[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7A9P17CTPR4E2Z5NXC63BXT":3,"skill-same:01M7A9P17CTPR4E2Z5NXC63BXT":102},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":32,"commitSha":33,"skillDirectory":34,"forksCount":35,"repositoryHtmlUrl":36,"repositoryHomepageUrl":37,"ownerHtmlUrl":38,"ownerBlogUrl":39,"ownerDescription":40,"updatedAt":41,"install":42,"resources":55},"01M7A9P17CTPR4E2Z5NXC63BXT","nv-segment-ct-finetune",null,"Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.","themusebook\u002Fskills","themusebook","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F349927247?v=4","01M7A5C1RV5S3V5FF6ZACVDNRN","musebook","01M7A5C1SFC3KRZAXNN60XJ7KM",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: nv-segment-ct-finetune\ndescription: Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.\nlicense: Apache-2.0\nallowed-tools: Bash\nmetadata:\n  author: NVIDIA MedTech Team\n  tags:\n    - MedTech\n    - CT\n    - finetuning\n    - segmentation\n---\n\n# NV-Segment-CT Finetune\n\n## Purpose\n\n- Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.\n- Wraps the upstream MONAI bundle entrypoint; do not replace it with handwritten training or inference code.\n- Manifest inputs are `dataset_dir`, `datalist`, `target_anatomy`, `label_mapping`, `smoke`, `sanity`, `auto_seg`, and `skip_formal_eval`.\n- Manifest outputs are `finetuned_ckpt` and schema-checked `result_json`.\n\n## Instructions\n\n- Run `scripts\u002Frun_finetune.py`; do not patch files under `bundle\u002F` or upstream checkouts during normal skill use.\n- For standalone Bash, include the fresh-environment setup line before the wrapper; benchmark venvs start empty.\n- Run the committed script in place from the repo root. Do not copy this skill to a runtime directory, and do not use `rm` or cleanup commands in generated invocations.\n- If a host exposes `run_script`, use `run_script(\"scripts\u002Frun_finetune.py\", args=[...])`; otherwise run from the repo root.\n- For the shortest workflow check, use `--smoke`; for MSD Task06 Lung Tumor reproduction, use `--sanity`.\n- Read `references\u002Ftask06-and-results.md` only when you need Task06 reference details, output-field definitions, or manual bundle setup notes.\n\n## Available Scripts\n\n| Script | Purpose | Arguments |\n|---|---|---|\n| `scripts\u002Frun_finetune.py` | Primary entrypoint declared by `skill_manifest.yaml`; stages configs, runs MONAI, and writes `output.json`. | `[FIXTURE_OR_DATASET] --output-dir OUT_DIR [--smoke] [--sanity] [--auto-seg] [--dataset-dir DIR] [--datalist JSON] [--target-anatomy TEXT] [--label-mapping JSON] [--patch-size JSON]` |\n\n## Prerequisites\n\n- Python 3.10+ with CUDA-capable Torch for GPU runs.\n- Runtime packages from `skill_manifest.yaml`, especially `monai==1.4.0`, `numpy\u003C2`, `nibabel`, `scipy`, `typer`, `PyYAML`, `fire`, `pytorch-ignite`, `einops`, and `huggingface_hub`.\n- Optional environment variables: `CUDA_VISIBLE_DEVICES` restricts visible GPUs; `NPROC_PER_NODE` overrides GPU count and values `>=2` select multi-GPU mode for non-sanity runs.\n- Side effects: writes generated bundle configs under `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002F`, including `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fauto_override.json`, `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Ftrain_continual_task06_lung.json`, and `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fdfw_no_logging.json`; writes checkpoints\u002Fevidence under `--output-dir`, may cache model assets under `~\u002F.cache\u002Fhuggingface\u002F`, and may contact `https:\u002F\u002Fhuggingface.co` or `https:\u002F\u002Fraw.githubusercontent.com`.\n\nFresh environment setup:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub\n```\n\nKnown upstream compatibility constraints:\n\n- DFW Task06 reference: Python `3.10.16`, MONAI `1.4.0`, Torch `2.7.0+cu126`.\n- Use exact `monai==1.4.0` for smoke, sanity, and evidence runs; MONAI 1.5.x can crash the upstream finetune loss on boolean labels.\n- Do not float the dependency as `monai>=1.4,\u003C1.6` in generated commands.\n\n## Usage\n\nSmoke-scale workflow check:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub && \\\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  PATH_TO_DATASET \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nUse the staged dataset as `PATH_TO_DATASET`. For the micro fixture, use `skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro`. Smoke mode proves wiring, config generation, checkpoint loading, and runtime compatibility; it is not a quality bar.\n\nMSD Task06 Lung Tumor sanity reproduction:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  \u002Fpath\u002Fto\u002FTask06 \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\nThe sanity preset follows the single-GPU DFW recipe: fold-0 validation, label mapping `[[1, 23]]` for `lung tumor`, automatic class-prompt segmentation, patch `[128,128,128]`, 5 epochs, and original-spacing `configs\u002Fevaluate.json` scoring before and after training. Expected reference range is pretrained Dice about `0.6697`, training-best Dice about `0.6905`, and fine-tuned formal Dice about `0.6836`.\n\nUser-data finetune:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  --dataset-dir \u002Fpath\u002Fto\u002Fdataset \\\n  --datalist \u002Fpath\u002Fto\u002Fdatalist.json \\\n  --target-anatomy \"lung tumor\" \\\n  --auto-seg \\\n  --epochs 5 \\\n  --patch-size '[128,128,128]' \\\n  --output-dir runs\u002Fnvseg_user_finetune\n```\n\nUse `--label-mapping '[[1, 23]]'` when local label values are custom or the anatomy name is ambiguous.\n\n## Examples\n\nSmoke run on a staged tiny dataset:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  runs\u002Fwith_vs_without_nv\u002F_inputs\u002Fnv_segment_ct_finetune\u002Finput_dataset \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nTask06 sanity run on a local MSD cache:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  .workbench_data\u002Fdatasets\u002FTask06_Lung \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\n## Data Contract\n\n- Preferred layout: `dataset\u002FimagesTr\u002F*.nii.gz` and `dataset\u002FlabelsTr\u002F*.nii.gz`.\n- Labels must align one-to-one with images by basename.\n- The target label value must be present in the training labels.\n- Use a datalist when patient-level splitting matters. The bundle default `fold` is `0`, so `fold: 0` entries are validation and all other folds are training.\n- Every trained foreground label must map to an existing VISTA3D global class id from `bundle\u002Flabel_dict.json`; this skill cannot invent a new class.\n\n## Results\n\nCheck `output.json` in the run directory first:\n\n- `formal_pretrained_val_dice` and `formal_finetuned_val_dice`: original-spacing pre\u002Fpost scores when formal eval is enabled.\n- `training_start_val_dice`, `val_dice_per_epoch`, and `training_best_val_dice`: training-time validation trace.\n- `finetuned_ckpt_matches_pretrained_weights`: detects the epoch-0 checkpoint trap when `val_at_start=true`.\n- `recommended_ckpt`: checkpoint to keep. Do not blindly use the last epoch or `model_finetune.pt`.\n- `runtime.oom`, `runtime.peak_gpu_mb`, and phase logs: distinguish OOM, slow validation, and process failure.\n\nDecision rule: prefer formal original-spacing pre\u002Fpost scores when present; reject tensor-identical \"fine-tuned\" checkpoints for sanity recovery; treat `improved: false` as valid evidence rather than a wrapper failure.\n\n## Limitations\n\n- Thin wrapper. Training, validation, transforms, and checkpointing are delegated to the upstream bundle in `bundle\u002F`.\n- Reproduction record only: the successful five-epoch Task06 run used Python\n  `3.12.3`, PyTorch `2.12.0+cu130` with CUDA `13.0`, MONAI `1.4.0`, NumPy\n  `1.26.4`, PyTorch-Ignite `0.5.4`, NiBabel `5.4.2`, SciPy `1.16.0`, einops\n  `0.8.2`, Fire `0.7.1`, Hugging Face Hub `0.36.2`, Transformers `4.57.6`,\n  Typer `0.25.1`, PyYAML `6.0.3`, and MLflow `3.14.0` on one NVIDIA RTX 6000\n  Ada 48 GB GPU. These versions document the evidence environment; they are\n  not additional package constraints or a claim that other versions cannot\n  work.\n- The auto-derived plan is heuristic; caller-provided `--patch-size`, `--cache-rate`, `--epochs`, and `--learning-rate` win.\n- The Task06 sanity recipe intentionally forces single-GPU execution to match the DFW reference. Multi-GPU mode for other datasets requires host `torchrun` support.\n- The paired verifier is CPU-only and audits the evidence pack; it does not re-run GPU segmentation.\n- Not for clinical deployment, clinical interpretation, autonomous diagnosis, or regulatory submission.\n\n## Troubleshooting\n\n| Error | Cause | Fix |\n|---|---|---|\n| Missing dependency or import error | Runtime drift from `skill_manifest.yaml`. | Install the packages above or use the documented environment. |\n| Low Task06 pretrained Dice | Wrong config, wrong checkpoint, data split drift, or dependency drift. | Compare environment fields and staged configs before changing training logic. |\n| `model_finetune.pt` matches pretrained | `val_at_start=true` selected epoch 0 as best. | Use `recommended_ckpt`; treat sanity recovery as failed unless a changed checkpoint improves formal Dice. |\n| Missing formal Dice fields | Formal eval failed or was skipped. | Inspect `eval_pretrained.log`, `eval_finetuned.log`, and `metrics.csv`. |\n| GPU out of memory | Patch\u002Fcache settings too large. | Reduce `--patch-size`, lower `--cache-rate`, or reduce workers. |\n| No validation cases | Datalist lacks `fold: 0`. | Provide at least one validation entry. |\n\n## Verification\n\nRun the implemented verifier when quality gates matter:\n\n```bash\npython -m eval_engine.run_trusted skills\u002Fnv-segment-ct-finetune \\\n  --fixture skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro \\\n  --out runs\u002Fnvseg_trusted\n```\n","# NV-Segment-CT Finetune\n\n## Purpose\n\n- Used for smoke or dataset finetuning of NV-Segment-CT VISTA3D on CT NIfTI labels. Not for clinical validation.\n- Wraps the upstream MONAI bundle entrypoint; do not replace it with handwritten training or inference code.\n- Manifest inputs are `dataset_dir`, `datalist`, `target_anatomy`, `label_mapping`, `smoke`, `sanity`, `auto_seg`, and `skip_formal_eval`.\n- Manifest outputs are `finetuned_ckpt` and schema-checked `result_json`.\n\n## Instructions\n\n- Run `scripts\u002Frun_finetune.py`; do not patch files under `bundle\u002F` or upstream checkouts during normal skill use.\n- For standalone Bash, include the fresh-environment setup line before the wrapper; benchmark venvs start empty.\n- Run the committed script in place from the repo root. Do not copy this skill to a runtime directory, and do not use `rm` or cleanup commands in generated invocations.\n- If a host exposes `run_script`, use `run_script(\"scripts\u002Frun_finetune.py\", args=[...])`; otherwise run from the repo root.\n- For the shortest workflow check, use `--smoke`; for MSD Task06 Lung Tumor reproduction, use `--sanity`.\n- Read `references\u002Ftask06-and-results.md` only when you need Task06 reference details, output-field definitions, or manual bundle setup notes.\n\n## Available Scripts\n\n| Script | Purpose | Arguments |\n|---|---|---|\n| `scripts\u002Frun_finetune.py` | Primary entrypoint declared by `skill_manifest.yaml`; stages configs, runs MONAI, and writes `output.json`. | `[FIXTURE_OR_DATASET] --output-dir OUT_DIR [--smoke] [--sanity] [--auto-seg] [--dataset-dir DIR] [--datalist JSON] [--target-anatomy TEXT] [--label-mapping JSON] [--patch-size JSON]` |\n\n## Prerequisites\n\n- Python 3.10+ with CUDA-capable Torch for GPU runs.\n- Runtime packages from `skill_manifest.yaml`, especially `monai==1.4.0`, `numpy\u003C2`, `nibabel`, `scipy`, `typer`, `PyYAML`, `fire`, `pytorch-ignite`, `einops`, and `huggingface_hub`.\n- Optional environment variables: `CUDA_VISIBLE_DEVICES` restricts visible GPUs; `NPROC_PER_NODE` overrides GPU count and values `>=2` select multi-GPU mode for non-sanity runs.\n- Side effects: writes generated bundle configs under `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002F`, including `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fauto_override.json`, `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Ftrain_continual_task06_lung.json`, and `skills\u002Fnv-segment-ct-finetune\u002Fbundle\u002Fconfigs\u002Fdfw_no_logging.json`; writes checkpoints\u002Fevidence under `--output-dir`, may cache model assets under `~\u002F.cache\u002Fhuggingface\u002F`, and may contact `https:\u002F\u002Fhuggingface.co` or `https:\u002F\u002Fraw.githubusercontent.com`.\n\nFresh environment setup:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub\n```\n\nKnown upstream compatibility constraints:\n\n- DFW Task06 reference: Python `3.10.16`, MONAI `1.4.0`, Torch `2.7.0+cu126`.\n- Use exact `monai==1.4.0` for smoke, sanity, and evidence runs; MONAI 1.5.x can crash the upstream finetune loss on boolean labels.\n- Do not float the dependency as `monai>=1.4,\u003C1.6` in generated commands.\n\n## Usage\n\nSmoke-scale workflow check:\n\n```bash\npython -m pip install \"monai==1.4.0\" \"numpy\u003C2\" pytorch-ignite einops nibabel scipy typer PyYAML fire huggingface_hub && \\\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  PATH_TO_DATASET \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nUse the staged dataset as `PATH_TO_DATASET`. For the micro fixture, use `skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro`. Smoke mode proves wiring, config generation, checkpoint loading, and runtime compatibility; it is not a quality bar.\n\nMSD Task06 Lung Tumor sanity reproduction:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  \u002Fpath\u002Fto\u002FTask06 \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\nThe sanity preset follows the single-GPU DFW recipe: fold-0 validation, label mapping `[[1, 23]]` for `lung tumor`, automatic class-prompt segmentation, patch `[128,128,128]`, 5 epochs, and original-spacing `configs\u002Fevaluate.json` scoring before and after training. Expected reference range is pretrained Dice about `0.6697`, training-best Dice about `0.6905`, and fine-tuned formal Dice about `0.6836`.\n\nUser-data finetune:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  --dataset-dir \u002Fpath\u002Fto\u002Fdataset \\\n  --datalist \u002Fpath\u002Fto\u002Fdatalist.json \\\n  --target-anatomy \"lung tumor\" \\\n  --auto-seg \\\n  --epochs 5 \\\n  --patch-size '[128,128,128]' \\\n  --output-dir runs\u002Fnvseg_user_finetune\n```\n\nUse `--label-mapping '[[1, 23]]'` when local label values are custom or the anatomy name is ambiguous.\n\n## Examples\n\nSmoke run on a staged tiny dataset:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  runs\u002Fwith_vs_without_nv\u002F_inputs\u002Fnv_segment_ct_finetune\u002Finput_dataset \\\n  --smoke \\\n  --patch-size '[64,64,64]' \\\n  --output-dir runs\u002Fnvseg_smoke\n```\n\nTask06 sanity run on a local MSD cache:\n\n```bash\npython skills\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py \\\n  .workbench_data\u002Fdatasets\u002FTask06_Lung \\\n  --sanity \\\n  --output-dir runs\u002Fnvseg_task06_sanity\n```\n\n## Data Contract\n\n- Preferred layout: `dataset\u002FimagesTr\u002F*.nii.gz` and `dataset\u002FlabelsTr\u002F*.nii.gz`.\n- Labels must align one-to-one with images by basename.\n- The target label value must be present in the training labels.\n- Use a datalist when patient-level splitting matters. The bundle default `fold` is `0`, so `fold: 0` entries are validation and all other folds are training.\n- Every trained foreground label must map to an existing VISTA3D global class id from `bundle\u002Flabel_dict.json`; this skill cannot invent a new class.\n\n## Results\n\nCheck `output.json` in the run directory first:\n\n- `formal_pretrained_val_dice` and `formal_finetuned_val_dice`: original-spacing pre\u002Fpost scores when formal eval is enabled.\n- `training_start_val_dice`, `val_dice_per_epoch`, and `training_best_val_dice`: training-time validation trace.\n- `finetuned_ckpt_matches_pretrained_weights`: detects the epoch-0 checkpoint trap when `val_at_start=true`.\n- `recommended_ckpt`: checkpoint to keep. Do not blindly use the last epoch or `model_finetune.pt`.\n- `runtime.oom`, `runtime.peak_gpu_mb`, and phase logs: distinguish OOM, slow validation, and process failure.\n\nDecision rule: prefer formal original-spacing pre\u002Fpost scores when present; reject tensor-identical \"fine-tuned\" checkpoints for sanity recovery; treat `improved: false` as valid evidence rather than a wrapper failure.\n\n## Limitations\n\n- Thin wrapper. Training, validation, transforms, and checkpointing are delegated to the upstream bundle in `bundle\u002F`.\n- Reproduction record only: the successful five-epoch Task06 run used Python\n  `3.12.3`, PyTorch `2.12.0+cu130` with CUDA `13.0`, MONAI `1.4.0`, NumPy\n  `1.26.4`, PyTorch-Ignite `0.5.4`, NiBabel `5.4.2`, SciPy `1.16.0`, einops\n  `0.8.2`, Fire `0.7.1`, Hugging Face Hub `0.36.2`, Transformers `4.57.6`,\n  Typer `0.25.1`, PyYAML `6.0.3`, and MLflow `3.14.0` on one NVIDIA RTX 6000\n  Ada 48 GB GPU. These versions document the evidence environment; they are\n  not additional package constraints or a claim that other versions cannot\n  work.\n- The auto-derived plan is heuristic; caller-provided `--patch-size`, `--cache-rate`, `--epochs`, and `--learning-rate` win.\n- The Task06 sanity recipe intentionally forces single-GPU execution to match the DFW reference. Multi-GPU mode for other datasets requires host `torchrun` support.\n- The paired verifier is CPU-only and audits the evidence pack; it does not re-run GPU segmentation.\n- Not for clinical deployment, clinical interpretation, autonomous diagnosis, or regulatory submission.\n\n## Troubleshooting\n\n| Error | Cause | Fix |\n|---|---|---|\n| Missing dependency or import error | Runtime drift from `skill_manifest.yaml`. | Install the packages above or use the documented environment. |\n| Low Task06 pretrained Dice | Wrong config, wrong checkpoint, data split drift, or dependency drift. | Compare environment fields and staged configs before changing training logic. |\n| `model_finetune.pt` matches pretrained | `val_at_start=true` selected epoch 0 as best. | Use `recommended_ckpt`; treat sanity recovery as failed unless a changed checkpoint improves formal Dice. |\n| Missing formal Dice fields | Formal eval failed or was skipped. | Inspect `eval_pretrained.log`, `eval_finetuned.log`, and `metrics.csv`. |\n| GPU out of memory | Patch\u002Fcache settings too large. | Reduce `--patch-size`, lower `--cache-rate`, or reduce workers. |\n| No validation cases | Datalist lacks `fold: 0`. | Provide at least one validation entry. |\n\n## Verification\n\nRun the implemented verifier when quality gates matter:\n\n```bash\npython -m eval_engine.run_trusted skills\u002Fnv-segment-ct-finetune \\\n  --fixture skills\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro \\\n  --out runs\u002Fnvseg_trusted\n```\n",{"name":5,"license":15,"metadata":24,"description":7,"allowed-tools":31},{"tags":25,"author":30},[26,27,28,29],"MedTech","CT","finetuning","segmentation","NVIDIA MedTech Team","Bash","main","1c523aba5cdf867a79f078847075492d7778092a","skills\u002Fnvidia\u002Fnv-segment-ct-finetune",1,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills","https:\u002F\u002Fskills.x402.wtf","https:\u002F\u002Fgithub.com\u002Fthemusebook","https:\u002F\u002Fmusebook.trade","The official Github of musebook","2026-11-02 23:42:26",{"source":8,"skillName":5,"commands":43,"prompt":53,"zipAvailable":54},[44,47,50],{"tool":45,"command":46},"npx","npx skills add themusebook\u002Fskills --skill nv-segment-ct-finetune",{"tool":48,"command":49},"bunx","bunx skills add themusebook\u002Fskills --skill nv-segment-ct-finetune",{"tool":51,"command":52},"pnpm","pnpm dlx skills add themusebook\u002Fskills --skill nv-segment-ct-finetune","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002FSKILL.md",true,[56,62,66,70,74,80,84,88,94,98],{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"BENCHMARK.md","REFERENCE","参考文档",4033,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002FBENCHMARK.md",{"path":63,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":64,"sourceHtmlUrl":65},"evals\u002Fevals.json",1341,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fevals\u002Fevals.json",{"path":67,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":68,"sourceHtmlUrl":69},"fixtures\u002Fspleen_micro\u002Fdatalist.json",507,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Ffixtures\u002Fspleen_micro\u002Fdatalist.json",{"path":71,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Ftask06-and-results.md",3888,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Freferences\u002Ftask06-and-results.md",{"path":75,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":78,"sourceHtmlUrl":79},"scripts\u002Frun_finetune.py","SCRIPT","脚本",68176,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fscripts\u002Frun_finetune.py",{"path":81,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":82,"sourceHtmlUrl":83},"skill_manifest.yaml",12451,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fskill_manifest.yaml",{"path":85,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":86,"sourceHtmlUrl":87},"skill-card.md",3945,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fskill-card.md",{"path":89,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":92,"sourceHtmlUrl":93},"skill.oms.sig","OTHER","其他",6041,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fskill.oms.sig",{"path":95,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":96,"sourceHtmlUrl":97},"tests\u002Ftest_run_finetune.py",9752,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Ftests\u002Ftest_run_finetune.py",{"path":99,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":100,"sourceHtmlUrl":101},"validators\u002Foutput_schema.json",6842,"https:\u002F\u002Fgithub.com\u002Fthemusebook\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fnvidia\u002Fnv-segment-ct-finetune\u002Fvalidators\u002Foutput_schema.json",[103,112,119,126,133,140],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":108,"sourceHtmlUrl":109,"categories":110,"keywords":111,"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":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":118,"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":120,"name":121,"nameZh":6,"description":122,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":123,"categories":124,"keywords":125,"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":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":132,"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":134,"name":135,"nameZh":6,"description":136,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":137,"categories":138,"keywords":139,"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":141,"name":142,"nameZh":6,"description":143,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":144,"categories":145,"keywords":146,"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",[],[]]