# median-research-group/LibMTL

A PyTorch Library for Multi-Task Learning

Repository: https://github.com/median-research-group/LibMTL
Canonical: https://ross.abutalabs.com/products/libmtl
Language: Python
License: MIT
License Family: permissive
Topics: multitask-learning, multi-task-learning, pytorch, python, multi-domain-learning, deep-learning, mtl, mmoe, ple, multiobjective-optimization, multi-objective-optimization
Last push: 2025-05-14T15:06:50+00:00

## Health v2 (maintenance only)
Score: 42/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 21, release rhythm 35, longevity 100
- inputs: {"age_days": 1719, "days_push": 476, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2586, forks 232 (observed 2026-08-28T04:07:02.698008+00:00)

## What it is
LibMTL is an open-source PyTorch library for Multi-Task Learning (MTL). It provides implementations of many MTL architectures and gradient-based multi-objective optimization methods with a unified API.

## Use cases
- train multi-task learning models in pytorch
- compare MTL weighting algorithms like PCGrad and CAGrad
- implement MMoE or PLE architectures
- run multi-objective optimization for deep learning
- benchmark gradient manipulation methods for MTL

## When to choose
- you need a unified PyTorch framework for multi-task learning experiments
- you want to evaluate many recent MTL optimization methods without reimplementing them
- you need architectures like MMoE and PLE out of the box

## When to avoid
- you need single-task training only
- you work outside PyTorch
- you need production serving rather than research training

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning
- domain: machine-learning, deep-learning
- platform: python
- tags: pytorch, multi-task-learning, mtl, multiobjective-optimization, mmoe, ple

## Member repositories
- median-research-group/LibMTL (main) score 42

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:02.698008+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 2026-08-30T02:22:26.572113+00:00, confidence not recorded.
  - readme: https://github.com/median-research-group/LibMTL (fetched 2026-08-28T04:07:02.698008+00:00, sha 1453414a9238)
  - registry_pypi: https://pypi.org/pypi/libmtl/json (fetched 2026-08-29T10:04:48.023256+00:00, sha c814a1a392d4)
- Data as of 2026-08-30T08:39:29.467469+00:00.