# lucidrains/flamingo-pytorch

Implementation of 🦩 Flamingo, state-of-the-art few-shot visual question answering attention net out of Deepmind, in Pytorch

Repository: https://github.com/lucidrains/flamingo-pytorch
Canonical: https://ross.abutalabs.com/products/flamingo-pytorch
Language: Python
License: MIT
License Family: permissive
Topics: artificial-intelligence, attention-mechanism, deep-learning, transformers, visual-question-answering
Last push: 2022-10-18T21:44:45+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1588, "days_push": 1415, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1269, forks 66 (observed 2026-08-28T04:04:11.666961+00:00)

## What it is
A PyTorch implementation of DeepMind's Flamingo visual language model architecture, providing the Perceiver Resampler and Gated Cross-Attention blocks for few-shot visual question answering. It is a research building block meant to be inserted into a large language model so text can attend to image embeddings.

## Use cases
- implement the Flamingo architecture in PyTorch
- add visual question answering to a large language model
- resample image embeddings with a perceiver resampler
- build a multimodal vision-language model from scratch
- insert gated cross attention between text and media embeddings
- experiment with few-shot visual language model training

## When to choose
- you want a lightweight, pip-installable PyTorch implementation of Flamingo's core components
- you are building a custom multimodal LLM and need the perceiver resampler or gated cross-attention blocks
- you want readable research code to study or adapt the Flamingo architecture

## When to avoid
- you need a fully trained, ready-to-use visual question answering model with pretrained weights
- you want a complete production multimodal pipeline rather than architectural components
- you need a framework-agnostic or non-PyTorch solution

## Facets
- artifact type: library
- maturity: active
- function: deep-learning, machine-learning, llm-training, transformers
- domain: artificial-intelligence, deep-learning, large-language-models, computer-vision
- platform: python
- tags: flamingo, visual-question-answering, perceiver-resampler, gated-cross-attention, vision-language-model, research-implementation, pytorch, multimodal, natural-language-processing

## Member repositories
- lucidrains/flamingo-pytorch (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:11.666961+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 2026-08-30T05:03:32.467344+00:00, confidence not recorded.
  - readme: https://github.com/lucidrains/flamingo-pytorch (fetched 2026-08-28T04:04:11.666961+00:00, sha 05ba2deb3dda)
  - registry_pypi: https://pypi.org/pypi/flamingo-pytorch/json (fetched 2026-08-29T12:15:09.976476+00:00, sha 600b018ead33)
- Data as of 2026-08-30T08:39:29.467469+00:00.