# karpathy/neuraltalk

NeuralTalk is a Python+numpy project for learning Multimodal Recurrent Neural Networks that describe images with sentences.

Repository: https://github.com/karpathy/neuraltalk
Canonical: https://ross.abutalabs.com/products/neuraltalk
Language: Python
License Family: other
Last push: 2020-12-22T22:26:36+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 4297, "days_push": 2080, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5503, forks 1332 (observed 2026-08-28T04:09:21.726359+00:00)

## What it is
NeuralTalk is a Python+numpy implementation of Multimodal Recurrent Neural Networks that generate natural-language descriptions of images. It trains and evaluates CNN+RNN/LSTM image captioning models on Flickr8K, Flickr30K, and MSCOCO datasets.

## Use cases
- generate captions for images with a neural network
- train an image captioning model on MSCOCO
- learn how multimodal RNNs work
- evaluate image captions with BLEU score
- reproduce CNN+LSTM image description research

## When to choose
- you want to study the original educational implementation of image captioning RNNs
- you need a pure Python+numpy reference implementation without deep learning frameworks

## When to avoid
- you want to actually run or train image captioning in production - it is deprecated and slow
- you need GPU acceleration or CNN finetuning - use NeuralTalk2 instead
- you need Python 3 support - this requires Python 2.7

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, deep-learning, nlp, computer-vision, image-processing
- domain: deep-learning, computer-vision, machine-learning
- platform: python, cross-platform
- tags: image-captioning, recurrent-neural-networks, deprecated, educational, numpy, lstm, mscoco, natural-language-processing, linux

## Member repositories
- karpathy/neuraltalk (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:21.726359+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 2026-08-29T17:56:17.560939+00:00, confidence not recorded.
  - readme: https://github.com/karpathy/neuraltalk (fetched 2026-08-28T04:09:21.726359+00:00, sha 63a080857cf0)
- Data as of 2026-08-30T08:39:29.467469+00:00.