Trendoday

LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning

Multimodal tabular learning involves classification and regression on tables combining structured numerical and categorical features with unstructured text and images. These unstructured fields are co

  • Upvotes14
–Rank now
#3Peak rank
2Days on list
307Score · 7d
Nov 19, 2026First seen
Rank · 7 days

About this paper

LightTAR adapts dimensionality reduction to the prediction task by adding target-aware features from encoder layers via regularized partial least squares, improving multimodal tabular model performance with minimal runtime overhead.

Multimodal tabular learning involves classification and regression on tables combining structured numerical and categorical features with unstructured text and images. These unstructured fields are commonly represented by pretrained encoder embeddings reduced with principal component analysis (PCA). This reduction prioritizes input variance but may overlook information relevant to predicting the target. We introduce LightTAR, a simple, inexpensive method that adapts dimensionality reduction to the prediction task. LightTAR augments the standard PCA representation with a few target-aware features extracted jointly from the final and an intermediate encoder layer using regularized partial…

multimodal tabular learningpretrained encoder embeddingsprincipal component analysis (PCA)dimensionality reductiontarget-aware featuresintermediate encoder layerregularized partial least squares (PLS)tree ensemblesneural networkstabular foundation models
Upvotes
14
Authors
4
Published
Nov 12, 2026

Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.

Trending history

LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning was first seen on Hugging Face · Daily papers on Nov 19, 2026. Since then it has appeared in 13 checks on 2 different days. Its best position was #3 of 30, reached on Nov 20, 05:29 UTC. It dropped off the list after Nov 20, 11:29 UTC.

Rank history

Position on Hugging Face · Daily papers at every check (lower is better).

#1#11#20#30 11:2917:2923:2905:2911:29 Nov 19, 11:29 UTC · #8Nov 19, 13:29 UTC · #8Nov 19, 15:29 UTC · #8Nov 19, 17:29 UTC · #8Nov 19, 19:29 UTC · #8Nov 19, 21:29 UTC · #8Nov 19, 23:29 UTC · #8Nov 20, 01:29 UTC · #8Nov 20, 03:29 UTC · #8Nov 20, 05:29 UTC · #3Nov 20, 07:29 UTC · #7Nov 20, 09:29 UTC · #8Nov 20, 11:29 UTC · #9

Last 14 days

Best rank per day (UTC).

Nov 9today

Metric history

Platform metric as reported by Hugging Face.

071320 11:2917:2923:2905:2911:29 Nov 19, 11:29 UTC · 14Nov 19, 13:29 UTC · 15Nov 19, 15:29 UTC · 19Nov 19, 17:29 UTC · 19Nov 19, 19:29 UTC · 20Nov 19, 21:29 UTC · 20Nov 19, 23:29 UTC · 20Nov 20, 01:29 UTC · 20Nov 20, 03:29 UTC · 20Nov 20, 05:29 UTC · 20Nov 20, 07:29 UTC · 20Nov 20, 09:29 UTC · 20Nov 20, 11:29 UTC · 20

Questions

Is LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning trending on HF Daily Papers right now?

Not at the moment. It was last on the list on Nov 20, 11:29 UTC, after 2 days of trending.

What is the highest rank LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning reached?

#3 on Hugging Face · Daily papers, on Nov 20, 05:29 UTC. Trendoday checks this list every 120 minutes.

How long has LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning been trending?

It was first seen on Nov 19, 2026 and has been on the list on 2 days (13 checks) since.