AWS ML - keshavbaweja-git/guides GitHub Wiki
Ground Truth Plus
Helps customers create high quality training data sets.
No need to build labeling applications and manage labeling workforce.
Provides mutually agreed upon upfront SLA for label quality.
Provides a feedback option on labels through a review interface.
Inference Recommender
Automates load testing and optimizes model performance across machine learning (ML) instances
Reduces the time it takes to get ML models from development to production in a cost effective way
Provides recommendations for best price performant instance type and endpoint configuration for model deployment
Free to use, customer only pays for instance usage during testing
Training Compiler
Optimize training runs
Deep Learning Container, but at this time tested again Hugging Face NLP models
PyTorch or TensorFlow
Minimal code changes
Amazon Lex Automated Chatbot Designer
Extends Amazon Lex by analyzing thousands of lines of transcripts in a couple of hours to give you an automated initial bot design that includes common intents
Integrated with Amazon Connect Contact Lens
Amazon Textract AnalyzeID
API for Amazon Textract that will automatically extract relevant information from identification documents, such as driver’s licenses and passports, without the need for templates or configuration
Amazon Personalize
New recommenders optimized for retail and media & entertainment use cases
New recipe type called USER_SEGMENTATION which includes two new recipes, aws-item-affinity and aws-item-attribute, both of which can be used to identify users based on their preferences
Amazon Kendra - Kendra Experience Builder
Deploy a fully functional and customizable search experience in a few clicks, without any coding or ML experience.
Use an intuitive, visual workflow to build, customize, and launch a Kendra powered search application
Amazon Kendra - Kendra Search Analytics Dashboard
View quality and usability metrics associated with a Kendra powered search application
Custom Document Enrichment build a custom ingestion pipeline that can pre-process documents before they get indexed into Kendra