schema_version: 1
id: EIDORA/AlexNet_IN1k
name: AlexNet_IN1k
version: 1.0.0
model_family: AlexNet
backend: onnx
runtime:
  adapter: onnx_image
  execution_provider: CPUExecutionProvider
  model_path: model.onnx
  input_names:
  - pixel_values
  output_name: embedding
  onnxruntime:
    opset: 17
    tested_versions: '>=1.17,<2'
artifact:
  path: model.onnx
  sha256: 3e1277374f9da12a0c92084e19003dd541a792fccd5b4c0fd2bc4bfa54b28cc6
  package_size_bytes: 228020495
inputs:
- id: image
  modality: image
  label: Images
  required: true
  source_kind: media_source
  requirements:
    color_space: RGB
    layout: NCHW
    width: 224
    height: 224
preprocess:
  image:
    resize:
      mode: resize_shorter_edge_then_center_crop
      resize_size: 256
      crop_width: 224
      crop_height: 224
      interpolation: bilinear
      source: TorchVision AlexNet_Weights.IMAGENET1K_V1.transforms
    rescale: 1/255
    normalize:
      mean:
      - 0.485
      - 0.456
      - 0.406
      std:
      - 0.229
      - 0.224
      - 0.225
      inside_onnx: true
embedding:
  dimensions: 4096
  feature_type: embedding
  pooling: fc_head
  normalized: true
  similarity: cosine
  output_name: embedding
  dtype: float32
  shape:
  - batch
  - 4096
display:
  summary: 'Light: classic ImageNet visual embeddings for fast first-pass grouping
    on laptops.'
  compute_tier: light
  modality_labels:
  - image
  recommended_batch_size: 8
validation:
  fixtures:
  - id: image_tensor_001
    input_shape:
    - 1
    - 3
    - 224
    - 224
    expected_shape:
    - 1
    - 4096
    seed: 17
  checks:
    load_with: onnxruntime
    execution_provider: CPUExecutionProvider
    output_dtype: float32
    finite: true
    normalized_l2_range:
    - 0.99
    - 1.01
provenance:
  base_model: torchvision/alexnet-imagenet1k-v1
  source_repository: https://pytorch.org/vision/stable/models/generated/torchvision.models.alexnet.html
  original_model_name: AlexNet ImageNet
  original_model_url: https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
  authors:
  - Alex Krizhevsky
  - Ilya Sutskever
  - Geoffrey E. Hinton
  paper_title: ImageNet Classification with Deep Convolutional Neural Networks
  paper_url: https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
  upstream_license: BSD-3-Clause for TorchVision code; ImageNet-trained weights distributed
    by PyTorch under their documented model terms.
  training_data: ImageNet-1K supervised classification data.
  citation: "@inproceedings{krizhevsky2012imagenet,\n  title={ImageNet Classification\
    \ with Deep Convolutional Neural Networks},\n  author={Krizhevsky, Alex and Sutskever,\
    \ Ilya and Hinton, Geoffrey E.},\n  booktitle={Advances in Neural Information\
    \ Processing Systems},\n  year={2012}\n}\n"
  conversion_note: EIDORA produced this ONNX conversion and is not the original model
    creator.
  export_date: '2026-07-10'
  exporter_version: eidora-onnx-exporter 0.1.0
model_card:
  best_for:
  - Fast baseline visual grouping.
  - Small or exploratory image projects on ordinary laptops.
  - Regression testing the EIDORA ONNX package pipeline.
  not_ideal_for:
  - Fine-grained visual similarity where modern self-supervised models perform better.
  - Text, video, or audio inputs.
  - Production-quality semantic image retrieval when a stronger model is acceptable.
  limitations: AlexNet is an older supervised ImageNet model. It is useful as a lightweight
    baseline, but modern CLIP, DINOv2, SigLIP, or domain-specific models will usually
    produce stronger semantic similarity.
  license:
    id: bsd-3-clause
    attribution: Converted to ONNX for EIDORA from the TorchVision AlexNet ImageNet
      weights.
  huggingface:
    org: eidora
    repo_name: alexnet_imagenet1k_4096
    pipeline_tag: feature-extraction
    tags:
    - eidora
    - eidora-model-zoo
    - onnx
    - onnxruntime
    - embeddings
    - image
    - alexnet
    - imagenet
    - compute:light
    - modality:image
    datasets:
    - imagenet-1k
    metrics:
    - cosine-similarity