paranke commited on
Commit
fabf840
·
verified ·
1 Parent(s): 6049be1

Public release: CC BY-NC 4.0 licence, no access approval, updated links

Browse files
Files changed (1) hide show
  1. README.md +7 -6
README.md CHANGED
@@ -1,4 +1,5 @@
1
  ---
 
2
  pretty_name: Expanded9 original training inputs and labels
3
  task_categories:
4
  - image-segmentation
@@ -9,11 +10,11 @@ tags:
9
 
10
  # Expanded9 training data and labels
11
 
12
- Private review dataset for `expanded9-resunet128-v2-scratch-20260917`.
13
  113 representations: **112 training and one validation**, across 12 parent-scroll groups.
14
  **No PHerc0211 data or labels were used for training, fine-tuning, or validation.**
15
 
16
- - [Training code and usage](https://github.com/priyankaranke/vesuvius-expanded9-training)
17
  - [Expanded9 model (private)](https://huggingface.co/paranke/vesuvius-expanded9-resunet)
18
  - [Submission documents and image](https://drive.google.com/drive/folders/1bvKolL_TIE3k5Ixi1qlOiAmofQ6r2ocH)
19
  - [Methodology](https://drive.google.com/file/d/1LJaEokRQKgDLcSSWQGUBdPqEI2rjWnWe/view)
@@ -41,11 +42,11 @@ an independently certified ground-truth collection. The sole validation region
41
  is PHerc0841 with the recorded 2mm training exclusion. The final model is epoch10,
42
  74,160 updates; it was not selected by its validation score.
43
 
44
- Download with authenticated Hugging Face access, for example:
45
  `hf download paranke/vesuvius-expanded9-training-data --repo-type dataset --local-dir dataset`
46
 
47
- Keep private for prize review. No license is granted beyond underlying source
48
- terms; source URLs and attribution are preserved in the original manifests.
49
 
50
  ## Submission inference data (excluded from training)
51
 
@@ -57,4 +58,4 @@ Follow the [numbered GitHub guide](https://github.com/priyankaranke/vesuvius-exp
57
 
58
  The default preparation step restores the original compressed tile files from the byte packs; it does not render meshes or resample CT. Once downloaded, training needs no original workstation, historical cache, source-site login or W&B service. The historical source URLs remain provenance; the full input corpus is supplied here. Human-approved labels are preserved.
59
 
60
- Normal authorized Hub file downloads are sufficient. HF Pro and the paid private-dataset viewer are not required. Read access must cover this dataset and the model repository; private URLs alone do not grant permission. Code and native rendering helpers are supplied in the one linked GitHub repository. The final mesh and PNGs remain on Drive.
 
1
  ---
2
+ license: cc-by-nc-4.0
3
  pretty_name: Expanded9 original training inputs and labels
4
  task_categories:
5
  - image-segmentation
 
10
 
11
  # Expanded9 training data and labels
12
 
13
+ Training dataset for `expanded9-resunet128-v2-scratch-20260917`. It is also part of the training data of the Studio ink model behind the PHerc0211 First Letters submission ([code](https://github.com/priyankaranke/vesuvius-studio-multiscroll-cpt)).
14
  113 representations: **112 training and one validation**, across 12 parent-scroll groups.
15
  **No PHerc0211 data or labels were used for training, fine-tuning, or validation.**
16
 
17
+ - [Training code and usage](https://github.com/priyankaranke/vesuvius-expanded9-training) (private)
18
  - [Expanded9 model (private)](https://huggingface.co/paranke/vesuvius-expanded9-resunet)
19
  - [Submission documents and image](https://drive.google.com/drive/folders/1bvKolL_TIE3k5Ixi1qlOiAmofQ6r2ocH)
20
  - [Methodology](https://drive.google.com/file/d/1LJaEokRQKgDLcSSWQGUBdPqEI2rjWnWe/view)
 
42
  is PHerc0841 with the recorded 2mm training exclusion. The final model is epoch10,
43
  74,160 updates; it was not selected by its validation score.
44
 
45
+ Download (no login needed), for example:
46
  `hf download paranke/vesuvius-expanded9-training-data --repo-type dataset --local-dir dataset`
47
 
48
+ Released under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/), the licence of the Vesuvius Challenge scan data it is derived from.
49
+ Source URLs and attribution are preserved in the original manifests.
50
 
51
  ## Submission inference data (excluded from training)
52
 
 
58
 
59
  The default preparation step restores the original compressed tile files from the byte packs; it does not render meshes or resample CT. Once downloaded, training needs no original workstation, historical cache, source-site login or W&B service. The historical source URLs remain provenance; the full input corpus is supplied here. Human-approved labels are preserved.
60
 
61
+ Normal Hub file downloads are sufficient; no login is needed. The Expanded9 training repository and model linked above are private. The final mesh and PNGs remain on Drive.