Schnellstart: Verwenden der Groundedness-Erkennung (Vorschau)

Im folgenden Abschnitt wird eine Beispielanforderung mit cURL erläutert. Fügen Sie den nachstehenden Befehl in einen Text-Editor ein, und nehmen Sie folgende Änderungen vor.

1. Ersetzen Sie <endpoint> durch die Endpunkt-URL, die Ihrer Ressource zugeordnet ist.

2. Ersetzen Sie <your_subscription_key> durch einen der Schlüssel für Ihre Ressource.

3. Ersetzen Sie optional Felder "query" oder "text" im Text durch Ihren eigenen Text für die Analyse.

   curl --location --request POST '<endpoint>/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview' \
   --header 'Ocp-Apim-Subscription-Key: <your_subscription_key>' \
   --header 'Content-Type: application/json' \
   --data-raw '{
     "domain": "Generic",
     "task": "QnA",
     "qna": {
       "query": "How much does she currently get paid per hour at the bank?"
     },
     "text": "12/hour",
     "groundingSources": [
       "I'm 21 years old and I need to make a decision about the next two years of my life. Within a week. I currently work for a bank that requires strict sales goals to meet. IF they aren't met three times (three months) you're canned. They pay me 10/hour and it's not unheard of to get a raise in 6ish months. The issue is, **I'm not a salesperson**. That's not my personality. I'm amazing at customer service, I have the most positive customer service \"reports\" done about me in the short time I've worked here. A coworker asked \"do you ask for people to fill these out? you have a ton\". That being said, I have a job opportunity at Chase Bank as a part time teller. What makes this decision so hard is that at my current job, I get 40 hours and Chase could only offer me 20 hours/week. Drive time to my current job is also 21 miles **one way** while Chase is literally 1.8 miles from my house, allowing me to go home for lunch. I do have an apartment and an awesome roommate that I know wont be late on his portion of rent, so paying bills with 20hours a week isn't the issue. It's the spending money and being broke all the time.\n\nI previously worked at Wal-Mart and took home just about 400 dollars every other week. So I know i can survive on this income. I just don't know whether I should go for Chase as I could definitely see myself having a career there. I'm a math major likely going to become an actuary, so Chase could provide excellent opportunities for me **eventually**."
     ],
     "reasoning": false
   }'
   

Öffnen Sie ein Eingabeaufforderungsfenster, und führen Sie den cURL-Befehl aus.

Erstellen Sie eine neue Python-Datei namens quickstart.py. Öffnen Sie die Datei in Ihrem bevorzugten Editor oder Ihrer bevorzugten IDE.

1. Ersetzen Sie den Inhalt von quickstart.py durch den folgenden Code. Geben Sie Ihre Endpunkt-URL und ihren Schlüssel in die entsprechenden Felder ein. Ersetzen Sie optional Felder "query" oder "text" im Text durch Ihren eigenen Text für die Analyse.

   import http.client
   import json
   
   endpoint = "<your_custom_subdomain>.cognitiveservices.azure.com"
   conn = http.client.HTTPSConnection(endpoint)
   payload = json.dumps({
     "domain": "Generic",
     "task": "QnA",
     "qna": {
       "query": "How much does she currently get paid per hour at the bank?"
     },
     "text": "12/hour",
     "groundingSources": [
       "I'm 21 years old and I need to make a decision about the next two years of my life. Within a week. I currently work for a bank that requires strict sales goals to meet. IF they aren't met three times (three months) you're canned. They pay me 10/hour and it's not unheard of to get a raise in 6ish months. The issue is, **I'm not a salesperson**. That's not my personality. I'm amazing at customer service, I have the most positive customer service \"reports\" done about me in the short time I've worked here. A coworker asked \"do you ask for people to fill these out? you have a ton\". That being said, I have a job opportunity at Chase Bank as a part time teller. What makes this decision so hard is that at my current job, I get 40 hours and Chase could only offer me 20 hours/week. Drive time to my current job is also 21 miles **one way** while Chase is literally 1.8 miles from my house, allowing me to go home for lunch. I do have an apartment and an awesome roommate that I know wont be late on his portion of rent, so paying bills with 20hours a week isn't the issue. It's the spending money and being broke all the time.\n\nI previously worked at Wal-Mart and took home just about 400 dollars every other week. So I know i can survive on this income. I just don't know whether I should go for Chase as I could definitely see myself having a career there. I'm a math major likely going to become an actuary, so Chase could provide excellent opportunities for me **eventually**."
     ],
     "reasoning": false
   })
   headers = {
     'Ocp-Apim-Subscription-Key': '<your_subscription_key>',
     'Content-Type': 'application/json'
   }
   conn.request("POST", "/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview", payload, headers)
   res = conn.getresponse()
   data = res.read()
   print(data.decode("utf-8"))
   

   Wichtig

   Wir empfehlen die Microsoft Entra ID-Authentifizierung mit verwalteten Identitäten für Azure-Ressourcen , um zu vermeiden, dass Anmeldeinformationen mit Ihren Anwendungen gespeichert werden, die in der Cloud ausgeführt werden.

   Verwenden Sie API-Schlüssel mit Vorsicht. Fügen Sie den API-Schlüssel nicht direkt in Ihren Code ein, und machen Sie ihn nicht öffentlich zugänglich. Wenn Sie API-Schlüssel verwenden, speichern Sie sie sicher in Azure Key Vault, rotieren Sie die Schlüssel regelmäßig, und beschränken Sie den Zugriff auf Azure Key Vault mithilfe von rollenbasierter Zugriffssteuerung und Netzwerkzugriffsbeschränkungen. Weitere Informationen zur sicheren Verwendung von API-Schlüsseln in Ihren Apps finden Sie unter API-Schlüssel mit Azure Key Vault.

   Weitere Informationen zur Sicherheit von KI-Diensten finden Sie unter Authentifizieren von Anforderungen an Azure AI-Dienste.

2. Führen Sie die Anwendung mit Befehl python aus:

   python quickstart.py
   

   Die Ausgabe der Ergebnisse dauert einen Moment.

Im folgenden Abschnitt wird eine Beispielanforderung mit cURL erläutert. Fügen Sie den nachstehenden Befehl in einen Text-Editor ein, und nehmen Sie folgende Änderungen vor.

1. Ersetzen Sie <endpoint> durch die Endpunkt-URL, die Ihrer Ressource in der Azure KI Inhaltssicherheit zugeordnet ist.

2. Ersetzen Sie <your_subscription_key> durch einen der Schlüssel für Ihre Ressource.

3. Ersetzen Sie <your_OpenAI_endpoint> durch die Endpunkt-URL, die Ihrer Azure OpenAI-Ressource zugeordnet ist.

4. Ersetzen Sie <your_deployment_name> durch den Namen Ihrer Azure-OpenAI-Bereitstellung.

5. Ersetzen Sie optional Felder "query" oder "text" im Text durch Ihren eigenen Text für die Analyse.

   curl --location --request POST '<endpoint>/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview' \
   --header 'Ocp-Apim-Subscription-Key: <your_subscription_key>' \
   --header 'Content-Type: application/json' \
   --data-raw '{
     "domain": "Generic",
     "task": "QnA",
     "qna": {
       "query": "How much does she currently get paid per hour at the bank?"
     },
     "text": "12/hour",
     "groundingSources": [
       "I'm 21 years old and I need to make a decision about the next two years of my life. Within a week. I currently work for a bank that requires strict sales goals to meet. If they aren't met three times (three months) you're canned. They pay me 10/hour and it's not unheard of to get a raise in 6ish months. The issue is, **I'm not a salesperson**. That's not my personality. I'm amazing at customer service, I have the most positive customer service \"reports\" done about me in the short time I've worked here. A coworker asked \"do you ask for people to fill these out? you have a ton\". That being said, I have a job opportunity at Chase Bank as a part time teller. What makes this decision so hard is that at my current job, I get 40 hours and Chase could only offer me 20 hours/week. Drive time to my current job is also 21 miles **one way** while Chase is literally 1.8 miles from my house, allowing me to go home for lunch. I do have an apartment and an awesome roommate that I know wont be late on his portion of rent, so paying bills with 20hours a week isn't the issue. It's the spending money and being broke all the time.\n\nI previously worked at Wal-Mart and took home just about 400 dollars every other week. So I know i can survive on this income. I just don't know whether I should go for Chase as I could definitely see myself having a career there. I'm a math major likely going to become an actuary, so Chase could provide excellent opportunities for me **eventually**."
     ],
     "reasoning": true,
     "llmResource": {
           "resourceType": "AzureOpenAI",
           "azureOpenAIEndpoint": "<your_OpenAI_endpoint>",
           "azureOpenAIDeploymentName": "<your_deployment_name>"
   }'
   

6. Öffnen Sie ein Eingabeaufforderungsfenster, und führen Sie den cURL-Befehl aus.

Erstellen Sie eine neue Python-Datei namens quickstart.py. Öffnen Sie die Datei in Ihrem bevorzugten Editor oder Ihrer bevorzugten IDE.

1. Ersetzen Sie den Inhalt von quickstart.py durch den folgenden Code. Geben Sie Ihre Endpunkt-URL und ihren Schlüssel in die entsprechenden Felder ein. Geben Sie außerdem Ihren Azure OpenAI-Endpunkt und den Bereitstellungsnamen ein. Ersetzen Sie optional Felder "query" oder "text" im Text durch Ihren eigenen Text für die Analyse.

   import http.client
   import json
   
   conn = http.client.HTTPSConnection("<endpoint>/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview")
   payload = json.dumps({
     "domain": "Generic",
     "task": "QnA",
     "qna": {
       "query": "How much does she currently get paid per hour at the bank?"
     },
     "text": "12/hour",
     "groundingSources": [
       "I'm 21 years old and I need to make a decision about the next two years of my life. Within a week. I currently work for a bank that requires strict sales goals to meet. IF they aren't met three times (three months) you're canned. They pay me 10/hour and it's not unheard of to get a raise in 6ish months. The issue is, **I'm not a salesperson**. That's not my personality. I'm amazing at customer service, I have the most positive customer service \"reports\" done about me in the short time I've worked here. A coworker asked \"do you ask for people to fill these out? you have a ton\". That being said, I have a job opportunity at Chase Bank as a part time teller. What makes this decision so hard is that at my current job, I get 40 hours and Chase could only offer me 20 hours/week. Drive time to my current job is also 21 miles **one way** while Chase is literally 1.8 miles from my house, allowing me to go home for lunch. I do have an apartment and an awesome roommate that I know wont be late on his portion of rent, so paying bills with 20hours a week isn't the issue. It's the spending money and being broke all the time.\n\nI previously worked at Wal-Mart and took home just about 400 dollars every other week. So I know i can survive on this income. I just don't know whether I should go for Chase as I could definitely see myself having a career there. I'm a math major likely going to become an actuary, so Chase could provide excellent opportunities for me **eventually**."
     ],
     "reasoning": True
     "llmResource": {
      "resourceType": "AzureOpenAI",
      "azureOpenAIEndpoint": "<your_OpenAI_endpoint>",
      "azureOpenAIDeploymentName": "<your_deployment_name>"
     }
   })
   headers = {
     'Ocp-Apim-Subscription-Key': '<your_subscription_key>',
     'Content-Type': 'application/json'
   }
   conn.request("POST", "/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview", payload, headers)
   res = conn.getresponse()
   data = res.read()
   print(data.decode("utf-8"))
   

2. Führen Sie die Anwendung mit Befehl python aus:

   python quickstart.py
   

   Die Ausgabe der Ergebnisse dauert einen Moment.

In diesem Abschnitt wird eine Beispielanforderung mit cURL veranschaulicht. Ersetzen Sie die Platzhalter nach Bedarf:

• Ersetzen Sie <endpoint> durch Ihre Ressourcenendpunkt-URL.
• Ersetzen Sie <your_subscription_key> durch Ihren Abonnementschlüssel.
• Ersetzen Sie optional das Feld „Text“ durch den Text, den Sie analysieren möchten.

curl --location --request POST '<endpoint>/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview' \
--header 'Ocp-Apim-Subscription-Key: <your_subscription_key>' \
--header 'Content-Type: application/json' \
--data-raw '{
  "domain": "Generic",
  "task": "Summarization",
  "text": "The patient name is Kevin.",
  "groundingSources": [
    "The patient name is Jane."
  ],
  "correction": true,
  "llmResource": {
        "resourceType": "AzureOpenAI",
        "azureOpenAIEndpoint": "<your_OpenAI_endpoint>",
        "azureOpenAIDeploymentName": "<your_deployment_name>"
  }
}'

Erstellen Sie ein Python-Skript mit dem Namen quickstart.py, und fügen Sie den folgenden Code hinzu. Aktualisieren Sie die Endpunkt-URL und den Schlüssel entsprechend:

conn = http.client.HTTPSConnection("<endpoint>/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview")
payload = json.dumps({
  "domain": "Generic",
  "task": "Summarization",
  "text": "The patient name is Kevin.",
  "groundingSources": [
    "The patient name is Jane."
  ],
  "correction": True,
  "llmResource": {
   "resourceType": "AzureOpenAI",
   "azureOpenAIEndpoint": "<your_OpenAI_endpoint>",
   "azureOpenAIDeploymentName": "<your_deployment_name>"
  }
})
headers = {
  'Ocp-Apim-Subscription-Key': '<your_subscription_key>',
  'Content-Type': 'application/json'
}
conn.request("POST", "/contentsafety/text:detectGroundedness?api-version=2024-09-15-preview", payload, headers)
res = conn.getresponse()
data = res.read()
print(data.decode("utf-8"))