Chris Van Pelt vanpelt

Overview

In preperation for our 10 week course we want to offer up some resources for students interested in getting a head start.

Educational Resources

We have a series of online videos that assume no knowledge of deep learning or machine learning and cover all of the basic applications. Each video is around 20 minutes. If you want to get a feel for what the class will be like or brush up on the fundamentals, feel free to watch them at your own pace: https://www.wandb.com/classes/intro/overview. We will cover a lot of these topics in more depth in person.

The only real requirement for these classes is a solid understanding of python. If you can follow along with the videos, you will be in good shape. If not, you may want to brush up on your python with a book or an online course. The books Deep Learning with Python and Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems are excellent compliments to the class materials

Overview

The task at hand is to create a sortable tableview of issues listed in the tensorflow/tensorflow github repository using React, ApolloClient, and the GitHub GraphQL API.

MVP

The MVP should render the first page of issues with a hard coded access token. The look and feel is less important than the function, but still considered.

cml_iframe

To communicate with CrowdFlower run the following whenever events occur requiring data exchange:

window._cf.post("persist", {"key": "value"})

Option 1: CML iFrame

Include the following javascript snippet in the <head> of your custom interface:

<script src="https://crowdflower.github.io/construe/cml_iframe.js"></script>

Then there are two required calls for your interface to communicate with the CrowdFlower platform:

	#!/bin/bash

	# move all services to runit, was tricky to make this happen in docker without
	# overwriting cron / sshd
	echo "*** Copying services to runit"
	mv /home/wandb/service/* /etc/service/
	mv /home/wandb/wandb-logrotate /etc/logrotate.d/

	HOST=${HOST:-http://localhost:8080}

	import sagemaker
	import wandb
	tuner_name = "sagemaker-pytorch-190111-1056"
	tuner = sagemaker.HyperparameterTuner.attach(tuner_name)
	job_name = tuner.analytics().description()['TrainingJobDefinition']['StaticHyperParameters']['sagemaker_job_name']
	client = tuner.sagemaker_session.sagemaker_client

	api = wandb.Api()
	runs = api.runs("wandb/sm-pytorch-cifar", {"config.sagemaker_job_name": job_name})
	for run in runs:

	absl-py==0.2.2
	alembic==1.0.0
	argh==0.26.2
	astor==0.6.2
	async-generator==1.9
	backcall==0.1.0
	bleach==1.5.0
	certifi==2018.4.16
	chainer==4.1.0
	chardet==3.0.4

	from app import db
	from sqlalchemy import func
	from sqlalchemy.dialects.mysql import insert

	def upsert(model, insert_dict):
	"""model can be a db.Model or a table(), insert_dict should contain a primary or unique key."""
	inserted = insert(model).values(**insert_dict)
	upserted = inserted.on_duplicate_key_update(
	id=func.LAST_INSERT_ID(model.id), **{k: inserted.inserted[k]
	for k, v in insert_dict.items()})

	//The test question interface doesn't have require.js, this mimics minimal functionality
	var require = window.require \|\| function() {
	var args = Array.prototype.slice.call(arguments);
	var opts_or_list = args.shift(), opts, list
	if(Array.isArray(opts_or_list)) {
	opts = {}, list = opts_or_list
	} else {
	list = args.shift(), opts = opts_or_list
	}
	var libraries = ['//cdnjs.cloudflare.com/ajax/libs/twitter-bootstrap/2.3.2/js/bootstrap.min.js']

	{
	"shapes":[
	{"type": "predicted_box", "width": 100, "height": 100, "x": 100, "y": 100, "rotation": 180, "tag": "optional class", "probability": 0.8},
	{"type": "predicted_poly", "points": [100, 100, 150, 150, 100, 100], "tag": "optional class", "probability": 0.8},
	{"type": "predicted_dots", "dots": [{"x":100, "y":100, "tag": "optional class"}], "tag": "optional class", "probability": 0.8}
	]
	}