Observability for Portkey with Opik

1
import os
2
import getpass
3
4
if "PORTKEY_API_KEY" not in os.environ:
5
    os.environ["PORTKEY_API_KEY"] = getpass.getpass("Enter your Portkey API key: ")
6
7
if "PORTKEY_VIRTUAL_KEY" not in os.environ:
8
    os.environ["PORTKEY_VIRTUAL_KEY"] = getpass.getpass("Enter your Portkey virtual key: ")

1
import os
2
from opik.integrations.openai import track_openai
3
from openai import OpenAI
4
from portkey_ai import PORTKEY_GATEWAY_URL, createHeaders
5
6
client = OpenAI(
7
    api_key=os.environ["OPENAI_API_KEY"],
8
    base_url=PORTKEY_GATEWAY_URL,
9
    default_headers=createHeaders(
10
        api_key=os.environ["PORTKEY_API_KEY"],
11
        provider="@OPENAI_PROVIDER"
12
    )
13
)
14
15
# Wrap the client with Opik tracking
16
client = track_openai(client, project_name="portkey-integration-demo")
17
18
# Make a chat completion request
19
response = client.chat.completions.create(
20
    model="gpt-3.5-turbo",
21
    messages=[
22
        {"role": "system", "content": "You are a knowledgeable AI assistant."},
23
        {"role": "user", "content": "What is the largest city in France?"}
24
    ]
25
)
26
27
# Print the assistant's reply
28
print(response.choices[0].message.content)

1
import os
2
from opik import track
3
from opik.integrations.openai import track_openai
4
from openai import OpenAI
5
from portkey_ai import PORTKEY_GATEWAY_URL, createHeaders
6
7
# Create an OpenAI client configured for Portkey
8
client = OpenAI(
9
    api_key=os.environ["OPENAI_API_KEY"],
10
    base_url=PORTKEY_GATEWAY_URL,
11
    default_headers=createHeaders(
12
        api_key=os.environ["PORTKEY_API_KEY"],
13
        provider="@OPENAI_PROVIDER"
14
    )
15
)
16
17
# Wrap the client with Opik tracking
18
client = track_openai(client, project_name="portkey-integration-demo")
19
20
@track
21
def generate_response(prompt: str):
22
    response = client.chat.completions.create(
23
        model="gpt-3.5-turbo",
24
        messages=[
25
            {"role": "system", "content": "You are a knowledgeable AI assistant."},
26
            {"role": "user", "content": prompt}
27
        ]
28
    )
29
    return response.choices[0].message.content
30
31
@track
32
def refine_response(initial_response: str):
33
    response = client.chat.completions.create(
34
        model="gpt-4",
35
        messages=[
36
            {"role": "system", "content": "You enhance and polish text responses."},
37
            {"role": "user", "content": f"Please improve this response: {initial_response}"}
38
        ]
39
    )
40
    return response.choices[0].message.content
41
42
@track(project_name="portkey-integration-demo")
43
def generate_and_refine(prompt: str):
44
    # First LLM call: Generate initial response
45
    initial = generate_response(prompt)
46
    
47
    # Second LLM call: Refine the response
48
    refined = refine_response(initial)
49
    
50
    return refined
51
52
# Example usage
53
result = generate_and_refine("Explain quantum computing in simple terms.")