[openai_api_base support] - ft/Added openai OPENAI_API_BASE llm config support

ParseDark commented 3 weeks ago

Description

Some background: Coming from Asian countries with openai restrictions, we cannot directly use the services provided by the openai endpoint, so we have to use a transit service to use openai. So the base_url parameter was added so that users from China can use the products of the mem0 team.

The current PR adds the OPENAI_API_BASE parameter to the config and llm wrapper layers, ensuring that users can set the api_base in the environment variable.

Type of change

I'm not sure if this is a new feature. If it's not suitable, please tell me in the comment and I will modify it. Thanks 🙏

Please delete options that are not relevant.

[x] New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

Please delete options that are not relevant.

[x] Test Script (please provide)

Because I am a novice in Python, I cannot help you complete the existing ut. But I provide the script I tested locally. If you are willing, you can help me verify my changes locally. Thank you very much.

import os
from mem0 import Memory

# os set env variable
# os.environ["OPENAI_API_KEY"] = "sk-xxx"
# os.environ["OPENAI_API_BASE"] = "https://api.xxxx.xxx/v1"

config = {
    "llm": {
        "provider": "openai",
        "config": {
            "openai_base_url": "https://api.xxxx.xxx/v1",
            "api_key": "sk-xxx",
            "model": "gpt-4o-mini",
            "temperature": 0.2,
            "max_tokens": 1500,
        }
    },
    "embedder": {
        "provider": "openai",
        "config": {
            "model": "text-embedding-3-small",
            "openai_base_url": "https://api.xxxx.xxx/v1",
            "api_key": "sk-xxx"
        }
    }
}

m = Memory.from_config(config)
m.enable_graph = True
m.add("Likes to play cricket on weekends", user_id="alice", metadata={"category": "hobbies"})

related_memories = m.search(query="What are Alice's hobbies?", user_id="alice")
print(related_memories)

Some immature suggestions

For the design related to openai, I think the config should be further optimized. For example, according to the existing design, I have to modify the embedding config to support this feature. Can we share a base config for openai? This may be better.

Finally, thank the community for their contribution. Great product

interface OpenaiBaseConfig {
  api_base: string;
  api_key: string;
}

interface OpenaiChartConfig extends OpenaiBaseConfig  {
    top_k: float;
     temperature: float;
     model: string
}

interface OpenaiChartConfig extends OpenaiBaseConfig  {
    top_k: float;
     t: float;
     model: string
}

Checklist:

[x] My code follows the style guidelines of this project
[ ] I have performed a self-review of my own code
[ ] I have commented my code, particularly in hard-to-understand areas
[ ] I have made corresponding changes to the documentation
[ ] My changes generate no new warnings
[ ] I have added tests that prove my fix is effective or that my feature works
[ ] New and existing unit tests pass locally with my changes
[ ] Any dependent changes have been merged and published in downstream modules
[ ] I have checked my code and corrected any misspellings

Maintainer Checklist

[ ] closes #xxxx (Replace xxxx with the GitHub issue number)
[ ] Made sure Checks passed

CLAassistant commented 3 weeks ago

All committers have signed the CLA.

Dev-Khant commented 2 weeks ago

LGTM!!!

ParseDark commented 2 weeks ago

LGTM!!!

NP! And Willing works with you.

mem0ai / mem0