Update dataset card: add paper, project page, and GitHub links

README.md

@@ -1,45 +1,69 @@
 ---
 license: cc-by-4.0
-pretty_name: "FLOATBench: Wind Turbine Tower Damage"
 size_categories:
-  - 100K<n<1M
+- 100K<n<1M
 task_categories:
-  - tabular-regression
+- other
+pretty_name: 'FLOATBench: Wind Turbine Tower Damage'
 tags:
-  - floating-offshore-wind-turbine
-  - tower-fatigue
-  - 22-MW-wind-turbine
-  - IEA-22-reference-turbine
-  - tabular-dataset
-  - benchmark-dataset
+- floating-offshore-wind-turbine
+- tower-fatigue
+- 22-MW-wind-turbine
+- IEA-22-reference-turbine
+- tabular-dataset
+- benchmark-dataset
 configs:
-  - config_name: ref
-    data_files:
-      - split: train
-        path: ref/train_damage.csv
-      - split: test
-        path: ref/test_damage.csv
-  - config_name: opt1
-    data_files:
-      - split: train
-        path: opt1/train_damage.csv
-      - split: test
-        path: opt1/test_damage.csv
-  - config_name: opt2
-    data_files:
-      - split: train
-        path: opt2/train_damage.csv
-      - split: test
-        path: opt2/test_damage.csv
+- config_name: ref
+  data_files:
+  - split: train
+    path: ref/train_damage.csv
+  - split: test
+    path: ref/test_damage.csv
+- config_name: opt1
+  data_files:
+  - split: train
+    path: opt1/train_damage.csv
+  - split: test
+    path: opt1/test_damage.csv
+- config_name: opt2
+  data_files:
+  - split: train
+    path: opt2/train_damage.csv
+  - split: test
+    path: opt2/test_damage.csv
 ---
 
 # FLOATBench: Wind Turbine Tower Damage
 
-Tabular fatigue dataset for 22 MW floating offshore wind turbine
-(FOWT) towers. Contains 582,120 labelled tower section fatigue
-damage records across three tower geometries: the IEA-22 reference
-turbine baseline (`ref`) and two FLOAT-derived re-designs (`opt1`,
-`opt2`).
+[**Paper**](https://huggingface.co/papers/2605.25717) | [**Project Page**](https://joao97ribeiro.github.io/FLOATBench/) | [**GitHub**](https://github.com/Joao97ribeiro/FLOATBench)
+
+Tabular fatigue dataset for 22 MW floating offshore wind turbine (FOWT) towers. Contains 582,120 labelled tower section fatigue damage records across three tower geometries: the IEA-22 reference turbine baseline (`ref`) and two FLOAT-derived re-designs (`opt1`, `opt2`).
+
+FLOATBench is the first FOWT fatigue benchmark for tabular surrogate modeling, offering an evaluation protocol that generalizes to engineering surrogates defined over physical operating envelopes.
+
+## Sample Usage
+
+### Using the `datasets` library
+```python
+from datasets import load_dataset
+
+# Load the IEA-22 reference turbine baseline
+ds = load_dataset('DeCoDELab/FLOATBench', 'ref')
+print(ds)
+```
+
+### Using `pandas`
+```python
+import pandas as pd
+
+# Load from local CSVs (after downloading)
+train = pd.read_csv("ref/train_damage.csv")
+test  = pd.read_csv("ref/test_damage.csv")
+
+# Evaluate on the worst-case wind+wave extrapolation cell
+ex_ex = test[(test.wind_group == "Extrapolate") &
+             (test.wave_group == "Extrapolate")]
+```
 
 ## Layout
 
@@ -58,10 +82,7 @@ FLOATBench/
 
 ## Schema
 
-Columns appear in the order below. Each `*_id` grid index sits
-immediately before the value it indexes (`wind_speed_id` before
-`wind_speed`, `wave_hs_id` before `wave_hs`, `wave_tp_id` before
-`wave_tp`).
+Columns appear in the order below. Each `*_id` grid index sits immediately before the value it indexes (`wind_speed_id` before `wind_speed`, `wave_hs_id` before `wave_hs`, `wave_tp_id` before `wave_tp`).
 
 `data.csv` (16 cols):
 
@@ -72,8 +93,7 @@ section_id, section_height_m, section_radius_m, section_thickness_m,
 damage_weight, damage
 ```
 
-`train_damage.csv` / `test_damage.csv` (18 cols): same order, with
-`wind_group, wave_group` inserted right before `damage_weight`.
+`train_damage.csv` / `test_damage.csv` (18 cols): same order, with `wind_group, wave_group` inserted right before `damage_weight`.
 
 The tables below describe each column grouped by category.
 
@@ -120,31 +140,20 @@ The tables below describe each column grouped by category.
 | `damage`        | float | Miner-summed fatigue damage at the section (dimensionless)                             |
 | `damage_weight` | float | Probability of occurrence over the 25-year service life                                |
 
-Lifetime damage at a section is recovered as
-`sum(damage_i * damage_weight_i)` over all conditions.
+Lifetime damage at a section is recovered as `sum(damage_i * damage_weight_i)` over all conditions.
 
 ## Regime-aware split
 
-The recommended train/test partition is **regime-aware**: an
-alpha-shape over the joint wind/wave operating envelope partitions
-test points into `In-train` / `Interpolate` / `Extrapolate` regimes
-on both the wind and wave axes, populating all nine cells of the
-3×3 wind×wave regime grid. Per tower:
+The recommended train/test partition is **regime-aware**: an alpha-shape over the joint wind/wave operating envelope partitions test points into `In-train` / `Interpolate` / `Extrapolate` regimes on both the wind and wave axes, populating all nine cells of the 3×3 wind×wave regime grid. Per tower:
 
 | Subset | Rows    | Conditions | Description                                  |
 |--------|---------|------------|----------------------------------------------|
 | Train  | 51,840  | 288        | All `In-train`/`In-train` cell               |
 | Test   | 142,200 | 790        | Spans the remaining 8 wind×wave regime cells |
 
-Train rows carry `wind_group = wave_group = In-train` by
-construction. Test rows carry the assigned regime labels so the
-9-cell evaluation can be run directly.
-
 ### Reproducing the split from grid IDs
 
-The partition is **fully determined by the integer grid IDs**
-(`wind_speed_id`, `wave_hs_id`, `wave_tp_id`) shipped on every row.
-A row is in train iff its three IDs all fall in the train sets:
+The partition is **fully determined by the integer grid IDs** (`wind_speed_id`, `wave_hs_id`, `wave_tp_id`) shipped on every row. A row is in train iff its three IDs all fall in the train sets:
 
 ```python
 TRAIN_WS_IDS = {2, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13, 14,
@@ -157,49 +166,20 @@ is_train = (df.wind_speed_id.isin(TRAIN_WS_IDS)
             & df.wave_tp_id.isin(TRAIN_TP_IDS))
 ```
 
-Train cells: 18 × 4 × 4 = 288. Total grid: 22 × 7 × 7 = 1,078.
-
-The alpha-shape regime labels (`wind_group`, `wave_group`) are
-derived from the train set's joint wind–wave envelope. Reproducing
-them and producing diagnostic plots requires the FLOATBench code
-repo: <https://github.com/Joao97ribeiro/FLOATBench>
-
-```bash
-python scripts/split/run.py --flagfile=scripts/split/config.cfg \
-       --dataset_dir=/path/to/FLOATBench-dataset
-```
-
-This regenerates `train_damage.csv` / `test_damage.csv` byte-for-byte
-identical to the shipped files, plus a `split_metadata.json` and
-plots of the partition and train spacing.
-
-## Quickstart
-
-```python
-import pandas as pd
-
-train = pd.read_csv("ref/train_damage.csv")
-test  = pd.read_csv("ref/test_damage.csv")
-
-# evaluate on the worst-case wind+wave extrapolation cell
-ex_ex = test[(test.wind_group == "Extrapolate") &
-             (test.wave_group == "Extrapolate")]
+## Citation
+
+```bibtex
+@misc{ribeiro2026floatbenchdatasetbenchmarkfloating,
+      title={FLOATBench: A Dataset and Benchmark for Floating Offshore Wind Turbine Tower Fatigue},
+      author={João Alves Ribeiro and Bruno Alves Ribeiro and Francisco Pimenta and Sérgio M. O. Tavares and Faez Ahmed},
+      year={2026},
+      eprint={2605.25717},
+      archivePrefix={arXiv},
+      primaryClass={cs.AI},
+      url={https://arxiv.org/abs/2605.25717},
+}
 ```
 
 ## License
 
-Released under [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).
-
-## Authors
-
-- João Alves Ribeiro (corresponding), Massachusetts Institute of
-  Technology, `jpar@mit.edu`
-- Bruno Alves Ribeiro, Brown University
-- Francisco Pimenta, University of Porto
-- Sérgio M. O. Tavares, University of Aveiro
-- Faez Ahmed, Massachusetts Institute of Technology, `faez@mit.edu`
-
-## Contact
-
-For questions or issues with the dataset, contact the corresponding
-author João Alves Ribeiro at `jpar@mit.edu`.
+Released under [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).