Two news papers, one in French and one in English to symbolize the SageMaker text processing algorithm Sequence-to-sequence which performs machione translation of languages

Sequence-to-Sequence Algorithm

ByMichael Stainsbury Last updated date:

SageMaker Sequence-to-Sequence algorithm is used for machine translation of languages. The algorithm takes the input sequence of tokens, for example French words, and outputs the translation as a sequence of English words. As well as translation, Sequence-to-Sequence can be used to summarize a document and convert speech to text. Sequence-to-Sequence is a Supervised Learning algorithm.

Attributes

Problem attributeDescription
Data types and formatText
Learning paradigm or domainTextual analysis, Supervised Learning
Problem typeMachine translation
Use case examplesConvert audio files to text, Summarize a long text corpus, Convert text from one language to other.

Training

Sequence-to Sequence requires the input record to be in recordIO-protobuf with integer values only. Training input files:

  • train.rec
  • val.rec
  • vocab.src.json
  • vocab.trg.json

Model artifacts and inference

DescriptionArtifacts
Learning paradigmSupervised learning
Request formatJSON, recordIO-protobuf
ResultSame format as the choice of request format used
Batch request formatJSON lines
Batch resultJSON lines

Processing environment

Sequence-to-Sequence can only use a single GPU instance, however the single instance may contain multiple GPUs.

Video: Amazon SageMaker’s Built-in Algorithm Webinar Series: Sequence2Sequence

This is a one hour video from AWS.

https://www.youtube.com/watch?v=pZIV5NWfGIU&t=2043s

Credits

French and English newspapers photo by Markus Spiske on Unsplash

Similar Posts

SageMaker text processing algorithms

SageMaker text processing algorithms

ByMichael Stainsbury

There are four SageMaker text processing algorithms: BlazingText, LDA, NTM and Sequence-to-sequence. BlazingText converts text to numeric vectors. LDA and NTM identify topics in text documents and Sequence-to-sequence provides machine translation of languages. Each algorithm has it’s own section and embedded video. These revision notes are part of subdomain 3.2 Select the appropriate model(s) for…

How to evaluate Machine Learning models

How to evaluate Machine Learning models

ByMichael Stainsbury

Evaluating Machine Learning models is the last stage before deploying a model to production. We evaluate Machine Learning models to confirm that they are performing as expected and that they are good enough for the task they were created for. The evaluation stage is performed after model training is finished. Different techniques are used depending…

SageMaker image processing algorithms

SageMaker image processing algorithms

ByMichael Stainsbury

There are three built-in SageMaker image processing algorithms. They are all Supervised Learning algorithms and so have to be trained using labelled data. Each one analyzes images in a different way and returns different inference data for downstream processing. SageMaker’s three built-in image processing algorithms each have their own way of visualizing real word objects….

Factorization Machines Algorithm

Factorization Machines Algorithm

ByMichael Stainsbury

The Factorization Machines Algorithm has two modes: Classification and Regression. Classification is a binary method that returns either one or zero and a label which is a number. The Regression mode returns the predicted value. Factorization Machines are a good choice for high dimensional, sparse datasets. Common uses are web page click prediction and item…