What exactly has TabPFN learned to do?
Fuente:
arXiv
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910824481685504 |
|---|---|
| author | McCarter, Calvin |
| author_facet | McCarter, Calvin |
| contents | TabPFN [Hollmann et al., 2023], a Transformer model pretrained to perform in-context learning on fresh tabular classification problems, was presented at the last ICLR conference. To better understand its behavior, we treat it as a black-box function approximator generator and observe its generated function approximations on a varied selection of training datasets. Exploring its learned inductive biases in this manner, we observe behavior that is at turns either brilliant or baffling. We conclude this post with thoughts on how these results might inform the development, evaluation, and application of prior-data fitted networks (PFNs) in the future. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_08978 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | What exactly has TabPFN learned to do? McCarter, Calvin Machine Learning TabPFN [Hollmann et al., 2023], a Transformer model pretrained to perform in-context learning on fresh tabular classification problems, was presented at the last ICLR conference. To better understand its behavior, we treat it as a black-box function approximator generator and observe its generated function approximations on a varied selection of training datasets. Exploring its learned inductive biases in this manner, we observe behavior that is at turns either brilliant or baffling. We conclude this post with thoughts on how these results might inform the development, evaluation, and application of prior-data fitted networks (PFNs) in the future. |
| title | What exactly has TabPFN learned to do? |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2502.08978 |