Skip to content

📊 LLM Support

GENIUS-LLM adopts a Protocol-Oriented Architecture. We utilize the OpenAI Python library not as a provider-specific lock-in, but as a universal adapter for the industry-standard completion protocol.

1. Supported Models Matrix

The following table explicitly lists the models verified with the current release of GENIUS-LLM. Compatibility is achieved by simply updating the url and model fields in data_config.yaml.

Model Category Representative Models Protocol Status
Official OpenAI Models GPT-4o, GPT-4 Turbo, GPT-3.5 Native Support
OpenAI-Compatible Models DeepSeek (V3/Chat), Qwen, Moonshot (Kimi), Zhipu AI, Llama 3 (via vLLM) Universal OpenAI-Style Protocol

By adhering to the OpenAI-standardized API format, GENIUS-LLM provides seamless integration across a wide range of providers. Users can switch between high-performance proprietary models and cost-effective local or regional alternatives (like DeepSeek-V3) without changing a single line of core code.

2. Decoupled Architecture

Our src/model/llm_model.py is designed with zero hardcoded endpoints. The internal logic treats the LLM as a black-box service defined by three parameters in the configuration:

  1. api_key: User-provided credential.
  2. url: The entry point for the API (e.g. https://api.deepseek.com/v1 or https://api.openai.com/v1 or anyother API endpoints you want to use).
  3. model: The specific model identifier.

Why use the openai library?

The openai Python client has evolved into a standardized wrapper for RESTful LLM APIs. By leveraging this library, GENIUS-LLM benefits from: * Robust Streaming: Stable handling of long-context gene function descriptions. * Automatic Retries: Built-in exponential backoff for scientific computing stability. * Format Consistency: Ensures that the JSON output from different providers (DeepSeek vs. GPT) is parsed identically.

3. How to Switch Models (Step-by-Step)

To switch from the default configuration to a different provider, modify your config/data_config.yaml as follows:

Example: Switching to a (via vLLM)

model:
    url: https://api.openai.com/v1 # Here is the base URL for the API; you can modify it according to your specific LLM configuration.
    api_key: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx # Replace your api_key
    model: GPT-4o #The model you want to use