tigershen23

This is being recorded, so we are going to upload it to Disco after. So if you need to stretch your legs or grab a cup of tea, feel free to do so. And if you're sharing this on social media, tag Dribble. We will repost everything that you guys share, okay? Without further ado, I would love to introduce our first Q&A guest speaker of this course week one. We have got Lenora, a senior product designer based in Miami. She's worked on design systems at Skillshare and Salesforce and is a super talented force to be reckoned with in the design systems world. So we're so grateful and happy that she is with us today. Start asking your questions. I already see one in there. And without further ado, we will get Lenora up. Hello everyone. Hi, you're right. I have worked on Salesforce and on Skillshare. I am now with Digital Ocean working through their systems and their design, so it's been awesome working with them. I think about four months. So yeah, nice. I guess beginning steps of working on a new system, which is awe

tigershen23

This is being recorded, so we are going to upload it to Disco after. So if you need to stretch your legs or grab a cup of tea, feel free to do so. And if you're sharing this on social media, tag Dribble. We will repost everything that you guys share, okay? Without further ado, I would love to introduce our first Q&A guest speaker of this course week one. We have got Lenora, a senior product designer based in Miami. She's worked on design systems at Skillshare and Salesforce and is a super talented force to be reckoned with in the design systems world. So we're so grateful and happy that she is with us today. Start asking your questions. I already see one in there. And without further ado, we will get Lenora up. Hello everyone. Hi, you're right. I have worked on Salesforce and on Skillshare. I am now with Digital Ocean working through their systems and their design, so it's been awesome working with them. I think about four months. So yeah, nice. I guess beginning steps of working on a new system, which is awe

tigershen23

• Thor::Util#snake_case: well-isolated test cases. Clear and thorough about expected behavior with potentially ambiguous input
• HTTParty.normalize_base_uri: describes behavior well for a method that might be confusing at first
• Virtus::Attribute#==: good use of intermediate variable for clarity. tests are collectively exhaustive within the scope of the method
• AASM::Core::Event#fire: good isolation of complex behavior; good use of mocks and doubles. when/then/expect division is clear (BDD). good use of context
• [Money.from_amount](https://gi

tigershen23

adam_params = {"lr": 0.0005}
optimizer = Adam(adam_params)
svi = SVI(model, guide, optimizer, loss=Trace_ELBO())

n_steps = 2501
for step in range(n_steps):
  svi.step(data)
  if step % 100 == 0:
    print_progress()

tigershen23

def print_progress():
  alphas = pyro.param("alphas")
  betas = pyro.param("betas")
  
  if torch.cuda.is_available():
    alphas.cuda()
    betas.cuda()

  means = alphas / (alphas + betas)
  normalized_means = means / torch.sum(means)

tigershen23

def guide(data):
  alphas = pyro.param('alphas', torch.tensor(6.).expand(3), constraint=constraints.positive)
  betas = pyro.param('betas', torch.tensor(10.).expand(3), constraint=constraints.positive) 

  pyro.sample('pay_probs', dist.Beta(alphas, betas).independent(1))

tigershen23

def model(data):
  alpha = torch.tensor(6.0)
  beta = torch.tensor(10.0)
  pay_probs = pyro.sample('pay_probs', dist.Beta(alpha, beta).expand(3).independent(1))
  normalized_pay_probs = pay_probs / torch.sum(pay_probs)

  with pyro.iarange('data_loop', len(data)):
    pyro.sample('obs', dist.Categorical(probs=normalized_pay_probs), obs=data)

tigershen23

#                 Tiger           Jason          James
data = torch.cat((torch.zeros(9), torch.ones(7), torch.empty(4).fill_(2.)))

tigershen23

pyro.enable_validation(True)
pyro.clear_param_store()

tigershen23

from __future__ import print_function
import math
import os
import torch
import torch.distributions.constraints as constraints
import pyro
from pyro.optim import Adam
from pyro.infer import SVI, Trace_ELBO
import pyro.distributions as dist