AttributeError: 'InstructorLLM' object has no attribute 'agenerate_prompt'

[02hao09]

[ ] I have checked the documentation and related resources and couldn't resolve my bug.

Describe the bug
AttributeError: 'InstructorLLM' object has no attribute 'agenerate_prompt'

Ragas version: 0.3.7
Python version: 3.10

Code to Reproduce
from ragas.llms.base import instructor_llm_factory, llm_factory, InstructorLLM
from ragas.embeddings import OpenAIEmbeddings

初始化 Azure OpenAI Client

azure_client = AzureOpenAI(
api_key=api_key,
api_version=api_version,
azure_endpoint=api_base
)

建立 RAGAS 可用的 LLM

llm = instructor_llm_factory(provider="openai",model="gpt-4o", client=azure_client)
#lm = llm_factory(model="gpt-4o")

embeddings = OpenAIEmbeddings(
model="text-embedding-ada-002",
client=azure_client
)

import asyncio
async def main():
context_precision_adapted_prompts = await context_precision.adapt_prompts(language="chinese", llm=llm)
print(context_precision_adapted_prompts)
asyncio.run(main())
print(context_precision)

Error trace

File "/myapps/test.py", line 40, in
asyncio.run(main())
File "/usr/local/lib/python3.10/asyncio/runners.py", line 44, in run
return loop.run_until_complete(main)
File "/usr/local/lib/python3.10/asyncio/base_events.py", line 641, in run_until_complete
return future.result()
File "/myapps/test.py", line 38, in main
context_precision_adapted_prompts = await context_precision.adapt_prompts(language="chinese", llm=llm)
File "/usr/local/lib/python3.10/site-packages/ragas/prompt/mixin.py", line 78, in adapt_prompts
adapted_prompt = await prompt.adapt(language, llm, adapt_instruction)
File "/usr/local/lib/python3.10/site-packages/ragas/prompt/pydantic_prompt.py", line 330, in adapt
translated_strings = await translate_statements_prompt.generate(
File "/usr/local/lib/python3.10/site-packages/ragas/prompt/pydantic_prompt.py", line 164, in generate
output_single = await self.generate_multiple(
File "/usr/local/lib/python3.10/site-packages/ragas/prompt/pydantic_prompt.py", line 234, in generate_multiple
resp = await langchain_llm.agenerate_prompt(
AttributeError: 'InstructorLLM' object has no attribute 'agenerate_prompt'

Expected behavior
context_precision.adapt_prompts(language="chinese", llm=llm) converts PROMPT into Chinese.

Additional context
Add any other context about the problem here.

[anistark]

Thanks for bringing this up @02hao09
We're updating some major changes in this and will take this into consideration as well.

While at it, the issue is supposedly elsewhere cause InstructorLLM is not a LangChain LLM.

Use a general-purpose Ragas LLM for adapt_prompts(), not InstructorLLM:

from openai import OpenAI
from ragas.llms import llm_factory

# For prompt adaptation, use llm_factory (general-purpose LLM)
llm = llm_factory("gpt-4o-mini", client=OpenAI(api_key="..."))

# Now this will work
await context_precision.adapt_prompts(language="chinese", llm=llm)

# InstructorLLM is for structured outputs:
from pydantic import BaseModel

class MyResponse(BaseModel):
    answer: str

response = instructor_llm.generate("...", response_model=MyResponse)

In short: InstructorLLM was never intended for general prompt adaptation. Use llm_factory() instead for that use case.

[github-actions]

It seems the issue was answered, closing this now.