Upload README.md with huggingface_hub

README.md

@@ -52,7 +52,7 @@ configs:
 
 # GeomCAD
 
-A procedural benchmark of executable CAD programs paired with watertight meshes, spanning four
+A procedural benchmark of executable CAD programs paired with triangle meshes, spanning four
 generative operations — **extrude**, **revolve**, **sweep**, and **loft** — at roughly one million
 samples per operation. Every numerical literal in every program is an integer in `[0, 1000]`, 1,001 levels, which
 makes exact parameter-level comparison (Parameter Exact Match) well defined rather than approximate.
@@ -133,9 +133,11 @@ also be exported to industrial B-Rep formats.
 
 ## Integer parameters
 
-Numeric literals are min–max normalised per script and rewritten as integers in `[0, 1000]`, and a
-validity gate constrains each solid to the canonical cube `[0, 1000]^3`, so parameter values and
-spatial dimensions share one scale. Physical dimensions reported elsewhere (for example in figures
+Numeric literals are min–max normalised per script and rewritten as integers in `[0, 1000]`, so
+every parameter a model must predict is drawn from one 1,001-level grid. The extent of the
+resulting solid is a consequence of the operation applied and is not itself constrained to that
+range; revolving a profile about an axis, for instance, carries it to the far side and spans about
+twice the profile radius by construction. Physical dimensions reported elsewhere (for example in figures
 comparing against float-parameter baselines) are the result of applying the per-script inverse map,
 and are therefore real-valued even though the program literals are not.
 
@@ -150,7 +152,10 @@ and are therefore real-valued even though the program literals are not.
 | **total** | **3,995,521** | **4,661** | **6,000** | **4,006,182** |
 
 Training and validation come from each operation's procedural pool; the test split is generated
-independently through the same pipeline, and no program is shared across splits. Validation is
+independently through the same pipeline. Overlap at the level of sample indices is exactly zero. At
+the stricter level of program text, 15 training rows reproduce a test program (revolve 1, sweep 14)
+and one reproduces a validation program; they are listed in `REPRODUCIBILITY.md` and removed by the
+audit shipped with the corpus. Validation is
 1,500 for extrude and smaller elsewhere because those held-out sets were carved from
 multi-operation training mixtures and only the rows belonging to each operation are kept here.