Match approved DINO minimal schema-v3 public package layout

README.md

@@ -1,99 +1,96 @@
 ---
-pipeline_tag: image-feature-extraction
+pipeline_tag: feature-extraction
 tags:
 - eidora
 - eidora-model-zoo
 - onnx
 - onnxruntime
-- image
 - embeddings
+- image
+- modality:image
 - alexnet
-- imagenet
 ---
 
 # AlexNet Image Light
 
-AlexNet Image Light is a light image embedding model kept mainly as a reproducible baseline for comparing how older CNN features organize cultural image collections.
-
-## Source and provenance
-
-- Base model/checkpoint: `torchvision/alexnet-imagenet1k-v1`
-- Source: https://pytorch.org/vision/stable/models/generated/torchvision.models.alexnet.html
-- Original model: AlexNet
-- Architecture paper: https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
-- Training data: ImageNet-1K
-
-## EIDORA package
-
-- Package schema: `3` (Desktop preparation implementation `1`)
-- Declared providers: `CPUExecutionProvider`
-- Provider declarations require certification in validation-report.json before publication.
-- Preparation runs in EIDORA; exposed defaults are recommended, alternatives are not equivalent-quality claims.
-- Modality: `image`
-- Embedding dimensions: `4096`
-- Normalized: `true`
-- Similarity: `cosine`
-
-## Preparation
-
-```yaml
-width: 224
-height: 224
-layout: NCHW
-resize: short_edge_crop
-interpolation: bilinear
-rescale: 0.00392156862745098
-mean:
-- 0.485
-- 0.456
-- 0.406
-std:
-- 0.229
-- 0.224
-- 0.225
-alpha_background:
-- 0
-- 0
-- 0
-short_edge: 256
-parameters:
-  resize:
-    default: short_edge_crop
-    choices:
-    - stretch
-    - fill_crop
-    - fit_pad
-    - short_edge_crop
-  interpolation:
-    default: bilinear
-    choices:
-    - nearest
-    - bilinear
-    - bicubic
-    - lanczos
-  crop_position:
-    default: center
-  short_edge:
-    default: 256
-    minimum: 224
-    maximum: 2048
-```
+AlexNet is a convolutional image model from the ImageNet classification era. This EIDORA adaptation extracts a 4096-dimensional feature vector from its classifier rather than returning class probabilities.
+
+## Training Data and Checkpoint Provenance
+
+- **Architecture:** AlexNet convolutional network with fully connected classifier.
+- **Original checkpoint:** [torchvision/alexnet-imagenet1k-v1](https://pytorch.org/vision/stable/models/generated/torchvision.models.alexnet.html).
+- **Weight variant:** `IMAGENET1K_V1` ([TorchVision download](https://download.pytorch.org/models/alexnet-owt-7be5be79.pth)).
+- **Training data:** ImageNet-1K
+- **Training method:** Supervised image classification.
+- **Source:** AlexNet (original author attribution retained).
+
+## Model Overview
+
+- **Model:** AlexNet Image Light
+- **EIDORA identifier:** `eidora/image-alexnet-imagenet-light`
+- **Modality:** Image
+- **Embedding dimensions:** 4096
+- **Embedding normalization:** L2-normalized
+- **Similarity measure:** Cosine similarity
+- **Package schema:** Version 3
+- **Model version:** `0.1.0`
+
+## EIDORA Use and Supported Modalities
+
+**Supported modality:** Images only.
+
+Researchers can use this package to:
+
+- Use a familiar CNN baseline when comparing image embedding models.
+- Explore simple visual similarity and clustering in an image collection.
+- Compare older CNN features with ResNet, DINO and CLIP.
+
+## Input and Preprocessing
+
+EIDORA prepares the image before inference:
+
+1. Convert to RGB.
+2. Resize the shorter side to 256 pixels, then take a center crop of **224 × 224**.
+3. Use **bilinear** interpolation. Use **truncate** dimension rounding.
+4. Rescale by `0.00392156862745098` and normalize each RGB channel.
+5. Arrange the tensor in `NCHW` layout and pass `float32` `pixel_values` with shape `[batch, 3, 224, 224]`.
+
+**Mean:** `[0.485, 0.456, 0.406]`  
+**Standard deviation:** `[0.229, 0.224, 0.225]`
 
-## Intended use
+Normalization is performed by EIDORA and should not be applied a second time. The defaults in `config.yaml` determine the documented embedding behavior; alternate settings may change the vectors.
 
-- Baseline comparisons against newer image embedding models.
-- Teaching, debugging, and validating the EIDORA ONNX conversion workflow.
-- Projects where compatibility and a simple architecture matter more than representation quality.
+## Output Embeddings
+
+The ONNX output is `float32` with shape `['batch', 4096]`. Each vector has **4096 dimensions**, is **L2-normalized**, and is intended for **cosine similarity** comparisons.
+
+EIDORA retains the pretrained image network and extracts features from the second fully connected layer after ReLU, followed by L2 normalization.
+
+## Runtime and Compatibility
+
+The package uses `model.onnx` and schema-v3 runtime declarations for image preparation and embedding output.
+
+It declares **`CPUExecutionProvider`**. Actual execution availability depends on the installed ONNX Runtime and EIDORA Desktop environment. `config.yaml` is authoritative for graph inputs, preprocessing, provider and embedding dimensions.
 
 ## Limitations
 
-AlexNet is an older supervised ImageNet model. Its embedding geometry reflects ImageNet categories and may not align with culturally meaningful distinctions. Use it mainly as a baseline and compare it with ResNet, DINO, and CLIP on the target collection.
+- The model was trained for ImageNet classification, not cultural or historical interpretation.
+- CNN feature proximity does not establish common authorship, identity or meaning.
+- Other image resize settings may produce different embeddings.
+
+Results should be evaluated against the intended research data and questions.
+
+## Research Context and References
+
+### Main paper
 
-## License and attribution
+**ImageNet Classification with Deep Convolutional Neural Networks**
 
-Converted from TorchVision AlexNet weights; verify pretrained-weight redistribution terms before publishing the ONNX artifact.
+Paper: https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
 
-## Citation
+The publication describes the upstream learning method and original model. The embeddings returned by EIDORA additionally depend on the specific preprocessing and output adaptation documented here.
+
+### Primary citation (BibTeX)
 
 ```bibtex
 @inproceedings{krizhevsky2012imagenet,
@@ -103,3 +100,22 @@ Converted from TorchVision AlexNet weights; verify pretrained-weight redistribut
   year={2012}
 }
 ```
+
+## License and Attribution
+
+The TorchVision software library uses BSD-3-Clause terms. This does not by itself assign those terms to the pretrained ImageNet weights. The linked TorchVision model documentation does not establish a separate license for these pretrained weights. No weight-license identifier is asserted here. The ONNX graph contains converted weights from TorchVision. EIDORA does not claim authorship of the original model.
+
+No separate upstream checkpoint `LICENSE` file was supplied in the captured EIDORA repository; that absence should not be interpreted as permission or prohibition.
+
+## Package Information
+
+- **Model repository:** `EIDORA/AlexNet_IN1k`
+- **Package version:** `0.1.0`
+- **Package schema:** `3`
+- **EIDORA configuration:** `config.yaml`
+- **ONNX graph(s):** `model.onnx`
+- **Supported modality:** Image
+- **License identifier:** `other`
+- **Declared provider:** CPUExecutionProvider
+
+The configuration defines the authoritative model-facing runtime contract. Other packaged assets, where present, do not introduce additional inference modalities.

config.yaml

@@ -11,30 +11,9 @@ bibtex: "@inproceedings{krizhevsky2012imagenet,\n  title={ImageNet Classificatio
 compatibility:
   min_eidora_version: 0.1.0
 artifacts:
-- path: conversion_report.json
-  sha256: a7f417fc4d8969b4682b96fda6ad6863cebab4f89adb48d1998bfde6b926abbd
-  role: asset
 - path: model.onnx
   sha256: deea3812842a09731d3a9ddc88dcace1049d1b39cafd25a902a39437b9b239a5
   role: graph
-- path: validation.yaml
-  sha256: 18b12ddef178c1cf77334dde9e101c228c4b8c88ade6ed2f0eef3fc4d8f2bbb4
-  role: asset
-- path: validation/embedding.npy
-  sha256: 990d9316a16850630befdd0ec9cfe5584e6ac441302d87fb68a7c4adde2cfa33
-  role: asset
-- path: validation/photo.png
-  sha256: 1d70a03fc99c16e3aa674dbb9c7a8d0c127b45e1cf416118131451932b2f2389
-  role: asset
-- path: validation/pixels.npy
-  sha256: f559af24acc89a03d8d1b1b59037d301fcb4d77ab8bea2e0e64d355a226abc8b
-  role: asset
-- path: validation/random_image.npy
-  sha256: 2e7b24b32c98c0ae8c05f508b9b037c6c3e51b0ba35405e5f2594604c3acbf48
-  role: asset
-- path: validation/source_parity.json
-  sha256: 7b5f9e26cae0535eebff345800ffab3c5aaf2d9b4f228a6e4a71cf2e8bbf9756
-  role: asset
 runtime:
   graph: model.onnx
   inputs:

conversion_report.json

@@ -1,18 +0,0 @@
-{
-  "converter": "torchvision_alexnet",
-  "embedding_dimensions": 4096,
-  "input_name": "pixel_values",
-  "opset": 17,
-  "output_name": "embedding",
-  "schema_version": 1,
-  "source_framework": "pytorch",
-  "source_vs_onnx": {
-    "atol": 0.0001,
-    "max_abs_error": 1.043081283569336e-07,
-    "mean_abs_error": 1.883620370790595e-09,
-    "onnxruntime_version": "1.30.0",
-    "rtol": 0.001,
-    "status": "passed"
-  },
-  "torch_version": "2.14.0+cu130"
-}

validation-report.json

@@ -1,26 +0,0 @@
-{
-  "desktop_version": "0.2.0",
-  "schema_version": 3,
-  "preparation_version": 2,
-  "providers": {
-    "CPUExecutionProvider": {
-      "status": "passed",
-      "fixture_runs": 19,
-      "scope": "session provider chain; CPU graph-node support allowed",
-      "environment": {
-        "onnxruntime": "1.31.0",
-        "platform": "Linux-7.0.0-38-generic-x86_64-with-glibc2.43",
-        "python": "3.13.16 (main, Oct  9 2026, 15:23:36) [Clang 22.1.3 ]"
-      }
-    }
-  },
-  "environment": {
-    "onnxruntime": "1.31.0",
-    "platform": "Linux-7.0.0-38-generic-x86_64-with-glibc2.43",
-    "python": "3.13.16 (main, Oct  9 2026, 15:23:36) [Clang 22.1.3 ]"
-  },
-  "publication_ready": true,
-  "validation_sha256": "df0c747d8cd0734a8dcad20da6afe6f74d437b51e431845f633577e714074167",
-  "contract_sha256": "65c20b19452a5cedc6874434d8379a280d6c9b4343a4deab6d9772d93ee8d6b0",
-  "package_sha256": "5dbe16d8f71ed8ce5c6507a74faeed81a02b3de3d98e1fe588c54ed4ad839093"
-}

validation.yaml

@@ -1,21 +0,0 @@
-diagnostic_fixtures:
-- id: random_image
-  path: validation/random_image.npy
-  input_shape:
-  - 1
-  - 3
-  - 224
-  - 224
-  expected_shape:
-  - 1
-  - 4096
-  seed: 17
-fixtures:
-- modality: image
-  path: validation/photo.png
-  expected_tensors:
-    pixel_values: validation/pixels.npy
-  expected_vectors: validation/embedding.npy
-note: Publication requires real-media fixtures with source-reference expected_tensors
-  and expected_vectors, copied using assets. Synthetic diagnostic fixtures do not
-  certify compatibility.

validation/embedding.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:990d9316a16850630befdd0ec9cfe5584e6ac441302d87fb68a7c4adde2cfa33
-size 16512

validation/pixels.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f559af24acc89a03d8d1b1b59037d301fcb4d77ab8bea2e0e64d355a226abc8b
-size 602240

validation/random_image.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2e7b24b32c98c0ae8c05f508b9b037c6c3e51b0ba35405e5f2594604c3acbf48
-size 602240

validation/source_parity.json

@@ -1,15 +0,0 @@
-{
-  "model": "alexnet",
-  "source_checkpoint": "IMAGENET1K_V1",
-  "image_sha256": "1d70a03fc99c16e3aa674dbb9c7a8d0c127b45e1cf416118131451932b2f2389",
-  "onnx_sha256": "deea3812842a09731d3a9ddc88dcace1049d1b39cafd25a902a39437b9b239a5",
-  "preprocessing_max_abs_error": 7.152557373046875e-07,
-  "embedding_max_abs_error": 1.7695128917694092e-07,
-  "cosine_similarity": 1.0000001192092896,
-  "output_shape": [
-    1,
-    4096
-  ],
-  "source": "TorchVision pretrained model",
-  "status": "passed"
-}