Exam DP-100 topic 3 question 29 discussion

Box 1 should be NO: only grid sampling does an exhaustive search https://docs.microsoft.com/en-us/azure/machine-learning/how-to-tune-hyperparameters#define-search-space

It should be: No, Yes, No, No

For those who are still thinking it must be Grid Sampling: All combination + Reduce computing resource , because "Research has shown that this method (Random Grid Sweep) yields the same results, but is more efficient computationally." I think D would be the best choice Referred from: Topic 1 Question 37 in DP-100 Credit: @lookaaaa

Based on the Hyperdrive configuration code snippet provided: 1. False - Using RandomParameterSampling does not test every possible parameter combination, only randomly selected values from the specified distributions. 2. True - The learning_rate parameter is configured to select values randomly from a normal distribution with mean 10 and standard deviation 3. 3. False - The keep_probability parameter uses a uniform distribution between 0.05 and 0.1, so it can take on any value in that range. 4. False - The number_of_hidden_layers parameter samples from a choice of discrete values from 3 to 5, not from a normal distribution.

No, Yes, No, No

No,Yes,No,No

was on exam 07/March/2023

No, Yes, Yes, Yes

on 03 Jan 2023

Please note that range() in Python is excluding the upper boundary value, see https://docs.microsoft.com/en-us/azure/machine-learning/how-to-tune-hyperparameters: command_job_for_sweep = command_job( batch_size=Choice(values=[16, 32, 64, 128]), number_of_hidden_layers=Choice(values=range(1,5)), command_job_for_sweep = command_job( batch_size=Choice(values=[16, 32, 64, 128]), number_of_hidden_layers=Choice(values=range(1,5)), ) In this case, batch_size one of the values [16, 32, 64, 128] and number_of_hidden_layers takes one of the values [1, 2, 3, 4].

Similar question on exam 18-06-22

It should be No, Yes,No, No. It is random sampling not grid sampling, so not all possible combinations will be tested (which would be impossible here anyway, since some of the parameters are coming from continuous distributions).

on exam 18/03/2022

on 7 Nov 2021, same question about RandomSample, but different question (very similar), be aware of different distributions and traits of RandomSample

on 19Oct2021

on exam 9/24/2021

On exam 2021/7/10