Practice Test

You need to identify the language of customer reviews by using the Azure AI Language SDK for Python. Which method should you call?

detect_language()

You are developing an application with the Azure AI Language SDK that must recognize named entities in 40 support tickets. You submit all 40 documents to the service at once using client.recognize_entities(), but your code produced an error. What happened?

NER only accepts batches of 5 or less per call

When using the OpenAI Responses API and a vision-enabled model, images can be provided only as base64-encoded image data.

False

Your company needs to reduce the cost and latency of a generative AI workload that performs simple, high-volume classification tasks. Which type of model should you deploy?

a small language model (SLM)

Byte-pair encoding (BPE) — the tokenization algorithm used by GPT models — does what?

Common character sequences get a single token ID; rare sequences are split into multiple tokens

You have a Microsoft Foundry project that has a generative AI model deployment. You need to ensure that responses generated by the model minimize costs and remain within a defined length. Which parameter should you configure?

Max Completion Tokens

A model that learns to classify support tickets after being shown thousands of labeled (ticket → category) examples is using which type of learning?

Supervised learning

You call recognize_entities() and need the most fine-grained classification of an entity, such as distinguishing a "City" within the broader "Location" category. Which entity property provides this?

entity.subcategory

To route each recognized PII entity to the correct location in the original document for redaction, you use the entity's offset and _______ properties.

length

The Speech service's recognize_once() method is suitable for transcribing a single utterance that ends when the speaker pauses

True

You call the Azure AI Language service through the REST API to detect language. Select the appropriate keys to complete the request body. Each key may be used once, more than once, or not at all. body = { " kind " : "LanguageDetection" , " analysisInput " : { "documents" : data }, " parameters " : { "modelVersion" : "latest" } }

kind; analysisInput; parameters

At its core, what is a Large Language Model (LLM)?

A next-token predictor that outputs a probability distribution and samples from it

When using the OpenAI Responses API and a vision-enabled model, you can include an image in a request by providing the image as ______

base64-encoded image data

Voice Live combines speech to text, reasoning, and text to speech into a single conversational experience

True

How are LLM API requests typically priced?

Per token of input AND output, with output usually 3-4× the input rate

You need to link entities mentioned in an article to a well-known knowledge base and retrieve a URL for each entity. Which Azure AI Language method should you call?

recognize_linked_entities()

You are developing an application that uses the Azure OpenAI Responses API and needs to maintain conversation state across multiple turns without resending the full history. Which parameter should you use?

previous_response_id

What does top_p (nucleus sampling) do?

Sorts tokens by probability, keeps the smallest set whose cumulative probability is at least p, renormalizes, and samples from that set

You need to build a solution that identifies the location of each vehicle in a traffic photo and returns a bounding box for each one. Which computer vision task should you use?

object detection

Hallucination is a consequence of a model always predicting a plausible next token.

True

A trainee asks a gpt model "what year was the Acme client onboarded?" and gets a confident but wrong answer. The most likely cause is:

The model completing a plausible pattern in the absence of grounded information

The role="system" message in a chat completions call is used to:

Set persistent behavior — persona, constraints, format, refusal rules

Evaluating model outcomes across demographic groups to reduce bias is an example of the Microsoft responsible AI principle of

fairness

You need to build an AI solution that produces new product images based on written descriptions provided by users. Which AI workload should you use?

image generation

You are using the Azure AI Language SDK to remove sensitive information, such as social security numbers and phone numbers, from customer records before storing them. Which method should you use?

recognize_pii_entities()

You are extracting text from scanned documents by using Azure AI Document Intelligence. You need to return only the raw text content without extracting any structured fields. Which prebuilt model ID should you use?

prebuilt-read

Setting temperature to 0 in the OpenAI / Azure OpenAI API guarantees identical output across two consecutive calls.

False

You are developing a Python application that calls a gpt-5 reasoning model deployed in Microsoft Foundry by using the Chat Completions API. The application fails with an HTTP 400 error whenever it sends a request. You need to resolve the error. Which parameter should you remove from the request?

temperature

In the Foundry playground, you can upload a local image and include text in the same message when prompting a multimodal model.

True

You have a Microsoft Foundry project that contains a generative AI model deployment. You test the model by using the Foundry playground. You need to develop an application that sends requests to the deployed model. Which information must the application include to call the model?

The model endpoint and authentication credentials

You are developing an application that processes voicemail recordings by using Azure Content Understanding in Foundry Tools. Which feature does Azure Content Understanding use to convert audio to text?

transcription

You are developing an AI-powered customer support application. Which task is an example of the Microsoft responsible AI principle of inclusiveness?

Design the interface to support multiple languages and screen readers.

The _________ is used for comparing and deploying a wide range of models for generative AI development in Microsoft Foundry.

Model catalog

An AI agent uses a generative AI model to establish actions based on user input.

True

A generative AI model produces output by predicting the next token based on patterns learned from the model's training data.

True

You are developing an application that converts text into spoken audio and saves the synthesized audio to a file by using Azure Speech in Foundry Tools. How should you complete the Python code below? import azure.cognitiveservices.speech as speechsdk … speech_config = speechsdk.SpeechConfig(subscription=key, region=region) audio_config = speechsdk.audio. [ answer ] synthesizer = speechsdk.SpeechSynthesizer( speech_config=speech_config, audio_config=audio_config ) …

AudioOutputConfig(filename="output.wav")

You need to extract printed and handwritten text from photographed receipts and get per-word confidence scores by using Azure AI Vision. Which visual feature should you request?

VisualFeatures.READ

You are calling the Azure AI Language service through the REST API. Which HTTP header should you use to authenticate the request with a resource key?

Ocp-Apim-Subscription-Key

You have a Microsoft Foundry project that contains a vision-enabled model deployment. You need to develop an application that sends a message containing text and an image URL. The solution must ensure the quickest response time. Which message structure should you include in the request?

a user message that includes both a text item and an image item in the content array

You have an Azure subscription. You need to use Azure Content Understanding in Foundry Tools to extract structured data from invoices. What should you provision?

a Microsoft Foundry resource

As a rough estimate, one token corresponds to approximately ______ of English text.

4 characters

An AI workload that produces new content based on user input is an example of _______

generative AI

You are using the Azure Speech SDK to build a Python application that transcribes a single short audio file to text. Which method should you call on the SpeechRecognizer object?

recognize_once_async()

You are developing a Python application that transcribes a long meeting recording in real time and processes each finalized phrase as it becomes available.

Use start_continuous_recognition() and handle the recognized event.

Voice Live returns only transcribed text

False

Which of the following are part of Microsoft's six Responsible AI principles? Select all that apply.

Reliability and safety; Inclusiveness; Accountability; Privacy and security; Fairness; Transparency

______ defines which fields to extract when analyzing content by using Azure Content Understanding in Foundry Tools.

A schema

For a brainstorming or creative drafting task, the most appropriate range for the temperature parameter is:

0.7 - 0.9

An LLM API call is stateless — the model has no memory between requests, so the full conversation history must be re-sent on every call.

True

You need to authenticate to an Azure AI service by using Microsoft Entra ID (keyless) instead of a resource key. Which credential should you use in the Python SDK?

DefaultAzureCredential()

You are developing a web app that processes invoices to calculate expenses. You need to extract structured fields, including nested values, from the invoices by using a defined schema. What should you use?

an analyzer in Azure Content Understanding in Foundry Tools

You are choosing a text analysis approach. The requirement is to summarize long documents into a few key sentences without training a custom model. Which Azure AI Language capability should you use?

extractive summarization

A small language model (SLM) such as the Phi family requires less compute than a typical large language model (LLM).

True

You need to synthesize speech in which specific words are emphasized, pauses are inserted, and the speaking rate changes mid-sentence. Which method should you call?

synthesizer.speak_ssml_async(ssml).get()

When you initialize the OpenAI client to call a model deployed in Microsoft Foundry, the value that you pass to the model parameter is the _______

model deployment name

You need to generate new images from text prompts in Microsoft Foundry. Which type of model should you deploy from the model catalog?

a diffusion model

You need to determine the overall opinion, and the opinions about specific aspects such as "food" and "service", in restaurant reviews by using the Azure AI Language SDK.

Call analyze_sentiment() with show_opinion_mining=True.

You are developing a voice application that listens for spoken commands and converts them into text by using Azure Speech in Foundry Tools. How should you complete the Python code below? import azure.cognitiveservices.speech as speechsdk … speech_config = speechsdk.SpeechConfig(subscription=key, region=region) recognizer = speechsdk.SpeechRecognizer(speech_config=speech_config) [ answer ]

recognizer.recognize_once()

You are developing an application that analyze invoices by using Azure Content Understanding in Foundry Tools. You need to ensure that the application retrieves the analysis results after processing completes. How should you complete the Python code below? poller = client.begin_analyze( analyzer_id= "invoice" , input_url=url ) result = poller. [answer] ()

result

You need to convert written customer notifications into natural-sounding spoken audio that can be played over a phone system. Which Azure Speech in Foundry Tools capability should you use?

speech synthesis

What innovation introduced by the Transformer architecture powers modern LLMs?

Self-attention over the input sequence

Voice Live requires you to separately implement speech to text and text to speech services

False

To reduce the risk of biased or non-inclusive language in image captions, you can set the _______ parameter when calling Azure AI Vision.

gender_neutral_caption

Your company processes customer support emails. You need to implement an AI solution that automatically identifies mentions of people, organizations, and locations in the emails. Which text analysis technique should you use?

Named Entity Recognition (NER)

What is a context window in a GenAI model?

The fixed token budget covering input AND output for a single LLM call

You are building a ticket classifier that must produce the same category for the same input. Which temperature should you use to best acheive this?

0

Azure Content Understanding in Foundry Tools results are returned in the JSON format.

True

You detect the language of ambiguous short strings and want to bias detection toward United States content. Select the answer that correctly completes the code below: result = client.detect_language( documents=documents, [ answer ] = "us" )[ 0 ] print (result.primary_language.iso6391_name)

country_hint

You are building a speech-to-text feature that captures audio from the computer's microphone in real time rather than from a file. How should you configure the AudioConfig ?

speechsdk.audio.AudioConfig(use_default_microphone=True)

When content is submitted to Azure Content Understanding in Foundry Tools, the analysis is _________.

asynchronous

You need to synthesize speech that emphasizes certain words, inserts pauses, and controls the speaking rate. What allows the speech API to do this?

a Speech Synthesis Markup Language (SSML) document

You are using the Azure Speech SDK to develop a Python application that supports real-time spoken conversations. Which Azure Speech class should you use to configure the connection to the Azure Speech service?

SpeechConfig

You are developing a Python application that uses Azure AI Vision to generate a description of an image located at a public URL. Which method should you call on the ImageAnalysisClient ?

analyze_from_url()

Azure AI Language sentiment analysis can only return a sentiment of positive, neutral, or negative

False

An AI agent can call tools in a loop to take actions on behalf of a user.

True

You need to build an AI solution that creates marketing email drafts based on a short description of a product and its target audience. Which AI workload should you use?

generative AI

Match each Azure AI service to the best-suited task — Extract key phrases and named entities from support tickets:

Azure AI Language

Match each Azure AI service to the best-suited task — Convert a written notification into spoken audio:

Azure AI Speech

Match each Azure AI service to the best-suited task — Generate a caption for a product photo:

Azure AI Vision

Match each Azure AI service to the best-suited task — Extract the invoice total and vendor name from a scanned PDF:

Azure AI Document Intelligence

Match each Azure AI SDK client to its correct authentication pattern — TextAnalyticsClient (Azure AI Language):

AzureKeyCredential(key)

Match each Azure AI SDK client to its correct authentication pattern — OpenAI client for a Foundry model:

AzureOpenAI(azure_endpoint=…, azure_ad_token_provider=…) using DefaultAzureCredential

Match each Azure AI SDK client to its correct authentication pattern — SpeechRecognizer (Azure Speech):

SpeechConfig(subscription=key, region=region)