multabench/embeddings-tile

MulTaBench text embeddings (TILE) Part of MulTaBench, a benchmark for multimodal tabular learning. Frozen TILE embeddings of the MulTaBench text datasets (Llama-3.1-8B, first 14 layers), cached so the TILE models' leaderboard runs don't re-encode them. TILE reads whole rows, so each <dataset>/embeddings.npz maps every distinct serialized row to one mean-pooled 4096-d vector per string column, the text columns and the ones stored as categories (float16; PCA is fitted per run, on… See the full description on the dataset page: https://huggingface.co/datasets/multabench/embeddings-tile.

<iframe src="https://silytics.si/embed/item/a15eb16f-1bdf-4335-93bc-264b4830739a" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 7 November 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.