# yandex-research/tabm

(ICLR 2025) TabM: Advancing Tabular Deep Learning With Parameter-Efficient Ensembling

Repository: https://github.com/yandex-research/tabm
Canonical: https://ross.abutalabs.com/products/tabm
Homepage: https://arxiv.org/abs/2410.24210
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-11-10T07:18:37+00:00

## Health v2 (maintenance only)
Score: 36/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 51, release rhythm 10, longevity 47
- inputs: {"age_days": 671, "days_push": 296, "days_rel": 384, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1099, forks 98 (observed 2026-08-28T04:03:35.039422+00:00)

## What it is
TabM is a PyTorch-based deep learning model for tabular data that efficiently imitates an ensemble of MLPs through parameter-efficient ensembling. It is the official implementation of an ICLR 2025 paper and ships as a Python package plus paper reproduction code.

## Use cases
- train a neural network on tabular data
- improve on gradient boosting for tabular prediction
- build an efficient ensemble of MLPs for classification or regression
- win Kaggle competitions with tabular datasets
- apply deep learning to large tabular datasets with millions of rows
- find a strong deep learning baseline for tabular benchmarks

## When to choose
- you need strong accuracy on tabular data with better efficiency than deep ensembles or transformer-based tabular models
- you want a practical tabular DL model that scales to large datasets (10M+ rows)
- you want a competitive alternative or complement to GBDT models like XGBoost

## When to avoid
- you need a quick baseline and GBDT already meets your accuracy and latency needs
- your data is not tabular (images, text, audio)
- you need a model with extensive interpretability tooling

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning
- domain: machine-learning, data-science
- platform: python
- tags: tabular-data, pytorch, ensembling, mlp, kaggle, iclr-2025, supervised-learning

## Member repositories
- yandex-research/tabm (main) score 36

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:35.039422+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:46:11.908548+00:00, confidence not recorded.
  - readme: https://github.com/yandex-research/tabm (fetched 2026-08-28T04:03:35.039422+00:00, sha 0eb3e4df6c0d)
  - homepage: https://arxiv.org/abs/2410.24210 (fetched 2026-08-29T12:49:41.002074+00:00, sha 48b31ef89eaf)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T12:49:41.049623+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T12:49:41.146334+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T12:49:41.203963+00:00, sha a1f16f915a9a)
  - registry_pypi: https://pypi.org/pypi/tabm/json (fetched 2026-08-29T12:49:41.252880+00:00, sha 33e5b3bfeeed)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T12:49:41.081709+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.