[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78DC62X9MY2FD8NYXF563RH":3,"skill-same:01M78DC62X9MY2FD8NYXF563RH":150},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":32,"commitSha":33,"skillDirectory":34,"forksCount":14,"repositoryHtmlUrl":35,"repositoryHomepageUrl":36,"ownerHtmlUrl":37,"ownerBlogUrl":38,"ownerDescription":39,"updatedAt":40,"install":41,"resources":54},"01M78DC62X9MY2FD8NYXF563RH","tao-train-grounding-dino",null,"Grounding DINO for open-set object detection. Combines DINO-style detection with a BERT text encoder for language-guided detection — detects objects described by text prompts without a fixed class vocabulary. Use when training, evaluating, exporting, quantizing, or running inference for a TAO Grounding DINO model. Trigger phrases include \"train Grounding DINO\", \"open-vocabulary detection\", \"text-prompted detector\", \"language-guided object detection\".","Solizardking\u002Fskillhub-main","Solizardking","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F108096810?v=4","01M786KDB10AYV042N0F9BD5B5","Clawd","01M786NYYXSZK1D05WM8JQPC48",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-train-grounding-dino\ndescription: Grounding DINO for open-set object detection. Combines DINO-style detection with a BERT text encoder for\n  language-guided detection — detects objects described by text prompts without a fixed class vocabulary. Use when training,\n  evaluating, exporting, quantizing, or running inference for a TAO Grounding DINO model. Trigger phrases include \"train\n  Grounding DINO\", \"open-vocabulary detection\", \"text-prompted detector\", \"language-guided object detection\".\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- object\n- detection\n---\n\n# Grounding DINO\n\nGrounding DINO for open-set object detection. Combines DINO-style detection with BERT text encoder for language-guided detection. Detects objects described by text prompts without fixed class vocabulary.\n\nSet train.pretrained_model_path for full Grounding DINO weights or model.pretrained_backbone_path for backbone-only.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-grounding-dino.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:** object_detection\n- **Formats:** odvg, coco, raw\n- **Monitoring metric:** val_mAP50\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.test_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| inference | dataset.infer_data_sources.image_dir | inference_dataset | images.tar.gz | Yes |\n| inference | dataset.infer_data_sources.captions | workflow prompts | prompt list | Yes |\n| quantize | dataset.train_data_sources | train_datasets | image_dir: images.tar.gz, json_file: annotations_odvg.jsonl, label_map: annotations_odvg_labelmap.json | Yes |\n| quantize | dataset.val_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| quantize | dataset.quant_calibration_data_sources | calibration\u002Feval dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| train | dataset.train_data_sources | train_datasets | image_dir: images.tar.gz, json_file: annotations_odvg.jsonl, label_map: annotations_odvg_labelmap.json | Yes |\n| train | dataset.val_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n\nThe runner may source image archives as `images.tar.gz`, but direct local\nDocker TAO CLI specs must point `image_dir` to an extracted image directory.\nSkill metadata marks these archive-backed image sources with\n`runtime: extracted_folder` so a fresh runner can unpack the archive before\nlaunching TAO.\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_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.train_data_sources\": [{\"image_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_TRAIN}\u002Fannotations_odvg.jsonl\", \"label_map\": f\"{S3_TRAIN}\u002Fannotations_odvg_labelmap.json\"}],\n    \"dataset.val_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n\n**deploy\u002Fgen_trt_engine (use `references\u002Ftao-deploy-grounding-dino.md`):**\n```python\n{\n    \"gen_trt_engine.onnx_file\": \"\u003Cexported_onnx_uri>\",\n    \"gen_trt_engine.trt_engine\": \"\u003Coutput_engine_path>\",\n    \"gen_trt_engine.tensorrt.data_type\": \"FP16\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.infer_data_sources.image_dir\": [f\"{S3_EVAL}\u002Fimages.tar.gz\"],\n    \"dataset.infer_data_sources.captions\": [\n        \"fire extinguisher\",\n        \"cone\",\n        \"cart\",\n        \"forklift\"\n    ],\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.test_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train checkpoint or exported ONNX model>\",\n    \"dataset.train_data_sources\": [{\"image_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_TRAIN}\u002Fannotations_odvg.jsonl\", \"label_map\": f\"{S3_TRAIN}\u002Fannotations_odvg_labelmap.json\"}],\n    \"dataset.val_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n    \"dataset.quant_calibration_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n## Eval Dataset\n\nOptional. Validation uses COCO-format annotations for mAP even though training can use ODVG format.\n\n## Important Parameters\n\n- **model.backbone**: Default swin_tiny_224_1k. Also supports resnet_50 and other Swin variants. Swin generally performs better for grounding tasks.\n- **model.text_encoder_type**: BERT model for text encoding. Default bert-base-uncased. max_text_len defaults to 256.\n- **model.max_text_len**: Keep this aligned with the dataset label\u002Ftoken\n  position maps. Do not shrink it for smoke tests unless the corresponding\n  label maps are regenerated with the same length; otherwise validation can\n  fail with a matrix shape mismatch between token probabilities and position\n  maps.\n- **train.optim.lr**: Learning rate. Default 2e-4. lr_backbone 2e-5. Supports bf16 precision in addition to fp16\u002Ffp32.\n- **dataset.max_labels**: Maximum labels per image during training. Default 50. Increase for dense annotation datasets.\n- **model.num_queries**: Object queries. Default 900 (higher than DINO's 300) due to open-vocabulary nature.\n- **model.num_queries \u002F model.num_select**: Keep `num_queries` high enough\n  for the number of matched ODVG targets in a batch. Very small smoke values\n  such as 20 can fail during Hungarian target indexing on dense images; use at\n  least 100 for minimal Grounding DINO smoke runs unless the dataset is known\n  to have fewer objects per image.\n- **train.optim.lr_steps**: MultiStep LR schedule. Default [10].\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\nSame DDP\u002FFSDP behavior as DINO. Multi-node requires `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT` env vars set by orchestrator.\n\n## Export \u002F TRT Defaults\n\n- Export input: 960x544 (larger than other OD models), opset 17. Keep\n  Grounding-DINO export specs at the template export resolution for smoke tests;\n  reducing export to very small image sizes such as 128x128 can trigger a\n  PyTorch ONNX shape-inference assertion in the contrastive text head during\n  `torch.onnx.export`.\n- The parent PyTorch `grounding_dino` CLI supports `train`, `evaluate`,\n  `inference`, `export`, and `quantize`. Run TensorRT engine generation,\n  TensorRT inference, and TensorRT evaluation through `references\u002Ftao-deploy-grounding-dino.md`.\n- TRT data types: FP32, FP16 only — **INT8 is NOT supported**\n- TRT workspace: 8192 MB (8x larger than other OD models)\n- TRT max_batch_size: 4\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 4 GPU(s). 24GB+ (A100 recommended) VRAM per GPU. Grounding DINO is heavier than standard DINO due to the text encoder (BERT). 24GB+ GPU memory recommended. Reduce batch_size for 16GB GPUs.\n\n## Error Patterns\n\n**CUDA out of memory**: Reduce batch_size (4 -> 2 -> 1). The BERT text encoder adds significant memory overhead on top of the vision backbone.\n\n**Val annotation category IDs**: Validation annotations should have category IDs starting from 0 for correct loss computation. Use annotation format conversion if needed.\n\n**Text encoder loading error**: Ensure the container has access to download bert-base-uncased weights or provide a local path.\n\n**Quantize with a PyTorch checkpoint fails in TAO Toolkit 7.0.0-rc-226**:\nThe container's Grounding-DINO quantize script passes `cap_lists=None` when\nloading a checkpoint, which fails in `post_process.py`. ONNX quantization uses\nthe exported ONNX artifact and COCO calibration data, but the default rc-226\nPyTorch image also lacks the `modelopt.onnx.quantization` module. Treat this as\nan image\u002FSDK blocker, not a checkpoint resolver issue.\n\n**mat1 and mat2 shapes cannot be multiplied in `post_process.py`**: The text\ntoken length and label position maps are inconsistent, commonly because\n`model.max_text_len` was overridden below the default 256 while the dataset\nlabel maps still use 256-length position maps. Restore `model.max_text_len` or\nregenerate the label maps with the same length.\n\n**index is out of bounds for dimension 0 in `criterion.py`**: `model.num_queries`\nis too small for the matched ODVG targets in the current batch. Increase\n`model.num_queries` and keep `model.num_select` compatible with it.\n\n**NotADirectoryError with `images.tar.gz\u002F\u003Cimage>.jpg`**: The direct TAO CLI is\ntrying to traverse an archive path as a directory. Extract the archive and set\nthe relevant `image_dir` field to the extracted image folder; archive-backed\nskill data sources use `runtime: extracted_folder` for this reason.\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 `grounding_dino.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| 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.pretrained_backbone_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.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\nWhen selecting a Grounding-DINO checkpoint outside the SDK resolver, match the\nintended epoch\u002Fstep artifact exactly, for example\n`model_epoch_000_step_00046.pth`. The `gdino_model_latest.pth` symlink is valid\nonly when latest is explicitly requested. Carry structural model settings such\nas `model.backbone`, `model.num_queries`, `model.num_select`,\n`model.num_feature_levels`, `model.max_text_len`, and export input resolution\nforward into evaluate, inference, export, and deploy specs so checkpoint and\nengine shapes match.\n\n## Deployment\n\n- [tao-deploy-grounding-dino](references\u002Ftao-deploy-grounding-dino.md)\n","# Grounding DINO\n\nGrounding DINO for open-set object detection. Combines DINO-style detection with BERT text encoder for language-guided detection. Detects objects described by text prompts without fixed class vocabulary.\n\nSet train.pretrained_model_path for full Grounding DINO weights or model.pretrained_backbone_path for backbone-only.\n\nFor TAO Deploy TensorRT actions (`gen_trt_engine`, TensorRT `evaluate`, and TensorRT `inference`), read `references\u002Ftao-deploy-grounding-dino.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:** object_detection\n- **Formats:** odvg, coco, raw\n- **Monitoring metric:** val_mAP50\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.test_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| inference | dataset.infer_data_sources.image_dir | inference_dataset | images.tar.gz | Yes |\n| inference | dataset.infer_data_sources.captions | workflow prompts | prompt list | Yes |\n| quantize | dataset.train_data_sources | train_datasets | image_dir: images.tar.gz, json_file: annotations_odvg.jsonl, label_map: annotations_odvg_labelmap.json | Yes |\n| quantize | dataset.val_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| quantize | dataset.quant_calibration_data_sources | calibration\u002Feval dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n| train | dataset.train_data_sources | train_datasets | image_dir: images.tar.gz, json_file: annotations_odvg.jsonl, label_map: annotations_odvg_labelmap.json | Yes |\n| train | dataset.val_data_sources | eval_dataset | image_dir: images.tar.gz, json_file: annotations.json | No |\n\nThe runner may source image archives as `images.tar.gz`, but direct local\nDocker TAO CLI specs must point `image_dir` to an extracted image directory.\nSkill metadata marks these archive-backed image sources with\n`runtime: extracted_folder` so a fresh runner can unpack the archive before\nlaunching TAO.\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_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.train_data_sources\": [{\"image_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_TRAIN}\u002Fannotations_odvg.jsonl\", \"label_map\": f\"{S3_TRAIN}\u002Fannotations_odvg_labelmap.json\"}],\n    \"dataset.val_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n\n**deploy\u002Fgen_trt_engine (use `references\u002Ftao-deploy-grounding-dino.md`):**\n```python\n{\n    \"gen_trt_engine.onnx_file\": \"\u003Cexported_onnx_uri>\",\n    \"gen_trt_engine.trt_engine\": \"\u003Coutput_engine_path>\",\n    \"gen_trt_engine.tensorrt.data_type\": \"FP16\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.infer_data_sources.image_dir\": [f\"{S3_EVAL}\u002Fimages.tar.gz\"],\n    \"dataset.infer_data_sources.captions\": [\n        \"fire extinguisher\",\n        \"cone\",\n        \"cart\",\n        \"forklift\"\n    ],\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.test_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n\n**quantize (mandatory data sources):**\n```python\n{\n    \"quantize.model_path\": \"\u003Cselected train checkpoint or exported ONNX model>\",\n    \"dataset.train_data_sources\": [{\"image_dir\": f\"{S3_TRAIN}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_TRAIN}\u002Fannotations_odvg.jsonl\", \"label_map\": f\"{S3_TRAIN}\u002Fannotations_odvg_labelmap.json\"}],\n    \"dataset.val_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n    \"dataset.quant_calibration_data_sources\": {\"image_dir\": f\"{S3_EVAL}\u002Fimages.tar.gz\", \"json_file\": f\"{S3_EVAL}\u002Fannotations.json\"},\n}\n```\n## Eval Dataset\n\nOptional. Validation uses COCO-format annotations for mAP even though training can use ODVG format.\n\n## Important Parameters\n\n- **model.backbone**: Default swin_tiny_224_1k. Also supports resnet_50 and other Swin variants. Swin generally performs better for grounding tasks.\n- **model.text_encoder_type**: BERT model for text encoding. Default bert-base-uncased. max_text_len defaults to 256.\n- **model.max_text_len**: Keep this aligned with the dataset label\u002Ftoken\n  position maps. Do not shrink it for smoke tests unless the corresponding\n  label maps are regenerated with the same length; otherwise validation can\n  fail with a matrix shape mismatch between token probabilities and position\n  maps.\n- **train.optim.lr**: Learning rate. Default 2e-4. lr_backbone 2e-5. Supports bf16 precision in addition to fp16\u002Ffp32.\n- **dataset.max_labels**: Maximum labels per image during training. Default 50. Increase for dense annotation datasets.\n- **model.num_queries**: Object queries. Default 900 (higher than DINO's 300) due to open-vocabulary nature.\n- **model.num_queries \u002F model.num_select**: Keep `num_queries` high enough\n  for the number of matched ODVG targets in a batch. Very small smoke values\n  such as 20 can fail during Hungarian target indexing on dense images; use at\n  least 100 for minimal Grounding DINO smoke runs unless the dataset is known\n  to have fewer objects per image.\n- **train.optim.lr_steps**: MultiStep LR schedule. Default [10].\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\nSame DDP\u002FFSDP behavior as DINO. Multi-node requires `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT` env vars set by orchestrator.\n\n## Export \u002F TRT Defaults\n\n- Export input: 960x544 (larger than other OD models), opset 17. Keep\n  Grounding-DINO export specs at the template export resolution for smoke tests;\n  reducing export to very small image sizes such as 128x128 can trigger a\n  PyTorch ONNX shape-inference assertion in the contrastive text head during\n  `torch.onnx.export`.\n- The parent PyTorch `grounding_dino` CLI supports `train`, `evaluate`,\n  `inference`, `export`, and `quantize`. Run TensorRT engine generation,\n  TensorRT inference, and TensorRT evaluation through `references\u002Ftao-deploy-grounding-dino.md`.\n- TRT data types: FP32, FP16 only — **INT8 is NOT supported**\n- TRT workspace: 8192 MB (8x larger than other OD models)\n- TRT max_batch_size: 4\n\n## Hardware\n\nMinimum 1 GPU(s), recommended 4 GPU(s). 24GB+ (A100 recommended) VRAM per GPU. Grounding DINO is heavier than standard DINO due to the text encoder (BERT). 24GB+ GPU memory recommended. Reduce batch_size for 16GB GPUs.\n\n## Error Patterns\n\n**CUDA out of memory**: Reduce batch_size (4 -> 2 -> 1). The BERT text encoder adds significant memory overhead on top of the vision backbone.\n\n**Val annotation category IDs**: Validation annotations should have category IDs starting from 0 for correct loss computation. Use annotation format conversion if needed.\n\n**Text encoder loading error**: Ensure the container has access to download bert-base-uncased weights or provide a local path.\n\n**Quantize with a PyTorch checkpoint fails in TAO Toolkit 7.0.0-rc-226**:\nThe container's Grounding-DINO quantize script passes `cap_lists=None` when\nloading a checkpoint, which fails in `post_process.py`. ONNX quantization uses\nthe exported ONNX artifact and COCO calibration data, but the default rc-226\nPyTorch image also lacks the `modelopt.onnx.quantization` module. Treat this as\nan image\u002FSDK blocker, not a checkpoint resolver issue.\n\n**mat1 and mat2 shapes cannot be multiplied in `post_process.py`**: The text\ntoken length and label position maps are inconsistent, commonly because\n`model.max_text_len` was overridden below the default 256 while the dataset\nlabel maps still use 256-length position maps. Restore `model.max_text_len` or\nregenerate the label maps with the same length.\n\n**index is out of bounds for dimension 0 in `criterion.py`**: `model.num_queries`\nis too small for the matched ODVG targets in the current batch. Increase\n`model.num_queries` and keep `model.num_select` compatible with it.\n\n**NotADirectoryError with `images.tar.gz\u002F\u003Cimage>.jpg`**: The direct TAO CLI is\ntrying to traverse an archive path as a directory. Extract the archive and set\nthe relevant `image_dir` field to the extracted image folder; archive-backed\nskill data sources use `runtime: extracted_folder` for this reason.\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 `grounding_dino.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| 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.pretrained_backbone_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.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\nWhen selecting a Grounding-DINO checkpoint outside the SDK resolver, match the\nintended epoch\u002Fstep artifact exactly, for example\n`model_epoch_000_step_00046.pth`. The `gdino_model_latest.pth` symlink is valid\nonly when latest is explicitly requested. Carry structural model settings such\nas `model.backbone`, `model.num_queries`, `model.num_select`,\n`model.num_feature_levels`, `model.max_text_len`, and export input resolution\nforward into evaluate, inference, export, and deploy specs so checkpoint and\nengine shapes match.\n\n## Deployment\n\n- [tao-deploy-grounding-dino](references\u002Ftao-deploy-grounding-dino.md)\n",{"name":5,"tags":24,"license":15,"metadata":27,"description":7,"allowed-tools":30,"compatibility":31},[25,26],"object","detection",{"author":28,"version":29},"NVIDIA Corporation","0.1.0","Read Bash","Requires docker + nvidia-container-toolkit.","main","be3de0244dc1a57e323d8160787ebb40fb64930e","skills\u002Fnvidia\u002Ftao-train-grounding-dino","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main","https:\u002F\u002Fskills.x402.wtf","https:\u002F\u002Fgithub.com\u002FSolizardking","https:\u002F\u002Fmusebook.trade","Got a mile high and never came down","2026-10-08 23:20:13",{"source":8,"skillName":5,"commands":42,"prompt":52,"zipAvailable":53},[43,46,49],{"tool":44,"command":45},"npx","npx skills add Solizardking\u002Fskillhub-main --skill tao-train-grounding-dino",{"tool":47,"command":48},"bunx","bunx skills add Solizardking\u002Fskillhub-main --skill tao-train-grounding-dino",{"tool":50,"command":51},"pnpm","pnpm dlx skills add Solizardking\u002Fskillhub-main --skill tao-train-grounding-dino","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002FSKILL.md",true,[55,61,65,69,73,77,81,85,89,93,97,101,104,108,112,116,120,124,128,132,136,140,144],{"path":56,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":59,"sourceHtmlUrl":60},"BENCHMARK.md","REFERENCE","参考文档",4103,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002FBENCHMARK.md",{"path":62,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":63,"sourceHtmlUrl":64},"evals\u002Fevals.json",805,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fevals\u002Fevals.json",{"path":66,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fskill_info.yaml",6038,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fskill_info.yaml",{"path":70,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fspec_template_deploy_evaluate.yaml",375,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_deploy_evaluate.yaml",{"path":74,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fspec_template_deploy_gen_trt_engine.yaml",343,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_deploy_gen_trt_engine.yaml",{"path":78,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fspec_template_deploy_inference.yaml",417,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_deploy_inference.yaml",{"path":82,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":83,"sourceHtmlUrl":84},"references\u002Fspec_template_evaluate.yaml",3321,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_evaluate.yaml",{"path":86,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":87,"sourceHtmlUrl":88},"references\u002Fspec_template_export.yaml",3418,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_export.yaml",{"path":90,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":91,"sourceHtmlUrl":92},"references\u002Fspec_template_gen_trt_engine.yaml",3455,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_gen_trt_engine.yaml",{"path":94,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":95,"sourceHtmlUrl":96},"references\u002Fspec_template_inference.yaml",3401,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_inference.yaml",{"path":98,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":99,"sourceHtmlUrl":100},"references\u002Fspec_template_quantize.yaml",3173,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_quantize.yaml",{"path":102,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":99,"sourceHtmlUrl":103},"references\u002Fspec_template_train.yaml","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Fspec_template_train.yaml",{"path":105,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":106,"sourceHtmlUrl":107},"references\u002Ftao-deploy-grounding-dino.md",5389,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Ftao-deploy-grounding-dino.md",{"path":109,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":110,"sourceHtmlUrl":111},"references\u002Ftao-deploy-grounding-dino.skill_info.yaml",2930,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Freferences\u002Ftao-deploy-grounding-dino.skill_info.yaml",{"path":113,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":114,"sourceHtmlUrl":115},"schemas\u002Fevaluate.schema.json",57583,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Fevaluate.schema.json",{"path":117,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":118,"sourceHtmlUrl":119},"schemas\u002Fexport.schema.json",58148,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Fexport.schema.json",{"path":121,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":122,"sourceHtmlUrl":123},"schemas\u002Fgen_trt_engine.schema.json",59913,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Fgen_trt_engine.schema.json",{"path":125,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":126,"sourceHtmlUrl":127},"schemas\u002Finference.schema.json",58523,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Finference.schema.json",{"path":129,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":130,"sourceHtmlUrl":131},"schemas\u002Fmanifest.json",16353,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Fmanifest.json",{"path":133,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":134,"sourceHtmlUrl":135},"schemas\u002Fquantize.schema.json",54213,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Fquantize.schema.json",{"path":137,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":138,"sourceHtmlUrl":139},"schemas\u002Ftrain.schema.json",54207,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fschemas\u002Ftrain.schema.json",{"path":141,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":142,"sourceHtmlUrl":143},"skill-card.md",3940,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fskill-card.md",{"path":145,"resourceType":146,"resourceTypeDesc":147,"sizeBytes":148,"sourceHtmlUrl":149},"skill.oms.sig","OTHER","其他",9381,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Ftao-train-grounding-dino\u002Fskill.oms.sig",[151,159,166,173,180,188],{"skillNo":152,"name":153,"nameZh":6,"description":154,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":6,"sourceHtmlUrl":156,"categories":157,"keywords":158,"publishedAt":19},"01M78E0TGCAMNPS207BA3G07S7","zkrouter","Self-hosted OpenAI-compatible LLM router — smart tier-based routing, ZK-stamped routing decisions, mode overrides, free OpenRouter at install via the Birth bot. Save 60-80% on inference costs by routing to the cheapest capable model across 12+ OpenRouter models.","MIT","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzkrouter\u002FSKILL.md",[],[],{"skillNo":160,"name":161,"nameZh":6,"description":162,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":6,"sourceHtmlUrl":163,"categories":164,"keywords":165,"publishedAt":19},"01M78E0R1Z8BDT6EDTHY17MTNB","zk","For custom ZK Solana programs and privacy-preserving applications to prevent double spending. Guide to integrate rent-free nullifier PDAs for double-spend prevention.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzk\u002FSKILL.md",[],[],{"skillNo":167,"name":168,"nameZh":6,"description":169,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":6,"sourceHtmlUrl":170,"categories":171,"keywords":172,"publishedAt":19},"01M78E0P3ZTZ1YWF8MD5MM23QE","zk-omni-messaging","Zero-knowledge omnichain messaging between Robinhood Chain and Solana using CheshireZkOmniMessenger (msgType 4) with nullifier anti-replay and the zk-omni-relayer service.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzk-omni-messaging\u002FSKILL.md",[],[],{"skillNo":174,"name":175,"nameZh":6,"description":176,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":6,"sourceHtmlUrl":177,"categories":178,"keywords":179,"publishedAt":19},"01M78E0JXZ2PNB1896VXAGDBWD","zerion-wallet-data","Use when the user asks to check a wallet's portfolio, token balances, DeFi positions, PnL, transaction history, or NFTs. Provides interpreted on-chain data across 40+ EVM chains and Solana via x402 pay-per-request or API key — no raw RPC decoding needed.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fzerion-wallet-data\u002FSKILL.md",[],[],{"skillNo":181,"name":182,"nameZh":6,"description":183,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":184,"sourceHtmlUrl":185,"categories":186,"keywords":187,"publishedAt":19},"01M78E0G14FH9GRRTQYMKD5E59","youtube-clipper","YouTube 视频智能剪辑工具。下载视频和字幕，AI 分析生成精细章节（几分钟级别）， 用户选择片段后自动剪辑、翻译字幕为中英双语、烧录字幕到视频，并生成总结文案。 使用场景：当用户需要剪辑 YouTube 视频、生成短视频片段、制作双语字幕版本时。 关键词：视频剪辑、YouTube、字幕翻译、双语字幕、视频下载、clip video","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyoutube-clipper\u002FSKILL.md",[],[],{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":155,"licenseFileUrl":6,"sourceHtmlUrl":192,"categories":193,"keywords":194,"publishedAt":19},"01M78E0C2ZYA99CFFDNB7ZMK5S","yield-optimization","Multi-chain yield optimization via yield.xyz (StakeKit). Discover 2,988+ yield opportunities across 75+ blockchains, build deposit\u002Fwithdrawal transactions with shell scripts, and sign them with a MoonPay wallet. Use when the user wants to earn yield, find the best lending or staking rate, enter\u002Fexit a position, or check portfolio balances.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Fskillhub-main\u002Fblob\u002Fmain\u002Fskills\u002Fyield-optimization\u002FSKILL.md",[],[]]