Quickstart/Examples

1) Install the package into your environment (see [[Installation]])

2) Create a config file for the LLM or LLMs you want to use:

e.g. myconf.hjson

{
  llms: [
    { llm: "openai/gpt-4o"
      api_key_env: "OPENAI_KEY1"
      temperature: 0
      alias: "openai1"
    }
    { llm: "gemini/gemini-1.5-flash", temperature: 1, alias: "geminiflash1" }
    { llm: "gemini/gemini-1.5-flash", temperature: 0, alias: "geminiflash2" }
]
  providers: {
    gemini: {
      api_key_env: "GEMINI_KEY1"
    }
  }
}

3) In your code:

• import the class LLMS from llms_wrapper.llms
• read the configuration file
• instantiate the class with the configuration object
• create a messages object, optionally using the make_messages method
• invoke any of the configured LLMs using the query method by specifying the alias of the LLM and passing on the prompt messages. This can also optionally return the cost of the invocation and/or the full response object.
• in order to invoke all LLMs configured, you can iterate over all aliases using the list_alias method

from llms_wrapper.llms import LLMS
from llms_wrapper.config import read_config_file

# [....]
config = read_config_file("myconf.hjson")
llms = LLMS(config)

messages = llms.make_messages(query="What is a monoid?")
response = llms.query("geminiflash", messages, return_cost=True)
if response["ok"]:
    print("Got the answer:", response["answer"])
    print("Cost:", response["cost"])
else:
    print("Got an error:", response["error"])