Utils

labridge.interact.collect.utils

labridge.interact.collect.utils.acondition_analyze(llm, prompt, condition_true_word, **kwargs) async

Asynchronously choose from two conditions according to the input.

PARAMETER	DESCRIPTION
llm	The used LLM. TYPE: LLM
prompt	The prompt template. TYPE: PromptTemplate
condition_true_word	The word that the LLM is supposed to output in the True condition. TYPE: str
**kwargs	DEFAULT: {}

RETURNS	DESCRIPTION
bool	True condition or False. TYPE: bool

Source code in labridge/interact/collect/utils.py

async def acondition_analyze(
	llm: LLM,
	prompt: PromptTemplate,
	condition_true_word: str,
	**kwargs,
) -> bool:
	r"""
	Asynchronously choose from two conditions according to the input.

	Args:
		llm (LLM): The used LLM.
		prompt (PromptTemplate): The prompt template.
		condition_true_word (str): The word that the LLM is supposed to output in the True condition.
		**kwargs:

	Returns:
		bool: True condition or False.
	"""
	llm_response = await llm.apredict(
		prompt=prompt,
		**kwargs,
	)

	llm_str = filter(lambda x: x.isalpha(), [char for char in llm_response])
	llm_str = "".join(llm_str)
	return llm_str == condition_true_word

labridge.interact.collect.utils.condition_analyze(llm, prompt, condition_true_word, **kwargs)

Choose from two conditions according to the input.

PARAMETER	DESCRIPTION
llm	The used LLM. TYPE: LLM
prompt	The prompt template. TYPE: PromptTemplate
condition_true_word	The word that the LLM is supposed to output in the True condition. TYPE: str
**kwargs	DEFAULT: {}

RETURNS	DESCRIPTION
bool	True condition or False. TYPE: bool

Source code in labridge/interact/collect/utils.py

def condition_analyze(
	llm: LLM,
	prompt: PromptTemplate,
	condition_true_word: str,
	**kwargs,
) -> bool:
	r"""
	Choose from two conditions according to the input.

	Args:
		llm (LLM): The used LLM.
		prompt (PromptTemplate): The prompt template.
		condition_true_word (str): The word that the LLM is supposed to output in the True condition.
		**kwargs:

	Returns:
		bool: True condition or False.
	"""
	llm_response = llm.predict(
		prompt=prompt,
		**kwargs,
	)

	llm_str = filter(lambda x: x.isalpha(), [char for char in llm_response])
	llm_str = "".join(llm_str)
	return llm_str.lower() == condition_true_word.lower()