[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7A78BS3VT7Y919VMQQCF4SS":3,"skill-same:01M7A78BS3VT7Y919VMQQCF4SS":119},{"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":31,"commitSha":32,"skillDirectory":33,"forksCount":34,"repositoryHtmlUrl":35,"repositoryHomepageUrl":36,"ownerHtmlUrl":37,"ownerBlogUrl":38,"ownerDescription":39,"updatedAt":40,"install":41,"resources":54},"01M7A78BS3VT7Y919VMQQCF4SS","tao-train-mask2former",null,"Mask2Former for universal image segmentation (panoptic, instance, and semantic). Transformer-based with masked attention for high-quality segmentation results. Use when training, evaluating, exporting, quantizing, or running inference for a TAO Mask2Former model. Trigger phrases include \"train Mask2Former\", \"universal segmentation\", \"panoptic \u002F instance \u002F semantic segmentation\", \"masked-attention transformer segmenter\".","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\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-train-mask2former\ndescription: Mask2Former for universal image segmentation (panoptic, instance, and semantic). Transformer-based with\n  masked attention for high-quality segmentation results. Use when training, evaluating, exporting, quantizing, or running\n  inference for a TAO Mask2Former model. Trigger phrases include \"train Mask2Former\", \"universal segmentation\",\n  \"panoptic \u002F instance \u002F semantic segmentation\", \"masked-attention transformer segmenter\".\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- segmentation\n---\n\n# Mask2Former\n\nMask2Former for universal image segmentation (panoptic, instance, and semantic). Transformer-based with masked attention for high-quality segmentation results.\n\nSet model.backbone.pretrained_weights for Swin backbone weights.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-mask2former.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:** segmentation\n- **Formats:** coco_panoptic, coco\n- **Monitoring metric:** mIoU\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.train.type | train_datasets | coco_panoptic | No |\n| evaluate | dataset.val.type | eval_dataset | coco_panoptic | No |\n| evaluate | dataset.test.type | eval_dataset | coco_panoptic | No |\n| evaluate | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| evaluate | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| evaluate | dataset.train.instance_json | train_datasets | annotations.json | No |\n| evaluate | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| evaluate | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| evaluate | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| evaluate | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| evaluate | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| evaluate | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| evaluate | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| inference | dataset.train.type | train_datasets | coco_panoptic | No |\n| inference | dataset.val.type | eval_dataset | coco_panoptic | No |\n| inference | dataset.test.type | eval_dataset | coco_panoptic | No |\n| inference | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| inference | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| inference | dataset.train.instance_json | train_datasets | annotations.json | No |\n| inference | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| inference | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| inference | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| inference | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| inference | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| inference | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| inference | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.train.type | train_datasets | coco_panoptic | No |\n| quantize | dataset.val.type | eval_dataset | coco_panoptic | No |\n| quantize | dataset.test.type | eval_dataset | coco_panoptic | No |\n| quantize | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| quantize | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| quantize | dataset.train.instance_json | train_datasets | annotations.json | No |\n| quantize | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| quantize | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| quantize | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| quantize | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| quantize | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| quantize | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.quant_calibration_dataset.images_dir | train_datasets | images.tar.gz | No |\n| train | dataset.train.type | train_datasets | coco_panoptic | No |\n| train | dataset.val.type | eval_dataset | coco_panoptic | No |\n| train | dataset.test.type | eval_dataset | coco_panoptic | No |\n| train | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| train | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| train | dataset.train.instance_json | train_datasets | annotations.json | No |\n| train | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| train | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| train | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| train | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| train | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| train | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| train | dataset.test.img_dir | eval_dataset | images.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\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n```\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_gpus\": 1,\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"model.sem_seg_head.num_classes\": 133,\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.type\": \"coco_panoptic\",\n    \"dataset.val.type\": \"coco_panoptic\",\n    \"dataset.test.type\": \"coco_panoptic\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"model.sem_seg_head.num_classes\": 133,\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.type\": \"coco_panoptic\",\n    \"dataset.val.type\": \"coco_panoptic\",\n    \"dataset.test.type\": \"coco_panoptic\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**export:**\n```python\n{\n    \"export.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"export.onnx_file\": \"\u003Coutput ONNX path>\",\n    \"model.sem_seg_head.num_classes\": \"\u003Csame value used for train>\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"model.sem_seg_head.num_classes\": \"\u003Csame value used for train>\",\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train\u002Fexport artifact>\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.quant_calibration_dataset.images_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n}\n```\n## Eval Dataset\n\nOptional. Val data sources are part of the dataset config alongside train.\n\n## Important Parameters\n\n- **model.sem_seg_head.num_classes**: Number of segmentation classes. Default 200. Must match your annotation categories.\n- **model.backbone.swin.type**: Swin Transformer variant. Default tiny. Options include tiny, small, base, large.\n- **model.mode**: Segmentation mode. Default panoptic. Options: panoptic, instance, semantic.\n- **train.optim.lr**: Learning rate. Default 2e-4 (AdamW).\n- **dataset.train.batch_size**: Per-GPU batch size. Default 1. Mask2Former is memory-intensive due to per-pixel predictions.\n- **dataset.contiguous_id**: If true, set `model.sem_seg_head.num_classes`\n  to the number of label-map categories. If false, set\n  `model.sem_seg_head.num_classes` above the maximum raw category id and keep\n  the same setting for evaluate, inference, export, deploy, and quantize. The\n  COCO panoptic S3 sample has 133 categories with raw ids up to 200, so raw-id\n  validation uses `num_classes: 201`.\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.num_nodes` | Number of nodes | 1 |\n| `train.distributed_strategy` | `ddp` or `fsdp` | `ddp` |\n\n- Same DDP\u002FFSDP behavior as DINO (activation checkpoint aware)\n- FAN backbones auto-enable `sync_batchnorm`\n- `fsdp` forces FP16\n\n**Multi-node env vars** (set by orchestrator): `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT`, `NUM_GPU_PER_NODE`.\n\n## Export \u002F TRT Defaults\n\n- TRT data types: FP32, FP16 only — **INT8 is NOT supported**\n- The parent PyTorch `mask2former` CLI supports `train`, `evaluate`,\n  `inference`, `export`, and `quantize`; run TensorRT engine generation,\n  TensorRT inference, and TensorRT evaluation through `references\u002Ftao-deploy-mask2former.md`.\n  Export semantic ONNX (`model.mode: semantic`) when validating TensorRT\n  evaluation because the current deploy evaluator accepts semantic engines.\n- Keep export input dimensions compatible with the deploy templates. The\n  packaged default `export.input_width: 960` and `export.input_height: 544`\n  exports and builds a TensorRT engine successfully; shrinking export to tiny\n  validation-only sizes such as `128x128` can hit a PyTorch ONNX\n  `minus_one_pos != -1` shape-inference assertion before ONNX is produced.\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 4 GPU(s). 24GB+ (A100 recommended) VRAM per GPU. Mask2Former is memory-heavy. batch_size=1 is the default for good reason. Multi-GPU recommended for reasonable training speed.\n\n## Error Patterns\n\n**CUDA out of memory**: batch_size is already 1 by default. Reduce image resolution in augmentation config or use a smaller Swin variant.\n\n**Panoptic vs instance format mismatch**: Ensure you provide the correct annotation format matching model.mode setting.\n\n**Deploy schema error for top-level `dataset.type`**: TAO Deploy uses\n`dataset.val.type` and `dataset.test.type`. Do not put `dataset.type` at the\ntop level of Mask2Former deploy specs.\n\n**Export ONNX shape assertion at very small resolution**: If export fails with\n`minus_one_pos != -1` from PyTorch ONNX shape inference, restore the template\nexport dimensions (`960x544`) before retrying deploy validation. Keep training\nand evaluation image sizes small when needed for quick smoke tests, but do not\ncarry those tiny dimensions into export unless the target shape has been\nverified.\n\n**Quantize checkpoint load error**: Older PyTorch images can fail\ncheckpoint-based `mask2former quantize` because the runtime quantize script\npasses `experiment_spec` to `Mask2formerPlModule.load_from_checkpoint` instead\nof the required `cfg` argument. Images with the quantize fix support the default\n`torchao` checkpoint flow. ONNX quantization still requires\n`backend: modelopt.onnx`, `mode: static_ptq`, a fixed\n`dataset.test.target_size`, and an image that includes\n`modelopt.onnx.quantization`.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nInference mappings from TAO Core `mask2former.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.trt_engine` | `parent_model` | model file inferred from the parent job results folder |\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| gen_trt_engine | `encryption_key` | `key` | encryption key |\n| gen_trt_engine | `gen_trt_engine.onnx_file` | `parent_model` | model file inferred from the parent job results folder |\n| gen_trt_engine | `gen_trt_engine.trt_engine` | `create_engine_file` | output TensorRT engine path |\n| gen_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 | `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| train | `encryption_key` | `key` | encryption key |\n| train | `model.backbone.pretrained_weights` | `{'link': 'https:\u002F\u002Fgithub.com\u002FSwinTransformer\u002Fstorage\u002Freleases\u002Fdownload\u002Fv1.0.8\u002Fswin_tiny_patch4_window7_224_22k.pth', 'destination_path': '\u002Fptm\u002Fmask2former\u002Fswin_tiny_patch4_window7_224_22k\u002Fswin_tiny_patch4_window7_224_22k.pth'}` | {'link': 'https:\u002F\u002Fgithub.com\u002FSwinTransformer\u002Fstorage\u002Freleases\u002Fdownload\u002Fv1.0.8\u002Fswin_tiny_patch4_window7_224_22k.pth', 'destination_path': '\u002Fptm\u002Fmask2former\u002Fswin_tiny_patch4_window7_224_22k\u002Fswin_tiny_patch4_window7_224_22k.pth'} |\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\nWhen selecting a Mask2Former checkpoint outside the SDK resolver, match the\nintended epoch\u002Fstep artifact exactly, for example\n`model_epoch_000_step_00100.pth`. The `mask2former_model_latest.pth` symlink\nis valid only when latest is explicitly requested.\n\n## Deployment\n\n- [tao-deploy-mask2former](references\u002Ftao-deploy-mask2former.md)\n","# Mask2Former\n\nMask2Former for universal image segmentation (panoptic, instance, and semantic). Transformer-based with masked attention for high-quality segmentation results.\n\nSet model.backbone.pretrained_weights for Swin backbone weights.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-mask2former.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:** segmentation\n- **Formats:** coco_panoptic, coco\n- **Monitoring metric:** mIoU\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.train.type | train_datasets | coco_panoptic | No |\n| evaluate | dataset.val.type | eval_dataset | coco_panoptic | No |\n| evaluate | dataset.test.type | eval_dataset | coco_panoptic | No |\n| evaluate | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| evaluate | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| evaluate | dataset.train.instance_json | train_datasets | annotations.json | No |\n| evaluate | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| evaluate | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| evaluate | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| evaluate | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| evaluate | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| evaluate | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| evaluate | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| inference | dataset.train.type | train_datasets | coco_panoptic | No |\n| inference | dataset.val.type | eval_dataset | coco_panoptic | No |\n| inference | dataset.test.type | eval_dataset | coco_panoptic | No |\n| inference | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| inference | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| inference | dataset.train.instance_json | train_datasets | annotations.json | No |\n| inference | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| inference | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| inference | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| inference | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| inference | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| inference | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| inference | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.train.type | train_datasets | coco_panoptic | No |\n| quantize | dataset.val.type | eval_dataset | coco_panoptic | No |\n| quantize | dataset.test.type | eval_dataset | coco_panoptic | No |\n| quantize | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| quantize | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| quantize | dataset.train.instance_json | train_datasets | annotations.json | No |\n| quantize | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| quantize | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| quantize | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| quantize | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| quantize | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| quantize | dataset.test.img_dir | eval_dataset | images.tar.gz | No |\n| quantize | dataset.quant_calibration_dataset.images_dir | train_datasets | images.tar.gz | No |\n| train | dataset.train.type | train_datasets | coco_panoptic | No |\n| train | dataset.val.type | eval_dataset | coco_panoptic | No |\n| train | dataset.test.type | eval_dataset | coco_panoptic | No |\n| train | dataset.train.img_dir | train_datasets | images.tar.gz | No |\n| train | dataset.label_map | train_datasets | coco_panoptic: label_map_panoptic.json; *: label_map.json | No |\n| train | dataset.train.instance_json | train_datasets | annotations.json | No |\n| train | dataset.train.panoptic_json | train_datasets | annotations_panoptic.json | No |\n| train | dataset.train.panoptic_dir | train_datasets | images_panoptic.tar.gz | No |\n| train | dataset.val.img_dir | eval_dataset | images.tar.gz | No |\n| train | dataset.val.instance_json | eval_dataset | annotations.json | No |\n| train | dataset.val.panoptic_json | eval_dataset | annotations_panoptic.json | No |\n| train | dataset.val.panoptic_dir | eval_dataset | images_panoptic.tar.gz | No |\n| train | dataset.test.img_dir | eval_dataset | images.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\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n```\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_gpus\": 1,\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"model.sem_seg_head.num_classes\": 133,\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.type\": \"coco_panoptic\",\n    \"dataset.val.type\": \"coco_panoptic\",\n    \"dataset.test.type\": \"coco_panoptic\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"model.sem_seg_head.num_classes\": 133,\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.type\": \"coco_panoptic\",\n    \"dataset.val.type\": \"coco_panoptic\",\n    \"dataset.test.type\": \"coco_panoptic\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**export:**\n```python\n{\n    \"export.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"export.onnx_file\": \"\u003Coutput ONNX path>\",\n    \"model.sem_seg_head.num_classes\": \"\u003Csame value used for train>\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"model.sem_seg_head.num_classes\": \"\u003Csame value used for train>\",\n    \"dataset.contiguous_id\": True,\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train\u002Fexport artifact>\",\n    \"dataset.train.img_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Flabel_map_panoptic.json\",\n    \"dataset.train.instance_json\": f\"{S3_TRAIN}\u002Fannotations.json\",\n    \"dataset.train.panoptic_json\": f\"{S3_TRAIN}\u002Fannotations_panoptic.json\",\n    \"dataset.train.panoptic_dir\": f\"{S3_TRAIN}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.val.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.val.instance_json\": f\"{S3_EVAL}\u002Fannotations.json\",\n    \"dataset.val.panoptic_json\": f\"{S3_EVAL}\u002Fannotations_panoptic.json\",\n    \"dataset.val.panoptic_dir\": f\"{S3_EVAL}\u002Fimages_panoptic.tar.gz\",\n    \"dataset.test.img_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\",\n    \"dataset.quant_calibration_dataset.images_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\",\n}\n```\n## Eval Dataset\n\nOptional. Val data sources are part of the dataset config alongside train.\n\n## Important Parameters\n\n- **model.sem_seg_head.num_classes**: Number of segmentation classes. Default 200. Must match your annotation categories.\n- **model.backbone.swin.type**: Swin Transformer variant. Default tiny. Options include tiny, small, base, large.\n- **model.mode**: Segmentation mode. Default panoptic. Options: panoptic, instance, semantic.\n- **train.optim.lr**: Learning rate. Default 2e-4 (AdamW).\n- **dataset.train.batch_size**: Per-GPU batch size. Default 1. Mask2Former is memory-intensive due to per-pixel predictions.\n- **dataset.contiguous_id**: If true, set `model.sem_seg_head.num_classes`\n  to the number of label-map categories. If false, set\n  `model.sem_seg_head.num_classes` above the maximum raw category id and keep\n  the same setting for evaluate, inference, export, deploy, and quantize. The\n  COCO panoptic S3 sample has 133 categories with raw ids up to 200, so raw-id\n  validation uses `num_classes: 201`.\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.num_nodes` | Number of nodes | 1 |\n| `train.distributed_strategy` | `ddp` or `fsdp` | `ddp` |\n\n- Same DDP\u002FFSDP behavior as DINO (activation checkpoint aware)\n- FAN backbones auto-enable `sync_batchnorm`\n- `fsdp` forces FP16\n\n**Multi-node env vars** (set by orchestrator): `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT`, `NUM_GPU_PER_NODE`.\n\n## Export \u002F TRT Defaults\n\n- TRT data types: FP32, FP16 only — **INT8 is NOT supported**\n- The parent PyTorch `mask2former` CLI supports `train`, `evaluate`,\n  `inference`, `export`, and `quantize`; run TensorRT engine generation,\n  TensorRT inference, and TensorRT evaluation through `references\u002Ftao-deploy-mask2former.md`.\n  Export semantic ONNX (`model.mode: semantic`) when validating TensorRT\n  evaluation because the current deploy evaluator accepts semantic engines.\n- Keep export input dimensions compatible with the deploy templates. The\n  packaged default `export.input_width: 960` and `export.input_height: 544`\n  exports and builds a TensorRT engine successfully; shrinking export to tiny\n  validation-only sizes such as `128x128` can hit a PyTorch ONNX\n  `minus_one_pos != -1` shape-inference assertion before ONNX is produced.\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 4 GPU(s). 24GB+ (A100 recommended) VRAM per GPU. Mask2Former is memory-heavy. batch_size=1 is the default for good reason. Multi-GPU recommended for reasonable training speed.\n\n## Error Patterns\n\n**CUDA out of memory**: batch_size is already 1 by default. Reduce image resolution in augmentation config or use a smaller Swin variant.\n\n**Panoptic vs instance format mismatch**: Ensure you provide the correct annotation format matching model.mode setting.\n\n**Deploy schema error for top-level `dataset.type`**: TAO Deploy uses\n`dataset.val.type` and `dataset.test.type`. Do not put `dataset.type` at the\ntop level of Mask2Former deploy specs.\n\n**Export ONNX shape assertion at very small resolution**: If export fails with\n`minus_one_pos != -1` from PyTorch ONNX shape inference, restore the template\nexport dimensions (`960x544`) before retrying deploy validation. Keep training\nand evaluation image sizes small when needed for quick smoke tests, but do not\ncarry those tiny dimensions into export unless the target shape has been\nverified.\n\n**Quantize checkpoint load error**: Older PyTorch images can fail\ncheckpoint-based `mask2former quantize` because the runtime quantize script\npasses `experiment_spec` to `Mask2formerPlModule.load_from_checkpoint` instead\nof the required `cfg` argument. Images with the quantize fix support the default\n`torchao` checkpoint flow. ONNX quantization still requires\n`backend: modelopt.onnx`, `mode: static_ptq`, a fixed\n`dataset.test.target_size`, and an image that includes\n`modelopt.onnx.quantization`.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nInference mappings from TAO Core `mask2former.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.trt_engine` | `parent_model` | model file inferred from the parent job results folder |\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| gen_trt_engine | `encryption_key` | `key` | encryption key |\n| gen_trt_engine | `gen_trt_engine.onnx_file` | `parent_model` | model file inferred from the parent job results folder |\n| gen_trt_engine | `gen_trt_engine.trt_engine` | `create_engine_file` | output TensorRT engine path |\n| gen_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 | `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| train | `encryption_key` | `key` | encryption key |\n| train | `model.backbone.pretrained_weights` | `{'link': 'https:\u002F\u002Fgithub.com\u002FSwinTransformer\u002Fstorage\u002Freleases\u002Fdownload\u002Fv1.0.8\u002Fswin_tiny_patch4_window7_224_22k.pth', 'destination_path': '\u002Fptm\u002Fmask2former\u002Fswin_tiny_patch4_window7_224_22k\u002Fswin_tiny_patch4_window7_224_22k.pth'}` | {'link': 'https:\u002F\u002Fgithub.com\u002FSwinTransformer\u002Fstorage\u002Freleases\u002Fdownload\u002Fv1.0.8\u002Fswin_tiny_patch4_window7_224_22k.pth', 'destination_path': '\u002Fptm\u002Fmask2former\u002Fswin_tiny_patch4_window7_224_22k\u002Fswin_tiny_patch4_window7_224_22k.pth'} |\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\nWhen selecting a Mask2Former checkpoint outside the SDK resolver, match the\nintended epoch\u002Fstep artifact exactly, for example\n`model_epoch_000_step_00100.pth`. The `mask2former_model_latest.pth` symlink\nis valid only when latest is explicitly requested.\n\n## Deployment\n\n- [tao-deploy-mask2former](references\u002Ftao-deploy-mask2former.md)\n",{"name":5,"tags":24,"license":15,"metadata":26,"description":7,"allowed-tools":29,"compatibility":30},[25],"segmentation",{"author":27,"version":28},"NVIDIA Corporation","0.1.0","Read Bash","Requires docker + nvidia-container-toolkit.","main","1c523aba5cdf867a79f078847075492d7778092a","public\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former",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":42,"prompt":52,"zipAvailable":53},[43,46,49],{"tool":44,"command":45},"npx","npx skills add themusebook\u002Fskills --skill tao-train-mask2former",{"tool":47,"command":48},"bunx","bunx skills add themusebook\u002Fskills --skill tao-train-mask2former",{"tool":50,"command":51},"pnpm","pnpm dlx skills add themusebook\u002Fskills --skill tao-train-mask2former","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002FSKILL.md",true,[55,61,65,69,73,77,80,84,88,92,96,100,103,107,111,115],{"path":56,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":59,"sourceHtmlUrl":60},"BENCHMARK.md","REFERENCE","参考文档",4085,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002FBENCHMARK.md",{"path":62,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":63,"sourceHtmlUrl":64},"metadata.json",644,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Fmetadata.json",{"path":66,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fskill_info.yaml",9771,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fskill_info.yaml",{"path":70,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fspec_template_deploy_evaluate.yaml",709,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_deploy_evaluate.yaml",{"path":74,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fspec_template_deploy_gen_trt_engine.yaml",247,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_deploy_gen_trt_engine.yaml",{"path":78,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":71,"sourceHtmlUrl":79},"references\u002Fspec_template_deploy_inference.yaml","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_deploy_inference.yaml",{"path":81,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":82,"sourceHtmlUrl":83},"references\u002Fspec_template_evaluate.yaml",3756,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_evaluate.yaml",{"path":85,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":86,"sourceHtmlUrl":87},"references\u002Fspec_template_export.yaml",3831,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_export.yaml",{"path":89,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":90,"sourceHtmlUrl":91},"references\u002Fspec_template_gen_trt_engine.yaml",3912,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_gen_trt_engine.yaml",{"path":93,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":94,"sourceHtmlUrl":95},"references\u002Fspec_template_inference.yaml",3757,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_inference.yaml",{"path":97,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":98,"sourceHtmlUrl":99},"references\u002Fspec_template_quantize.yaml",3630,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_quantize.yaml",{"path":101,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":98,"sourceHtmlUrl":102},"references\u002Fspec_template_train.yaml","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Fspec_template_train.yaml",{"path":104,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":105,"sourceHtmlUrl":106},"references\u002Ftao-deploy-mask2former.md",5887,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Ftao-deploy-mask2former.md",{"path":108,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":109,"sourceHtmlUrl":110},"references\u002Ftao-deploy-mask2former.skill_info.yaml",3729,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Freferences\u002Ftao-deploy-mask2former.skill_info.yaml",{"path":112,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":113,"sourceHtmlUrl":114},"skill-card.md",3898,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Fskill-card.md",{"path":116,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":117,"sourceHtmlUrl":118},"verification.json",3830,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Ftao-train-mask2former\u002Fverification.json",[120,129,136,143,150,157],{"skillNo":121,"name":122,"nameZh":6,"description":123,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":125,"sourceHtmlUrl":126,"categories":127,"keywords":128,"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":130,"name":131,"nameZh":6,"description":132,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":135,"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":137,"name":138,"nameZh":6,"description":139,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":6,"sourceHtmlUrl":140,"categories":141,"keywords":142,"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":144,"name":145,"nameZh":6,"description":146,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":6,"sourceHtmlUrl":147,"categories":148,"keywords":149,"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":151,"name":152,"nameZh":6,"description":153,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":6,"sourceHtmlUrl":154,"categories":155,"keywords":156,"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":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":124,"licenseFileUrl":6,"sourceHtmlUrl":161,"categories":162,"keywords":163,"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",[],[]]