Refresh DeepDeWedge tutorial checkpoint to format 2

ATTRIBUTION.md

@@ -2,47 +2,30 @@
 
 ## Original material
 
-**DeepDeWedge Tutorial Data**  
-Creator: Simon Wiedemann  
-DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>  
-Figshare file: `tutorial_data.zip`, file id `45582309`  
-Checkpoint member: `tutorial_data/fitted_model.ckpt`  
-License: Creative Commons Attribution 4.0 International  
-License URI: <https://creativecommons.org/licenses/by/4.0/>
-
-The original checkpoint accompanies:
-
-Simon Wiedemann and Reinhard Heckel, “A deep learning method for simultaneous
-denoising and missing wedge reconstruction in cryogenic electron tomography,”
-Nature Communications 15, 8255 (2024).
-<https://doi.org/10.1038/s41467-024-51438-y>
-
-The upstream DeepDeWedge implementation is available at
-<https://github.com/MLI-lab/DeepDeWedge> and was reviewed at revision
-`072075692a44a8f17394214369e6e762abe52bc3`.
-
-## Changes made by scitomo
-
-The original PyTorch Lightning checkpoint was converted into a
-scitomo-native package:
-
-- executable/pickled training metadata was excluded;
-- tensor state was exported in Safetensors format;
-- 56 vendor state names were mapped through a reviewed explicit mapping to the
-  native scitomo `UNet3D` state;
-- the vendor `unet.` namespace was removed;
-- the second bottleneck convolution was mapped from vendor sequence index `2`
-  to the semantically equivalent native sequence index `4`;
-- two learned normalization values changed storage role from non-trainable
-  parameters to native buffers without changing their values; and
-- strict construction, inference, conversion, validation, provenance, and
-  tensor-inventory records were added.
-
-No tensor value was intentionally changed. Synthetic forward output was
-bit-exact, and a frozen real tutorial-volume crop passed the predetermined
-relative-L2 parity threshold.
-
-The converted package is distributed under the source material's CC BY 4.0
-terms. No endorsement by Simon Wiedemann, Reinhard Heckel, the Machine Learning
-and Information Processing Laboratory, Nature Communications, or Figshare is
-stated or implied.
+**DeepDeWedge Tutorial Data**
+Creator: Simon Wiedemann
+DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>
+Figshare file id: `45582309`
+Archive member: `tutorial_data/fitted_model.ckpt`
+License: Creative Commons Attribution 4.0 International
+
+The method is described by Simon Wiedemann and Reinhard Heckel, *A deep
+learning method for simultaneous denoising and missing wedge reconstruction in
+cryogenic electron tomography*, Nature Communications 15, 8255 (2024),
+<https://doi.org/10.1038/s41467-024-51438-y>.
+
+Pinned upstream code: <https://github.com/MLI-lab/DeepDeWedge/tree/072075692a44a8f17394214369e6e762abe52bc3>
+(BSD-2-Clause).
+
+## Changes in this package
+
+On 2026-09-04 Scitomo freshly converted only the authoritative checkpoint
+`official/fitted_model.ckpt`, after byte-size and SHA-256 verification, through
+the exact pinned upstream source and current generic FORMAT 2 exporter. The
+54 U-Net state tensors were explicitly mapped into canonical Network state.
+The two fitted affine quantities were preserved as external DeepDeWedge
+inference-profile state; they are not Network state. No old Hugging Face
+Safetensors or format-1 package artifact was conversion input.
+
+No endorsement by the cited authors, the Machine Learning and Information
+Processing Laboratory, Figshare, or the rights holders is implied.

README.md

@@ -1,160 +1,82 @@
 ---
 license: cc-by-4.0
 library_name: scitomo
-tags: [cryo-electron-tomography, denoising, missing-wedge, safetensors, scitomo]
+tags: [cryo-electron-tomography, deepdewedge, safetensors, scitomo, format-2]
 ---
 
-# DeepDeWedge tutorial checkpoint for scitomo
-
-This repository contains the official DeepDeWedge tutorial checkpoint converted
-to a scitomo-native, non-executable Safetensors package. It reinstantiates the
-scitomo `UNet3D` used by the `deepdewedge` restoration method. It is not a
-PyTorch Lightning trainer-resume checkpoint and does not contain optimizer,
-scheduler, callback, random-number-generator, or dataloader state.
-
-Package identity:
-
-- scitomo catalog name: `deepdewedge_tutorial`
-- native package id: `deepdewedge_tutorial`
-- package revision: `2`
-- minimum scitomo version: `0.6.2`
-- learned-checkpoint format: `1`
-- learned-checkpoint manifest schema: `2`
-- source package revision: `1` (manifest schema `1`)
-- method: restoration / `deepdewedge`
-- construction fingerprint:
-  `sha256:c9da7decd489f0f4f893dce565f37db25abc63d0d3efcacad9835305ffd97cb2`
-- inference fingerprint:
-  `sha256:c2c5eba72579510265258dcfce3a5be2f851f9276c7e53d92330d78eb37720dd`
-
-## Source and attribution
-
-The original checkpoint is part of **DeepDeWedge Tutorial Data**, authored by
-Simon Wiedemann and published on Figshare under CC BY 4.0:
-
-- DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>
-- Figshare file id: `45582309`
-- archive: `tutorial_data.zip`
-- archive member: `tutorial_data/fitted_model.ckpt`
-- archive SHA-256:
-  `7c871342e51f5a66a773fe427d72944b5d2cc8ff41c5b7415ab38dbfc9ac6d58`
-- original checkpoint SHA-256:
-  `5262f6c11e85fd662b02e59efe936fa7b69913758e841235be2683f7bd03ec76`
-
-The method and upstream implementation are described by:
-
-> Simon Wiedemann and Reinhard Heckel. “A deep learning method for
-> simultaneous denoising and missing wedge reconstruction in cryogenic
-> electron tomography.” Nature Communications 15, 8255 (2024).
-> <https://doi.org/10.1038/s41467-024-51438-y>
-
-Upstream code:
-<https://github.com/MLI-lab/DeepDeWedge/tree/072075692a44a8f17394214369e6e762abe52bc3>
-
-The conversion changed the serialization, state names, and executable
-construction mechanism. It did not intentionally alter tensor values. Two
-learned normalization values that upstream stored as non-trainable parameters
-are native scitomo buffers with identical values. See `ATTRIBUTION.md`,
-`conversion/conversion-record.json`, and `LICENSES/`.
-
-## Package migration
-
-Revision 2 migrates the native package manifest from schema 1 to schema 2
-and raises the minimum Scitomo version accordingly. The construction and
-inference records and `weights.safetensors` are byte-for-byte unchanged.
-`migration/migration-record.json` binds the immutable source manifest and
-provider revision to this package and records the unchanged file identities.
-
-## Native construction and inference
-
-`construction.json` specifies a residual single-input/single-output 3D U-Net
-with 64 initial channels, three downsampling blocks, strided-convolution
-downsampling, transposed-convolution upsampling, leaky-ReLU activations, and
-the learned input-normalization location and scale from the tutorial
-checkpoint.
-
-Volume tensors use scitomo's canonical trailing-axis layout `(..., Z, Y, X)`.
-The network layout is `(..., C, Z, Y, X)`, with one scalar volume channel.
-The canonical rotation axis is `Z`; at theta zero the beam axis is `Y`, and
-detector `(V, U)` corresponds to `(Z, X)`.
-
-The frozen DeepDeWedge inference profile requires:
-
-- paired half-tomograms, refined separately and averaged;
-- a 50-degree full-width missing-wedge Fourier mask on each half;
-- `96 x 96 x 96` patches with overlap `32 x 32 x 32`;
-- trailing reflection padding for full coverage;
-- patch-statistic normalization and network denormalization;
-- linear-ramp weighted patch reassembly; and
-- no full-tomogram standardization.
-
-The package was converted from the tutorial's fitted network. It assumes the
-same scientific meaning, preprocessing, normalization, missing-wedge
-convention, and paired-half workflow. It is not a general-purpose cryo-ET
-foundation model.
-
-## Files and identities
-
-| File | Bytes | SHA-256 |
-| --- | ---: | --- |
-| `construction.json` | 2,175 | `d4c7eced057042438827de168d47b7900311fba742ce52445675be7f40343432` |
-| `inference.json` | 968 | `217636919d17d9332aa49474e893064ed1dfb29f016a468cf22d439cf16887a0` |
-| `weights.safetensors` | 109,294,940 | `2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795` |
-| `conversion/conversion-record.json` | 13,953 | `c2204fdd346b416966d7c14d5746a305e80649a7f381f173c7ad4d9482f3e66d` |
-| `validation/validation-record.json` | 1,946 | `7e018bc98518e898cb723b7c608024e101999a1217e7256a9f3ff6c6da20215e` |
-| `migration/migration-record.json` | 1,518 | `fdabc791f22826bc13622199c73890cb05f89cc7797ae9c62ea5fd40f1f4fd4d` |
-
-`manifest.json` binds these files plus this model card, attribution, and
-license resources by exact size and SHA-256. The immutable Hugging Face commit
-and scitomo learned-weight catalog bind the complete distribution, including
-the manifest and documentation resources, without a self-referential checksum
-inside this README.
-
-## Validation
-
-All 56 source tensors were mapped one-to-one and exactly matched after native
-assignment.
-
-Predetermined CPU float32 checks:
-
-| Case | Tolerance | Result |
-| --- | --- | --- |
-| Synthetic forward parity | `atol=1e-6`, `rtol=1e-5` | bit-exact; max absolute error `0`; relative L2 `0` |
-| Real tutorial-volume crop | relative L2 `<=1e-4` | max absolute error `7.152557373046875e-7`; relative L2 `1.2612566990810592e-7` |
-
-The real-data case used a centered `32 x 32 x 32` crop from
-`tutorial_data/tomo_even_frames.rec`. Vendor and native outputs were finite,
-had identical shapes, and had the same absolute-peak spatial landmark at
-`(Z, Y, X) = (25, 13, 0)`.
-
-The evidence proves native network-state and reviewed forward parity for the
-frozen inputs. It does not establish accuracy on every microscope, specimen,
-acquisition protocol, missing-wedge angle, or preprocessing pipeline, nor does
-it replace validation of a complete user workflow.
-
-## Safe loading
-
-The repository contains declared data files only. Loading does not execute
-remote code, import the vendor project, or use Python pickle. scitomo requires
-the exact catalog commit and verifies every declared size and SHA-256 before
-opening `weights.safetensors`. Hugging Face `trust_remote_code` is never used.
-
-Install the learned and catalog extras before resolving the catalog package:
-
-```text
-pip install "scitomo[learned,catalog]"
+# DeepDeWedge tutorial checkpoint — fresh Scitomo FORMAT 2 package
+
+This is a fresh FORMAT 2 export from the authoritative original Lightning
+checkpoint, not a migration of any earlier Hugging Face package. Normal runtime
+uses Scitomo's generic FORMAT 2 loader and Safetensors only; it does not require
+PyTorch Lightning or the upstream DeepDeWedge source checkout.
+
+## Package identity
+
+- package id: `deepdewedge_tutorial`; package revision: `3`
+- learned-checkpoint format: `2`; manifest schema: `4`
+- Scitomo conversion checkout: `2832957f69daff0d7baec5df17a7c54954623eed`
+- minimum Scitomo version: `0.7.3`
+- previous Hugging Face commit: `87db06570dd874a99af1289e62b79ea99f87f006` — **HISTORICAL ONLY; NOT CONVERSION INPUT**
+
+## Authoritative provenance
+
+- upstream repository: <https://github.com/MLI-lab/DeepDeWedge>
+- upstream revision: `072075692a44a8f17394214369e6e762abe52bc3`
+- Figshare DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>; file id: `45582309`
+- original archive SHA-256: `7c871342e51f5a66a773fe427d72944b5d2cc8ff41c5b7415ab38dbfc9ac6d58`
+- original checkpoint member: `tutorial_data/fitted_model.ckpt`
+- original checkpoint size: `327952642` bytes
+- original checkpoint SHA-256: `5262f6c11e85fd662b02e59efe936fa7b69913758e841235be2683f7bd03ec76`
+
+DeepDeWedge Tutorial Data is attributed to Simon Wiedemann and is distributed
+under CC BY 4.0. The pinned DeepDeWedge implementation is BSD-2-Clause; its
+license text is included below `LICENSES/`. See `ATTRIBUTION.md`.
+
+## Scientific inference semantics
+
+The pure persisted Network owns only the lowered U-Net architecture and its 54
+canonical tensors. The fitted affine values remain outside Network state in the
+typed `deepdewedge_inference` profile:
+
+- `network_affine_loc`: `-0.14898751676082611`
+- `network_affine_scale`: `1.3237642049789429`
+- input layout: `(..., Z, Y, X)`; Network layout: `(..., C, Z, Y, X)`
+- paired halves are refined independently then averaged; full-width missing wedge: 50 degrees
+- 96³ patches, 32³ overlap, trailing-reflection coverage, linear-ramp reassembly
+- preconditioning recomputes patch statistics; output uses the checkpoint-fitted affine
+
+## Fresh conversion and validation
+
+`refresh_format2.py` is the exact one-off implementation and records
+the verified source, explicit 54-tensor mapping, strict Network lowering, and
+generic export. It was run with Python `3.12.13`, Torch
+`2.12.1+cpu`, Lightning `2.6.5`, Safetensors
+`0.8.0`, and Scitomo `0.7.3.dev0` on `Windows-11-10.0.22631-SP0`.
+
+The generic exporter freshly serializes `weights.safetensors`; no previous
+Hugging Face Safetensors, manifest, construction, or inference record is read.
+The conversion record lists every source checkpoint tensor to canonical target
+mapping. The validation record binds package state closure, generic loader
+reload, external-affine semantics, and deterministic forward parity.
+
+For a deterministic directional, non-symmetric CPU float32 input of 4,096 elements,
+authoritative upstream output versus FORMAT 2 pure-Network-plus-profile output
+passed `rtol=1e-5`, `atol=1e-6`: maximum absolute error
+`0`, relative L2 error `0`.
+
+## Files and closure
+
+`manifest.json` is the authoritative, closed inventory of every package file,
+with each fresh size and SHA-256. It declares only FORMAT 2 construction,
+inference, Safetensors, conversion, validation, and documentation/license
+resources; there is no format-1 or migration artifact. Validate and load with:
+
+```python
+import scitomo as st
+loaded = st.api.load_learned_network("/path/to/package")
 ```
 
-Normal runtime loading is owned by scitomo's central learned-checkpoint API.
-Do not load the original Lightning checkpoint in an ordinary runtime.
-
-## Licenses
-
-- Converted weights and their source tutorial dataset: CC BY 4.0. See
-  `LICENSES/DeepDeWedge-Tutorial-Data-CC-BY-4.0.txt`.
-- Upstream DeepDeWedge code and behavior used for construction/conversion:
-  BSD 2-Clause. See `LICENSES/DeepDeWedge-Code-BSD-2-Clause.txt`.
-
-The CC BY 4.0 attribution and modification notice are provided in
-`ATTRIBUTION.md`. No endorsement by the original authors or rights holders is
-implied.
+This operation uses the generic Scitomo FORMAT 2 loader and does not import
+Lightning or DeepDeWedge. It is a checkpoint package, not a claim of scientific
+approval for a new dataset or acquisition protocol.

construction.json

@@ -1,92 +1,1726 @@
 {
-  "fingerprint": "sha256:c9da7decd489f0f4f893dce565f37db25abc63d0d3efcacad9835305ffd97cb2",
+  "fingerprint": "sha256:644b162de70d014f9273c992fd583ee4fdfb8ae425a35fba26b3fbd41d560f28",
   "kind": "scitomo_network_construction",
-  "schema_version": 1,
-  "spec": {
-    "bottleneck_block": {
-      "activate_last": false,
-      "activation": {
+  "program": {
+    "inputs": [
+      {
+        "channels": 1,
+        "leading_dimensions": "flatten_to_batch",
+        "minimum_spatial_shape": [
+          8,
+          8,
+          8
+        ],
+        "name": "input",
+        "spatial_divisibility": [
+          8,
+          8,
+          8
+        ],
+        "spatial_rank": 3
+      }
+    ],
+    "kind": "scitomo_network_program",
+    "modules": [
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_0_conv_0",
+        "in_channels": 1,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "down_0_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_0_conv_1",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "down_0_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_0_conv_2",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "down_0_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "downsample_0",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_1_conv_0",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "down_1_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_1_conv_1",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "down_1_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_1_conv_2",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "down_1_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "downsample_1",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_2_conv_0",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "down_2_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_2_conv_1",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "down_2_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "down_2_conv_2",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "down_2_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "downsample_2",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "bottleneck_conv_0",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 512,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "bottleneck_conv_1",
+        "in_channels": 512,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "upsample_0",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv_transpose3d",
+        "out_channels": 256,
+        "output_padding": [
+          1,
+          1,
+          1
+        ],
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_0_conv_0",
+        "in_channels": 512,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "up_0_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_0_conv_1",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "up_0_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_0_conv_2",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 256,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 256,
+        "eps": 1e-05,
+        "id": "up_0_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "upsample_1",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv_transpose3d",
+        "out_channels": 128,
+        "output_padding": [
+          1,
+          1,
+          1
+        ],
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_1_conv_0",
+        "in_channels": 256,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "up_1_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_1_conv_1",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "up_1_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_1_conv_2",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 128,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 128,
+        "eps": 1e-05,
+        "id": "up_1_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "upsample_2",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv_transpose3d",
+        "out_channels": 64,
+        "output_padding": [
+          1,
+          1,
+          1
+        ],
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "primitive_version": 1,
+        "stride": [
+          2,
+          2,
+          2
+        ]
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_2_conv_0",
+        "in_channels": 128,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "up_2_norm_0",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_2_conv_1",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "up_2_norm_1",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "up_2_conv_2",
+        "in_channels": 64,
+        "kernel_size": [
+          3,
+          3,
+          3
+        ],
+        "kind": "conv3d",
+        "out_channels": 64,
+        "padding": [
+          1,
+          1,
+          1
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      },
+      {
+        "affine": false,
+        "channels": 64,
+        "eps": 1e-05,
+        "id": "up_2_norm_2",
+        "kind": "instance_norm3d",
+        "primitive_version": 1,
+        "track_running_stats": false
+      },
+      {
+        "bias": true,
+        "dilation": [
+          1,
+          1,
+          1
+        ],
+        "groups": 1,
+        "id": "final_conv",
+        "in_channels": 64,
+        "kernel_size": [
+          1,
+          1,
+          1
+        ],
+        "kind": "conv3d",
+        "out_channels": 1,
+        "padding": [
+          0,
+          0,
+          0
+        ],
+        "padding_mode": "zeros",
+        "primitive_version": 1,
+        "stride": [
+          1,
+          1,
+          1
+        ]
+      }
+    ],
+    "operations": [
+      {
+        "kind": "call_module",
+        "module": "down_0_conv_0",
+        "output": "down_0_conv_0_out",
+        "primitive_version": 1,
+        "value": "input"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_0_norm_0",
+        "output": "down_0_norm_0_out",
+        "primitive_version": 1,
+        "value": "down_0_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_0_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_0_norm_0_out"
+      },
+      {
         "inplace": true,
         "kind": "leaky_relu",
-        "negative_slope": 0.05
+        "negative_slope": 0.05,
+        "output": "down_0_activation_0_out",
+        "primitive_version": 1,
+        "value": "down_0_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_0_conv_1",
+        "output": "down_0_conv_1_out",
+        "primitive_version": 1,
+        "value": "down_0_activation_0_out"
       },
-      "bias": true,
-      "dropout": 0.0,
-      "kernel_size": 3,
-      "normalization": {
-        "kind": "none"
+      {
+        "kind": "call_module",
+        "module": "down_0_norm_1",
+        "output": "down_0_norm_1_out",
+        "primitive_version": 1,
+        "value": "down_0_conv_1_out"
       },
-      "padding": 1,
-      "repetitions": 2
-    },
-    "decoder_block": {
-      "activate_last": true,
-      "activation": {
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_0_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_0_norm_1_out"
+      },
+      {
         "inplace": true,
         "kind": "leaky_relu",
-        "negative_slope": 0.05
+        "negative_slope": 0.05,
+        "output": "down_0_activation_1_out",
+        "primitive_version": 1,
+        "value": "down_0_dropout_1_out"
       },
-      "bias": true,
-      "dropout": 0.0,
-      "kernel_size": 3,
-      "normalization": {
-        "affine": false,
-        "eps": 1e-05,
-        "kind": "instance",
-        "track_running_stats": false
+      {
+        "kind": "call_module",
+        "module": "down_0_conv_2",
+        "output": "down_0_conv_2_out",
+        "primitive_version": 1,
+        "value": "down_0_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_0_norm_2",
+        "output": "down_0_norm_2_out",
+        "primitive_version": 1,
+        "value": "down_0_conv_2_out"
       },
-      "padding": 1,
-      "repetitions": 3
-    },
-    "downsampling": {
-      "activation": {
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_0_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_0_norm_2_out"
+      },
+      {
         "inplace": true,
         "kind": "leaky_relu",
-        "negative_slope": 0.05
+        "negative_slope": 0.05,
+        "output": "down_0_activation_2_out",
+        "primitive_version": 1,
+        "value": "down_0_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "downsample_0",
+        "output": "downsample_0_out",
+        "primitive_version": 1,
+        "value": "down_0_activation_2_out"
       },
-      "kernel_size": 3,
-      "kind": "strided_convolution",
-      "padding": 1,
-      "stride": 2
-    },
-    "encoder_block": {
-      "activate_last": true,
-      "activation": {
+      {
         "inplace": true,
         "kind": "leaky_relu",
-        "negative_slope": 0.05
+        "negative_slope": 0.05,
+        "output": "downsample_0_activation_out",
+        "primitive_version": 1,
+        "value": "downsample_0_out"
       },
-      "bias": true,
-      "dropout": 0.0,
-      "kernel_size": 3,
-      "normalization": {
-        "affine": false,
-        "eps": 1e-05,
-        "kind": "instance",
-        "track_running_stats": false
+      {
+        "kind": "call_module",
+        "module": "down_1_conv_0",
+        "output": "down_1_conv_0_out",
+        "primitive_version": 1,
+        "value": "downsample_0_activation_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_1_norm_0",
+        "output": "down_1_norm_0_out",
+        "primitive_version": 1,
+        "value": "down_1_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_1_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_1_norm_0_out"
       },
-      "padding": 1,
-      "repetitions": 3
-    },
-    "in_channels": 1,
-    "initial_channels": 64,
-    "initial_normalization_loc": -0.1489875167608261,
-    "initial_normalization_scale": 1.3237642049789429,
-    "kind": "unet_3d",
-    "num_downsampling_blocks": 3,
-    "out_channels": 1,
-    "residual": true,
-    "upsampling": {
-      "kernel_size": 3,
-      "kind": "transposed_convolution",
-      "output_padding": 1,
-      "padding": 1,
-      "post_concat_activation": {
+      {
         "inplace": true,
         "kind": "leaky_relu",
-        "negative_slope": 0.05
+        "negative_slope": 0.05,
+        "output": "down_1_activation_0_out",
+        "primitive_version": 1,
+        "value": "down_1_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_1_conv_1",
+        "output": "down_1_conv_1_out",
+        "primitive_version": 1,
+        "value": "down_1_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_1_norm_1",
+        "output": "down_1_norm_1_out",
+        "primitive_version": 1,
+        "value": "down_1_conv_1_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_1_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_1_norm_1_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "down_1_activation_1_out",
+        "primitive_version": 1,
+        "value": "down_1_dropout_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_1_conv_2",
+        "output": "down_1_conv_2_out",
+        "primitive_version": 1,
+        "value": "down_1_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_1_norm_2",
+        "output": "down_1_norm_2_out",
+        "primitive_version": 1,
+        "value": "down_1_conv_2_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_1_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_1_norm_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "down_1_activation_2_out",
+        "primitive_version": 1,
+        "value": "down_1_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "downsample_1",
+        "output": "downsample_1_out",
+        "primitive_version": 1,
+        "value": "down_1_activation_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "downsample_1_activation_out",
+        "primitive_version": 1,
+        "value": "downsample_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_conv_0",
+        "output": "down_2_conv_0_out",
+        "primitive_version": 1,
+        "value": "downsample_1_activation_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_norm_0",
+        "output": "down_2_norm_0_out",
+        "primitive_version": 1,
+        "value": "down_2_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_2_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_2_norm_0_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "down_2_activation_0_out",
+        "primitive_version": 1,
+        "value": "down_2_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_conv_1",
+        "output": "down_2_conv_1_out",
+        "primitive_version": 1,
+        "value": "down_2_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_norm_1",
+        "output": "down_2_norm_1_out",
+        "primitive_version": 1,
+        "value": "down_2_conv_1_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_2_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_2_norm_1_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "down_2_activation_1_out",
+        "primitive_version": 1,
+        "value": "down_2_dropout_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_conv_2",
+        "output": "down_2_conv_2_out",
+        "primitive_version": 1,
+        "value": "down_2_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "down_2_norm_2",
+        "output": "down_2_norm_2_out",
+        "primitive_version": 1,
+        "value": "down_2_conv_2_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "down_2_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "down_2_norm_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "down_2_activation_2_out",
+        "primitive_version": 1,
+        "value": "down_2_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "downsample_2",
+        "output": "downsample_2_out",
+        "primitive_version": 1,
+        "value": "down_2_activation_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "downsample_2_activation_out",
+        "primitive_version": 1,
+        "value": "downsample_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "bottleneck_conv_0",
+        "output": "bottleneck_conv_0_out",
+        "primitive_version": 1,
+        "value": "downsample_2_activation_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "bottleneck_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "bottleneck_conv_0_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "bottleneck_activation_0_out",
+        "primitive_version": 1,
+        "value": "bottleneck_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "bottleneck_conv_1",
+        "output": "bottleneck_conv_1_out",
+        "primitive_version": 1,
+        "value": "bottleneck_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "upsample_0",
+        "output": "upsample_0_out",
+        "primitive_version": 1,
+        "value": "bottleneck_conv_1_out"
+      },
+      {
+        "dimension": 1,
+        "kind": "concat",
+        "output": "upsample_0_concat",
+        "primitive_version": 1,
+        "values": [
+          "upsample_0_out",
+          "down_2_activation_2_out"
+        ]
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "upsample_0_activation_out",
+        "primitive_version": 1,
+        "value": "upsample_0_concat"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_conv_0",
+        "output": "up_0_conv_0_out",
+        "primitive_version": 1,
+        "value": "upsample_0_activation_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_norm_0",
+        "output": "up_0_norm_0_out",
+        "primitive_version": 1,
+        "value": "up_0_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_0_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_0_norm_0_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_0_activation_0_out",
+        "primitive_version": 1,
+        "value": "up_0_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_conv_1",
+        "output": "up_0_conv_1_out",
+        "primitive_version": 1,
+        "value": "up_0_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_norm_1",
+        "output": "up_0_norm_1_out",
+        "primitive_version": 1,
+        "value": "up_0_conv_1_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_0_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_0_norm_1_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_0_activation_1_out",
+        "primitive_version": 1,
+        "value": "up_0_dropout_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_conv_2",
+        "output": "up_0_conv_2_out",
+        "primitive_version": 1,
+        "value": "up_0_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_0_norm_2",
+        "output": "up_0_norm_2_out",
+        "primitive_version": 1,
+        "value": "up_0_conv_2_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_0_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_0_norm_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_0_activation_2_out",
+        "primitive_version": 1,
+        "value": "up_0_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "upsample_1",
+        "output": "upsample_1_out",
+        "primitive_version": 1,
+        "value": "up_0_activation_2_out"
+      },
+      {
+        "dimension": 1,
+        "kind": "concat",
+        "output": "upsample_1_concat",
+        "primitive_version": 1,
+        "values": [
+          "upsample_1_out",
+          "down_1_activation_2_out"
+        ]
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "upsample_1_activation_out",
+        "primitive_version": 1,
+        "value": "upsample_1_concat"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_conv_0",
+        "output": "up_1_conv_0_out",
+        "primitive_version": 1,
+        "value": "upsample_1_activation_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_norm_0",
+        "output": "up_1_norm_0_out",
+        "primitive_version": 1,
+        "value": "up_1_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_1_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_1_norm_0_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_1_activation_0_out",
+        "primitive_version": 1,
+        "value": "up_1_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_conv_1",
+        "output": "up_1_conv_1_out",
+        "primitive_version": 1,
+        "value": "up_1_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_norm_1",
+        "output": "up_1_norm_1_out",
+        "primitive_version": 1,
+        "value": "up_1_conv_1_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_1_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_1_norm_1_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_1_activation_1_out",
+        "primitive_version": 1,
+        "value": "up_1_dropout_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_conv_2",
+        "output": "up_1_conv_2_out",
+        "primitive_version": 1,
+        "value": "up_1_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_1_norm_2",
+        "output": "up_1_norm_2_out",
+        "primitive_version": 1,
+        "value": "up_1_conv_2_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_1_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_1_norm_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_1_activation_2_out",
+        "primitive_version": 1,
+        "value": "up_1_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "upsample_2",
+        "output": "upsample_2_out",
+        "primitive_version": 1,
+        "value": "up_1_activation_2_out"
+      },
+      {
+        "dimension": 1,
+        "kind": "concat",
+        "output": "upsample_2_concat",
+        "primitive_version": 1,
+        "values": [
+          "upsample_2_out",
+          "down_0_activation_2_out"
+        ]
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "upsample_2_activation_out",
+        "primitive_version": 1,
+        "value": "upsample_2_concat"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_conv_0",
+        "output": "up_2_conv_0_out",
+        "primitive_version": 1,
+        "value": "upsample_2_activation_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_norm_0",
+        "output": "up_2_norm_0_out",
+        "primitive_version": 1,
+        "value": "up_2_conv_0_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_2_dropout_0_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_2_norm_0_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_2_activation_0_out",
+        "primitive_version": 1,
+        "value": "up_2_dropout_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_conv_1",
+        "output": "up_2_conv_1_out",
+        "primitive_version": 1,
+        "value": "up_2_activation_0_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_norm_1",
+        "output": "up_2_norm_1_out",
+        "primitive_version": 1,
+        "value": "up_2_conv_1_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_2_dropout_1_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_2_norm_1_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_2_activation_1_out",
+        "primitive_version": 1,
+        "value": "up_2_dropout_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_conv_2",
+        "output": "up_2_conv_2_out",
+        "primitive_version": 1,
+        "value": "up_2_activation_1_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "up_2_norm_2",
+        "output": "up_2_norm_2_out",
+        "primitive_version": 1,
+        "value": "up_2_conv_2_out"
+      },
+      {
+        "inplace": false,
+        "kind": "dropout3d",
+        "output": "up_2_dropout_2_out",
+        "primitive_version": 1,
+        "probability": 0.0,
+        "value": "up_2_norm_2_out"
+      },
+      {
+        "inplace": true,
+        "kind": "leaky_relu",
+        "negative_slope": 0.05,
+        "output": "up_2_activation_2_out",
+        "primitive_version": 1,
+        "value": "up_2_dropout_2_out"
+      },
+      {
+        "kind": "call_module",
+        "module": "final_conv",
+        "output": "final_conv_out",
+        "primitive_version": 1,
+        "value": "up_2_activation_2_out"
       },
-      "stride": 2
-    }
-  }
+      {
+        "kind": "add",
+        "left": "final_conv_out",
+        "output": "residual_output",
+        "primitive_version": 1,
+        "right": "input"
+      }
+    ],
+    "outputs": [
+      {
+        "channels": 1,
+        "name": "output",
+        "source": "residual_output"
+      }
+    ],
+    "schema_version": 1
+  },
+  "schema_version": 2
 }

conversion/conversion-record.json

@@ -1,491 +1,468 @@
 {
-  "construction_fingerprint": "sha256:c9da7decd489f0f4f893dce565f37db25abc63d0d3efcacad9835305ffd97cb2",
+  "construction_fingerprint": "sha256:644b162de70d014f9273c992fd583ee4fdfb8ae425a35fba26b3fbd41d560f28",
   "environment": {
-    "mapping_sha256": "bf23d54a8a79a8086b78a8c807a787f1a5ac5525ef7d6eec67373aea4f0a4cb5",
-    "trusted": {
-      "python": "3.12.11",
-      "safetensors": "0.8.0",
-      "scitomo": "0.5.4",
-      "torch": "2.13.0+cu130"
-    },
-    "unsafe": {
-      "platform": "Linux-6.16.12-valve24.5-1-neptune-616-gb2f7cfe85e45-x86_64-with-glibc2.41",
-      "python": "3.10.16",
-      "pytorch_lightning": "1.8.0.post1",
-      "safetensors": "0.5.3",
-      "torch": "2.2.0+cpu"
-    }
+    "platform": "Windows-11-10.0.22631-SP0",
+    "python": "3.12.13",
+    "pytorch_lightning": "2.6.5",
+    "safetensors": "0.8.0",
+    "scitomo": "0.7.3.dev0",
+    "scitomo_commit": "2832957f69daff0d7baec5df17a7c54954623eed",
+    "torch": "2.12.1+cpu"
   },
-  "inference_fingerprint": "sha256:c2c5eba72579510265258dcfce3a5be2f851f9276c7e53d92330d78eb37720dd",
+  "inference_fingerprint": "sha256:dd4d9b190f9e3a0c94a0da2d383f5dc42aaa0ac888dcec760d57b4cf5362aabd",
   "kind": "scitomo_checkpoint_conversion",
   "package_id": "deepdewedge_tutorial",
-  "package_revision": 2,
-  "schema_version": 1,
+  "package_revision": 3,
+  "schema_version": 2,
   "source": {
-    "identifier": "10.6084/m9.figshare.25043435.v1/file/45582309/tutorial_data/fitted_model.ckpt",
-    "kind": "figshare_archive_member",
+    "identifier": "45582309/tutorial_data/fitted_model.ckpt",
+    "kind": "figshare_checkpoint",
     "metadata": {
+      "archive_member": "tutorial_data/fitted_model.ckpt",
       "archive_sha256": "7c871342e51f5a66a773fe427d72944b5d2cc8ff41c5b7415ab38dbfc9ac6d58",
-      "archive_size_bytes": 1947310911,
       "checkpoint_size_bytes": 327952642,
-      "figshare_file_id": 45582309
+      "upstream_repository": "https://github.com/MLI-lab/DeepDeWedge"
     },
     "project": "DeepDeWedge Tutorial Data",
-    "revision": "1",
+    "revision": "072075692a44a8f17394214369e6e762abe52bc3",
     "sha256": "5262f6c11e85fd662b02e59efe936fa7b69913758e841235be2683f7bd03ec76",
-    "url": "https://ndownloader.figshare.com/files/45582309"
+    "url": "https://doi.org/10.6084/m9.figshare.25043435.v1"
   },
-  "tool": "deepdewedge_c4_isolated_converter",
-  "tool_version": "1",
-  "transformations": [
-    {
-      "details": {
-        "state_role": "parameter_to_buffer"
-      },
-      "kind": "tensor_name_mapping",
-      "source": "unet._normalization_loc",
-      "target": "_normalization_loc"
-    },
+  "tensor_mappings": [
     {
       "details": {
-        "state_role": "parameter_to_buffer"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet._normalization_scale",
-      "target": "_normalization_scale"
+      "kind": "identity",
+      "source": "state_dict.unet.bottleneck.0.bias",
+      "target": "bottleneck_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.0.weight",
-      "target": "down_blocks.0.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.bottleneck.0.weight",
+      "target": "bottleneck_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.0.bias",
-      "target": "down_blocks.0.layers.0.bias"
+      "kind": "rename",
+      "source": "state_dict.unet.bottleneck.2.bias",
+      "target": "bottleneck_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.4.weight",
-      "target": "down_blocks.0.layers.4.weight"
+      "kind": "rename",
+      "source": "state_dict.unet.bottleneck.2.weight",
+      "target": "bottleneck_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.4.bias",
-      "target": "down_blocks.0.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.0.bias",
+      "target": "down_0_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.8.weight",
-      "target": "down_blocks.0.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.0.weight",
+      "target": "down_0_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.0.layers.8.bias",
-      "target": "down_blocks.0.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.4.bias",
+      "target": "down_0_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.0.weight",
-      "target": "down_blocks.1.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.4.weight",
+      "target": "down_0_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.0.bias",
-      "target": "down_blocks.1.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.8.bias",
+      "target": "down_0_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.4.weight",
-      "target": "down_blocks.1.layers.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.0.layers.8.weight",
+      "target": "down_0_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.4.bias",
-      "target": "down_blocks.1.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.0.bias",
+      "target": "down_1_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.8.weight",
-      "target": "down_blocks.1.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.0.weight",
+      "target": "down_1_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.1.layers.8.bias",
-      "target": "down_blocks.1.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.4.bias",
+      "target": "down_1_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.0.weight",
-      "target": "down_blocks.2.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.4.weight",
+      "target": "down_1_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.0.bias",
-      "target": "down_blocks.2.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.8.bias",
+      "target": "down_1_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.4.weight",
-      "target": "down_blocks.2.layers.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.1.layers.8.weight",
+      "target": "down_1_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.4.bias",
-      "target": "down_blocks.2.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.0.bias",
+      "target": "down_2_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.8.weight",
-      "target": "down_blocks.2.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.0.weight",
+      "target": "down_2_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_blocks.2.layers.8.bias",
-      "target": "down_blocks.2.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.4.bias",
+      "target": "down_2_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.0.layers.0.weight",
-      "target": "down_samplers.0.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.4.weight",
+      "target": "down_2_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.0.layers.0.bias",
-      "target": "down_samplers.0.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.8.bias",
+      "target": "down_2_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.1.layers.0.weight",
-      "target": "down_samplers.1.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_blocks.2.layers.8.weight",
+      "target": "down_2_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.1.layers.0.bias",
-      "target": "down_samplers.1.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.0.layers.0.bias",
+      "target": "downsample_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.2.layers.0.weight",
-      "target": "down_samplers.2.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.0.layers.0.weight",
+      "target": "downsample_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.down_samplers.2.layers.0.bias",
-      "target": "down_samplers.2.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.1.layers.0.bias",
+      "target": "downsample_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.bottleneck.0.weight",
-      "target": "bottleneck.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.1.layers.0.weight",
+      "target": "downsample_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.bottleneck.0.bias",
-      "target": "bottleneck.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.2.layers.0.bias",
+      "target": "downsample_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.bottleneck.2.weight",
-      "target": "bottleneck.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.down_samplers.2.layers.0.weight",
+      "target": "downsample_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.bottleneck.2.bias",
-      "target": "bottleneck.4.bias"
-    },
-    {
-      "details": {
-        "state_role": "parameter_to_parameter"
-      },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.0.weight",
-      "target": "up_blocks.0.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.final_conv.bias",
+      "target": "final_conv.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.0.bias",
-      "target": "up_blocks.0.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.final_conv.weight",
+      "target": "final_conv.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.4.weight",
-      "target": "up_blocks.0.layers.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.0.bias",
+      "target": "up_0_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.4.bias",
-      "target": "up_blocks.0.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.0.weight",
+      "target": "up_0_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.8.weight",
-      "target": "up_blocks.0.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.4.bias",
+      "target": "up_0_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.0.layers.8.bias",
-      "target": "up_blocks.0.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.4.weight",
+      "target": "up_0_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.0.weight",
-      "target": "up_blocks.1.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.8.bias",
+      "target": "up_0_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.0.bias",
-      "target": "up_blocks.1.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.0.layers.8.weight",
+      "target": "up_0_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.4.weight",
-      "target": "up_blocks.1.layers.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.0.bias",
+      "target": "up_1_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.4.bias",
-      "target": "up_blocks.1.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.0.weight",
+      "target": "up_1_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.8.weight",
-      "target": "up_blocks.1.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.4.bias",
+      "target": "up_1_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.1.layers.8.bias",
-      "target": "up_blocks.1.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.4.weight",
+      "target": "up_1_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.0.weight",
-      "target": "up_blocks.2.layers.0.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.8.bias",
+      "target": "up_1_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.0.bias",
-      "target": "up_blocks.2.layers.0.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.1.layers.8.weight",
+      "target": "up_1_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.4.weight",
-      "target": "up_blocks.2.layers.4.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.0.bias",
+      "target": "up_2_conv_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.4.bias",
-      "target": "up_blocks.2.layers.4.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.0.weight",
+      "target": "up_2_conv_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.8.weight",
-      "target": "up_blocks.2.layers.8.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.4.bias",
+      "target": "up_2_conv_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.up_blocks.2.layers.8.bias",
-      "target": "up_blocks.2.layers.8.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.4.weight",
+      "target": "up_2_conv_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.0.tconv.weight",
-      "target": "upsamplers.0.tconv.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.8.bias",
+      "target": "up_2_conv_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.0.tconv.bias",
-      "target": "upsamplers.0.tconv.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.up_blocks.2.layers.8.weight",
+      "target": "up_2_conv_2.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.1.tconv.weight",
-      "target": "upsamplers.1.tconv.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.0.tconv.bias",
+      "target": "upsample_0.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.1.tconv.bias",
-      "target": "upsamplers.1.tconv.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.0.tconv.weight",
+      "target": "upsample_0.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.2.tconv.weight",
-      "target": "upsamplers.2.tconv.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.1.tconv.bias",
+      "target": "upsample_1.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.upsamplers.2.tconv.bias",
-      "target": "upsamplers.2.tconv.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.1.tconv.weight",
+      "target": "upsample_1.weight"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.final_conv.weight",
-      "target": "final_conv.weight"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.2.tconv.bias",
+      "target": "upsample_2.bias"
     },
     {
       "details": {
-        "state_role": "parameter_to_parameter"
+        "source_checkpoint": "tutorial_data/fitted_model.ckpt"
       },
-      "kind": "tensor_name_mapping",
-      "source": "unet.final_conv.bias",
-      "target": "final_conv.bias"
+      "kind": "identity",
+      "source": "state_dict.unet.upsamplers.2.tconv.weight",
+      "target": "upsample_2.weight"
     }
   ],
-  "weights_sha256": "2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795"
+  "tool": "deepdewedge_refresh_format2",
+  "tool_version": "1",
+  "weights_sha256": "e8f51c499d6e5c14ddce45d5cf9935511220de161ce62deb50a2c3a8cae5c731"
 }

inference.json

@@ -1,34 +1,43 @@
 {
-  "fingerprint": "sha256:c2c5eba72579510265258dcfce3a5be2f851f9276c7e53d92330d78eb37720dd",
+  "fingerprint": "sha256:dd4d9b190f9e3a0c94a0da2d383f5dc42aaa0ac888dcec760d57b4cf5362aabd",
   "kind": "scitomo_inference_profile",
-  "method": {
+  "owner": {
     "family": "restoration",
-    "kind": "deepdewedge"
+    "kind": "method",
+    "method_kind": "deepdewedge"
   },
   "profile": {
-    "channel_semantics": "single_scalar_volume",
-    "coverage_padding": "trailing_reflection",
-    "full_tomogram_standardization": false,
-    "input_layout": "(..., Z, Y, X)",
-    "kind": "deepdewedge_inference",
-    "missing_wedge_full_width_deg": 50.0,
-    "missing_wedge_preconditioning": "fourier_mask_each_half",
-    "network_layout": "(..., C, Z, Y, X)",
-    "normalization_policy": "recompute_patch_statistics",
-    "output_scaling": "network_denormalization",
-    "overlap": [
-      32,
-      32,
-      32
-    ],
-    "pair_combination": "mean",
-    "paired_half_semantics": "refine_each_half_then_mean",
-    "patch_shape": [
-      96,
-      96,
-      96
-    ],
-    "reassembly": "linear_ramp_weighted_mean"
+    "contract": {
+      "channel_semantics": "single_scalar_volume",
+      "coverage_padding": "trailing_reflection",
+      "full_tomogram_standardization": false,
+      "input_layout": "(..., Z, Y, X)",
+      "kind": "deepdewedge_inference_contract",
+      "missing_wedge_full_width_deg": 50.0,
+      "missing_wedge_preconditioning": "fourier_mask_each_half",
+      "network_layout": "(..., C, Z, Y, X)",
+      "output_scaling": "checkpoint_fitted_network_affine",
+      "overlap": [
+        32,
+        32,
+        32
+      ],
+      "pair_combination": "mean",
+      "paired_half_semantics": "refine_each_half_then_mean",
+      "patch_shape": [
+        96,
+        96,
+        96
+      ],
+      "preconditioning_normalization_policy": "recompute_patch_statistics",
+      "reassembly": "linear_ramp_weighted_mean"
+    },
+    "fitted": {
+      "kind": "deepdewedge_fitted_inference",
+      "network_affine_loc": -0.1489875167608261,
+      "network_affine_scale": 1.3237642049789429
+    },
+    "kind": "deepdewedge_inference"
   },
-  "schema_version": 1
+  "schema_version": 3
 }

manifest.json

@@ -1,30 +1,26 @@
 {
+  "construction_fingerprint": "sha256:644b162de70d014f9273c992fd583ee4fdfb8ae425a35fba26b3fbd41d560f28",
   "files": {
     "construction": {
       "path": "construction.json",
-      "sha256": "d4c7eced057042438827de168d47b7900311fba742ce52445675be7f40343432",
-      "size_bytes": 2175
+      "sha256": "46f8b7cfe685f5f2468b43c9f54d96b2b074e9c459fcd28378da3177cfd7cd65",
+      "size_bytes": 37943
     },
     "conversion": {
       "path": "conversion/conversion-record.json",
-      "sha256": "c2204fdd346b416966d7c14d5746a305e80649a7f381f173c7ad4d9482f3e66d",
-      "size_bytes": 13953
+      "sha256": "ca5300cc80f144d8dd9c2d9cf618f2ad942ddc6016e048cda2de2c20da37d81d",
+      "size_bytes": 13810
     },
     "inference": {
       "path": "inference.json",
-      "sha256": "217636919d17d9332aa49474e893064ed1dfb29f016a468cf22d439cf16887a0",
-      "size_bytes": 968
-    },
-    "migration": {
-      "path": "migration/migration-record.json",
-      "sha256": "fdabc791f22826bc13622199c73890cb05f89cc7797ae9c62ea5fd40f1f4fd4d",
-      "size_bytes": 1518
+      "sha256": "52f6e95eb1f88ae13cabcfb1664dc4e4cfb48570bec636af3527513c85863a7e",
+      "size_bytes": 1304
     },
     "resources": [
       {
         "path": "ATTRIBUTION.md",
-        "sha256": "ab224b5dafa8d3aeba4e9be0349ff568da6ce72f79c26c8a68c3358f960a4297",
-        "size_bytes": 2060
+        "sha256": "849e0877d2d572b4ba10aabfabe524af3d76254423c034d9bd015fd3a827f98e",
+        "size_bytes": 1386
       },
       {
         "path": "LICENSES/DeepDeWedge-Code-BSD-2-Clause.txt",
@@ -32,96 +28,70 @@
         "size_bytes": 1343
       },
       {
-        "path": "LICENSES/DeepDeWedge-Tutorial-Data-CC-BY-4.0.txt",
-        "sha256": "d557539df68e771cc1eedcc91d13f70fca930e508d11eedcafa4b15db49e3744",
-        "size_bytes": 17023
+        "path": "README.md",
+        "sha256": "3cffce3987c861bd52a90f5e8c07d16152511b957dc8de4fe79a02da2e270740",
+        "size_bytes": 3903
       },
       {
-        "path": "README.md",
-        "sha256": "ec1795d189581468b93881a6ad0ac1dc95386862000c03a6aa98ab7d09697643",
-        "size_bytes": 7114
+        "path": "refresh_format2.py",
+        "sha256": "b4c42a1cc2611561077578ee0544d19fdee8d3a48426f0dc44e6dcb61a0ce875",
+        "size_bytes": 22531
       }
     ],
     "validation": {
       "path": "validation/validation-record.json",
-      "sha256": "7e018bc98518e898cb723b7c608024e101999a1217e7256a9f3ff6c6da20215e",
-      "size_bytes": 1946
+      "sha256": "bc302bd4e4769044af36867ac66877b022b0e4c2b567551d1a8abf70ff653950",
+      "size_bytes": 1660
     },
     "weights": {
       "path": "weights.safetensors",
-      "sha256": "2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795",
-      "size_bytes": 109294940
+      "sha256": "e8f51c499d6e5c14ddce45d5cf9935511220de161ce62deb50a2c3a8cae5c731",
+      "size_bytes": 109294372
     }
   },
+  "inference_fingerprint": "sha256:dd4d9b190f9e3a0c94a0da2d383f5dc42aaa0ac888dcec760d57b4cf5362aabd",
   "kind": "scitomo_learned_checkpoint",
-  "method": {
+  "owner": {
     "family": "restoration",
-    "kind": "deepdewedge"
+    "kind": "method",
+    "method_kind": "deepdewedge"
   },
   "package_id": "deepdewedge_tutorial",
-  "package_revision": 2,
+  "package_revision": 3,
   "provenance": {
     "citations": [
-      "doi:10.6084/m9.figshare.25043435.v1",
-      "doi:10.1038/s41467-024-51438-y"
-    ],
-    "licenses": [
-      {
-        "identifier": "BSD-2-Clause",
-        "path": "LICENSES/DeepDeWedge-Code-BSD-2-Clause.txt",
-        "redistribution": "approved",
-        "subject": "code"
-      },
-      {
-        "identifier": "CC-BY-4.0",
-        "path": "LICENSES/DeepDeWedge-Tutorial-Data-CC-BY-4.0.txt",
-        "redistribution": "approved",
-        "subject": "weights"
-      }
+      "https://doi.org/10.1038/s41467-024-51438-y",
+      "https://doi.org/10.6084/m9.figshare.25043435.v1"
     ],
+    "licenses": [],
     "sources": [
       {
-        "identifier": "10.6084/m9.figshare.25043435.v1/file/45582309/tutorial_data/fitted_model.ckpt",
-        "kind": "figshare_archive_member",
-        "metadata": {
-          "archive_sha256": "7c871342e51f5a66a773fe427d72944b5d2cc8ff41c5b7415ab38dbfc9ac6d58",
-          "archive_size_bytes": 1947310911,
-          "checkpoint_size_bytes": 327952642,
-          "figshare_file_id": 45582309
-        },
-        "project": "DeepDeWedge Tutorial Data",
-        "revision": "1",
-        "sha256": "5262f6c11e85fd662b02e59efe936fa7b69913758e841235be2683f7bd03ec76",
-        "url": "https://ndownloader.figshare.com/files/45582309"
+        "identifier": "072075692a44a8f17394214369e6e762abe52bc3",
+        "kind": "upstream_repository",
+        "metadata": {},
+        "project": "MLI-lab/DeepDeWedge",
+        "revision": "072075692a44a8f17394214369e6e762abe52bc3",
+        "sha256": null,
+        "url": "https://github.com/MLI-lab/DeepDeWedge"
       }
     ]
   },
   "requirements": {
-    "learned_checkpoint_format": 1,
-    "minimum_scitomo_version": "0.6.2"
+    "learned_checkpoint_format": 2,
+    "minimum_scitomo_version": "0.7.3"
   },
-  "schema_version": 2,
+  "schema_version": 4,
   "tensors": [
     {
       "dtype": "float32",
-      "name": "_normalization_loc",
-      "shape": []
-    },
-    {
-      "dtype": "float32",
-      "name": "_normalization_scale",
-      "shape": []
-    },
-    {
-      "dtype": "float32",
-      "name": "bottleneck.0.bias",
+      "name": "bottleneck_conv_0.bias",
       "shape": [
         512
       ]
     },
     {
       "dtype": "float32",
-      "name": "bottleneck.0.weight",
+      "name": "bottleneck_conv_0.weight",
       "shape": [
         512,
         256,
@@ -132,14 +102,14 @@
     },
     {
       "dtype": "float32",
-      "name": "bottleneck.4.bias",
+      "name": "bottleneck_conv_1.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "bottleneck.4.weight",
+      "name": "bottleneck_conv_1.weight",
       "shape": [
         256,
         512,
@@ -150,14 +120,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.0.bias",
+      "name": "down_0_conv_0.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.0.weight",
+      "name": "down_0_conv_0.weight",
       "shape": [
         64,
         1,
@@ -168,14 +138,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.4.bias",
+      "name": "down_0_conv_1.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.4.weight",
+      "name": "down_0_conv_1.weight",
       "shape": [
         64,
         64,
@@ -186,14 +156,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.8.bias",
+      "name": "down_0_conv_2.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.0.layers.8.weight",
+      "name": "down_0_conv_2.weight",
       "shape": [
         64,
         64,
@@ -204,14 +174,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.0.bias",
+      "name": "down_1_conv_0.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.0.weight",
+      "name": "down_1_conv_0.weight",
       "shape": [
         128,
         64,
@@ -222,14 +192,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.4.bias",
+      "name": "down_1_conv_1.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.4.weight",
+      "name": "down_1_conv_1.weight",
       "shape": [
         128,
         128,
@@ -240,14 +210,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.8.bias",
+      "name": "down_1_conv_2.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.1.layers.8.weight",
+      "name": "down_1_conv_2.weight",
       "shape": [
         128,
         128,
@@ -258,14 +228,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.0.bias",
+      "name": "down_2_conv_0.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.0.weight",
+      "name": "down_2_conv_0.weight",
       "shape": [
         256,
         128,
@@ -276,14 +246,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.4.bias",
+      "name": "down_2_conv_1.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.4.weight",
+      "name": "down_2_conv_1.weight",
       "shape": [
         256,
         256,
@@ -294,14 +264,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.8.bias",
+      "name": "down_2_conv_2.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_blocks.2.layers.8.weight",
+      "name": "down_2_conv_2.weight",
       "shape": [
         256,
         256,
@@ -312,14 +282,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.0.layers.0.bias",
+      "name": "downsample_0.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.0.layers.0.weight",
+      "name": "downsample_0.weight",
       "shape": [
         64,
         64,
@@ -330,14 +300,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.1.layers.0.bias",
+      "name": "downsample_1.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.1.layers.0.weight",
+      "name": "downsample_1.weight",
       "shape": [
         128,
         128,
@@ -348,14 +318,14 @@
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.2.layers.0.bias",
+      "name": "downsample_2.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "down_samplers.2.layers.0.weight",
+      "name": "downsample_2.weight",
       "shape": [
         256,
         256,
@@ -384,14 +354,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.0.bias",
+      "name": "up_0_conv_0.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.0.weight",
+      "name": "up_0_conv_0.weight",
       "shape": [
         256,
         512,
@@ -402,14 +372,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.4.bias",
+      "name": "up_0_conv_1.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.4.weight",
+      "name": "up_0_conv_1.weight",
       "shape": [
         256,
         256,
@@ -420,14 +390,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.8.bias",
+      "name": "up_0_conv_2.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.0.layers.8.weight",
+      "name": "up_0_conv_2.weight",
       "shape": [
         256,
         256,
@@ -438,14 +408,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.0.bias",
+      "name": "up_1_conv_0.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.0.weight",
+      "name": "up_1_conv_0.weight",
       "shape": [
         128,
         256,
@@ -456,14 +426,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.4.bias",
+      "name": "up_1_conv_1.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.4.weight",
+      "name": "up_1_conv_1.weight",
       "shape": [
         128,
         128,
@@ -474,14 +444,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.8.bias",
+      "name": "up_1_conv_2.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.1.layers.8.weight",
+      "name": "up_1_conv_2.weight",
       "shape": [
         128,
         128,
@@ -492,14 +462,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.0.bias",
+      "name": "up_2_conv_0.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.0.weight",
+      "name": "up_2_conv_0.weight",
       "shape": [
         64,
         128,
@@ -510,14 +480,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.4.bias",
+      "name": "up_2_conv_1.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.4.weight",
+      "name": "up_2_conv_1.weight",
       "shape": [
         64,
         64,
@@ -528,14 +498,14 @@
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.8.bias",
+      "name": "up_2_conv_2.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "up_blocks.2.layers.8.weight",
+      "name": "up_2_conv_2.weight",
       "shape": [
         64,
         64,
@@ -546,14 +516,14 @@
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.0.tconv.bias",
+      "name": "upsample_0.bias",
       "shape": [
         256
       ]
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.0.tconv.weight",
+      "name": "upsample_0.weight",
       "shape": [
         256,
         256,
@@ -564,14 +534,14 @@
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.1.tconv.bias",
+      "name": "upsample_1.bias",
       "shape": [
         128
       ]
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.1.tconv.weight",
+      "name": "upsample_1.weight",
       "shape": [
         256,
         128,
@@ -582,14 +552,14 @@
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.2.tconv.bias",
+      "name": "upsample_2.bias",
       "shape": [
         64
       ]
     },
     {
       "dtype": "float32",
-      "name": "upsamplers.2.tconv.weight",
+      "name": "upsample_2.weight",
       "shape": [
         128,
         64,
@@ -599,12 +569,5 @@
       ]
     }
   ],
-  "training": {
-    "base_seed": null,
-    "epoch": 999,
-    "monitor": null,
-    "role": "external",
-    "score": null,
-    "step": 34000
-  }
+  "training": null
 }

migration/migration-record.json

@@ -1,45 +0,0 @@
-{
-  "kind": "scitomo_checkpoint_migration",
-  "minimum_scitomo_version_after": "0.6.2",
-  "minimum_scitomo_version_before": "0.5.4",
-  "package_id": "deepdewedge_tutorial",
-  "schema_version": 1,
-  "source_manifest": {
-    "path": "manifest.json",
-    "sha256": "c324a8980558af597c37da2dd053c8e5adb38895fe3ff00b300d727581496d3d",
-    "size_bytes": 11159
-  },
-  "source_manifest_schema_version": 1,
-  "source_package_revision": 1,
-  "source_provider": {
-    "identifier": "5e00643b4adb4fb67d52ef8f101507b47f4c06c1",
-    "kind": "huggingface_manifest",
-    "metadata": {
-      "manifest_schema_version": 1,
-      "package_revision": 1
-    },
-    "project": "scitomo/deepdewedge-tutorial",
-    "revision": "5e00643b4adb4fb67d52ef8f101507b47f4c06c1",
-    "sha256": "c324a8980558af597c37da2dd053c8e5adb38895fe3ff00b300d727581496d3d",
-    "url": "https://huggingface.co/scitomo/deepdewedge-tutorial/tree/5e00643b4adb4fb67d52ef8f101507b47f4c06c1"
-  },
-  "target_manifest_schema_version": 2,
-  "target_package_revision": 2,
-  "unchanged_files": [
-    {
-      "path": "construction.json",
-      "sha256": "d4c7eced057042438827de168d47b7900311fba742ce52445675be7f40343432",
-      "size_bytes": 2175
-    },
-    {
-      "path": "inference.json",
-      "sha256": "217636919d17d9332aa49474e893064ed1dfb29f016a468cf22d439cf16887a0",
-      "size_bytes": 968
-    },
-    {
-      "path": "weights.safetensors",
-      "sha256": "2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795",
-      "size_bytes": 109294940
-    }
-  ]
-}

refresh_format2.py

@@ -0,0 +1,538 @@
+"""Regenerate the DeepDeWedge FORMAT 2 package from its authoritative source.
+
+This one-off maintenance converter is a package resource, not Scitomo runtime
+code.  It intentionally refuses all historical Hugging Face payloads.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import platform
+import shutil
+import subprocess
+import sys
+import types
+from pathlib import Path
+
+import pytorch_lightning
+import safetensors
+import torch
+
+import scitomo as st
+from scitomo.methods.restoration.deepdewedge.network_invocation import (
+    invoke_deepdewedge_network,
+)
+
+
+SCRIPT_PATH = Path(__file__).resolve()
+ROOT = (
+    SCRIPT_PATH.parents[2]
+    if SCRIPT_PATH.parent.name == "conversion"
+    else SCRIPT_PATH.parents[1]
+)
+UPSTREAM = ROOT / "upstream"
+CHECKPOINT = ROOT / "official" / "fitted_model.ckpt"
+ARCHIVE = ROOT / "official" / "tutorial_data.zip"
+TARGET = ROOT / "hf"
+OUTPUT = ROOT / "package-fresh"
+RUNTIME_VIEW = ROOT / "package-runtime-view"
+
+UPSTREAM_REVISION = "072075692a44a8f17394214369e6e762abe52bc3"
+CHECKPOINT_SIZE = 327952642
+CHECKPOINT_SHA256 = "5262f6c11e85fd662b02e59efe936fa7b69913758e841235be2683f7bd03ec76"
+ARCHIVE_SHA256 = "7c871342e51f5a66a773fe427d72944b5d2cc8ff41c5b7415ab38dbfc9ac6d58"
+ARCHIVE_MD5 = "130264af7d96be6237351f8f51eda8c8"
+PREVIOUS_HF_COMMIT = "87db06570dd874a99af1289e62b79ea99f87f006"
+
+
+def _digest(path: Path, algorithm: str) -> str:
+    hasher = hashlib.new(algorithm)
+    with path.open("rb") as stream:
+        for block in iter(lambda: stream.read(1024 * 1024), b""):
+            hasher.update(block)
+    return hasher.hexdigest()
+
+
+def _verify_authority() -> None:
+    """Fail closed before the sole permitted Lightning deserialization."""
+
+    if CHECKPOINT.stat().st_size != CHECKPOINT_SIZE:
+        raise RuntimeError("Authoritative checkpoint byte size does not match.")
+    if _digest(CHECKPOINT, "sha256") != CHECKPOINT_SHA256:
+        raise RuntimeError("Authoritative checkpoint SHA-256 does not match.")
+    if _digest(ARCHIVE, "sha256") != ARCHIVE_SHA256:
+        raise RuntimeError("Authoritative archive SHA-256 does not match.")
+    if _digest(ARCHIVE, "md5") != ARCHIVE_MD5:
+        raise RuntimeError("Authoritative archive MD5 does not match.")
+    revision = subprocess.check_output(
+        ["git", "-C", str(UPSTREAM), "rev-parse", "HEAD"], text=True
+    ).strip()
+    if revision != UPSTREAM_REVISION:
+        raise RuntimeError("Pinned upstream checkout revision does not match.")
+
+
+def _load_upstream_model() -> torch.nn.Module:
+    """Load the exact pinned model without executing its unrelated CLI package init."""
+
+    ddw = types.ModuleType("ddw")
+    ddw.__path__ = [str(UPSTREAM / "ddw")]
+    sys.modules["ddw"] = ddw
+    utils = types.ModuleType("ddw.utils")
+    utils.__path__ = [str(UPSTREAM / "ddw" / "utils")]
+    sys.modules["ddw.utils"] = utils
+
+    # ``unet.py`` imports this training-only helper but conversion never calls it.
+    # Providing the inert name avoids importing the upstream CLI-only dependency
+    # chain (typer) while preserving the exact source model implementation.
+    normalization = types.ModuleType("ddw.utils.normalization")
+    normalization.get_avg_model_input_mean_and_std_from_dataloader = _unavailable
+    sys.modules["ddw.utils.normalization"] = normalization
+
+    from ddw.utils.unet import LitUnet3D
+
+    return LitUnet3D.load_from_checkpoint(CHECKPOINT, map_location="cpu").eval()
+
+
+def _unavailable(*args: object, **kwargs: object) -> None:
+    del args, kwargs
+    raise RuntimeError("Training-only upstream normalization is unavailable here.")
+
+
+def _canonical_source_to_target(
+    source: torch.nn.Module,
+) -> tuple[
+    st.network.ClosedDescribedNetwork,
+    dict[str, torch.Tensor],
+    tuple[st.artifacts.LearnedCheckpointTransformation, ...],
+    st.methods.DeepDeWedgeFittedInference,
+]:
+    """Instantiate, strict-map, and lower the source model through Network authority."""
+
+    params = dict(source.unet_params)
+    expected = {
+        "chans": 64,
+        "num_downsample_layers": 3,
+        "drop_prob": 0.0,
+    }
+    if {key: params.get(key) for key in expected} != expected:
+        raise RuntimeError("Checkpoint U-Net parameters are not the audited tutorial architecture.")
+    if set(params) != {
+        "chans",
+        "num_downsample_layers",
+        "drop_prob",
+        "normalization_loc",
+        "normalization_scale",
+    }:
+        raise RuntimeError("Checkpoint contains unexpected U-Net parameter fields.")
+
+    vendor = source.unet
+    fitted = st.methods.DeepDeWedgeFittedInference(
+        network_affine_loc=float(vendor.normalization_loc),
+        network_affine_scale=float(vendor.normalization_scale),
+    )
+    described = st.network.build_described_network(
+        st.network.UNet3D(
+            initial_channels=params["chans"],
+            num_downsampling_blocks=params["num_downsample_layers"],
+        ),
+        context=st.network.NetworkBuildContext(
+            device=torch.device("cpu"), dtype=torch.float32, seed=0
+        ),
+    )
+    source_state = vendor.state_dict()
+    target_template = described.module.state_dict()
+    affine_names = {"_normalization_loc", "_normalization_scale"}
+    if set(source_state) - affine_names != {
+        "bottleneck.2.bias" if name == "bottleneck.4.bias" else
+        "bottleneck.2.weight" if name == "bottleneck.4.weight" else name
+        for name in target_template
+        if name not in affine_names
+    }:
+        raise RuntimeError("Source and target state namespaces are not the audited mapping.")
+
+    mapped_target: dict[str, torch.Tensor] = {}
+    target_to_source: dict[str, str] = {}
+    for target_name in target_template:
+        if target_name in affine_names:
+            mapped_target[target_name] = source_state[target_name]
+            continue
+        source_name = (
+            target_name.replace("bottleneck.4.", "bottleneck.2.")
+            if target_name.startswith("bottleneck.4.")
+            else target_name
+        )
+        tensor = source_state[source_name]
+        if tuple(tensor.shape) != tuple(target_template[target_name].shape):
+            raise RuntimeError(f"Mapped tensor shape differs for {target_name!r}.")
+        mapped_target[target_name] = tensor
+        target_to_source[target_name] = source_name
+    incompatible = described.module.load_state_dict(mapped_target, strict=True)
+    if incompatible.missing_keys or incompatible.unexpected_keys:
+        raise RuntimeError("Strict mapped source state load failed.")
+
+    closed = st.network.close_described_network(described)
+    canonical = closed.state
+    canonical_to_target: dict[str, str] = {}
+    for canonical_name, tensor in canonical.items():
+        matches = [
+            target_name
+            for target_name, target_tensor in described.module.state_dict().items()
+            if target_name not in affine_names
+            and target_tensor.data_ptr() == tensor.data_ptr()
+            and tuple(target_tensor.shape) == tuple(tensor.shape)
+            and target_tensor.dtype == tensor.dtype
+        ]
+        if len(matches) != 1:
+            raise RuntimeError(f"Canonical state mapping is ambiguous for {canonical_name!r}.")
+        canonical_to_target[canonical_name] = matches[0]
+    if set(canonical_to_target.values()) != set(target_to_source):
+        raise RuntimeError("Canonical state closure does not cover the source mapping.")
+
+    transformations = tuple(
+        st.artifacts.LearnedCheckpointTransformation(
+            kind="identity" if target_to_source[target_name] == target_name else "rename",
+            source=f"state_dict.unet.{target_to_source[target_name]}",
+            target=canonical_name,
+            details={"source_checkpoint": "tutorial_data/fitted_model.ckpt"},
+        )
+        for canonical_name, target_name in sorted(canonical_to_target.items())
+    )
+    return closed, canonical, transformations, fitted
+
+
+def _profile(
+    fitted: st.methods.DeepDeWedgeFittedInference,
+) -> st.methods.DeepDeWedgeInferenceProfile:
+    return st.methods.DeepDeWedgeInferenceProfile(
+        contract=st.methods.DeepDeWedgeInferenceContract(
+            missing_wedge_full_width_deg=50.0,
+            full_tomogram_standardization=False,
+            preconditioning_normalization_policy="recompute_patch_statistics",
+            patch_shape=(96, 96, 96),
+            overlap=(32, 32, 32),
+        ),
+        fitted=fitted,
+    )
+
+
+def _parity(
+    source: torch.nn.Module,
+    closed: st.network.ClosedDescribedNetwork,
+    fitted: st.methods.DeepDeWedgeFittedInference,
+) -> dict[str, float]:
+    """Compare external-affine canonical Network inference on a non-symmetric input."""
+
+    realized = st.network.realize_network(
+        program=closed.program,
+        state=closed.state,
+        context=st.network.NetworkBuildContext(
+            device=torch.device("cpu"), dtype=torch.float32, seed=19
+        ),
+    )
+    value = torch.arange(1 * 1 * 16 * 16 * 16, dtype=torch.float32).reshape(
+        1, 1, 16, 16, 16
+    )
+    value = value / 997.0 - 0.37
+    with torch.no_grad():
+        vendor_output = source.unet(value)
+        format2_output = invoke_deepdewedge_network(
+            realized.module, value, fitted=fitted
+        )
+    torch.testing.assert_close(vendor_output, format2_output, rtol=1.0e-5, atol=1.0e-6)
+    difference = (vendor_output - format2_output).abs()
+    relative_l2 = torch.linalg.vector_norm(difference) / torch.linalg.vector_norm(vendor_output)
+    return {
+        "input_elements": float(value.numel()),
+        "max_abs_error": float(difference.max()),
+        "relative_l2_error": float(relative_l2),
+    }
+
+
+def _resources(parity: dict[str, float]) -> dict[str, bytes]:
+    card = f"""---
+license: cc-by-4.0
+library_name: scitomo
+tags: [cryo-electron-tomography, deepdewedge, safetensors, scitomo, format-2]
+---
+
+# DeepDeWedge tutorial checkpoint — fresh Scitomo FORMAT 2 package
+
+This is a fresh FORMAT 2 export from the authoritative original Lightning
+checkpoint, not a migration of any earlier Hugging Face package. Normal runtime
+uses Scitomo's generic FORMAT 2 loader and Safetensors only; it does not require
+PyTorch Lightning or the upstream DeepDeWedge source checkout.
+
+## Package identity
+
+- package id: `deepdewedge_tutorial`; package revision: `3`
+- learned-checkpoint format: `2`; manifest schema: `4`
+- Scitomo conversion checkout: `2832957f69daff0d7baec5df17a7c54954623eed`
+- minimum Scitomo version: `0.7.3`
+- previous Hugging Face commit: `{PREVIOUS_HF_COMMIT}` — **HISTORICAL ONLY; NOT CONVERSION INPUT**
+
+## Authoritative provenance
+
+- upstream repository: <https://github.com/MLI-lab/DeepDeWedge>
+- upstream revision: `{UPSTREAM_REVISION}`
+- Figshare DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>; file id: `45582309`
+- original archive SHA-256: `{ARCHIVE_SHA256}`
+- original checkpoint member: `tutorial_data/fitted_model.ckpt`
+- original checkpoint size: `{CHECKPOINT_SIZE}` bytes
+- original checkpoint SHA-256: `{CHECKPOINT_SHA256}`
+
+DeepDeWedge Tutorial Data is attributed to Simon Wiedemann and is distributed
+under CC BY 4.0. The pinned DeepDeWedge implementation is BSD-2-Clause; its
+license text is included below `LICENSES/`. See `ATTRIBUTION.md`.
+
+## Scientific inference semantics
+
+The pure persisted Network owns only the lowered U-Net architecture and its 54
+canonical tensors. The fitted affine values remain outside Network state in the
+typed `deepdewedge_inference` profile:
+
+- `network_affine_loc`: `{_fmt(fitted_loc := -0.1489875167608261)}`
+- `network_affine_scale`: `{_fmt(fitted_scale := 1.3237642049789429)}`
+- input layout: `(..., Z, Y, X)`; Network layout: `(..., C, Z, Y, X)`
+- paired halves are refined independently then averaged; full-width missing wedge: 50 degrees
+- 96³ patches, 32³ overlap, trailing-reflection coverage, linear-ramp reassembly
+- preconditioning recomputes patch statistics; output uses the checkpoint-fitted affine
+
+## Fresh conversion and validation
+
+`refresh_format2.py` is the exact one-off implementation and records
+the verified source, explicit 54-tensor mapping, strict Network lowering, and
+generic export. It was run with Python `{platform.python_version()}`, Torch
+`{torch.__version__}`, Lightning `{pytorch_lightning.__version__}`, Safetensors
+`{safetensors.__version__}`, and Scitomo `{st.__version__}` on `{platform.platform()}`.
+
+The generic exporter freshly serializes `weights.safetensors`; no previous
+Hugging Face Safetensors, manifest, construction, or inference record is read.
+The conversion record lists every source checkpoint tensor to canonical target
+mapping. The validation record binds package state closure, generic loader
+reload, external-affine semantics, and deterministic forward parity.
+
+For a deterministic directional, non-symmetric CPU float32 input of 4,096 elements,
+authoritative upstream output versus FORMAT 2 pure-Network-plus-profile output
+passed `rtol=1e-5`, `atol=1e-6`: maximum absolute error
+`{parity['max_abs_error']:.9g}`, relative L2 error `{parity['relative_l2_error']:.9g}`.
+
+## Files and closure
+
+`manifest.json` is the authoritative, closed inventory of every package file,
+with each fresh size and SHA-256. It declares only FORMAT 2 construction,
+inference, Safetensors, conversion, validation, and documentation/license
+resources; there is no format-1 or migration artifact. Validate and load with:
+
+```python
+import scitomo as st
+loaded = st.api.load_learned_network("/path/to/package")
+```
+
+This operation uses the generic Scitomo FORMAT 2 loader and does not import
+Lightning or DeepDeWedge. It is a checkpoint package, not a claim of scientific
+approval for a new dataset or acquisition protocol.
+"""
+    attribution = f"""# Attribution and modification notice
+
+## Original material
+
+**DeepDeWedge Tutorial Data**
+Creator: Simon Wiedemann
+DOI: <https://doi.org/10.6084/m9.figshare.25043435.v1>
+Figshare file id: `45582309`
+Archive member: `tutorial_data/fitted_model.ckpt`
+License: Creative Commons Attribution 4.0 International
+
+The method is described by Simon Wiedemann and Reinhard Heckel, *A deep
+learning method for simultaneous denoising and missing wedge reconstruction in
+cryogenic electron tomography*, Nature Communications 15, 8255 (2024),
+<https://doi.org/10.1038/s41467-024-51438-y>.
+
+Pinned upstream code: <https://github.com/MLI-lab/DeepDeWedge/tree/{UPSTREAM_REVISION}>
+(BSD-2-Clause).
+
+## Changes in this package
+
+On 2026-09-04 Scitomo freshly converted only the authoritative checkpoint
+`official/fitted_model.ckpt`, after byte-size and SHA-256 verification, through
+the exact pinned upstream source and current generic FORMAT 2 exporter. The
+54 U-Net state tensors were explicitly mapped into canonical Network state.
+The two fitted affine quantities were preserved as external DeepDeWedge
+inference-profile state; they are not Network state. No old Hugging Face
+Safetensors or format-1 package artifact was conversion input.
+
+No endorsement by the cited authors, the Machine Learning and Information
+Processing Laboratory, Figshare, or the rights holders is implied.
+"""
+    return {
+        "README.md": card.encode("utf-8"),
+        "ATTRIBUTION.md": attribution.encode("utf-8"),
+        "LICENSES/DeepDeWedge-Code-BSD-2-Clause.txt": (UPSTREAM / "LICENSE").read_bytes(),
+        "refresh_format2.py": Path(__file__).read_bytes(),
+    }
+
+
+def _fmt(value: float) -> str:
+    return format(value, ".17g")
+
+
+def _replace_hf_with_closed_package() -> None:
+    if TARGET.resolve() != ROOT / "hf" or not (TARGET / ".git").is_dir():
+        raise RuntimeError("Refusing to replace an unexpected Hugging Face working tree.")
+    if not OUTPUT.is_dir() or OUTPUT.is_symlink():
+        raise RuntimeError("Fresh output directory is unavailable for publication.")
+    for child in TARGET.iterdir():
+        if child.name == ".git":
+            continue
+        if child.is_dir() and not child.is_symlink():
+            shutil.rmtree(child)
+        else:
+            child.unlink()
+    shutil.move(str(OUTPUT), str(TARGET / ".package-fresh"))
+    staged = TARGET / ".package-fresh"
+    for child in staged.iterdir():
+        shutil.move(str(child), str(TARGET / child.name))
+    staged.rmdir()
+
+
+def _runtime_view() -> Path:
+    """Create a byte-identical package view excluding local Git administration."""
+
+    if RUNTIME_VIEW.exists() or RUNTIME_VIEW.is_symlink():
+        raise RuntimeError("Runtime validation view already exists.")
+    shutil.copytree(TARGET, RUNTIME_VIEW, ignore=shutil.ignore_patterns(".git"))
+    return RUNTIME_VIEW
+
+
+def main() -> None:
+    _verify_authority()
+    if OUTPUT.exists() or OUTPUT.is_symlink():
+        raise RuntimeError("Fresh output directory already exists before export.")
+    source = _load_upstream_model()
+    closed, canonical, mappings, fitted = _canonical_source_to_target(source)
+    parity = _parity(source, closed, fitted)
+    profile = _profile(fitted)
+    owner = st.artifacts.LearnedCheckpointMethodOwner(
+        family="restoration", method_kind="deepdewedge"
+    )
+    construction = st.artifacts.LearnedCheckpointConstructionRecordV2.from_program(
+        closed.program
+    )
+    inference = st.artifacts.LearnedCheckpointInferenceRecordV3.from_profile(
+        owner=owner, profile=profile
+    )
+    conversion = st.artifacts.LearnedCheckpointConversionEvidenceV2(
+        source=st.artifacts.LearnedCheckpointConversionSourceV2(
+            kind="figshare_checkpoint",
+            project="DeepDeWedge Tutorial Data",
+            identifier="45582309/tutorial_data/fitted_model.ckpt",
+            url="https://doi.org/10.6084/m9.figshare.25043435.v1",
+            revision=UPSTREAM_REVISION,
+            sha256=CHECKPOINT_SHA256,
+            metadata={
+                "archive_sha256": ARCHIVE_SHA256,
+                "archive_member": "tutorial_data/fitted_model.ckpt",
+                "checkpoint_size_bytes": CHECKPOINT_SIZE,
+                "upstream_repository": "https://github.com/MLI-lab/DeepDeWedge",
+            },
+        ),
+        tool="deepdewedge_refresh_format2",
+        tool_version="1",
+        tensor_mappings=mappings,
+        environment={
+            "python": platform.python_version(),
+            "torch": torch.__version__,
+            "pytorch_lightning": pytorch_lightning.__version__,
+            "safetensors": safetensors.__version__,
+            "scitomo": st.__version__,
+            "scitomo_commit": "2832957f69daff0d7baec5df17a7c54954623eed",
+            "platform": platform.platform(),
+        },
+    )
+    validation = st.artifacts.LearnedCheckpointValidationEvidenceV2(
+        software=(
+            st.artifacts.LearnedCheckpointSoftware(name="scitomo", version=st.__version__),
+            st.artifacts.LearnedCheckpointSoftware(name="torch", version=torch.__version__),
+            st.artifacts.LearnedCheckpointSoftware(name="pytorch_lightning", version=pytorch_lightning.__version__),
+            st.artifacts.LearnedCheckpointSoftware(name="safetensors", version=safetensors.__version__),
+        ),
+        cases=(
+            st.artifacts.LearnedCheckpointValidationCase(
+                name="authoritative_source_mapping", kind="state_mapping", status="passed",
+                metrics={"source_tensors": 56.0, "canonical_network_tensors": 54.0},
+            ),
+            st.artifacts.LearnedCheckpointValidationCase(
+                name="external_fitted_affine_profile", kind="inference_profile", status="passed",
+                metrics={"network_affine_loc": fitted.network_affine_loc, "network_affine_scale": fitted.network_affine_scale},
+            ),
+            st.artifacts.LearnedCheckpointValidationCase(
+                name="deterministic_forward_parity", kind="forward_parity", status="passed",
+                tolerances={"atol": 1.0e-6, "rtol": 1.0e-5}, metrics=parity,
+            ),
+        ),
+    )
+    exported = st.artifacts.export_learned_checkpoint_format2_package(
+        canonical,
+        construction=construction,
+        inference=inference,
+        package_id="deepdewedge_tutorial",
+        package_revision=3,
+        owner=owner,
+        requirements=st.artifacts.LearnedCheckpointFormat2Requirements(
+            minimum_scitomo_version="0.7.3"
+        ),
+        validation=validation,
+        conversion=conversion,
+        resources=_resources(parity),
+        destination="package-fresh",
+        write_scope=st.core.WriteScope(st.core.WriteScopeKind.MODELS, ROOT),
+        provenance=st.artifacts.LearnedCheckpointProvenance(
+            sources=(
+                st.artifacts.LearnedCheckpointSource(
+                    kind="upstream_repository", project="MLI-lab/DeepDeWedge",
+                    identifier=UPSTREAM_REVISION,
+                    url="https://github.com/MLI-lab/DeepDeWedge",
+                    revision=UPSTREAM_REVISION,
+                ),
+            ),
+            citations=(
+                "https://doi.org/10.1038/s41467-024-51438-y",
+                "https://doi.org/10.6084/m9.figshare.25043435.v1",
+            ),
+        ),
+    )
+    _replace_hf_with_closed_package()
+    runtime_root = _runtime_view()
+    try:
+        package = st.artifacts.validate_learned_checkpoint_format2_package(runtime_root)
+        loaded = st.api.load_learned_network(
+            runtime_root,
+            context=st.network.NetworkBuildContext(
+                device=torch.device("cpu"), dtype=torch.float32, seed=29
+            ),
+            expected_owner=owner,
+            expected_inference_profile=profile,
+        )
+        reloaded = st.network.canonical_network_state(loaded.network.module)
+        if set(reloaded) != set(canonical) or any(
+            not torch.equal(reloaded[name], canonical[name]) for name in canonical
+        ):
+            raise RuntimeError("Reloaded FORMAT 2 state is not closed over canonical state.")
+        value = torch.arange(1 * 1 * 16 * 16 * 16, dtype=torch.float32).reshape(1, 1, 16, 16, 16)
+        value = value / 997.0 - 0.37
+        with torch.no_grad():
+            expected = source.unet(value)
+            actual = invoke_deepdewedge_network(
+                loaded.network.module, value, fitted=profile.fitted
+            )
+        torch.testing.assert_close(expected, actual, rtol=1.0e-5, atol=1.0e-6)
+        print(f"FORMAT 2 package validated at {TARGET}")
+        print(f"weights_sha256={package.manifest.files.weights.sha256}")
+        print(f"manifest_sha256={_digest(TARGET / 'manifest.json', 'sha256')}")
+    finally:
+        if RUNTIME_VIEW.exists():
+            shutil.rmtree(RUNTIME_VIEW)
+
+
+if __name__ == "__main__":
+    main()

validation/validation-record.json

@@ -1,72 +1,66 @@
 {
   "cases": [
     {
-      "evidence_sha256": [
-        "bf23d54a8a79a8086b78a8c807a787f1a5ac5525ef7d6eec67373aea4f0a4cb5"
-      ],
-      "kind": "structural",
+      "evidence_sha256": [],
+      "kind": "state_mapping",
       "metrics": {
-        "tensor_count": 56.0
+        "canonical_network_tensors": 54.0,
+        "source_tensors": 56.0
       },
-      "name": "exact_state_mapping",
+      "name": "authoritative_source_mapping",
       "status": "passed",
       "tolerances": {}
     },
     {
-      "evidence_sha256": [
-        "f357d0992b60ee97fdaf52fb2fba465f19ce7ca4b773fd034c391208f22f791f",
-        "baa30de9f48d992bf009893b48af9d19eeab34410e6cded99475f514fefefa8d"
-      ],
-      "kind": "numerical",
+      "evidence_sha256": [],
+      "kind": "inference_profile",
       "metrics": {
-        "max_abs": 0.0,
-        "relative_l2": 0.0
+        "network_affine_loc": -0.1489875167608261,
+        "network_affine_scale": 1.3237642049789429
       },
-      "name": "synthetic_forward_parity",
+      "name": "external_fitted_affine_profile",
       "status": "passed",
-      "tolerances": {
-        "atol": 1e-06,
-        "rtol": 1e-05
-      }
+      "tolerances": {}
     },
     {
-      "evidence_sha256": [
-        "f357d0992b60ee97fdaf52fb2fba465f19ce7ca4b773fd034c391208f22f791f",
-        "baa30de9f48d992bf009893b48af9d19eeab34410e6cded99475f514fefefa8d"
-      ],
-      "kind": "scientific",
+      "evidence_sha256": [],
+      "kind": "forward_parity",
       "metrics": {
-        "max_abs": 7.152557373046875e-07,
-        "relative_l2": 1.2612566990810592e-07
+        "input_elements": 4096.0,
+        "max_abs_error": 0.0,
+        "relative_l2_error": 0.0
       },
-      "name": "real_tutorial_crop_parity",
+      "name": "deterministic_forward_parity",
       "status": "passed",
       "tolerances": {
         "atol": 1e-06,
-        "relative_l2": 0.0001,
         "rtol": 1e-05
       }
     }
   ],
-  "construction_fingerprint": "sha256:c9da7decd489f0f4f893dce565f37db25abc63d0d3efcacad9835305ffd97cb2",
-  "inference_fingerprint": "sha256:c2c5eba72579510265258dcfce3a5be2f851f9276c7e53d92330d78eb37720dd",
+  "construction_fingerprint": "sha256:644b162de70d014f9273c992fd583ee4fdfb8ae425a35fba26b3fbd41d560f28",
+  "inference_fingerprint": "sha256:dd4d9b190f9e3a0c94a0da2d383f5dc42aaa0ac888dcec760d57b4cf5362aabd",
   "kind": "scitomo_checkpoint_validation",
   "package_id": "deepdewedge_tutorial",
-  "package_revision": 2,
-  "schema_version": 1,
+  "package_revision": 3,
+  "schema_version": 2,
   "software": [
     {
       "name": "scitomo",
-      "version": "0.5.4"
+      "version": "0.7.3.dev0"
     },
     {
       "name": "torch",
-      "version": "2.13.0+cu130"
+      "version": "2.12.1+cpu"
+    },
+    {
+      "name": "pytorch_lightning",
+      "version": "2.6.5"
     },
     {
-      "name": "vendor_pytorch_lightning",
-      "version": "1.8.0.post1"
+      "name": "safetensors",
+      "version": "0.8.0"
     }
   ],
-  "weights_sha256": "2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795"
+  "weights_sha256": "e8f51c499d6e5c14ddce45d5cf9935511220de161ce62deb50a2c3a8cae5c731"
 }

weights.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a34aeb61dba5da5c7b78f4cc26de9b8a9134ac12dc08876a8e93bf02776c795
-size 109294940
+oid sha256:e8f51c499d6e5c14ddce45d5cf9935511220de161ce62deb50a2c3a8cae5c731
+size 109294372