Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,34 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# SIMSHIFT — Inspection Samples
|
| 2 |
+
|
| 3 |
+
This repository hosts a single representative simulation per task from
|
| 4 |
+
the four SIMSHIFT datasets, for reviewer inspection only. The full
|
| 5 |
+
datasets are hosted separately at
|
| 6 |
+
[simshift/SIMSHIFT_data](https://huggingface.co/datasets/simshift/SIMSHIFT_data).
|
| 7 |
+
|
| 8 |
+
These files are **not** intended for benchmarking, training, or any
|
| 9 |
+
quantitative evaluation. They exist solely to let reviewers verify file
|
| 10 |
+
format, schema, mesh structure, and output fields without downloading
|
| 11 |
+
the full datasets (the heatsink set in particular is ~520 GB).
|
| 12 |
+
|
| 13 |
+
## Contents
|
| 14 |
+
|
| 15 |
+
One sample file per task, in the same HDF5 format and schema as the
|
| 16 |
+
full datasets:
|
| 17 |
+
|
| 18 |
+
| File | Task |
|
| 19 |
+
|---------------------|-----------------------|
|
| 20 |
+
| `rolling_sample.h5` | Hot Rolling |
|
| 21 |
+
| `forming_sample.h5` | Sheet Metal Forming |
|
| 22 |
+
| `motor_sample.h5` | Electric Motor Design |
|
| 23 |
+
| `heatsink_sample.h5`| Heatsink Design |
|
| 24 |
+
|
| 25 |
+
## How the samples were generated
|
| 26 |
+
|
| 27 |
+
Each file contains a single simulation drawn from the corresponding
|
| 28 |
+
full dataset.The simulation is stored exactly as
|
| 29 |
+
it appears in the full release, using the same HDF5 structure, field
|
| 30 |
+
names, and units.
|
| 31 |
+
|
| 32 |
+
## License
|
| 33 |
+
|
| 34 |
+
MIT, identical to the main SIMSHIFT release.
|