Verified project record

ludwig-ai/ludwig

ludwig is a declarative deep learning framework that lets users train, fine-tune, and deploy AI models from LLMs to tabular data using a YAML configuration file without writing boilerplate Python code. It supports mixed input modalities including text, numbers, images, audio, and timeseries.

Project overview

By replacing training loops, data pipelines, and evaluation boilerplate with a single declarative YAML configuration, ludwig removes the need to write custom Python for each new model, making multi-modal and LLM fine-tuning accessible through a CLI, Python API, or library call.

Project type: Model Development · Model Runtime
Use cases: Data Analysis
Deployment: Refer to project documentation
License: Apache-2.0

Best for

Developers, data teams, and AI engineers who prefer to define models and training pipelines declaratively rather than writing custom training loops.
Teams that need to work across both LLM fine-tuning and multimodal or tabular model training within a single framework.

Key capabilities

Supports supervised fine-tuning (SFT), alignment training (DPO, KTO, ORPO, GRPO), and various PEFT adapters (LoRA, DoRA, VeRA) for large language models.
Train models on mixed input modalities including text, numbers, images, audio, and timeseries with a variety of encoders and combiners.

Getting started

Install the framework via pip with a specific extras group such as ludwig[full] or ludwig[llm], then run ludwig train --config model.yaml --dataset my_data.csv to start training from a YAML configuration and a dataset.

Evidence and sources

README: Ludwig is a **declarative deep learning framework** that lets you train, fine-tune, and deploy AI models — from LLM fine-tuning to tabular classification — using a YAML config fil…
README: ludwig train --config model.yaml --dataset my_data.csv
README: export HUGGING_FACE_HUB_TOKEN="<your_token>" ludwig train --config model.yaml --dataset "ludwig://alpaca"
README: pip install ludwig # core pip install ludwig[full] # all optional dependencies pip install ludwig[llm] # LLM fine-tuning only
README: Ludwig supports the full LLM fine-tuning spectrum: | Technique | Config key | | --------------------------------- | ---------------------------------------------------------------…

Reviewed at: 2026-07-28T20:36:50Z