amaiya/ktrain

Health v2 · maintenance only

Adoption not part of the score

1268 stars · 260 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

ktrain is a lightweight Python wrapper around TensorFlow Keras that provides pre-canned, low-code models for text, vision, graph, and tabular data. It aims to make deep learning accessible with only a few lines of code for tasks like text classification, NER, and image classification.

Use cases

• classify text with BERT in a few lines of code
• train a named entity recognition model
• fine-tune transformer models for text regression
• build image classifiers without writing Keras boilerplate
• apply graph neural networks to node classification
• train models on tabular data easily
• learn deep learning as a beginner with simple API

When to choose

• you want fast, low-code deep learning on TensorFlow Keras
• you need pre-canned models for text, vision, graph, or tabular data
• you are a newcomer who wants accessible ML workflows
• you want fastai-like ergonomics in the TensorFlow ecosystem

When to avoid

• you need PyTorch instead of TensorFlow
• you require full low-level control over model architecture and training loops
• you need generative question-answering, which moved to the OnPrem.LLM package
• you need production-scale distributed training features

Facets

library · maturity active

machine-learning deep-learning nlp computer-vision llm-training machine-learning deep-learning computer-vision data-science python keras tensorflow text-classification ner graph-neural-networks tabular-data low-code-ml fastai-inspired natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="amaiya/ktrain")