Card, datasheet and Croissant: learned-result wording; notebook manifest cites its sources commit

Documents and executed notebooks for data commit 2ab44d899652c7fb99c425488e84a7d565963547.

DATASHEET.md

@@ -15,7 +15,7 @@ and accompanies its [Croissant metadata](https://huggingface.co/datasets/thalesg
 
 The release receipts summarise PASSAGE v1.0 as follows:
 
-PASSAGE v1.0 contains 4,400,000 core scenes and 26,400,000 objective/obstacle variants; 1,792 separate preview rows do not add independent training observations. Terrain and labels use persisted float32 inputs decoded before authoritative float64 graph solves. Missing/nodata source crops are rejected. All 26,400,000 variants are exact routes; start and goal cells that the obstacles disconnect are rejected by the generator. No learned model has been trained or evaluated on PASSAGE v1.0.
+PASSAGE v1.0 contains 4,400,000 core scenes and 26,400,000 objective/obstacle variants; 1,792 separate preview rows do not add independent training observations. Terrain and labels use persisted float32 inputs decoded before authoritative float64 graph solves. Missing/nodata source crops are rejected. All 26,400,000 variants are exact routes; start and goal cells that the obstacles disconnect are rejected by the generator. No learned-model result on PASSAGE v1.0 has been published yet.
 
 The paper's learned-model results are historical results of the original study, reported as obtained; not re-run on the released dataset. This datasheet documents a computational
 annotation and qualification process. It is not an operational deployment
@@ -268,7 +268,7 @@ that archive was redacted (see `release-receipts/privacy-redactions.json`);
 
 ### Has the dataset been used for any tasks already?
 
-No learned model has been trained or evaluated on PASSAGE v1.0. The paper's
+No learned-model result on PASSAGE v1.0 has been published yet. The paper's
 learned results (dense path-mask segmentation and cost-to-go regression) are
 historical results of the original study, reported as obtained;
 not re-run on the released dataset.

README.md

@@ -5846,7 +5846,7 @@ the `passage-eval` evaluator are in the
 > float32 inputs decoded before authoritative float64 graph solves.
 > Missing/nodata source crops are rejected. All 26,400,000 variants are exact
 > routes; start and goal cells that the obstacles disconnect are rejected by
-> the generator. No learned model has been trained or evaluated on PASSAGE v1.0.
+> the generator. No learned-model result on PASSAGE v1.0 has been published yet.
 
 The paper's learned-model results are historical results of the original
 study, reported as obtained; not re-run on the released dataset.

croissant.json

@@ -54,7 +54,7 @@
   "@id": "https://huggingface.co/datasets/thalesgroup/passage",
   "name": "PASSAGE",
   "alternateName": "PASSAGE v1.0",
-  "description": "PASSAGE v1.0 contains 4,400,000 core scenes and 26,400,000 objective/obstacle variants; 1,792 separate preview rows do not add independent training observations. Terrain and labels use persisted float32 inputs decoded before authoritative float64 graph solves. Missing/nodata source crops are rejected. All 26,400,000 variants are exact routes; start and goal cells that the obstacles disconnect are rejected by the generator. No learned model has been trained or evaluated on PASSAGE v1.0.",
+  "description": "PASSAGE v1.0 contains 4,400,000 core scenes and 26,400,000 objective/obstacle variants; 1,792 separate preview rows do not add independent training observations. Terrain and labels use persisted float32 inputs decoded before authoritative float64 graph solves. Missing/nodata source crops are rejected. All 26,400,000 variants are exact routes; start and goal cells that the obstacles disconnect are rejected by the generator. No learned-model result on PASSAGE v1.0 has been published yet.",
   "version": "1.0.0",
   "url": "https://huggingface.co/datasets/thalesgroup/passage",
   "sameAs": "https://github.com/ThalesGroup/passage",
@@ -59948,7 +59948,7 @@
     "Static 2D grid tasks with procedural obstacles and endpoints; no vehicle dynamics, clearance, weather, airspace, regulation or mission demand is modelled.",
     "Labels are exact only under the frozen graph contract (8-connectivity, destination-only diagonals, fixed 30 m edges, float64 solves on decoded float32 inputs). Other conventions change optima; tied optima can differ in geometry.",
     "Six variants per scene and scenes from the same 5x5 degree block or across resolutions can share terrain; treat them as correlated, not independent, samples. The 1,792 preview rows duplicate admitted scenes.",
-    "No learned model has been trained or evaluated on PASSAGE v1.0. The paper's learned-model results are historical results of the original study, reported as obtained; not re-run on the released dataset.",
+    "No learned-model result on PASSAGE v1.0 has been published yet. The paper's learned-model results are historical results of the original study, reported as obtained; not re-run on the released dataset.",
     "Not suitable for operational flight planning, search-and-rescue dispatch or other safety-of-life decisions; the dataset does not certify deployment or replace certification processes."
   ],
   "rai:dataUseCases": [

notebook-artifacts/manifest.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d2da42eca99a2c5655f7512cc2e740ce95be745ba091cffa31fe8a94d88ca439
-size 67619
+oid sha256:f3eaaf43b552a31b402fc599a3009d13206e2e68be2b0c8d306aa42da2fb138e
+size 67732

release-receipts/publication-metadata.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8cd352521a12d9dea696816bb0e7ade2698efde3e42156147dcd2494508e1dd5
+oid sha256:9dea29dd4f36eb6911020ae53d89c9b31c54371b02eb6eb82af905cf9d1a9744
 size 6835

release-receipts/ready-upload.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba83cd952f05fec96afd9169c7ad8ffe19ce54bf38d9dad21cb3666230796355
+oid sha256:57f950e9d5c7d051fe689c21762872548576c4d01aa81d9d0b9cb0c9ebab47b1
 size 1458925