manifest.yaml

onescience_manifest_version: "0.1"
resource_type: dataset

resource:
  id: OneScience/ShapeNetCar
  name: ShapeNetCar
  domain: cfd
  domain_tags:
    - cfd
    - external_aerodynamics
    - car_design
  task: flow_field_prediction
  task_tags:
    - car_design
    - pressure_prediction
    - velocity_prediction
    - drag_coefficient_evaluation
  modalities:
    - unstructured_mesh
    - point_cloud
    - graph
  input_formats:
    - npy
    - vtk
    - txt
  output_formats:
    - npy
    - vtk
    - png
  summary: ShapeNetCar 汽车外流场数据集，包含原始训练数据、预处理图数据、归一化统计量和可视化参考文件。

platform_resource:
  primary:
    platform: modelscope
    repo_id: OneScience/ShapeNetCar
    repo_type: dataset
    url: https://modelscope.cn/datasets/OneScience/ShapeNetCar
    revision: main
    readme_path: README.md
    manifest_path: manifest.yaml
  mirrors: []
  access:
    visibility: public
    license: null

website_integration:
  enabled: true
  click_target:
    platform: modelscope
    resource_url: https://modelscope.cn/datasets/OneScience/ShapeNetCar
  llm_handoff:
    readme_required: true
    manifest_required: true
    download_readme_first: true
    resolve_related_models: true
    default_run_goal: data_validation

runtime:
  enabled: true
  onescience_domain: cfd
  min_onescience_version: null
  supported_execution:
    - local
    - local_slurm
    - remote_slurm
    - scnet_mcp

onescience:
  repo: https://gitee.com/onescience-ai/onescience
  official_links:
    gitee:
      doc: https://gitee.com/onescience-ai/onescience-doc
      onescience: https://gitee.com/onescience-ai/onescience
      skills: https://gitee.com/onescience-ai/oneskills
    github:
      doc: https://github.com/onescience-ai/OneScience-doc
      onescience: https://github.com/onescience-ai/OneScience
      skills: https://github.com/onescience-ai/oneskills
  install:
    required_by_default: false
    command: bash install.sh cfd
  source_paths: []
  compatibility:
    examples_path: examples/cfd/Transolver-Car-Design
    status: examples_compatible

runtime_package:
  kind: dataset_package
  package_root: .
  standard_layout:
    workdir: .
    config_dir: metadata
    checkpoint_dir: null
    data_dir: data
    output_dir: null
  apply_policy:
    mode: direct_use
    target: session_workdir
    overwrite: true
    protect_installed_onescience: true
  entry_files:
    - metadata/schema.yaml
    - scripts/validate_shapenetcar_data.py
  entrypoints:
    preflight: scripts/validate_shapenetcar_data.py
    inference: null
    train: null
    finetune: null
    evaluate: null
    visualize: null
    deploy: null

files:
  model_files: []
  dataset_files:
    - id: mlcfd_data_root
      path: data/mlcfd_data
      role: train_split
      description_zh: ShapeNetCar 数据根目录，包含 training_data、preprocessed_data、stats、图表原始数据和可视化参考图。
      required: true
      required_for: [preflight, train, evaluate, inference, visualize]
      format: directory
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: data/mlcfd_data
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: data/mlcfd_data
      size: 18G
      checksum: metadata/file_manifest.sha256
      large_file: true
    - id: training_data
      path: data/mlcfd_data/training_data
      role: train_split
      description_zh: 原始训练数组和部分 VTK/文本数据，按 param0 到 param8 组织。
      required: true
      required_for: [train, evaluate, inference]
      format: npy_directory
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: data/mlcfd_data/training_data
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: data/mlcfd_data/training_data
      size: null
      checksum: metadata/file_manifest.sha256
      large_file: true
    - id: preprocessed_data
      path: data/mlcfd_data/preprocessed_data
      role: train_split
      description_zh: Transolver 数据管道直接读取的预处理图样本，样本内包含 x、y、pos、surf 和 edge_index。
      required: true
      required_for: [preflight, train, evaluate, inference]
      format: npy_directory
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: data/mlcfd_data/preprocessed_data
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: data/mlcfd_data/preprocessed_data
      size: null
      checksum: metadata/file_manifest.sha256
      large_file: true
    - id: normalization_stats
      path: data/mlcfd_data/stats
      role: normalization_stats
      description_zh: 输入输出均值和标准差，模型训练、推理和反归一化评测均需要。
      required: true
      required_for: [preflight, train, evaluate, inference]
      format: npy_directory
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: data/mlcfd_data/stats
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: data/mlcfd_data/stats
      size: null
      checksum: metadata/file_manifest.sha256
      large_file: false
    - id: schema
      path: metadata/schema.yaml
      role: schema
      description_zh: 数据字段、shape、dtype 和目录模式说明。
      required: true
      required_for: [preflight]
      format: yaml
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: metadata/schema.yaml
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: metadata/schema.yaml
      size: null
      checksum: null
      large_file: false
  config_files: []
  sample_files:
    - id: validation_script
      path: scripts/validate_shapenetcar_data.py
      role: sample
      description_zh: 数据集读取验证脚本，会读取统计文件、训练数组和一个预处理样本，检查 shape、dtype 和可读性。
      required: true
      required_for: [preflight]
      format: py
      source:
        platform: modelscope
        repo_id: OneScience/ShapeNetCar
        repo_type: dataset
        path: scripts/validate_shapenetcar_data.py
        revision: main
        url: null
        download_method: command_ref
        command_ref: commands.download.download_dataset_package
      mirrors: []
      local_path: scripts/validate_shapenetcar_data.py
      size: null
      checksum: null
      large_file: false

dataset:
  format:
    - npy
    - vtk
    - txt
    - xlsx
    - png
  sample_unit: 一个汽车几何样本的预处理图目录，包含 x.npy、y.npy、pos.npy、surf.npy 和 edge_index.npy。
  schema:
    path: metadata/schema.yaml
  splits:
    train: data/mlcfd_data/training_data/param0..param8
    validation: 由 Transolver-Car-Design 配置 datapipe.data.splits.fold_id 决定
    test: 由 Transolver-Car-Design 配置 datapipe.data.splits.fold_id 决定
  statistics:
    normalization:
      - data/mlcfd_data/stats/mean_in.npy
      - data/mlcfd_data/stats/std_in.npy
      - data/mlcfd_data/stats/mean_out.npy
      - data/mlcfd_data/stats/std_out.npy

relations:
  required_datasets: []
  optional_datasets: []
  compatible_models:
    - id: OneScience/Transolver-Car-Design/mlcfd_data
      role: train_data
      required_for: [preflight, train, inference, evaluate, visualize]
      resource_ref:
        platform: modelscope
        repo_id: OneScience/Transolver-Car-Design/mlcfd_data
        repo_type: model
        url: https://modelscope.cn/models/OneScience/Transolver-Car-Design/mlcfd_data
        revision: main
        readme_path: README.md
        manifest_path: manifest.yaml
      expected_local_path: session_workdir/data/mlcfd_data
      adapter: null

run_matrix:
  scenarios:
    - name: data_validation
      default: true
      capability: preflight
      description_zh: 验证 ShapeNetCar 数据目录结构、统计文件和预处理样本可读。
      required_datasets:
        - id: OneScience/ShapeNetCar
          role: train_data
          local_path: data/mlcfd_data
      required_model_files: []
      required_dataset_files:
        - data/mlcfd_data
        - metadata/schema.yaml
        - scripts/validate_shapenetcar_data.py
      command_refs:
        - commands.preflight.validate_data
      expected_outputs:
        - "[OK] ShapeNetCar data validation passed"

capabilities:
  inference: false
  train: false
  finetune: false
  evaluate: false
  visualize: false
  deploy: false
  data_validation: true

commands:
  download:
    - name: download_dataset_package
      target: dataset
      repo_id: OneScience/ShapeNetCar
      source: .
      local_path: .
      required: true
      required_for: [preflight, train, evaluate, inference, visualize]
      command: modelscope download --dataset OneScience/ShapeNetCar --local_dir .
  preflight:
    - name: validate_data
      description: 读取并验证数据目录结构、统计文件、训练数组和预处理图样本。
      cwd: .
      command: python scripts/validate_shapenetcar_data.py --data-root data/mlcfd_data
      required_files:
        - data/mlcfd_data
        - metadata/schema.yaml
        - scripts/validate_shapenetcar_data.py
      required_env:
        - python
        - numpy
      expected_outputs:
        - path: stdout
          type: text
          contains: "[OK] ShapeNetCar data validation passed"
      success_criteria:
        - exit_code == 0
  prepare: []
  inference: []
  train: []
  finetune: []
  evaluate: []
  visualize: []
  deploy: []

expected_outputs:
  - path: stdout
    type: text
    produced_by: commands.preflight.validate_data
    description_zh: 输出 ShapeNetCar data validation passed、训练参数目录数量、预处理样本数量和已检查样本路径。

diagnostics:
  - error_pattern: "data root not found"
    likely_cause: 数据集未下载到当前目录，或当前 cwd 不是数据集仓库根目录。
    action: 执行 modelscope download --dataset OneScience/ShapeNetCar --local_dir .，然后在数据集仓库根目录运行验证脚本。
  - error_pattern: "missing directory"
    likely_cause: data/mlcfd_data 内缺少 training_data、preprocessed_data 或 stats。
    action: 检查下载是否完整，必要时重新下载数据集仓库。
  - error_pattern: "stats shape mismatch"
    likely_cause: 归一化统计文件损坏或版本不匹配。
    action: 用 metadata/file_manifest.sha256 对比文件大小和 SHA256，重新下载不一致的文件。
  - error_pattern: "schema mismatch"
    likely_cause: 预处理样本不是 Transolver-Car-Design 期望的 ShapeNetCar 图数据格式。
    action: 确认使用的是 OneScience/ShapeNetCar，而不是其他车辆 CFD 数据。
  - error_pattern: "cannot read"
    likely_cause: npy 文件损坏或权限不足。
    action: 检查文件权限、磁盘完整性和 SHA256。

domain_extension:
  cfd:
    geometry_type: car_body
    mesh_type: unstructured_graph
    variables:
      - velocity_x
      - velocity_y
      - velocity_z
      - pressure
      - drag_coefficient
    metrics:
      - relative_l2_pressure
      - relative_l2_velocity
      - rmse_pressure
      - rmse_velocity
      - mean_relative_cd_error
      - spearman_correlation

large_resource:
  enabled: true
  total_size: 18G
  file_count: 10967
  download_policy: on_demand
  default_download: full_for_validation
  minimal_files:
    - data/mlcfd_data/stats
    - data/mlcfd_data/training_data
    - data/mlcfd_data/preprocessed_data
  full_download:
    command: modelscope download --dataset OneScience/ShapeNetCar --local_dir .
  partial_download:
    command: null