Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Parallelize router initialization #7648

Open
wants to merge 3 commits into
base: main
Choose a base branch
from

Conversation

noobHappylife
Copy link

Title

Use ProcessPoolExecutor to parallelize set_model_list in Router initialization

Relevant issues

Related to #7647

Type

🆕 New Feature

Changes

Use ProcessPoolExecutor to parallelize set_model_list in Router initialization

[REQUIRED]

A test script to test the router creation time

import time
from litellm import Router

# Function to create a list of model configurations
def generate_model_configs(num_models):
    model_configs = []
    for i in range(num_models):
        model_config = {
            "model_name": f"model_{i}",
            "litellm_params": {
                "model": f"openai/model_{i}",
                "api_key": "your-api-key",
                "api_base": "https://api.openai.com/v1"
            }
        }
        model_configs.append(model_config)
    return model_configs

# Function to test router creation time
def test_router_creation(num_models):
    model_configs = generate_model_configs(num_models)

    start_time = time.time()
    router = Router(model_list=model_configs)
    end_time = time.time()

    creation_time = end_time - start_time
    print(f"Router num models: {len(router.model_list)}")
    print(f"Time to create router with {num_models} models: {creation_time:.4f} seconds")

# Test with varying number of models
model_counts = [10, 20, 30, 40, 50]
for count in model_counts:
    test_router_creation(count)

Copy link

vercel bot commented Jan 9, 2025

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name Status Preview Comments Updated (UTC)
litellm ✅ Ready (Inspect) Visit Preview 💬 Add feedback Jan 11, 2025 7:10am

@krrishdholakia
Copy link
Contributor

krrishdholakia commented Jan 10, 2025

Hey @noobHappylife this is a great PR. Can you add some testing for your change.

Possibly in this file -

async def test_router_init():

we typically add a unit test for all functions in router.py

@noobHappylife
Copy link
Author

Hey @noobHappylife this is a great PR. Can you add some testing for your change.

Possibly in this file -

async def test_router_init():

we typically add a unit test for all functions in router.py

Apology, would you suggest what kind of unit test to be included, since this change affects the initiation step, and the checks has passed (I assumed includes the test_router_init). As this change are not supposed to alter any of current behavior (regarding input/output).

@krrishdholakia
Copy link
Contributor

@noobHappylife add a test for your helper utils to just confirm they work as expected.

this ensures any future changes/refactor don't introduce regressions

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants