karpathy/neuraltalk

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

5503 stars · 1332 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

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

library · maturity abandoned

machine-learning deep-learning nlp computer-vision image-processing deep-learning computer-vision machine-learning python cross-platform image-captioning recurrent-neural-networks deprecated educational numpy lstm mscoco natural-language-processing linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="karpathy/neuraltalk")