Skip to content
This repository has been archived by the owner on Nov 16, 2023. It is now read-only.

QA Quickstart #139

Merged
merged 7 commits into from
Jul 3, 2019
Merged
Show file tree
Hide file tree
Changes from 6 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
1 change: 1 addition & 0 deletions scenarios/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@ The following summarizes each scenario of the best practice notebooks. Each scen
|---| ------------------------ | -------------------------------------------- | ------------------- |
|[Text Classification](scenarios/text_classification) |Topic Classification|en, zh, ar|BERT|
|[Named Entity Recognition](scenarios/named_entity_recognition) |Wikipedia NER | en, zh |BERT|
|[Question Answering](scenarios/question_answering) |SQuAD | en |BiDAF|
|[Sentence Similarity](scenarios/sentence_similarity) |STS Benchmark |en|Representation: TF-IDF, Word Embeddings, Doc Embeddings<br>Metrics: Cosine Similarity, Word Mover's Distance|
|[Embeddings](scenarios/embeddings)| Custom Embeddings Training|en|Word2Vec<br>fastText<br>GloVe|

Expand Down
16 changes: 16 additions & 0 deletions scenarios/question_answering/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,16 @@
# Question Answering (QA)

This folder contains examples and best practices, written in Jupyter notebooks, for building question answering models. These models can be used in a wide variety of applications, such as search engines, and virtual assistants.

## What is Question Answering?

Question Answering is a classical NLP task which consists of determining the relevant "answer" (snippet of text out of a provided passage) that answers a user's "question". This task is a subset of Machine Comprehension, or measuring how well a machine comprehends a passage of text. The Stanford Question Answering Dataset ([SQuAD](https://rajpurkar.github.io/SQuAD-explorer/)) leaderboard displays the state-of-the-art models in this space. Traditional QA models are varients of Bidirectional Recurrent Neural (BRNN) Networks.

## Summary

The following summarizes each notebook for Question Answering. Each notebook provides more details and guiding in principles on building state of the art models.

|Notebook|Runs Local|Description|
|---|---|---|
|[Deployed QA System in Under 20 minutes](question_answering_system_bidaf_quickstart.ipynb)| No| Learn how to deploy a QA system in under 20 minutes using Azure Container Instances (ACI) and a popular AllenNLP pre-trained model called BiDAF.