[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7962XM91XAHX0X22SMZWHTV":3,"skill-same:01M7962XM91XAHX0X22SMZWHTV":70},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M7962XM91XAHX0X22SMZWHTV","tensorflow-yolov4-tflite",null,"Guides TensorFlow YOLOv3\u002FYOLOv4 conversion, inference, training data, evaluation, TensorRT, and Android TFLite workflows for hunglc007\u002Ftensorflow-yolov4-tflite.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"MIT","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tensorflow-yolov4-tflite\ndescription: \"Guides TensorFlow YOLOv3\u002FYOLOv4 conversion, inference, training\n  data, evaluation, TensorRT, and Android TFLite workflows for\n  hunglc007\u002Ftensorflow-yolov4-tflite.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# tensorflow-yolov4-tflite\n\nUse this repo skill when the user needs operational guidance for the\n`hunglc007\u002Ftensorflow-yolov4-tflite` project: YOLOv3\u002FYOLOv4 and tiny variants\nimplemented in TensorFlow 2, Darknet `.weights` conversion to SavedModel,\nTFLite or TF-TRT export, image\u002Fvideo detection, COCO\u002FVOC annotation preparation,\ntraining, mAP\u002FFPS checks, or the bundled Android TFLite demo.\n\n## First checks\n\n1. Confirm the user is working with a target checkout of this repository or a\n   project copied from it. The source scripts use repository-relative paths such\n   as `.\u002Fdata\u002Fclasses\u002Fcoco.names`; run commands from the target checkout root or\n   pass absolute paths where a flag supports them.\n2. Read [references\u002Frepo-provenance.md](references\u002Frepo-provenance.md) when\n   checking whether this skill matches a checkout before refreshing it.\n3. Read [references\u002Fcompatibility.md](references\u002Fcompatibility.md) before\n   installing dependencies. The upstream pins are old: TensorFlow 2.3 era,\n   Python 3.8 is the safest reconstruction target, modern protobuf must often\n   be pinned below 3.20, and GPU\u002FTensorRT needs a matching CUDA 10.x-era stack.\n4. Run [scripts\u002Fcheck_environment.py](scripts\u002Fcheck_environment.py) against the\n   user's target checkout for a non-destructive import\u002Fconfig\u002Fbackend probe.\n\nMinimal environment probe after installation:\n\n```bash\npython - \u003C\u003C'PY'\nimport tensorflow as tf, cv2\nfrom core.config import cfg\nfrom core import utils\nprint(\"tensorflow\", tf.__version__)\nprint(\"opencv\", cv2.__version__)\nprint(\"class_file\", cfg.YOLO.CLASSES)\nprint(\"classes\", len(utils.read_class_names(cfg.YOLO.CLASSES)))\nprint(\"gpus\", tf.config.experimental.list_physical_devices(\"GPU\"))\nPY\n```\n\nIf this fails, go to [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md)\nbefore changing the task plan.\n\n## Route by task\n\n- **Convert models or export formats**: use\n  [sub-skills\u002Fmodel-conversion\u002FSKILL.md](sub-skills\u002Fmodel-conversion\u002FSKILL.md)\n  for Darknet `.weights` to TensorFlow SavedModel, TFLite float32\u002Ffloat16\u002Fint8,\n  TF-TRT conversion, supported flags, artifact checks, and conversion failures.\n- **Run detection, evaluation, or benchmarks**: use\n  [sub-skills\u002Finference-evaluation\u002FSKILL.md](sub-skills\u002Finference-evaluation\u002FSKILL.md)\n  for image\u002Fvideo inference, TFLite inference, mAP file generation, FPS tests,\n  thresholds, output files, and post-processing checks.\n- **Prepare data or train**: use\n  [sub-skills\u002Ftraining-data\u002FSKILL.md](sub-skills\u002Ftraining-data\u002FSKILL.md) for\n  COCO\u002FVOC annotation formats, class\u002Fanchor config, dataset validation, training\n  stages, freeze layers, checkpoints, and dataset-related errors.\n- **Use the Android demo**: use\n  [sub-skills\u002Fandroid-deployment\u002FSKILL.md](sub-skills\u002Fandroid-deployment\u002FSKILL.md)\n  for TFLite asset placement, Gradle\u002FTFLite versions, YOLOv4 classifier\n  constants, GPU\u002FNNAPI delegate choices, and Android build\u002Frun troubleshooting.\n\n## Repo operating model\n\n- This is a script-style repository, not an installable Python distribution.\n  Future agents should install the dependencies in an isolated environment and\n  add the target checkout root to `PYTHONPATH` when running scripts from another\n  working directory.\n- `core.config.cfg` owns the default class file, anchors, strides, train\u002Ftest\n  annotation paths, thresholds, image size, batch size, and epoch counts.\n- The main Python entry points are `save_model.py`, `convert_tflite.py`,\n  `convert_trt.py`, `detect.py`, `detectvideo.py`, `evaluate.py`,\n  `benchmarks.py`, and `train.py`. The sub-skills distill their flags and safe\n  command-building steps; do not assume a command is safe to run until weights,\n  datasets, output paths, and backend requirements are checked.\n- Full YOLOv4 weights, COCO datasets, TensorRT conversion, Android builds, and\n  training are network-, data-, hardware-, or time-intensive. Treat them as\n  explicit user-approved actions, not default verification steps.\n\n## Bundled helpers\n\n- [scripts\u002Fcheck_environment.py](scripts\u002Fcheck_environment.py): safe environment\n  and target-checkout probe shared by all workflows.\n- [sub-skills\u002Fmodel-conversion\u002Fscripts\u002Fplan_conversion.py](sub-skills\u002Fmodel-conversion\u002Fscripts\u002Fplan_conversion.py):\n  generates conversion command plans without running TensorFlow conversion.\n- [sub-skills\u002Finference-evaluation\u002Fscripts\u002Fplan_inference.py](sub-skills\u002Finference-evaluation\u002Fscripts\u002Fplan_inference.py):\n  generates inference\u002Fevaluation\u002Fbenchmark command plans and validates common\n  option combinations.\n- [sub-skills\u002Ftraining-data\u002Fscripts\u002Fvalidate_annotation_line.py](sub-skills\u002Ftraining-data\u002Fscripts\u002Fvalidate_annotation_line.py):\n  validates converted COCO\u002FVOC annotation-line syntax using tiny examples.\n- [sub-skills\u002Fandroid-deployment\u002Fscripts\u002Fcheck_android_assets.py](sub-skills\u002Fandroid-deployment\u002Fscripts\u002Fcheck_android_assets.py):\n  checks Android asset naming, labels, and model-file presence for a target app.\n\n## Do not use this skill when\n\n- The user is asking about a different YOLO implementation such as Darknet,\n  Ultralytics, MMDetection, YOLOP, or DeepStream-Yolo and does not need this\n  repository's TensorFlow\u002FTFLite scripts.\n- The task is only generic object-detection theory, annotation labeling policy,\n  or Android camera app development unrelated to this repo's TFLite assets.\n- The user wants Creator-mode skill production or refresh work; route that to\n  the appropriate Creator workflow instead of using this operating skill.\n","# tensorflow-yolov4-tflite\n\nUse this repo skill when the user needs operational guidance for the\n`hunglc007\u002Ftensorflow-yolov4-tflite` project: YOLOv3\u002FYOLOv4 and tiny variants\nimplemented in TensorFlow 2, Darknet `.weights` conversion to SavedModel,\nTFLite or TF-TRT export, image\u002Fvideo detection, COCO\u002FVOC annotation preparation,\ntraining, mAP\u002FFPS checks, or the bundled Android TFLite demo.\n\n## First checks\n\n1. Confirm the user is working with a target checkout of this repository or a\n   project copied from it. The source scripts use repository-relative paths such\n   as `.\u002Fdata\u002Fclasses\u002Fcoco.names`; run commands from the target checkout root or\n   pass absolute paths where a flag supports them.\n2. Read [references\u002Frepo-provenance.md](references\u002Frepo-provenance.md) when\n   checking whether this skill matches a checkout before refreshing it.\n3. Read [references\u002Fcompatibility.md](references\u002Fcompatibility.md) before\n   installing dependencies. The upstream pins are old: TensorFlow 2.3 era,\n   Python 3.8 is the safest reconstruction target, modern protobuf must often\n   be pinned below 3.20, and GPU\u002FTensorRT needs a matching CUDA 10.x-era stack.\n4. Run [scripts\u002Fcheck_environment.py](scripts\u002Fcheck_environment.py) against the\n   user's target checkout for a non-destructive import\u002Fconfig\u002Fbackend probe.\n\nMinimal environment probe after installation:\n\n```bash\npython - \u003C\u003C'PY'\nimport tensorflow as tf, cv2\nfrom core.config import cfg\nfrom core import utils\nprint(\"tensorflow\", tf.__version__)\nprint(\"opencv\", cv2.__version__)\nprint(\"class_file\", cfg.YOLO.CLASSES)\nprint(\"classes\", len(utils.read_class_names(cfg.YOLO.CLASSES)))\nprint(\"gpus\", tf.config.experimental.list_physical_devices(\"GPU\"))\nPY\n```\n\nIf this fails, go to [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md)\nbefore changing the task plan.\n\n## Route by task\n\n- **Convert models or export formats**: use\n  [sub-skills\u002Fmodel-conversion\u002FSKILL.md](sub-skills\u002Fmodel-conversion\u002FSKILL.md)\n  for Darknet `.weights` to TensorFlow SavedModel, TFLite float32\u002Ffloat16\u002Fint8,\n  TF-TRT conversion, supported flags, artifact checks, and conversion failures.\n- **Run detection, evaluation, or benchmarks**: use\n  [sub-skills\u002Finference-evaluation\u002FSKILL.md](sub-skills\u002Finference-evaluation\u002FSKILL.md)\n  for image\u002Fvideo inference, TFLite inference, mAP file generation, FPS tests,\n  thresholds, output files, and post-processing checks.\n- **Prepare data or train**: use\n  [sub-skills\u002Ftraining-data\u002FSKILL.md](sub-skills\u002Ftraining-data\u002FSKILL.md) for\n  COCO\u002FVOC annotation formats, class\u002Fanchor config, dataset validation, training\n  stages, freeze layers, checkpoints, and dataset-related errors.\n- **Use the Android demo**: use\n  [sub-skills\u002Fandroid-deployment\u002FSKILL.md](sub-skills\u002Fandroid-deployment\u002FSKILL.md)\n  for TFLite asset placement, Gradle\u002FTFLite versions, YOLOv4 classifier\n  constants, GPU\u002FNNAPI delegate choices, and Android build\u002Frun troubleshooting.\n\n## Repo operating model\n\n- This is a script-style repository, not an installable Python distribution.\n  Future agents should install the dependencies in an isolated environment and\n  add the target checkout root to `PYTHONPATH` when running scripts from another\n  working directory.\n- `core.config.cfg` owns the default class file, anchors, strides, train\u002Ftest\n  annotation paths, thresholds, image size, batch size, and epoch counts.\n- The main Python entry points are `save_model.py`, `convert_tflite.py`,\n  `convert_trt.py`, `detect.py`, `detectvideo.py`, `evaluate.py`,\n  `benchmarks.py`, and `train.py`. The sub-skills distill their flags and safe\n  command-building steps; do not assume a command is safe to run until weights,\n  datasets, output paths, and backend requirements are checked.\n- Full YOLOv4 weights, COCO datasets, TensorRT conversion, Android builds, and\n  training are network-, data-, hardware-, or time-intensive. Treat them as\n  explicit user-approved actions, not default verification steps.\n\n## Bundled helpers\n\n- [scripts\u002Fcheck_environment.py](scripts\u002Fcheck_environment.py): safe environment\n  and target-checkout probe shared by all workflows.\n- [sub-skills\u002Fmodel-conversion\u002Fscripts\u002Fplan_conversion.py](sub-skills\u002Fmodel-conversion\u002Fscripts\u002Fplan_conversion.py):\n  generates conversion command plans without running TensorFlow conversion.\n- [sub-skills\u002Finference-evaluation\u002Fscripts\u002Fplan_inference.py](sub-skills\u002Finference-evaluation\u002Fscripts\u002Fplan_inference.py):\n  generates inference\u002Fevaluation\u002Fbenchmark command plans and validates common\n  option combinations.\n- [sub-skills\u002Ftraining-data\u002Fscripts\u002Fvalidate_annotation_line.py](sub-skills\u002Ftraining-data\u002Fscripts\u002Fvalidate_annotation_line.py):\n  validates converted COCO\u002FVOC annotation-line syntax using tiny examples.\n- [sub-skills\u002Fandroid-deployment\u002Fscripts\u002Fcheck_android_assets.py](sub-skills\u002Fandroid-deployment\u002Fscripts\u002Fcheck_android_assets.py):\n  checks Android asset naming, labels, and model-file presence for a target app.\n\n## Do not use this skill when\n\n- The user is asking about a different YOLO implementation such as Darknet,\n  Ultralytics, MMDetection, YOLOP, or DeepStream-Yolo and does not need this\n  repository's TensorFlow\u002FTFLite scripts.\n- The task is only generic object-detection theory, annotation labeling policy,\n  or Android camera app development unrelated to this repo's TFLite assets.\n- The user wants Creator-mode skill production or refresh work; route that to\n  the appropriate Creator workflow instead of using this operating skill.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill tensorflow-yolov4-tflite",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill tensorflow-yolov4-tflite",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill tensorflow-yolov4-tflite","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002FSKILL.md",[46,52,56,60,64],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fcompatibility.md","REFERENCE","参考文档",4138,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002Freferences\u002Fcompatibility.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Frepo-provenance.md",2863,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002Freferences\u002Frepo-provenance.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Frepo-routing-metadata.json",555,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002Freferences\u002Frepo-routing-metadata.json",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Ftroubleshooting.md",5051,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002Freferences\u002Ftroubleshooting.md",{"path":65,"resourceType":66,"resourceTypeDesc":67,"sizeBytes":68,"sourceHtmlUrl":69},"scripts\u002Fcheck_environment.py","SCRIPT","脚本",4255,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-yolov4-tflite\u002Fscripts\u002Fcheck_environment.py",[71,79,86,93,100,107],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":113,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]