-
Notifications
You must be signed in to change notification settings - Fork 5
Home
- Azure Functions V2 with Python is currently in public preview with Linux Consumption Plan.
- Currently cold start is 10-12 secs. Improvements are planned.
- Exploring Functions on AKS to use GPU's https://medium.com/@asavaritayal/azure-functions-on-kubernetes-75486225dac0
This sample demonstrates LDA topic modeling of Gutenberg books using Gensim/NLTK/Azure Python Storage SDK/ PyLDAVis Python libraries.
curl -X POST -H 'Content-Type:application/json' -d '{ "container_name" : "janeausten", "num_topics" : 20 }' https://gutenbergery.azurewebsites.net/api/TrainBookcode=mhu/Ihigx/0wgEtuyRybGDXRSah0vJ3wdsGT7rd2MMuuZOMCPFauqw==https://gutenbergbooks.blob.core.windows.net/janeaustenmodels/ldamodel.html
Performs the following tasks:
- On HttpTrigger, loads dataset from blob, cleans dataset (removes newlines and carriage returns etc)
- Sentence tokenizes it and removes stopwords (NLTK)
- Builds a dictionary using this data and feeds it to Gensim for topic modeling
- Runs 15 passes of modeling , num_topics we need is 20 (this is just a human readable number)
- Uses the model, dictionary, corpus and visualizes topics using PyLDAVis
- Supposed to lemmatize the words (see Gaps/Issues below)
- Outputs the visualized HTML (PyLDAVIS)
Performs the following
Loads a pretrained Keras model from a local function app zip blob. Accepts image URL's and uses Keras model for inferencing. Takes in URL's of images that show camping gear, uses Keras model to predict classes for the gear.
curl -X POST -H 'Content-Type:application/json' -d '{"urls":["https://i.pinimg.com/originals/ab/66/90/ab669021ae492a7a53e3e7bcb8abf160.jpg","https://i.pinimg.com/236x/3b/c8/a6/3bc8a639ed669f6d9bf029bcf433d3fd--backpacking-packs-hydration-pack.jpg"]}' https://campidentify.azurewebsites.net/api/CampingGear
Performs the following
Loads an inception V3 uncompressed pretrained model Accepts image URL's and infers the image using a simple HTTP Trigger.
