Datasets:
Tasks:
Tabular Regression
Modalities:
Tabular
Languages:
English
Size:
10K - 100K
Tags:
cfd
openfoam
surrogate-modeling
scientific-computing
scientific-machine-learning
physics-informed-neural-networks
License:
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -126,7 +126,7 @@ The geometry is controlled by 28 parameters sampled via a scrambled Sobol sequen
|
|
| 126 |
|
| 127 |
### Failed samples and error codes
|
| 128 |
|
| 129 |
-
The `params` configuration provides three splits: `train` (
|
| 130 |
|
| 131 |
| `target_0` value | Meaning |
|
| 132 |
|---|---|
|
|
|
|
| 126 |
|
| 127 |
### Failed samples and error codes
|
| 128 |
|
| 129 |
+
The `params` configuration provides three splits: `train` (≈8,050 samples) and `test` (≈895 samples) — a 90/10 split with seed=42 over successful simulations — plus `failed` (≈1,055 samples) containing all unsuccessful runs. Failed samples are indicated by sentinel values in the target columns:
|
| 130 |
|
| 131 |
| `target_0` value | Meaning |
|
| 132 |
|---|---|
|