Return to page

H2O Hydrogen Torch

Democratize State-of-the-Art Deep Learning

hydrogen-torch-full-c-3-x3x.png hydrogen-torch-full-c-3-x

H2O Hydrogen Torch is democratizing AI by simplifying and streamlining the process of making state-of-the-art deep learning models for all data scientists, from the novice to the expert. Deep learning is able to learn and make intelligent decisions on its own, even with limited and unstructured data available. H2O Hydrogen Torch unlocks value from this unstructured data to help teams understand it at scale. For the enterprise, H2O Hydrogen achieves AI transformation by changing the way they deliver value to both customers and internal teams.  

Deep Learning models prove more powerful and flexible than classical machine learning in solving complex problems such as natural language processing (NLP) and computer vision. However, making deep learning models that can address diverse problem types requires extensive data science expertise and time. H2O Hydrogen Torch solves these challenges by providing an intuitive user interface and highly optimized training routines, developed by the world’s top Grandmaster data scientists, for common deep learning problems. With H2O Hydrogen Torch, world-class deep learning models can be rapidly trained and deployed on the H2O AI Cloud to solve the most critical problems that businesses face today.

Grow Revenue

Optimize e-commerce web search and inventory

Build question-answering systems specific to the business

Mitigate Risk

Improve diagnostic imagery detection for patient care

Locate critical information from medical transcripts

Optimize Operations

Detect required pieces during assembly process

Categorize incoming emails by urgency

Personalize Experiences

Predict customer satisfaction from audio transcriptions

Simplify texts containing domain-specific terms

How Data Scientists use H2O Hydrogen Torch


Classification and Regression

Object Detection

Semantic Segmentation

Metric Learning


Classification and Regression

Token Classification

Span prediction

Sequence to Sequence

Metric Learning


Live stream object detection

Video Segmentation

Audio Audio


Classification and regression


Democratizes Deep Learning

Train state-of-the-art models with no coding needed

Deploy models easily to external Python environments or directly to H2O MLOps


Quickly experiment with hyperparameters to tune model performance


Ready-to-use library of problem types that continuously expands based on Data Science trends



Train deep learning models using training techniques developed by Kaggle Grandmasters 

Achieve results by using the best neural network architectures and transfer learning 

Upload your data to H2O Hydrogen Torch using our web interface, allowing you to handle a wide variety of unstructured data, including: 

  • Text (eg, transcriptions, emails)  with multi-lingual and domain-specific support

  • Images (data from photos and image analysis) 

  • Video (data captured into video that can be segmented into frames for analysis)

  • Audio (speech and any audio recordings)

Set up and manage deep learning experiments to maximize your model development. 

  • Choose model architecture and most relevant  tuning parameters 

  • Quick start options provide the most important settings

  • Expert settings provide a wide range of training approaches

Select the training set and problem type within H2O Hydrogen Torch, and it will automatically learn the data and create models

  • AI-ML engine that uses multiple computer vision and NLP algorithms for diverse AI tasks (eg, entity recognition, grouping and set identification)

  • Optimize neural network and perform hyperparameter tuning without the need to change any code 

  • Inspect and analyze your models within the tool.

  • Track progress during model training, assessing model accuracy and analyzing predictions of the model

  • Reveal errors in the data that supply ideas on how to improve the model

H2O Hydrogen Torch packages all the models automatically and provides a format which can be published to: 

  • MLOPS, which is part of the H2O AI Hybrid Cloud

  • Your cloud or on-premises environment of choice

Apply deep learning models made with H2O Hydrogen Torch across a variety of business problems by: 

  • Developing use-case specific apps built with H2O Wave 

  • Importing into existing applications via APIs. 

Watch to get an overview of H2O Hydrogen Torch.

Learn about natural language processing (NLP) use cases, and how to rapidly build an NLP model.


Prediction Insights that give Data Scientists ideas for how to improve models for next iterations.

Streamlined model deployment native to the H2O AI Cloud that can also be imported to other model serving platforms

Comprehensive set of techniques for deep learning model tuning selected from the best practices of Kaggle Grandmasters


Meets novice Data Scientists where they are with a no-code UI and fosters growth opportunities through practical experience and documentation


Multi GPU training support

Supported Deep Learning Problem Types

Text Data and Natural Language Processing (NLP) 

Text Classification and Regression

Predict customer satisfaction from transcribed phone calls and categorize incoming emails requesting customer support to send to relevant departments

Token Classification

Extract entities such as disease names from medical text with multi-lingual support 

Span Prediction

Find relevant information from medical transcripts and build a company-specific question-answering system


Simplify text that contains domain-specific terms and summarize text for better understanding of its content

Metric Learning

Detect fake reviews that display similarities and find similar questions in user forums

Image and Video Data

Image Classification and Regression

Predict images for abnormalities in medical X-rays and classify types of landscapes from drone photos. 

Object Detection

Detect vehicles from traffic and drone cameras and find objects required for assembly line processes.

Semantic Segmentation

Locate objects of interest in medical images, segment roads from video dashcams and remove objects from the background

Instance Segmentation

Detect and delineate distinct objects of interest in biological images, segment individual cars, pedestrians, or other objects from a video captured using a dashcam

Metric Learning

Find similar images to those in a training database, such as unauthorized use of a brand logo or duplicate items on a retail website

Featured Success Story

quotation mark

H2O Hydrogen Torch has been a key enabler in helping us operationalize machine learning for shifting data. We can get from a new dataset to a deployed model and updated tables in our data warehouse in a couple of days instead of weeks.”


Get Started

H2O Hydrogen Torch is available in the H2O AI Cloud. Signup for the 90-day free trial and start exploring.