[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AABDPPHBPNRME0JMM9R40B":3,"skill-same:01M7AABDPPHBPNRME0JMM9R40B":116},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":32,"commitSha":33,"skillDirectory":34,"forksCount":35,"repositoryHtmlUrl":36,"repositoryHomepageUrl":37,"ownerHtmlUrl":38,"ownerBlogUrl":39,"ownerDescription":40,"updatedAt":41,"install":42,"resources":55},"01M7AABDPPHBPNRME0JMM9R40B","tao-train-reid",null,"Person re-identification (ReID). Learns discriminative embeddings to match the same person across different camera views, based on metric learning. Use when training, evaluating, exporting, or running inference for a TAO person re-identification model. Trigger phrases include \"train ReID\", \"person re-identification\", \"cross-camera person matching\", \"ReID embeddings\", \"person re-id\".","themusebook\u002Fskills","themusebook","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F349927247?v=4","01M7A5C1RV5S3V5FF6ZACVDNRN","musebook","01M7A5C1SFC3KRZAXNN60XJ7KM",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-train-reid\ndescription: Person re-identification (ReID). Learns discriminative embeddings to match the same person across different\n  camera views, based on metric learning. Use when training, evaluating, exporting, or running inference for a TAO person\n  re-identification model. Trigger phrases include \"train ReID\", \"person re-identification\", \"cross-camera person matching\",\n  \"ReID embeddings\", \"person re-id\".\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- re\n- identification\n---\n\n# Re-Identification\n\nPerson re-identification. Learns discriminative embeddings to match the same person across different camera views. Metric learning based.\n\nSet model.pretrained_model_path for pretrained weights.\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## Supported Actions\n\nThe packaged Re-Identification PyT CLI supports `train`, `evaluate`, `inference`, `export`, and `default_specs`. This model skill exposes the runnable user actions `train`, `evaluate`, `inference`, and `export`; resume\u002Fretrain is performed through `train` with `train.resume_training_checkpoint_path`.\n\nDo not advertise or synthesize `dataset_convert`, `deploy`, `prune`, `quantize`, `gen_trt_engine`, or standalone `retrain` for this model unless the packaged model skill and real CLI add those actions.\n\n## Training Requirements\n\n- **Dataset type:** re_identification\n- **Formats:** default\n- **Monitoring metric:** cmc_rank_1, maximize\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | evaluate.test_dataset | train_datasets | sample_test.tar.gz | No |\n| evaluate | evaluate.query_dataset | train_datasets | sample_query.tar.gz | No |\n| inference | inference.test_dataset | train_datasets | sample_test.tar.gz | No |\n| inference | inference.query_dataset | train_datasets | sample_query.tar.gz | No |\n| train | dataset.train_dataset_dir | train_datasets | sample_train.tar.gz | No |\n| train | dataset.test_dataset_dir | train_datasets | sample_test.tar.gz | No |\n| train | dataset.query_dataset_dir | train_datasets | sample_query.tar.gz | No |\n\n### Typical Spec Overrides\n\nData source overrides are **mandatory for every action** — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in `spec_overrides`.\n\n```python\nS3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nCHECKPOINT = \"\u002Fresults\u002F{train_job_id}\u002Fresults_dir\u002Fmodel_epoch_000_step_00099.pth\"\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.num_classes\": 100,\n    \"dataset.num_workers\": 4,\n    \"dataset.batch_size\": 16,\n    \"dataset.num_instances\": 4,\n    \"dataset.train_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_train.tar.gz\",\n    \"dataset.test_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"dataset.query_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n}\n```\n\n**resume train (mandatory checkpoint):**\n```python\n{\n    \"train.num_epochs\": 31,\n    \"train.resume_training_checkpoint_path\": CHECKPOINT,\n    \"dataset.num_classes\": 100,\n    \"dataset.batch_size\": 16,\n    \"dataset.num_instances\": 4,\n    \"dataset.train_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_train.tar.gz\",\n    \"dataset.test_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"dataset.query_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n}\n```\n\n**evaluate (mandatory data sources and checkpoint):**\n```python\n{\n    \"evaluate.test_dataset\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"evaluate.query_dataset\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n    \"evaluate.checkpoint\": CHECKPOINT,\n    \"evaluate.output_cmc_curve_plot\": \"\u002Fresults\u002F{evaluate_job_id}\u002Fresults_dir\u002Fcmc_curve.png\",\n    \"evaluate.output_sampled_matches_plot\": \"\u002Fresults\u002F{evaluate_job_id}\u002Fresults_dir\u002Fsampled_matches.png\",\n}\n```\n\n**export (mandatory checkpoint and output):**\n```python\n{\n    \"export.checkpoint\": CHECKPOINT,\n    \"export.onnx_file\": \"\u002Fresults\u002F{export_job_id}\u002Fresults_dir\u002Freid.onnx\",\n}\n```\n\n**inference (mandatory data sources and checkpoint):**\n```python\n{\n    \"inference.test_dataset\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"inference.query_dataset\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n    \"inference.checkpoint\": CHECKPOINT,\n    \"inference.output_file\": \"\u002Fresults\u002F{inference_job_id}\u002Fresults_dir\u002Freid_inference.json\",\n}\n```\n\nFor export and inference, provide explicit file paths for `export.onnx_file` and `inference.output_file`. For evaluate, provide explicit file paths for `evaluate.output_cmc_curve_plot` and `evaluate.output_sampled_matches_plot`. Keep these as spec values or `spec_params` mappings; do not declare them as file outputs in `skill_info.yaml` for local Docker until the runner distinguishes files from folders during output pre-creation.\n\n## Eval Dataset\n\nRequired. Evaluation requires test and query datasets for retrieval-based metrics (CMC, mAP).\n\n## Important Parameters\n\n- **dataset.num_classes**: Number of identities. Default 751. Must match the number of unique identities in training data.\n- **model.backbone**: Default resnet_50.\n- **optim.base_lr**: Base learning rate. Default 3.5e-4.\n- **dataset.batch_size**: Per-GPU batch size. Default 64. Re-ID benefits from large batches for better triplet\u002Fcontrastive sampling.\n- **dataset.num_instances**: Number of instances per identity in a batch. Controls sampling strategy for metric learning.\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\n- Multi-GPU strategy: `ddp_find_unused_parameters_true`\n- `sync_batchnorm` is always enabled\n- Precision forced to FP16 (`16-mixed`)\n- No explicit `num_nodes` config — single-node oriented\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 2 GPU(s). 16GB+ VRAM per GPU. Re-ID models are relatively lightweight but benefit from large batch sizes for metric learning.\n\n## Error Patterns\n\n**num_classes mismatch**: Ensure dataset.num_classes equals the number of unique identity folders in the training set.\n\n**Invalid triplet batch shape**: `dataset.batch_size` must be compatible with `dataset.num_instances` so each mini-batch can be reshaped for hard-example mining. For local AutoML smoke runs, keep `dataset.batch_size` fixed to a known valid multiple such as 16 with `dataset.num_instances: 4`, and tune `train.optim.base_lr` instead of unconstrained batch size.\n\n**Query\u002Fgallery mismatch**: Query and test (gallery) datasets must share the same identity namespace.\n\n**PyTorch 2.6 checkpoint load failure on checkpoint consumers**: Current Re-ID\ncheckpoints include OmegaConf containers. For checkpoints produced by the same\ntrusted TAO train\u002FAutoML workflow, set\n`TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD=1` in downstream resume, evaluate, inference,\nand export job env vars so Lightning\u002FPyTorch can load the full checkpoint. Do\nnot use this env var for untrusted checkpoints.\n\n**AutoML metric extraction**: Re-ID train status files report retrieval KPIs such as `cmc_rank_1`, `cmc_rank_5`, `cmc_rank_10`, and `mAP`, plus train loss. Default AutoML train launches must optimize `cmc_rank_1` with `direction: maximize`; do not use `val_loss` as the metric for this model.\n\n**Checkpoint handoff**: Use the checkpoint resolver on the best AutoML child job's `results_dir\u002Ftrain\u002F` folder and select the action-appropriate `model_epoch_*.pth` checkpoint. Re-ID also writes `reid_model_latest.pth`, but that is a latest symlink and should only be used when a caller explicitly requests latest. Preserve the same dataset identity count and query\u002Fgallery archives for downstream actions.\n\n**Default spec generation**: The packaged `default_specs` CLI action does not\nconsume the normal `-e \u003Cspec.yaml>` experiment file for `results_dir`. Invoke it\nwith a Hydra override such as\n`re_identification default_specs results_dir=\u002Fworkspace\u002Frun\u002Fresults\u002Fdefault_specs`.\nPassing only `-e` leaves `cfg.results_dir` unset and fails with\n`MissingMandatoryValue: results_dir`.\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 `re_identification.config.json`:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\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.output_cmc_curve_plot` | `create_evaluate_cmc_plot_reid` | ReID CMC plot path |\n| evaluate | `evaluate.output_sampled_matches_plot` | `create_evaluate_matches_plot_reid` | ReID sampled matches plot path |\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| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `inference.output_file` | `create_inference_result_file_reid` | ReID inference JSON path |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `model.pretrained_model_path` | `ptm_if_no_resume_model` | PTM when no resume checkpoint exists |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.resume_training_checkpoint_path` | `resume_model` | model file inferred from the current job results folder |\n\nFor `parent_model` or `parent_model_folder`, pass the upstream train\u002Fexport\u002FAutoML child job id as `parent_job_id`. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to `config.json` and do not patch generated runner scripts to guess checkpoint paths.\n","# Re-Identification\n\nPerson re-identification. Learns discriminative embeddings to match the same person across different camera views. Metric learning based.\n\nSet model.pretrained_model_path for pretrained weights.\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## Supported Actions\n\nThe packaged Re-Identification PyT CLI supports `train`, `evaluate`, `inference`, `export`, and `default_specs`. This model skill exposes the runnable user actions `train`, `evaluate`, `inference`, and `export`; resume\u002Fretrain is performed through `train` with `train.resume_training_checkpoint_path`.\n\nDo not advertise or synthesize `dataset_convert`, `deploy`, `prune`, `quantize`, `gen_trt_engine`, or standalone `retrain` for this model unless the packaged model skill and real CLI add those actions.\n\n## Training Requirements\n\n- **Dataset type:** re_identification\n- **Formats:** default\n- **Monitoring metric:** cmc_rank_1, maximize\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | evaluate.test_dataset | train_datasets | sample_test.tar.gz | No |\n| evaluate | evaluate.query_dataset | train_datasets | sample_query.tar.gz | No |\n| inference | inference.test_dataset | train_datasets | sample_test.tar.gz | No |\n| inference | inference.query_dataset | train_datasets | sample_query.tar.gz | No |\n| train | dataset.train_dataset_dir | train_datasets | sample_train.tar.gz | No |\n| train | dataset.test_dataset_dir | train_datasets | sample_test.tar.gz | No |\n| train | dataset.query_dataset_dir | train_datasets | sample_query.tar.gz | No |\n\n### Typical Spec Overrides\n\nData source overrides are **mandatory for every action** — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in `spec_overrides`.\n\n```python\nS3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nCHECKPOINT = \"\u002Fresults\u002F{train_job_id}\u002Fresults_dir\u002Fmodel_epoch_000_step_00099.pth\"\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.num_classes\": 100,\n    \"dataset.num_workers\": 4,\n    \"dataset.batch_size\": 16,\n    \"dataset.num_instances\": 4,\n    \"dataset.train_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_train.tar.gz\",\n    \"dataset.test_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"dataset.query_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n}\n```\n\n**resume train (mandatory checkpoint):**\n```python\n{\n    \"train.num_epochs\": 31,\n    \"train.resume_training_checkpoint_path\": CHECKPOINT,\n    \"dataset.num_classes\": 100,\n    \"dataset.batch_size\": 16,\n    \"dataset.num_instances\": 4,\n    \"dataset.train_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_train.tar.gz\",\n    \"dataset.test_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"dataset.query_dataset_dir\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n}\n```\n\n**evaluate (mandatory data sources and checkpoint):**\n```python\n{\n    \"evaluate.test_dataset\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"evaluate.query_dataset\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n    \"evaluate.checkpoint\": CHECKPOINT,\n    \"evaluate.output_cmc_curve_plot\": \"\u002Fresults\u002F{evaluate_job_id}\u002Fresults_dir\u002Fcmc_curve.png\",\n    \"evaluate.output_sampled_matches_plot\": \"\u002Fresults\u002F{evaluate_job_id}\u002Fresults_dir\u002Fsampled_matches.png\",\n}\n```\n\n**export (mandatory checkpoint and output):**\n```python\n{\n    \"export.checkpoint\": CHECKPOINT,\n    \"export.onnx_file\": \"\u002Fresults\u002F{export_job_id}\u002Fresults_dir\u002Freid.onnx\",\n}\n```\n\n**inference (mandatory data sources and checkpoint):**\n```python\n{\n    \"inference.test_dataset\": f\"{S3_TRAIN}\u002Fsample_test.tar.gz\",\n    \"inference.query_dataset\": f\"{S3_TRAIN}\u002Fsample_query.tar.gz\",\n    \"inference.checkpoint\": CHECKPOINT,\n    \"inference.output_file\": \"\u002Fresults\u002F{inference_job_id}\u002Fresults_dir\u002Freid_inference.json\",\n}\n```\n\nFor export and inference, provide explicit file paths for `export.onnx_file` and `inference.output_file`. For evaluate, provide explicit file paths for `evaluate.output_cmc_curve_plot` and `evaluate.output_sampled_matches_plot`. Keep these as spec values or `spec_params` mappings; do not declare them as file outputs in `skill_info.yaml` for local Docker until the runner distinguishes files from folders during output pre-creation.\n\n## Eval Dataset\n\nRequired. Evaluation requires test and query datasets for retrieval-based metrics (CMC, mAP).\n\n## Important Parameters\n\n- **dataset.num_classes**: Number of identities. Default 751. Must match the number of unique identities in training data.\n- **model.backbone**: Default resnet_50.\n- **optim.base_lr**: Base learning rate. Default 3.5e-4.\n- **dataset.batch_size**: Per-GPU batch size. Default 64. Re-ID benefits from large batches for better triplet\u002Fcontrastive sampling.\n- **dataset.num_instances**: Number of instances per identity in a batch. Controls sampling strategy for metric learning.\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\n- Multi-GPU strategy: `ddp_find_unused_parameters_true`\n- `sync_batchnorm` is always enabled\n- Precision forced to FP16 (`16-mixed`)\n- No explicit `num_nodes` config — single-node oriented\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 2 GPU(s). 16GB+ VRAM per GPU. Re-ID models are relatively lightweight but benefit from large batch sizes for metric learning.\n\n## Error Patterns\n\n**num_classes mismatch**: Ensure dataset.num_classes equals the number of unique identity folders in the training set.\n\n**Invalid triplet batch shape**: `dataset.batch_size` must be compatible with `dataset.num_instances` so each mini-batch can be reshaped for hard-example mining. For local AutoML smoke runs, keep `dataset.batch_size` fixed to a known valid multiple such as 16 with `dataset.num_instances: 4`, and tune `train.optim.base_lr` instead of unconstrained batch size.\n\n**Query\u002Fgallery mismatch**: Query and test (gallery) datasets must share the same identity namespace.\n\n**PyTorch 2.6 checkpoint load failure on checkpoint consumers**: Current Re-ID\ncheckpoints include OmegaConf containers. For checkpoints produced by the same\ntrusted TAO train\u002FAutoML workflow, set\n`TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD=1` in downstream resume, evaluate, inference,\nand export job env vars so Lightning\u002FPyTorch can load the full checkpoint. Do\nnot use this env var for untrusted checkpoints.\n\n**AutoML metric extraction**: Re-ID train status files report retrieval KPIs such as `cmc_rank_1`, `cmc_rank_5`, `cmc_rank_10`, and `mAP`, plus train loss. Default AutoML train launches must optimize `cmc_rank_1` with `direction: maximize`; do not use `val_loss` as the metric for this model.\n\n**Checkpoint handoff**: Use the checkpoint resolver on the best AutoML child job's `results_dir\u002Ftrain\u002F` folder and select the action-appropriate `model_epoch_*.pth` checkpoint. Re-ID also writes `reid_model_latest.pth`, but that is a latest symlink and should only be used when a caller explicitly requests latest. Preserve the same dataset identity count and query\u002Fgallery archives for downstream actions.\n\n**Default spec generation**: The packaged `default_specs` CLI action does not\nconsume the normal `-e \u003Cspec.yaml>` experiment file for `results_dir`. Invoke it\nwith a Hydra override such as\n`re_identification default_specs results_dir=\u002Fworkspace\u002Frun\u002Fresults\u002Fdefault_specs`.\nPassing only `-e` leaves `cfg.results_dir` unset and fails with\n`MissingMandatoryValue: results_dir`.\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 `re_identification.config.json`:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\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.output_cmc_curve_plot` | `create_evaluate_cmc_plot_reid` | ReID CMC plot path |\n| evaluate | `evaluate.output_sampled_matches_plot` | `create_evaluate_matches_plot_reid` | ReID sampled matches plot path |\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| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `inference.output_file` | `create_inference_result_file_reid` | ReID inference JSON path |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `model.pretrained_model_path` | `ptm_if_no_resume_model` | PTM when no resume checkpoint exists |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.resume_training_checkpoint_path` | `resume_model` | model file inferred from the current job results folder |\n\nFor `parent_model` or `parent_model_folder`, pass the upstream train\u002Fexport\u002FAutoML child job id as `parent_job_id`. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to `config.json` and do not patch generated runner scripts to guess checkpoint paths.\n",{"name":5,"tags":24,"license":15,"metadata":27,"description":7,"allowed-tools":30,"compatibility":31},[25,26],"re","identification",{"author":28,"version":29},"NVIDIA Corporation","0.1.0","Read Bash","Requires docker + nvidia-container-toolkit.","main","1c523aba5cdf867a79f078847075492d7778092a","skills\u002Fnvidia\u002Ftao-train-reid",1,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills","https:\u002F\u002Fskills.x402.wtf","https:\u002F\u002Fgithub.com\u002Fthemusebook","https:\u002F\u002Fmusebook.trade","The official Github of musebook","2026-11-02 23:42:26",{"source":8,"skillName":5,"commands":43,"prompt":53,"zipAvailable":54},[44,47,50],{"tool":45,"command":46},"npx","npx skills add themusebook\u002Fskills --skill tao-train-reid",{"tool":48,"command":49},"bunx","bunx skills add themusebook\u002Fskills --skill tao-train-reid",{"tool":51,"command":52},"pnpm","pnpm dlx skills add themusebook\u002Fskills --skill tao-train-reid","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002FSKILL.md",true,[56,62,66,70,74,78,82,86,90,94,98,102,106,110],{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"BENCHMARK.md","REFERENCE","参考文档",4030,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002FBENCHMARK.md",{"path":63,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":64,"sourceHtmlUrl":65},"evals\u002Fevals.json",752,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fevals\u002Fevals.json",{"path":67,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002Fskill_info.yaml",4441,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Freferences\u002Fskill_info.yaml",{"path":71,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Fspec_template_evaluate.yaml",2448,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Freferences\u002Fspec_template_evaluate.yaml",{"path":75,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":76,"sourceHtmlUrl":77},"references\u002Fspec_template_export.yaml",2292,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Freferences\u002Fspec_template_export.yaml",{"path":79,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":80,"sourceHtmlUrl":81},"references\u002Fspec_template_inference.yaml",2405,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Freferences\u002Fspec_template_inference.yaml",{"path":83,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":84,"sourceHtmlUrl":85},"references\u002Fspec_template_train.yaml",2221,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Freferences\u002Fspec_template_train.yaml",{"path":87,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":88,"sourceHtmlUrl":89},"schemas\u002Fevaluate.schema.json",38303,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fschemas\u002Fevaluate.schema.json",{"path":91,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":92,"sourceHtmlUrl":93},"schemas\u002Fexport.schema.json",35732,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fschemas\u002Fexport.schema.json",{"path":95,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":96,"sourceHtmlUrl":97},"schemas\u002Finference.schema.json",37946,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fschemas\u002Finference.schema.json",{"path":99,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":100,"sourceHtmlUrl":101},"schemas\u002Fmanifest.json",6616,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fschemas\u002Fmanifest.json",{"path":103,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":104,"sourceHtmlUrl":105},"schemas\u002Ftrain.schema.json",34511,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fschemas\u002Ftrain.schema.json",{"path":107,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":108,"sourceHtmlUrl":109},"skill-card.md",4060,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fskill-card.md",{"path":111,"resourceType":112,"resourceTypeDesc":113,"sizeBytes":114,"sourceHtmlUrl":115},"skill.oms.sig","OTHER","其他",7025,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-reid\u002Fskill.oms.sig",[117,126,133,140,147,154],{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":122,"sourceHtmlUrl":123,"categories":124,"keywords":125,"publishedAt":19},"01M7AB34SPKVSDRRG009PM1CZH","youtube-clipper","YouTube 视频智能剪辑工具。下载视频和字幕，AI 分析生成精细章节（几分钟级别）， 用户选择片段后自动剪辑、翻译字幕为中英双语、烧录字幕到视频，并生成总结文案。 使用场景：当用户需要剪辑 YouTube 视频、生成短视频片段、制作双语字幕版本时。 关键词：视频剪辑、YouTube、字幕翻译、双语字幕、视频下载、clip video","MIT","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FSKILL.md",[],[],{"skillNo":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":132,"publishedAt":19},"01M7AB18BH6N7KR7939HRNWVMQ","viem-integration","Integrate EVM blockchains using viem. Use when user says \"read blockchain data\", \"send transaction\", \"interact with smart contract\", \"connect to Ethereum\", \"use viem\", \"use wagmi\", \"wallet integration\", \"viem setup\", or mentions blockchain\u002FEVM development with TypeScript.","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fviem-integration\u002FSKILL.md",[],[],{"skillNo":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":139,"publishedAt":19},"01M7AB12GHGP48WCNXDQQ9222H","v4-security-foundations","Security-first Uniswap v4 hook development. Use when user mentions \"v4 hooks\", \"hook security\", \"PoolManager\", \"beforeSwap\", \"afterSwap\", or asks about V4 hook best practices, vulnerabilities, or audit requirements.","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fv4-security-foundations\u002FSKILL.md",[],[],{"skillNo":141,"name":142,"nameZh":6,"description":143,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":6,"sourceHtmlUrl":144,"categories":145,"keywords":146,"publishedAt":19},"01M7AB0XM3PSR7V3Q2ZRR0Q0FB","v4-sdk-integration","App-layer SDK guide for building swap and liquidity experiences directly with the Uniswap v4 SDK. Use when user asks about \"v4 sdk\", \"uniswap v4\", \"v4 swap\", \"v4 liquidity\", \"PoolManager\", \"V4Planner\", \"StateView\", \"PositionManager\", \"pool state\", \"v4 position\", \"uniswap sdk\", or when building swap\u002Fliquidity UX directly with SDKs rather than via the Trading API.","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fv4-sdk-integration\u002FSKILL.md",[],[],{"skillNo":148,"name":149,"nameZh":6,"description":150,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":6,"sourceHtmlUrl":151,"categories":152,"keywords":153,"publishedAt":19},"01M7AB0SGA93ASA588DJW4P9WT","v4-hook-generator","Generate Uniswap v4 hook contracts via OpenZeppelin MCP. Use when building custom swap logic, async swaps, hook-owned liquidity, custom curves, dynamic fees, MEV protection, limit orders, or oracle hooks.","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fv4-hook-generator\u002FSKILL.md",[],[],{"skillNo":155,"name":156,"nameZh":6,"description":157,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":121,"licenseFileUrl":6,"sourceHtmlUrl":158,"categories":159,"keywords":160,"publishedAt":19},"01M7AAZYH75H4CS03WSW2EZ8DM","testing","For testing with Light Protocol programs and clients on localnet, devnet, and mainnet validation.","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftesting\u002FSKILL.md",[],[]]