Send human evals via API

If you are using the Logging API, you can pass the positive_feedback field in the request to record if the user liked the output.

positive_feedback is a boolean field that indicates if the user liked the output.

payload = {
    "model": "gpt-4o",
    # ...
    "positive_feedback": True, # means the user liked the output
}

If you are using the LLM proxy, you can pass the positive_feedback field in the request to record if the user liked the output.

positive_feedback is a boolean field that indicates if the user liked the output. If you’re using other LLM provider SDKs, see the integration docs here.

{ 
"model": "gpt-4o",
# ...
"positive_feedback": True, # means the user liked the output
}

In most cases, you will want to update an existing log with the positive_feedback field. In this case, you can use the Update Log API.

https://api.keywordsai.co/api/request-logs/batch-update/

Here’s the example that updates the positive_feedback field for a log with id xxxxxx. P.S. You can find the log id in the unique_id field in the response of the Get logs or get it directly from the Logs in the UI.

import requests

url = "https://api.keywordsai.co/api/request-logs/batch-update/"
api_key = "YOUR_KEY" # Replace with your actual Keywords AI API key
data = {
    "logs": [
        {   "unique_id": "xxxxxx",
            "positive_feedback": True
            # other fields to update
        }
        # other logs to update
    ]
}
headers = {
    "Authorization": f"Bearer {api_key}",
    "Content-Type": "application/json"
}

response = requests.patch(url, headers=headers, json=data)
print(response.json())