Quickstart
Browse files
README.md
CHANGED
@@ -6,36 +6,78 @@ pipeline_tag: tabular-classification
|
|
6 |
---
|
7 |
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
---
|
7 |
|
8 |
|
9 |
+
<p align="center">
|
10 |
+
<img src="https://raw.githubusercontent.com/alanarazi7/TabSTAR/main/figures/tabstar_logo.png" alt="TabSTAR Logo" width="50%">
|
11 |
+
</p>
|
12 |
+
|
13 |
+
---
|
14 |
+
|
15 |
+
## Install
|
16 |
+
|
17 |
+
To fit a pretrained TabSTAR model to your own dataset, install the package:
|
18 |
+
|
19 |
+
```bash
|
20 |
+
pip install tabstar
|
21 |
+
```
|
22 |
+
|
23 |
+
---
|
24 |
+
|
25 |
+
## Quickstart Example
|
26 |
+
|
27 |
+
```python
|
28 |
+
from importlib.resources import files
|
29 |
+
import pandas as pd
|
30 |
+
from sklearn.metrics import classification_report
|
31 |
+
from sklearn.model_selection import train_test_split
|
32 |
+
|
33 |
+
from tabstar.tabstar_model import TabSTARClassifier
|
34 |
+
|
35 |
+
# Load sample data
|
36 |
+
csv_path = files("tabstar").joinpath("resources", "imdb.csv")
|
37 |
+
x = pd.read_csv(csv_path)
|
38 |
+
y = x.pop('Genre_is_Drama')
|
39 |
+
|
40 |
+
# Split train/test
|
41 |
+
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.1)
|
42 |
+
|
43 |
+
# Initialize and train
|
44 |
+
tabstar = TabSTARClassifier()
|
45 |
+
tabstar.fit(x_train, y_train)
|
46 |
+
|
47 |
+
# Predict and evaluate
|
48 |
+
y
|
49 |
+
```
|
50 |
+
|
51 |
+
---
|
52 |
+
|
53 |
+
# 📚 TabSTAR: A Foundation Tabular Model With Semantically Target-Aware Representations
|
54 |
+
|
55 |
+
**Repository:** [alanarazi7/TabSTAR](https://github.com/alanarazi7/TabSTAR)
|
56 |
+
|
57 |
+
**Paper:** [TabSTAR: A Foundation Tabular Model With Semantically Target-Aware Representations](https://arxiv.org/abs/2505.18125)
|
58 |
+
|
59 |
+
**License:** MIT © Alan Arazi et al.
|
60 |
+
|
61 |
+
---
|
62 |
+
|
63 |
+
## Abstract
|
64 |
+
|
65 |
+
> While deep learning has achieved remarkable success across many domains, it
|
66 |
+
> has historically underperformed on tabular learning tasks, which remain
|
67 |
+
> dominated by gradient boosting decision trees (GBDTs). However, recent
|
68 |
+
> advancements are paving the way for Tabular Foundation Models, which can
|
69 |
+
> leverage real-world knowledge and generalize across diverse datasets,
|
70 |
+
> particularly when the data contains free-text. Although incorporating language
|
71 |
+
> model capabilities into tabular tasks has been explored, most existing methods
|
72 |
+
> utilize static, target-agnostic textual representations, limiting their
|
73 |
+
> effectiveness. We introduce TabSTAR: a Foundation Tabular Model with
|
74 |
+
> Semantically Target-Aware Representations. TabSTAR is designed to enable
|
75 |
+
> transfer learning on tabular data with textual features, with an architecture
|
76 |
+
> free of dataset-specific parameters. It unfreezes a pretrained text encoder and
|
77 |
+
> takes as input target tokens, which provide the model with the context needed
|
78 |
+
> to learn task-specific embeddings. TabSTAR achieves state-of-the-art
|
79 |
+
> performance for both medium- and large-sized datasets across known benchmarks
|
80 |
+
> of classification tasks with text features, and its pretraining phase exhibits
|
81 |
+
> scaling laws in the number of datasets, offering a pathway for further
|
82 |
+
> performance improvements.
|
83 |
+
|