Customs LLMs

[Pierrelefort]

Do you have any plans to add configuration options that would allow the use of custom LLMs in future versions ?

[uripeled2]

I recently published a package llm-client that can be very helpful in enabling the support to run other LLM models, including OpenAI, Google, AI21, HuggingfaceHub, Aleph Alpha, Anthropic, Local models with transformers.

[steffenslavetinsky]

I stumbled over the issue yesterday when I wanted to use promptimize for a custom model. I am not sure wether or not I complete got your question but isn't this already possible by passing a custom prompt_executor to a PromptCase?

even a simple lambda function instead of an LLM

from promptimize.prompt_cases import PromptCase
from promptimize import evals

simple_prompts = [
    PromptCase(
        "hello there!",
        lambda x: evals.any_word(x.response, ["hi", "hello"]),
        prompt_executor=lambda _x: "hi there!",
    ),
    PromptCase(
        lambda x: evals.all_words(x.response, ["zappa", "hendrix"]),
        weight=2,
        prompt_executor=lambda _x: "zappa, hendrix, clapton, page, beck, ...",
    ),
]