# digantamisra98/Mish

Official Repository for "Mish: A Self Regularized Non-Monotonic Neural Activation Function" [BMVC 2020]

Repository: https://github.com/digantamisra98/Mish
Canonical: https://ross.abutalabs.com/products/mish
Homepage: https://www.bmva-archive.org.uk/bmvc/2020/assets/papers/0928.pdf
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: deep-learning, activation-functions, mathematics, neural-networks, computer-vision, bmvc, bmvc20, object-detection, image-classification
Last push: 2026-07-20T10:05:16+00:00

## Health v2 (maintenance only)
Score: 74/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 93, release rhythm 35, longevity 100
- inputs: {"age_days": 2672, "days_push": 44, "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 1299, forks 128 (observed 2026-08-28T04:04:17.398631+00:00)

## What it is
Official repository for the Mish activation function, a self-regularized non-monotonic neural activation function published at BMVC 2020. It provides reference implementations, notebooks, and links to optimized variants for use in deep learning models.

## Use cases
- use mish activation function in pytorch
- improve image classification accuracy with activation functions
- find non-monotonic activation function for object detection
- compare activation functions like swish relu and mish
- implement mish in a neural network
- cite the mish paper for research

## When to choose
- you want to experiment with the Mish activation function in deep learning models
- you need the official reference implementation and paper citations
- you are doing computer vision research on activation functions

## When to avoid
- you need a production-optimized CUDA implementation (use mish-cuda instead)
- you need a maintained general-purpose deep learning framework
- you only need standard activations like ReLU built into your framework

## Facets
- artifact type: library
- maturity: maintenance
- function: deep-learning, machine-learning, math
- domain: deep-learning, computer-vision, machine-learning
- platform: python
- tags: activation-function, neural-networks, bmvc-2020, pytorch, research-paper, gpu

## Member repositories
- digantamisra98/Mish (main) score 74

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:17.398631+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 2026-08-30T04:53:25.703177+00:00, confidence not recorded.
  - readme: https://github.com/digantamisra98/Mish (fetched 2026-08-28T04:04:17.398631+00:00, sha 37e484d54141)
- Data as of 2026-08-30T08:39:29.467469+00:00.