the supported character set. This determines the output vocabulary size.",{"type":45,"tag":375,"props":2087,"children":2088},{},[2089,2094],{"type":45,"tag":379,"props":2090,"children":2091},{},[2092],{"type":39,"value":2093},"model.backbone",{"type":39,"value":2095},": Default ResNet.",{"type":45,"tag":375,"props":2097,"children":2098},{},[2099,2104],{"type":45,"tag":379,"props":2100,"children":2101},{},[2102],{"type":39,"value":2103},"model.prediction",{"type":39,"value":2105},": Decoder type. CTC or Attn (attention-based).",{"type":45,"tag":375,"props":2107,"children":2108},{},[2109,2114],{"type":45,"tag":379,"props":2110,"children":2111},{},[2112],{"type":39,"value":2113},"train.optim.lr",{"type":39,"value":2115},": Learning rate. Default 1.0 (Adadelta optimizer). High default is specific to Adadelta.",{"type":45,"tag":375,"props":2117,"children":2118},{},[2119,2124],{"type":45,"tag":379,"props":2120,"children":2121},{},[2122],{"type":39,"value":2123},"dataset.batch_size",{"type":39,"value":2125},": Per-GPU batch size. Default 16.",{"type":45,"tag":118,"props":2127,"children":2129},{"id":2128},"multi-gpu-multi-node",[2130],{"type":39,"value":2131},"Multi-GPU \u002F Multi-Node",{"type":45,"tag":53,"props":2133,"children":2134},{},[2135,2140,2142,2147],{"type":45,"tag":379,"props":2136,"children":2137},{},[2138],{"type":39,"value":2139},"Launch method:",{"type":39,"value":2141}," Lightning-managed (single ",{"type":45,"tag":69,"props":2143,"children":2145},{"className":2144},[],[2146],{"type":39,"value":1409},{"type":39,"value":2148}," process, Lightning spawns workers).",{"type":45,"tag":414,"props":2150,"children":2151},{},[2152,2172],{"type":45,"tag":418,"props":2153,"children":2154},{},[2155],{"type":45,"tag":422,"props":2156,"children":2157},{},[2158,2162,2167],{"type":45,"tag":426,"props":2159,"children":2160},{},[2161],{"type":39,"value":435},{"type":45,"tag":426,"props":2163,"children":2164},{},[2165],{"type":39,"value":2166},"Description",{"type":45,"tag":426,"props":2168,"children":2169},{},[2170],{"type":39,"value":2171},"Default",{"type":45,"tag":452,"props":2173,"children":2174},{},[2175,2197,2222],{"type":45,"tag":422,"props":2176,"children":2177},{},[2178,2187,2192],{"type":45,"tag":459,"props":2179,"children":2180},{},[2181],{"type":45,"tag":69,"props":2182,"children":2184},{"className":2183},[],[2185],{"type":39,"value":2186},"train.num_gpus",{"type":45,"tag":459,"props":2188,"children":2189},{},[2190],{"type":39,"value":2191},"Number of GPUs",{"type":45,"tag":459,"props":2193,"children":2194},{},[2195],{"type":39,"value":2196},"1",{"type":45,"tag":422,"props":2198,"children":2199},{},[2200,2209,2214],{"type":45,"tag":459,"props":2201,"children":2202},{},[2203],{"type":45,"tag":69,"props":2204,"children":2206},{"className":2205},[],[2207],{"type":39,"value":2208},"train.gpu_ids",{"type":45,"tag":459,"props":2210,"children":2211},{},[2212],{"type":39,"value":2213},"GPU device indices",{"type":45,"tag":459,"props":2215,"children":2216},{},[2217],{"type":45,"tag":1416,"props":2218,"children":2219},{},[2220],{"type":39,"value":2221},"0",{"type":45,"tag":422,"props":2223,"children":2224},{},[2225,2234,2239],{"type":45,"tag":459,"props":2226,"children":2227},{},[2228],{"type":45,"tag":69,"props":2229,"children":2231},{"className":2230},[],[2232],{"type":39,"value":2233},"train.distributed_strategy",{"type":45,"tag":459,"props":2235,"children":2236},{},[2237],{"type":39,"value":2238},"Strategy name",{"type":45,"tag":459,"props":2240,"children":2241},{},[2242],{"type":45,"tag":69,"props":2243,"children":2245},{"className":2244},[],[2246],{"type":39,"value":2247},"auto",{"type":45,"tag":371,"props":2249,"children":2250},{},[2251,2270,2283],{"type":45,"tag":375,"props":2252,"children":2253},{},[2254,2256,2261,2263,2268],{"type":39,"value":2255},"Strategy: ",{"type":45,"tag":69,"props":2257,"children":2259},{"className":2258},[],[2260],{"type":39,"value":2247},{"type":39,"value":2262}," for single-GPU, reads ",{"type":45,"tag":69,"props":2264,"children":2266},{"className":2265},[],[2267],{"type":39,"value":2233},{"type":39,"value":2269}," from config when multi-GPU",{"type":45,"tag":375,"props":2271,"children":2272},{},[2273,2275,2281],{"type":39,"value":2274},"No explicit ",{"type":45,"tag":69,"props":2276,"children":2278},{"className":2277},[],[2279],{"type":39,"value":2280},"num_nodes",{"type":39,"value":2282}," in train script — single-node oriented",{"type":45,"tag":375,"props":2284,"children":2285},{},[2286],{"type":39,"value":2287},"Lightweight model, single GPU typically sufficient",{"type":45,"tag":118,"props":2289,"children":2291},{"id":2290},"hardware",[2292],{"type":39,"value":2293},"Hardware",{"type":45,"tag":53,"props":2295,"children":2296},{},[2297],{"type":39,"value":2298},"Minimum 1 GPU(s), recommended 1 GPU(s). 8GB+ VRAM per GPU. OCR text recognition is lightweight. Single GPU is typically sufficient.",{"type":45,"tag":118,"props":2300,"children":2302},{"id":2301},"error-patterns",[2303],{"type":39,"value":2304},"Error Patterns",{"type":45,"tag":53,"props":2306,"children":2307},{},[2308,2313],{"type":45,"tag":379,"props":2309,"children":2310},{},[2311],{"type":39,"value":2312},"dataset_convert required",{"type":39,"value":2314},": If using raw images + gt files, run dataset_convert first to produce LMDB format.",{"type":45,"tag":53,"props":2316,"children":2317},{},[2318,2323,2325,2331,2333,2338,2339,2344,2346,2352,2354,2359,2360,2365,2367,2372,2373,2378],{"type":45,"tag":379,"props":2319,"children":2320},{},[2321],{"type":39,"value":2322},"dataset_convert output folder",{"type":39,"value":2324},": Direct ",{"type":45,"tag":69,"props":2326,"children":2328},{"className":2327},[],[2329],{"type":39,"value":2330},"ocrnet dataset_convert",{"type":39,"value":2332}," writes ",{"type":45,"tag":69,"props":2334,"children":2336},{"className":2335},[],[2337],{"type":39,"value":1394},{"type":39,"value":184},{"type":45,"tag":69,"props":2340,"children":2342},{"className":2341},[],[2343],{"type":39,"value":1401},{"type":39,"value":2345}," directly under ",{"type":45,"tag":69,"props":2347,"children":2349},{"className":2348},[],[2350],{"type":39,"value":2351},"dataset_convert.results_dir",{"type":39,"value":2353},". Use that folder itself for ",{"type":45,"tag":69,"props":2355,"children":2357},{"className":2356},[],[2358],{"type":39,"value":868},{"type":39,"value":200},{"type":45,"tag":69,"props":2361,"children":2363},{"className":2362},[],[2364],{"type":39,"value":894},{"type":39,"value":2366},", quantize, and retrain inputs. SDK-backed runs may wrap the same LMDB folder inside job artifact directories; resolve the actual folder containing ",{"type":45,"tag":69,"props":2368,"children":2370},{"className":2369},[],[2371],{"type":39,"value":1394},{"type":39,"value":184},{"type":45,"tag":69,"props":2374,"children":2376},{"className":2375},[],[2377],{"type":39,"value":1401},{"type":39,"value":1247},{"type":45,"tag":53,"props":2380,"children":2381},{},[2382,2387],{"type":45,"tag":379,"props":2383,"children":2384},{},[2385],{"type":39,"value":2386},"GT file BOM",{"type":39,"value":2388},": 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.",{"type":45,"tag":53,"props":2390,"children":2391},{},[2392,2397],{"type":45,"tag":379,"props":2393,"children":2394},{},[2395],{"type":39,"value":2396},"Character list mismatch",{"type":39,"value":2398},": All characters in training data must be present in the character_list file.",{"type":45,"tag":53,"props":2400,"children":2401},{},[2402,2407,2409,2415,2416,2422,2424,2429],{"type":45,"tag":379,"props":2403,"children":2404},{},[2405],{"type":39,"value":2406},"Export\u002Fprune output fields required",{"type":39,"value":2408},": ",{"type":45,"tag":69,"props":2410,"children":2412},{"className":2411},[],[2413],{"type":39,"value":2414},"export.onnx_file",{"type":39,"value":184},{"type":45,"tag":69,"props":2417,"children":2419},{"className":2418},[],[2420],{"type":39,"value":2421},"prune.pruned_file",{"type":39,"value":2423}," must be writable output paths. These are declared in ",{"type":45,"tag":69,"props":2425,"children":2427},{"className":2426},[],[2428],{"type":39,"value":166},{"type":39,"value":2430}," so SDK-backed model runs can create the paths automatically.",{"type":45,"tag":53,"props":2432,"children":2433},{},[2434,2439,2441,2446,2447,2452,2453,2458,2459,2464,2465,2470,2471,2476,2477,2482,2484,2489,2491,2496,2497,2503],{"type":45,"tag":379,"props":2435,"children":2436},{},[2437],{"type":39,"value":2438},"TensorRT lives in deploy",{"type":39,"value":2440},": The PyT OCRNet CLI exposes ",{"type":45,"tag":69,"props":2442,"children":2444},{"className":2443},[],[2445],{"type":39,"value":463},{"type":39,"value":200},{"type":45,"tag":69,"props":2448,"children":2450},{"className":2449},[],[2451],{"type":39,"value":82},{"type":39,"value":200},{"type":45,"tag":69,"props":2454,"children":2456},{"className":2455},[],[2457],{"type":39,"value":354},{"type":39,"value":200},{"type":45,"tag":69,"props":2460,"children":2462},{"className":2461},[],[2463],{"type":39,"value":90},{"type":39,"value":200},{"type":45,"tag":69,"props":2466,"children":2468},{"className":2467},[],[2469],{"type":39,"value":815},{"type":39,"value":200},{"type":45,"tag":69,"props":2472,"children":2474},{"className":2473},[],[2475],{"type":39,"value":863},{"type":39,"value":1281},{"type":45,"tag":69,"props":2478,"children":2480},{"className":2479},[],[2481],{"type":39,"value":1085},{"type":39,"value":2483},", but not ",{"type":45,"tag":69,"props":2485,"children":2487},{"className":2486},[],[2488],{"type":39,"value":74},{"type":39,"value":2490},". Use ",{"type":45,"tag":69,"props":2492,"children":2494},{"className":2493},[],[2495],{"type":39,"value":98},{"type":39,"value":184},{"type":45,"tag":69,"props":2498,"children":2500},{"className":2499},[],[2501],{"type":39,"value":2502},"deploy\u002Fskill_info.yaml",{"type":39,"value":2504}," for TensorRT engine generation and TensorRT-backed evaluate\u002Finference.",{"type":45,"tag":118,"props":2506,"children":2508},{"id":2507},"spec-param-parent-model-inference",[2509],{"type":39,"value":2510},"Spec Param \u002F Parent Model Inference",{"type":45,"tag":53,"props":2512,"children":2513},{},[2514,2516,2522,2524,2530,2532,2538],{"type":39,"value":2515},"Model-specific inference mappings belong in this MD file, not in ",{"type":45,"tag":69,"props":2517,"children":2519},{"className":2518},[],[2520],{"type":39,"value":2521},"config.json",{"type":39,"value":2523},". Generated runners should read this section and apply the mappings with SDK helpers before ",{"type":45,"tag":69,"props":2525,"children":2527},{"className":2526},[],[2528],{"type":39,"value":2529},"create_job()",{"type":39,"value":2531},". This mirrors the old microservices ",{"type":45,"tag":69,"props":2533,"children":2535},{"className":2534},[],[2536],{"type":39,"value":2537},"infer_params.py",{"type":39,"value":2539}," flow.",{"type":45,"tag":53,"props":2541,"children":2542},{},[2543,2545,2551],{"type":39,"value":2544},"Inference mappings from TAO Core ",{"type":45,"tag":69,"props":2546,"children":2548},{"className":2547},[],[2549],{"type":39,"value":2550},"ocrnet.config.json",{"type":39,"value":2552},":",{"type":45,"tag":414,"props":2554,"children":2555},{},[2556,2581],{"type":45,"tag":418,"props":2557,"children":2558},{},[2559],{"type":45,"tag":422,"props":2560,"children":2561},{},[2562,2566,2571,2576],{"type":45,"tag":426,"props":2563,"children":2564},{},[2565],{"type":39,"value":430},{"type":45,"tag":426,"props":2567,"children":2568},{},[2569],{"type":39,"value":2570},"Spec Field",{"type":45,"tag":426,"props":2572,"children":2573},{},[2574],{"type":39,"value":2575},"Inference Function",{"type":45,"tag":426,"props":2577,"children":2578},{},[2579],{"type":39,"value":2580},"Meaning",{"type":45,"tag":452,"props":2582,"children":2583},{},[2584,2614,2644,2673,2701,2730,2757,2784,2811,2840,2867,2894,2922,2952,2982,3009,3036,3063,3091,3118,3145,3172,3199,3228,3255,3282,3309,3336,3363,3390,3417,3444,3471,3501],{"type":45,"tag":422,"props":2585,"children":2586},{},[2587,2591,2600,2609],{"type":45,"tag":459,"props":2588,"children":2589},{},[2590],{"type":39,"value":463},{"type":45,"tag":459,"props":2592,"children":2593},{},[2594],{"type":45,"tag":69,"props":2595,"children":2597},{"className":2596},[],[2598],{"type":39,"value":2599},"results_dir",{"type":45,"tag":459,"props":2601,"children":2602},{},[2603],{"type":45,"tag":69,"props":2604,"children":2606},{"className":2605},[],[2607],{"type":39,"value":2608},"output_dir",{"type":45,"tag":459,"props":2610,"children":2611},{},[2612],{"type":39,"value":2613},"current job results directory",{"type":45,"tag":422,"props":2615,"children":2616},{},[2617,2621,2630,2639],{"type":45,"tag":459,"props":2618,"children":2619},{},[2620],{"type":39,"value":82},{"type":45,"tag":459,"props":2622,"children":2623},{},[2624],{"type":45,"tag":69,"props":2625,"children":2627},{"className":2626},[],[2628],{"type":39,"value":2629},"encryption_key",{"type":45,"tag":459,"props":2631,"children":2632},{},[2633],{"type":45,"tag":69,"props":2634,"children":2636},{"className":2635},[],[2637],{"type":39,"value":2638},"key",{"type":45,"tag":459,"props":2640,"children":2641},{},[2642],{"type":39,"value":2643},"encryption key",{"type":45,"tag":422,"props":2645,"children":2646},{},[2647,2651,2659,2668],{"type":45,"tag":459,"props":2648,"children":2649},{},[2650],{"type":39,"value":82},{"type":45,"tag":459,"props":2652,"children":2653},{},[2654],{"type":45,"tag":69,"props":2655,"children":2657},{"className":2656},[],[2658],{"type":39,"value":596},{"type":45,"tag":459,"props":2660,"children":2661},{},[2662],{"type":45,"tag":69,"props":2663,"children":2665},{"className":2664},[],[2666],{"type":39,"value":2667},"parent_model",{"type":45,"tag":459,"props":2669,"children":2670},{},[2671],{"type":39,"value":2672},"model file inferred from the parent job results folder",{"type":45,"tag":422,"props":2674,"children":2675},{},[2676,2680,2689,2697],{"type":45,"tag":459,"props":2677,"children":2678},{},[2679],{"type":39,"value":82},{"type":45,"tag":459,"props":2681,"children":2682},{},[2683],{"type":45,"tag":69,"props":2684,"children":2686},{"className":2685},[],[2687],{"type":39,"value":2688},"evaluate.trt_engine",{"type":45,"tag":459,"props":2690,"children":2691},{},[2692],{"type":45,"tag":69,"props":2693,"children":2695},{"className":2694},[],[2696],{"type":39,"value":2667},{"type":45,"tag":459,"props":2698,"children":2699},{},[2700],{"type":39,"value":2672},{"type":45,"tag":422,"props":2702,"children":2703},{},[2704,2708,2716,2725],{"type":45,"tag":459,"props":2705,"children":2706},{},[2707],{"type":39,"value":82},{"type":45,"tag":459,"props":2709,"children":2710},{},[2711],{"type":45,"tag":69,"props":2712,"children":2714},{"className":2713},[],[2715],{"type":39,"value":1063},{"type":45,"tag":459,"props":2717,"children":2718},{},[2719],{"type":45,"tag":69,"props":2720,"children":2722},{"className":2721},[],[2723],{"type":39,"value":2724},"pruned_model",{"type":45,"tag":459,"props":2726,"children":2727},{},[2728],{"type":39,"value":2729},"parent pruned model",{"type":45,"tag":422,"props":2731,"children":2732},{},[2733,2737,2745,2753],{"type":45,"tag":459,"props":2734,"children":2735},{},[2736],{"type":39,"value":82},{"type":45,"tag":459,"props":2738,"children":2739},{},[2740],{"type":45,"tag":69,"props":2741,"children":2743},{"className":2742},[],[2744],{"type":39,"value":2599},{"type":45,"tag":459,"props":2746,"children":2747},{},[2748],{"type":45,"tag":69,"props":2749,"children":2751},{"className":2750},[],[2752],{"type":39,"value":2608},{"type":45,"tag":459,"props":2754,"children":2755},{},[2756],{"type":39,"value":2613},{"type":45,"tag":422,"props":2758,"children":2759},{},[2760,2764,2772,2780],{"type":45,"tag":459,"props":2761,"children":2762},{},[2763],{"type":39,"value":354},{"type":45,"tag":459,"props":2765,"children":2766},{},[2767],{"type":45,"tag":69,"props":2768,"children":2770},{"className":2769},[],[2771],{"type":39,"value":2629},{"type":45,"tag":459,"props":2773,"children":2774},{},[2775],{"type":45,"tag":69,"props":2776,"children":2778},{"className":2777},[],[2779],{"type":39,"value":2638},{"type":45,"tag":459,"props":2781,"children":2782},{},[2783],{"type":39,"value":2643},{"type":45,"tag":422,"props":2785,"children":2786},{},[2787,2791,2799,2807],{"type":45,"tag":459,"props":2788,"children":2789},{},[2790],{"type":39,"value":354},{"type":45,"tag":459,"props":2792,"children":2793},{},[2794],{"type":45,"tag":69,"props":2795,"children":2797},{"className":2796},[],[2798],{"type":39,"value":645},{"type":45,"tag":459,"props":2800,"children":2801},{},[2802],{"type":45,"tag":69,"props":2803,"children":2805},{"className":2804},[],[2806],{"type":39,"value":2667},{"type":45,"tag":459,"props":2808,"children":2809},{},[2810],{"type":39,"value":2672},{"type":45,"tag":422,"props":2812,"children":2813},{},[2814,2818,2826,2835],{"type":45,"tag":459,"props":2815,"children":2816},{},[2817],{"type":39,"value":354},{"type":45,"tag":459,"props":2819,"children":2820},{},[2821],{"type":45,"tag":69,"props":2822,"children":2824},{"className":2823},[],[2825],{"type":39,"value":2414},{"type":45,"tag":459,"props":2827,"children":2828},{},[2829],{"type":45,"tag":69,"props":2830,"children":2832},{"className":2831},[],[2833],{"type":39,"value":2834},"create_onnx_file",{"type":45,"tag":459,"props":2836,"children":2837},{},[2838],{"type":39,"value":2839},"output ONNX path",{"type":45,"tag":422,"props":2841,"children":2842},{},[2843,2847,2855,2863],{"type":45,"tag":459,"props":2844,"children":2845},{},[2846],{"type":39,"value":354},{"type":45,"tag":459,"props":2848,"children":2849},{},[2850],{"type":45,"tag":69,"props":2851,"children":2853},{"className":2852},[],[2854],{"type":39,"value":2599},{"type":45,"tag":459,"props":2856,"children":2857},{},[2858],{"type":45,"tag":69,"props":2859,"children":2861},{"className":2860},[],[2862],{"type":39,"value":2608},{"type":45,"tag":459,"props":2864,"children":2865},{},[2866],{"type":39,"value":2613},{"type":45,"tag":422,"props":2868,"children":2869},{},[2870,2874,2882,2890],{"type":45,"tag":459,"props":2871,"children":2872},{},[2873],{"type":39,"value":665},{"type":45,"tag":459,"props":2875,"children":2876},{},[2877],{"type":45,"tag":69,"props":2878,"children":2880},{"className":2879},[],[2881],{"type":39,"value":2629},{"type":45,"tag":459,"props":2883,"children":2884},{},[2885],{"type":45,"tag":69,"props":2886,"children":2888},{"className":2887},[],[2889],{"type":39,"value":2638},{"type":45,"tag":459,"props":2891,"children":2892},{},[2893],{"type":39,"value":2643},{"type":45,"tag":422,"props":2895,"children":2896},{},[2897,2901,2909,2917],{"type":45,"tag":459,"props":2898,"children":2899},{},[2900],{"type":39,"value":665},{"type":45,"tag":459,"props":2902,"children":2903},{},[2904],{"type":45,"tag":69,"props":2905,"children":2907},{"className":2906},[],[2908],{"type":39,"value":697},{"type":45,"tag":459,"props":2910,"children":2911},{},[2912],{"type":45,"tag":69,"props":2913,"children":2915},{"className":2914},[],[2916],{"type":39,"value":2667},{"type":45,"tag":459,"props":2918,"children":2919},{},[2920],{"type":39,"value":2921},"ONNX file inferred from the parent export job results folder",{"type":45,"tag":422,"props":2923,"children":2924},{},[2925,2929,2938,2947],{"type":45,"tag":459,"props":2926,"children":2927},{},[2928],{"type":39,"value":665},{"type":45,"tag":459,"props":2930,"children":2931},{},[2932],{"type":45,"tag":69,"props":2933,"children":2935},{"className":2934},[],[2936],{"type":39,"value":2937},"gen_trt_engine.tensorrt.calibration.cal_cache_file",{"type":45,"tag":459,"props":2939,"children":2940},{},[2941],{"type":45,"tag":69,"props":2942,"children":2944},{"className":2943},[],[2945],{"type":39,"value":2946},"create_cal_cache",{"type":45,"tag":459,"props":2948,"children":2949},{},[2950],{"type":39,"value":2951},"calibration cache path",{"type":45,"tag":422,"props":2953,"children":2954},{},[2955,2959,2968,2977],{"type":45,"tag":459,"props":2956,"children":2957},{},[2958],{"type":39,"value":665},{"type":45,"tag":459,"props":2960,"children":2961},{},[2962],{"type":45,"tag":69,"props":2963,"children":2965},{"className":2964},[],[2966],{"type":39,"value":2967},"gen_trt_engine.trt_engine",{"type":45,"tag":459,"props":2969,"children":2970},{},[2971],{"type":45,"tag":69,"props":2972,"children":2974},{"className":2973},[],[2975],{"type":39,"value":2976},"create_engine_file",{"type":45,"tag":459,"props":2978,"children":2979},{},[2980],{"type":39,"value":2981},"output TensorRT engine path",{"type":45,"tag":422,"props":2983,"children":2984},{},[2985,2989,2997,3005],{"type":45,"tag":459,"props":2986,"children":2987},{},[2988],{"type":39,"value":665},{"type":45,"tag":459,"props":2990,"children":2991},{},[2992],{"type":45,"tag":69,"props":2993,"children":2995},{"className":2994},[],[2996],{"type":39,"value":2599},{"type":45,"tag":459,"props":2998,"children":2999},{},[3000],{"type":45,"tag":69,"props":3001,"children":3003},{"className":3002},[],[3004],{"type":39,"value":2608},{"type":45,"tag":459,"props":3006,"children":3007},{},[3008],{"type":39,"value":2613},{"type":45,"tag":422,"props":3010,"children":3011},{},[3012,3016,3024,3032],{"type":45,"tag":459,"props":3013,"children":3014},{},[3015],{"type":39,"value":90},{"type":45,"tag":459,"props":3017,"children":3018},{},[3019],{"type":45,"tag":69,"props":3020,"children":3022},{"className":3021},[],[3023],{"type":39,"value":2629},{"type":45,"tag":459,"props":3025,"children":3026},{},[3027],{"type":45,"tag":69,"props":3028,"children":3030},{"className":3029},[],[3031],{"type":39,"value":2638},{"type":45,"tag":459,"props":3033,"children":3034},{},[3035],{"type":39,"value":2643},{"type":45,"tag":422,"props":3037,"children":3038},{},[3039,3043,3051,3059],{"type":45,"tag":459,"props":3040,"children":3041},{},[3042],{"type":39,"value":90},{"type":45,"tag":459,"props":3044,"children":3045},{},[3046],{"type":45,"tag":69,"props":3047,"children":3049},{"className":3048},[],[3050],{"type":39,"value":795},{"type":45,"tag":459,"props":3052,"children":3053},{},[3054],{"type":45,"tag":69,"props":3055,"children":3057},{"className":3056},[],[3058],{"type":39,"value":2667},{"type":45,"tag":459,"props":3060,"children":3061},{},[3062],{"type":39,"value":2672},{"type":45,"tag":422,"props":3064,"children":3065},{},[3066,3070,3079,3087],{"type":45,"tag":459,"props":3067,"children":3068},{},[3069],{"type":39,"value":90},{"type":45,"tag":459,"props":3071,"children":3072},{},[3073],{"type":45,"tag":69,"props":3074,"children":3076},{"className":3075},[],[3077],{"type":39,"value":3078},"inference.trt_engine",{"type":45,"tag":459,"props":3080,"children":3081},{},[3082],{"type":45,"tag":69,"props":3083,"children":3085},{"className":3084},[],[3086],{"type":39,"value":2667},{"type":45,"tag":459,"props":3088,"children":3089},{},[3090],{"type":39,"value":2672},{"type":45,"tag":422,"props":3092,"children":3093},{},[3094,3098,3106,3114],{"type":45,"tag":459,"props":3095,"children":3096},{},[3097],{"type":39,"value":90},{"type":45,"tag":459,"props":3099,"children":3100},{},[3101],{"type":45,"tag":69,"props":3102,"children":3104},{"className":3103},[],[3105],{"type":39,"value":1063},{"type":45,"tag":459,"props":3107,"children":3108},{},[3109],{"type":45,"tag":69,"props":3110,"children":3112},{"className":3111},[],[3113],{"type":39,"value":2724},{"type":45,"tag":459,"props":3115,"children":3116},{},[3117],{"type":39,"value":2729},{"type":45,"tag":422,"props":3119,"children":3120},{},[3121,3125,3133,3141],{"type":45,"tag":459,"props":3122,"children":3123},{},[3124],{"type":39,"value":90},{"type":45,"tag":459,"props":3126,"children":3127},{},[3128],{"type":45,"tag":69,"props":3129,"children":3131},{"className":3130},[],[3132],{"type":39,"value":2599},{"type":45,"tag":459,"props":3134,"children":3135},{},[3136],{"type":45,"tag":69,"props":3137,"children":3139},{"className":3138},[],[3140],{"type":39,"value":2608},{"type":45,"tag":459,"props":3142,"children":3143},{},[3144],{"type":39,"value":2613},{"type":45,"tag":422,"props":3146,"children":3147},{},[3148,3152,3160,3168],{"type":45,"tag":459,"props":3149,"children":3150},{},[3151],{"type":39,"value":815},{"type":45,"tag":459,"props":3153,"children":3154},{},[3155],{"type":45,"tag":69,"props":3156,"children":3158},{"className":3157},[],[3159],{"type":39,"value":2629},{"type":45,"tag":459,"props":3161,"children":3162},{},[3163],{"type":45,"tag":69,"props":3164,"children":3166},{"className":3165},[],[3167],{"type":39,"value":2638},{"type":45,"tag":459,"props":3169,"children":3170},{},[3171],{"type":39,"value":2643},{"type":45,"tag":422,"props":3173,"children":3174},{},[3175,3179,3187,3195],{"type":45,"tag":459,"props":3176,"children":3177},{},[3178],{"type":39,"value":815},{"type":45,"tag":459,"props":3180,"children":3181},{},[3182],{"type":45,"tag":69,"props":3183,"children":3185},{"className":3184},[],[3186],{"type":39,"value":843},{"type":45,"tag":459,"props":3188,"children":3189},{},[3190],{"type":45,"tag":69,"props":3191,"children":3193},{"className":3192},[],[3194],{"type":39,"value":2667},{"type":45,"tag":459,"props":3196,"children":3197},{},[3198],{"type":39,"value":2672},{"type":45,"tag":422,"props":3200,"children":3201},{},[3202,3206,3214,3223],{"type":45,"tag":459,"props":3203,"children":3204},{},[3205],{"type":39,"value":815},{"type":45,"tag":459,"props":3207,"children":3208},{},[3209],{"type":45,"tag":69,"props":3210,"children":3212},{"className":3211},[],[3213],{"type":39,"value":2421},{"type":45,"tag":459,"props":3215,"children":3216},{},[3217],{"type":45,"tag":69,"props":3218,"children":3220},{"className":3219},[],[3221],{"type":39,"value":3222},"create_pth_file",{"type":45,"tag":459,"props":3224,"children":3225},{},[3226],{"type":39,"value":3227},"output PTH path",{"type":45,"tag":422,"props":3229,"children":3230},{},[3231,3235,3243,3251],{"type":45,"tag":459,"props":3232,"children":3233},{},[3234],{"type":39,"value":815},{"type":45,"tag":459,"props":3236,"children":3237},{},[3238],{"type":45,"tag":69,"props":3239,"children":3241},{"className":3240},[],[3242],{"type":39,"value":2599},{"type":45,"tag":459,"props":3244,"children":3245},{},[3246],{"type":45,"tag":69,"props":3247,"children":3249},{"className":3248},[],[3250],{"type":39,"value":2608},{"type":45,"tag":459,"props":3252,"children":3253},{},[3254],{"type":39,"value":2613},{"type":45,"tag":422,"props":3256,"children":3257},{},[3258,3262,3270,3278],{"type":45,"tag":459,"props":3259,"children":3260},{},[3261],{"type":39,"value":863},{"type":45,"tag":459,"props":3263,"children":3264},{},[3265],{"type":45,"tag":69,"props":3266,"children":3268},{"className":3267},[],[3269],{"type":39,"value":2629},{"type":45,"tag":459,"props":3271,"children":3272},{},[3273],{"type":45,"tag":69,"props":3274,"children":3276},{"className":3275},[],[3277],{"type":39,"value":2638},{"type":45,"tag":459,"props":3279,"children":3280},{},[3281],{"type":39,"value":2643},{"type":45,"tag":422,"props":3283,"children":3284},{},[3285,3289,3297,3305],{"type":45,"tag":459,"props":3286,"children":3287},{},[3288],{"type":39,"value":863},{"type":45,"tag":459,"props":3290,"children":3291},{},[3292],{"type":45,"tag":69,"props":3293,"children":3295},{"className":3294},[],[3296],{"type":39,"value":968},{"type":45,"tag":459,"props":3298,"children":3299},{},[3300],{"type":45,"tag":69,"props":3301,"children":3303},{"className":3302},[],[3304],{"type":39,"value":2667},{"type":45,"tag":459,"props":3306,"children":3307},{},[3308],{"type":39,"value":2672},{"type":45,"tag":422,"props":3310,"children":3311},{},[3312,3316,3324,3332],{"type":45,"tag":459,"props":3313,"children":3314},{},[3315],{"type":39,"value":863},{"type":45,"tag":459,"props":3317,"children":3318},{},[3319],{"type":45,"tag":69,"props":3320,"children":3322},{"className":3321},[],[3323],{"type":39,"value":2599},{"type":45,"tag":459,"props":3325,"children":3326},{},[3327],{"type":45,"tag":69,"props":3328,"children":3330},{"className":3329},[],[3331],{"type":39,"value":2608},{"type":45,"tag":459,"props":3333,"children":3334},{},[3335],{"type":39,"value":2613},{"type":45,"tag":422,"props":3337,"children":3338},{},[3339,3343,3351,3359],{"type":45,"tag":459,"props":3340,"children":3341},{},[3342],{"type":39,"value":989},{"type":45,"tag":459,"props":3344,"children":3345},{},[3346],{"type":45,"tag":69,"props":3347,"children":3349},{"className":3348},[],[3350],{"type":39,"value":2629},{"type":45,"tag":459,"props":3352,"children":3353},{},[3354],{"type":45,"tag":69,"props":3355,"children":3357},{"className":3356},[],[3358],{"type":39,"value":2638},{"type":45,"tag":459,"props":3360,"children":3361},{},[3362],{"type":39,"value":2643},{"type":45,"tag":422,"props":3364,"children":3365},{},[3366,3370,3378,3386],{"type":45,"tag":459,"props":3367,"children":3368},{},[3369],{"type":39,"value":989},{"type":45,"tag":459,"props":3371,"children":3372},{},[3373],{"type":45,"tag":69,"props":3374,"children":3376},{"className":3375},[],[3377],{"type":39,"value":1063},{"type":45,"tag":459,"props":3379,"children":3380},{},[3381],{"type":45,"tag":69,"props":3382,"children":3384},{"className":3383},[],[3385],{"type":39,"value":2667},{"type":45,"tag":459,"props":3387,"children":3388},{},[3389],{"type":39,"value":2672},{"type":45,"tag":422,"props":3391,"children":3392},{},[3393,3397,3405,3413],{"type":45,"tag":459,"props":3394,"children":3395},{},[3396],{"type":39,"value":989},{"type":45,"tag":459,"props":3398,"children":3399},{},[3400],{"type":45,"tag":69,"props":3401,"children":3403},{"className":3402},[],[3404],{"type":39,"value":2599},{"type":45,"tag":459,"props":3406,"children":3407},{},[3408],{"type":45,"tag":69,"props":3409,"children":3411},{"className":3410},[],[3412],{"type":39,"value":2608},{"type":45,"tag":459,"props":3414,"children":3415},{},[3416],{"type":39,"value":2613},{"type":45,"tag":422,"props":3418,"children":3419},{},[3420,3424,3432,3440],{"type":45,"tag":459,"props":3421,"children":3422},{},[3423],{"type":39,"value":1085},{"type":45,"tag":459,"props":3425,"children":3426},{},[3427],{"type":45,"tag":69,"props":3428,"children":3430},{"className":3429},[],[3431],{"type":39,"value":2629},{"type":45,"tag":459,"props":3433,"children":3434},{},[3435],{"type":45,"tag":69,"props":3436,"children":3438},{"className":3437},[],[3439],{"type":39,"value":2638},{"type":45,"tag":459,"props":3441,"children":3442},{},[3443],{"type":39,"value":2643},{"type":45,"tag":422,"props":3445,"children":3446},{},[3447,3451,3459,3467],{"type":45,"tag":459,"props":3448,"children":3449},{},[3450],{"type":39,"value":1085},{"type":45,"tag":459,"props":3452,"children":3453},{},[3454],{"type":45,"tag":69,"props":3455,"children":3457},{"className":3456},[],[3458],{"type":39,"value":2599},{"type":45,"tag":459,"props":3460,"children":3461},{},[3462],{"type":45,"tag":69,"props":3463,"children":3465},{"className":3464},[],[3466],{"type":39,"value":2608},{"type":45,"tag":459,"props":3468,"children":3469},{},[3470],{"type":39,"value":2613},{"type":45,"tag":422,"props":3472,"children":3473},{},[3474,3478,3487,3496],{"type":45,"tag":459,"props":3475,"children":3476},{},[3477],{"type":39,"value":1085},{"type":45,"tag":459,"props":3479,"children":3480},{},[3481],{"type":45,"tag":69,"props":3482,"children":3484},{"className":3483},[],[3485],{"type":39,"value":3486},"train.pretrained_model_path",{"type":45,"tag":459,"props":3488,"children":3489},{},[3490],{"type":45,"tag":69,"props":3491,"children":3493},{"className":3492},[],[3494],{"type":39,"value":3495},"ptm_if_no_resume_model",{"type":45,"tag":459,"props":3497,"children":3498},{},[3499],{"type":39,"value":3500},"PTM when no resume checkpoint exists",{"type":45,"tag":422,"props":3502,"children":3503},{},[3504,3508,3516,3525],{"type":45,"tag":459,"props":3505,"children":3506},{},[3507],{"type":39,"value":1085},{"type":45,"tag":459,"props":3509,"children":3510},{},[3511],{"type":45,"tag":69,"props":3512,"children":3514},{"className":3513},[],[3515],{"type":39,"value":1311},{"type":45,"tag":459,"props":3517,"children":3518},{},[3519],{"type":45,"tag":69,"props":3520,"children":3522},{"className":3521},[],[3523],{"type":39,"value":3524},"resume_model",{"type":45,"tag":459,"props":3526,"children":3527},{},[3528],{"type":39,"value":3529},"model file inferred from the current job results folder",{"type":45,"tag":53,"props":3531,"children":3532},{},[3533,3535,3540,3542,3548,3550,3556,3558,3563],{"type":39,"value":3534},"For ",{"type":45,"tag":69,"props":3536,"children":3538},{"className":3537},[],[3539],{"type":39,"value":2667},{"type":39,"value":3541}," or ",{"type":45,"tag":69,"props":3543,"children":3545},{"className":3544},[],[3546],{"type":39,"value":3547},"parent_model_folder",{"type":39,"value":3549},", pass the upstream train\u002Fexport\u002FAutoML child job id as ",{"type":45,"tag":69,"props":3551,"children":3553},{"className":3552},[],[3554],{"type":39,"value":3555},"parent_job_id",{"type":39,"value":3557},". The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to ",{"type":45,"tag":69,"props":3559,"children":3561},{"className":3560},[],[3562],{"type":39,"value":2521},{"type":39,"value":3564}," and do not patch generated runner scripts to guess checkpoint paths.",{"type":45,"tag":118,"props":3566,"children":3568},{"id":3567},"deployment",[3569],{"type":39,"value":3570},"Deployment",{"type":45,"tag":371,"props":3572,"children":3573},{},[3574],{"type":45,"tag":375,"props":3575,"children":3576},{},[3577],{"type":45,"tag":3578,"props":3579,"children":3580},"a",{"href":98},[3581],{"type":39,"value":3582},"tao-deploy-ocrnet",{"type":45,"tag":3584,"props":3585,"children":3586},"style",{},[3587],{"type":39,"value":3588},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":3590,"total":3746},[3591,3609,3624,3635,3647,3661,3674,3688,3701,3712,3726,3735],{"slug":3592,"name":3592,"fn":3593,"description":3594,"org":3595,"tags":3596,"stars":3606,"repoUrl":3607,"updatedAt":3608},"nemoclaw-user-guide","retrieve NemoClaw documentation and configuration","Guides human users' AI agents to the NemoClaw docs MCP server and canonical Fern documentation in Markdown form. Use when users ask how to install, configure, operate, troubleshoot, secure, or learn NemoClaw with an AI coding assistant. Trigger keywords - nemoclaw docs, use nemoclaw with ai agent, nemoclaw mcp docs, nemoclaw install help, nemoclaw quickstart, nemoclaw markdown docs, llms.txt, agent skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3597,3600,3603],{"name":3598,"slug":3599,"type":15},"Documentation","documentation",{"name":3601,"slug":3602,"type":15},"MCP","mcp",{"name":3604,"slug":3605,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":3610,"name":3610,"fn":3611,"description":3612,"org":3613,"tags":3614,"stars":3621,"repoUrl":3622,"updatedAt":3623},"mcore-build-and-dependency","manage Megatron-LM development environments","Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3615,3618,3619],{"name":3616,"slug":3617,"type":15},"Containers","containers",{"name":3570,"slug":3567,"type":15},{"name":3620,"slug":1409,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":3625,"name":3625,"fn":3626,"description":3627,"org":3628,"tags":3629,"stars":3621,"repoUrl":3622,"updatedAt":3634},"mcore-bump-base-image","update NVIDIA PyTorch base images","Bump the NVIDIA PyTorch base image (`nvcr.io\u002Fnvidia\u002Fpytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker\u002F.ngc_version.dev` and GitLab CI in `.gitlab\u002Fstages\u002F01.build.yml`), the post-bump CI loop (re-run functional tests, refresh golden values, mark broken tests), and the gotchas that bit PRs",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3630,3633],{"name":3631,"slug":3632,"type":15},"CI\u002FCD","ci-cd",{"name":3570,"slug":3567,"type":15},"2026-07-14T05:25:59.97109",{"slug":3636,"name":3636,"fn":3637,"description":3638,"org":3639,"tags":3640,"stars":3621,"repoUrl":3622,"updatedAt":3646},"mcore-cicd","manage CI\u002FCD pipelines for Megatron-LM","CI\u002FCD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request\u002FBRANCH ref — always dry-run and verify the destination first; never run against shared or protected branches), and CI failure investigation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3641,3642,3643],{"name":3631,"slug":3632,"type":15},{"name":3570,"slug":3567,"type":15},{"name":3644,"slug":3645,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":3648,"name":3648,"fn":3649,"description":3650,"org":3651,"tags":3652,"stars":3621,"repoUrl":3622,"updatedAt":3660},"mcore-create-issue","investigate CI failures and create issues","Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3653,3656,3657],{"name":3654,"slug":3655,"type":15},"Debugging","debugging",{"name":3644,"slug":3645,"type":15},{"name":3658,"slug":3659,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":3662,"name":3662,"fn":3663,"description":3664,"org":3665,"tags":3666,"stars":3621,"repoUrl":3622,"updatedAt":3673},"mcore-linting-and-formatting","lint and format Megatron-LM code","Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3667,3670],{"name":3668,"slug":3669,"type":15},"Best Practices","best-practices",{"name":3671,"slug":3672,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":3675,"name":3675,"fn":3676,"description":3677,"org":3678,"tags":3679,"stars":3621,"repoUrl":3622,"updatedAt":3687},"mcore-migrate-gpt-to-hybrid","migrate Megatron-LM models to HybridModel","Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3680,3683,3686],{"name":3681,"slug":3682,"type":15},"Machine Learning","machine-learning",{"name":3684,"slug":3685,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":3689,"name":3689,"fn":3690,"description":3691,"org":3692,"tags":3693,"stars":3621,"repoUrl":3622,"updatedAt":3700},"mcore-onboard-gb200-1node-tests","onboard functional tests for GB200","Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3694,3697],{"name":3695,"slug":3696,"type":15},"QA","qa",{"name":3698,"slug":3699,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":3702,"name":3702,"fn":3703,"description":3704,"org":3705,"tags":3706,"stars":3621,"repoUrl":3622,"updatedAt":3711},"mcore-run-on-slurm","launch distributed training jobs on SLURM","How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3707,3708],{"name":3570,"slug":3567,"type":15},{"name":3709,"slug":3710,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":3713,"name":3713,"fn":3714,"description":3715,"org":3716,"tags":3717,"stars":3621,"repoUrl":3622,"updatedAt":3725},"mcore-split-pr","split pull requests to reduce review load","Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3718,3721,3722],{"name":3719,"slug":3720,"type":15},"Code Review","code-review",{"name":3644,"slug":3645,"type":15},{"name":3723,"slug":3724,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":3727,"name":3727,"fn":3728,"description":3729,"org":3730,"tags":3731,"stars":3621,"repoUrl":3622,"updatedAt":3734},"mcore-testing","run and manage Megatron-LM tests","Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3732,3733],{"name":3695,"slug":3696,"type":15},{"name":3698,"slug":3699,"type":15},"2026-07-14T05:25:54.928983",{"slug":3736,"name":3736,"fn":3737,"description":3738,"org":3739,"tags":3740,"stars":3621,"repoUrl":3622,"updatedAt":3745},"nightly-sync","manage nightly main-to-dev sync workflows","Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3741,3744],{"name":3742,"slug":3743,"type":15},"Automation","automation",{"name":3631,"slug":3632,"type":15},"2026-07-30T05:29:03.275638",496,{"items":3748,"total":3844},[3749,3766,3776,3790,3800,3815,3830],{"slug":3750,"name":3750,"fn":3751,"description":3752,"org":3753,"tags":3754,"stars":20,"repoUrl":21,"updatedAt":3765},"accelerated-computing-cudf","accelerate data processing with cuDF","Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV\u002FParquet I\u002FO, nullable semantics, and multi-GPU DataFrame workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3755,3758,3761,3762],{"name":3756,"slug":3757,"type":15},"Data Analysis","data-analysis",{"name":3759,"slug":3760,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":3763,"slug":3764,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":3767,"name":3767,"fn":3768,"description":3769,"org":3770,"tags":3771,"stars":20,"repoUrl":21,"updatedAt":3775},"aiq-deploy","deploy and manage NVIDIA AI-Q infrastructure","Use when asked to install, deploy, run, validate, troubleshoot, or stop NVIDIA AI-Q Blueprint infrastructure.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3772,3773,3774],{"name":3570,"slug":3567,"type":15},{"name":3709,"slug":3710,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":3777,"name":3777,"fn":3778,"description":3779,"org":3780,"tags":3781,"stars":20,"repoUrl":21,"updatedAt":3789},"aiq-research","conduct deep research with AI-Q","Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3782,3785,3786],{"name":3783,"slug":3784,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":3787,"slug":3788,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":3791,"name":3791,"fn":3792,"description":3793,"org":3794,"tags":3795,"stars":20,"repoUrl":21,"updatedAt":3799},"amc-run-sample-calibration","run AMC sample dataset calibration","Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3796,3797,3798],{"name":3756,"slug":3757,"type":15},{"name":9,"slug":8,"type":15},{"name":3698,"slug":3699,"type":15},"2026-07-17T05:29:03.913266",{"slug":3801,"name":3801,"fn":3802,"description":3803,"org":3804,"tags":3805,"stars":20,"repoUrl":21,"updatedAt":3814},"amc-run-video-calibration","calibrate video datasets with AutoMagicCalib","Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API. Use when user has local MP4s and says 'calibrate my videos', 'run AMC on these videos', or similar. For RTSP\u002Flive streams, use amc-run-rtsp-calibration instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3806,3807,3810,3811],{"name":3742,"slug":3743,"type":15},{"name":3808,"slug":3809,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":3812,"slug":3813,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":3816,"name":3816,"fn":3817,"description":3818,"org":3819,"tags":3820,"stars":20,"repoUrl":21,"updatedAt":3829},"amc-setup-calibration-stack","deploy AutoMagicCalib microservice with Docker","Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3821,3822,3825,3826],{"name":3570,"slug":3567,"type":15},{"name":3823,"slug":3824,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":3827,"slug":3828,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":3831,"name":3831,"fn":3832,"description":3833,"org":3834,"tags":3835,"stars":20,"repoUrl":21,"updatedAt":3843},"cudaq-guide","develop quantum applications with CUDA-Q","CUDA-Q onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[3836,3837,3840],{"name":9,"slug":8,"type":15},{"name":3838,"slug":3839,"type":15},"Quantum Computing","quantum-computing",{"name":3841,"slug":3842,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305]