Add paper page link
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,27 +1,28 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
library_name: pytorch
|
|
|
|
| 4 |
pipeline_tag: tabular-regression
|
| 5 |
tags:
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
---
|
| 13 |
|
| 14 |
# LinearPFN
|
| 15 |
|
| 16 |
**Bayesian variable selection for linear models with interactions, in one forward pass.**
|
| 17 |
|
|
|
|
|
|
|
| 18 |
A transformer pretrained on a spike-and-slab prior over main effects and pairwise
|
| 19 |
interactions. Given a dataset, it returns posterior inclusion probabilities and
|
| 20 |
posterior-mean coefficients, with no MCMC and no per-dataset fitting.
|
| 21 |
|
| 22 |
Code and documentation: [github.com/schiekiera/LinearPFN](https://github.com/schiekiera/LinearPFN)
|
| 23 |
|
| 24 |
-
|
| 25 |
## Use
|
| 26 |
|
| 27 |
```bash
|
|
@@ -38,7 +39,6 @@ result = model.fit(X, y)
|
|
| 38 |
print(result.summary())
|
| 39 |
```
|
| 40 |
|
| 41 |
-
|
| 42 |
## Model
|
| 43 |
|
| 44 |
| | |
|
|
@@ -56,7 +56,6 @@ The outputs are the posterior under the training prior: linear main effects, pai
|
|
| 56 |
interactions, strong heredity, Gaussian noise. Missing values and categorical predictors
|
| 57 |
must be handled before fitting.
|
| 58 |
|
| 59 |
-
|
| 60 |
## Files
|
| 61 |
|
| 62 |
| | |
|
|
@@ -79,4 +78,4 @@ must be handled before fitting.
|
|
| 79 |
}
|
| 80 |
```
|
| 81 |
|
| 82 |
-
License: Apache 2.0.
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
library_name: pytorch
|
| 3 |
+
license: apache-2.0
|
| 4 |
pipeline_tag: tabular-regression
|
| 5 |
tags:
|
| 6 |
+
- variable-selection
|
| 7 |
+
- bayesian
|
| 8 |
+
- spike-and-slab
|
| 9 |
+
- prior-data-fitted-network
|
| 10 |
+
- in-context-learning
|
| 11 |
+
- tabular-regression
|
| 12 |
---
|
| 13 |
|
| 14 |
# LinearPFN
|
| 15 |
|
| 16 |
**Bayesian variable selection for linear models with interactions, in one forward pass.**
|
| 17 |
|
| 18 |
+
This model is described in [LinearPFN: Amortized Variable Selection for Linear Models with Interactions](https://huggingface.co/papers/2610.06580).
|
| 19 |
+
|
| 20 |
A transformer pretrained on a spike-and-slab prior over main effects and pairwise
|
| 21 |
interactions. Given a dataset, it returns posterior inclusion probabilities and
|
| 22 |
posterior-mean coefficients, with no MCMC and no per-dataset fitting.
|
| 23 |
|
| 24 |
Code and documentation: [github.com/schiekiera/LinearPFN](https://github.com/schiekiera/LinearPFN)
|
| 25 |
|
|
|
|
| 26 |
## Use
|
| 27 |
|
| 28 |
```bash
|
|
|
|
| 39 |
print(result.summary())
|
| 40 |
```
|
| 41 |
|
|
|
|
| 42 |
## Model
|
| 43 |
|
| 44 |
| | |
|
|
|
|
| 56 |
interactions, strong heredity, Gaussian noise. Missing values and categorical predictors
|
| 57 |
must be handled before fitting.
|
| 58 |
|
|
|
|
| 59 |
## Files
|
| 60 |
|
| 61 |
| | |
|
|
|
|
| 78 |
}
|
| 79 |
```
|
| 80 |
|
| 81 |
+
License: Apache 2.0.
|