ExamTopics Logo
Mail Us[email protected]
Menu
Microsoft Discussions
exam questions

Exam DP-100 All Questions

View all questions & answers for the DP-100 exam
Go to Exam

Exam DP-100 topic 3 question 81 discussion

Actual exam question from Microsoft's DP-100
Question #: 81
Topic #: 3
[All DP-100 Questions]

You are training machine learning models in Azure Machine Learning. You use Hyperdrive to tune the hyperparameters.
In previous model training and tuning runs, many models showed similar performance.
You need to select an early termination policy that meets the following requirements:
✑ accounts for the performance of all previous runs when evaluating the current run avoids comparing the current run with only the best performing run to date

Which two early termination policies should you use? Each correct answer presents part of the solution.
NOTE: Each correct selection is worth one point.

  • A. Median stopping
  • B. Bandit
  • C. Default
  • D. Truncation selection
Show Suggested Answer Hide Answer
Suggested Answer: AD 🗳️
by guddusao at July 13, 2021, 12:30 p.m.

Comments

Chosen Answer:
This is a voting comment (?). It is better to Upvote an existing comment if you don't have anything to add.
Switch to a voting comment New
Submit Cancel
dushmantha
Highly Voted 2 years, 10 months ago
Condition 1: account for all previous runs Condition 2: avoid comparing with only best performing run up to date 1) Condition 1: ok, Condition 2: ok (calculates running avg and its median at every step) 2) Condition 1: ok, Condition 2: no (slack transformed value is compared with previous best value) 3) Condition 1: no, Condition 2: no (no termination) 4) Condition 1: ok, Condition 2: ok (to get lowest performing runs need to account for all runs)
upvoted 25 times
...
phdykd
Most Recent 11 months, 2 weeks ago
A and D
upvoted 1 times
...
Peeking
1 year, 3 months ago
Selected Answer: AD
Median stopping is an early termination policy based on running averages of primary metrics reported by the runs. This policy computes running averages across all training runs and stops runs whose primary metric value is worse than the median of the averages. Truncation selection cancels a percentage of lowest performing runs at each evaluation interval. Runs are compared using the primary metric.
upvoted 2 times
...
phdykd
1 year, 4 months ago
So, the recommended early termination policies in this case are A (Median Stopping) and B (Bandit) because they both account for the performance of all previous runs when evaluating the current run and do not rely only on the best performing run to date.
upvoted 2 times
...
phdykd
1 year, 4 months ago
A. Median stopping B. Bandit
upvoted 2 times
...
therealola
2 years ago
On exam 18-06-22
upvoted 2 times
...
JTWang
2 years, 2 months ago
on exam 04/22/2022
upvoted 1 times
...
synapse
2 years, 3 months ago
Selected Answer: AD
AD see Dushmantha explain
upvoted 3 times
...
yuanxuan1
2 years, 4 months ago
Selected Answer: AD
answer is AD
upvoted 2 times
...
dija123
2 years, 6 months ago
Selected Answer: AD
I agree with AD
upvoted 4 times
...
tunaktunak
2 years, 7 months ago
On exam 26/11/2021
upvoted 2 times
...
VJPrakash
2 years, 10 months ago
It should be B and D(Truncate instead of Default). The default as per documentation means no termination policy.
upvoted 3 times
pancman
2 years, 2 months ago
The correct answer should be A and D.
upvoted 1 times
...
...
trickerk
2 years, 11 months ago
Given answers are correct. - Truncation cancels a percentage of lowest performing runs at each evaluation interval; - Bandit policy compares the value (Y + Y * slack_factor) to AUC value, and if smaller, cancels the run. So "Median stopping policy" and "Default" are correct answers. https://docs.microsoft.com/en-us/azure/machine-learning/how-to-tune-hyperparameters https://docs.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive.banditpolicy?view=azure-ml-py#definition
upvoted 3 times
trickerk
2 years, 11 months ago
"accounts for the performance of all previous runs"
upvoted 1 times
manualrg
1 year, 5 months ago
To apply truncation policy , a percentile must be computed, so indeed it uses "performance of all previous runs" IMHO
upvoted 1 times
...
...
pancman
2 years, 2 months ago
You're wrong. The default policy is no early termination. Therefore it doesn't satisfy the requirement in the question. The correct answer is median and truncation. (A and D)
upvoted 1 times
...
...
slash_nyk
2 years, 11 months ago
I take my words back. Median and Bandit look for best performing runs.. Truncation cancels at each interval
upvoted 2 times
...
slash_nyk
2 years, 11 months ago
Median and Truncation are the correct answers
upvoted 4 times
YipingRuan
2 years, 11 months ago
"Truncation selection cancels a percentage of lowest performing runs at [each evaluation interval]."
upvoted 1 times
...
...
guddusao
2 years, 11 months ago
I don't think default would be there. The right answer would be Median stopping policy and truncate selection policy both supports early termination policy.
upvoted 3 times
saurabh288
2 years, 11 months ago
Truncation selection doesn't stop the run.
upvoted 1 times
NickData90
2 years, 11 months ago
How does a "termination policy" not stop a run? If I look at the docs it clearly says that it looks at all runs and cancels a percentage of this each interval: https://docs.microsoft.com/en-us/python/api/azureml-train-core/azureml.train.hyperdrive.truncationselectionpolicy?view=azure-ml-py
upvoted 4 times
...
...
...
Community vote distribution
A (35%)
C (25%)
B (20%)
Other
Most Voted
A voting comment increases the vote count for the chosen answer by one.

Upvoting a comment with a selected answer will also increase the vote count towards that answer by one. So if you see a comment that you already agree with, you can upvote it instead of posting a new comment.

SaveCancel