9 ms·

I've been looking for something like this! Does it optimize the prompt template for LangChain only or is there a way I can get it to generate a raw system promp

• engineerick 2y ago
  Hello, I'm glad you find it useful. I aimed to create something that would serve a purpose. If you can provide me details about use case you are trying to solve, I may add a feature to llmdantic to support it. Right now: After initialize llmdantic you can get the prompt by running the following command: """ from llmdantic import LLMdantic, LLMdanticConfig from langchain_openai import ChatOpenAI llm = ChatOpenAI() config: LLMdanticConfig = LLMdanticConfig( objective="Summarize the text", inp_schema=SummarizeInput, out_schema=SummarizeOutput, retries=3, ) llmdantic = LLMdantic(llm=llm, config=config) input_data: SummarizeInput = SummarizeInput( text="The quick brown fox jumps over the lazy dog." ) prompt: str = llmdantic.prompt(input_data) """ But here you need to provide a langchain llm model. If you do not want to use langchain llm model, you can use the following code: """ from llmdantic.prompts.prompt_builder import LLMPromptBuilder from llmdantic.output_parsers.output_parser import LLMOutputParser output_parser: LLMOutputParser = LLMOutputParser(pydantic_object=SummarizeOutput) prompt_builder = LLMPromptBuilder( objective="Summarize the text", inp_model=SummarizeInput, out_model=SummarizeOutput, parser=output_parser, ) data: SummarizeInput = SummarizeInput(text="Some text to summarize") prompt = prompt_builder.build_template() print(prompt.format(input=data.model_dump())) """ But here still we use langchain for the prompt building. If you any questions, feel free to ask I will be happy to help you.