RAG.forward() missing 1 required positional argument: 'question'

[jeremierostan]

Hi,

qa = RAG() pred = qa("xxx?") print(f"Predicted Answer: {pred.answer}")

works with the following, but compiling does not. Can anyone help?

from dspy.datasets import DataLoader dl = DataLoader() JR_qa = dl.from_csv("/content/ai_integration_qa_no_headers.csv") splits = dl.train_test_split(JR_qa, train_size=15) #dataset` is a List of dspy.Example train_dataset = splits['train'] test_dataset = splits['test']

trainset = [x.with_inputs('question') for x in train_dataset] devset = [x.with_inputs('question') for x in test_dataset]

class GenerateAnswer(dspy.Signature): """Answer questions with short factoid answers."""

context = dspy.InputField(desc="may contain relevant facts")
question = dspy.InputField()
answer = dspy.OutputField(desc="often between 1 and 5 words")

class RAG(dspy.Module): def init(self, num_passages=3): super().init()

    self.retrieve = dspy.Retrieve(k=num_passages)
    self.generate_answer = dspy.ChainOfThought(GenerateAnswer)

def forward(self, question):
    context = self.retrieve(question).passages
    prediction = self.generate_answer(context=context, question=question)
    return dspy.Prediction(context=context, answer=prediction.answer)

from dspy.teleprompt import BootstrapFewShot

Validation logic: check that the predicted answer is correct.

Also check that the retrieved context does actually contain that answer.

def validate_context_and_answer(example, pred, trace=None): answer_EM = dspy.evaluate.answer_exact_match(example, pred) answer_PM = dspy.evaluate.answer_passage_match(example, pred) return answer_EM and answer_PM

Set up a basic teleprompter, which will compile our RAG program.

teleprompter = BootstrapFewShot(metric=validate_context_and_answer)

Compile!

compiled_rag = teleprompter.compile(RAG(), trainset=trainset) `

——>

TypeError Traceback (most recent call last) in <cell line: 14>() 12 13 # Compile! ---> 14 compiled_rag = teleprompter.compile(RAG(), trainset=trainset)

4 frames

/usr/local/lib/python3.10/dist-packages/dspy/primitives/program.py in call(self, *args, kwargs) 24 25 def call(self, *args, *kwargs): ---> 26 return self.forward(args, kwargs) 27 28 def named_predictors(self):

TypeError: RAG.forward() missing 1 required positional argument: 'question'

[mlederbauer]

Hi! I think you need to explicitly pass question, i.e.

qa = RAG()
answer = qa(question=“xxx?”).answer

Does this work for your case?

[jeremierostan]

The RAG itself works: qa = RAG() pred = qa("What is an AI dashboard?") print(f"Predicted Answer: {pred.answer}")

What does not work is compiling the module:

from dspy.teleprompt import BootstrapFewShot def validate_context_and_answer(example, pred, trace=None): answer_EM = dspy.evaluate.answer_exact_match(example, pred) answer_PM = dspy.evaluate.answer_passage_match(example, pred) return answer_EM and answer_PM teleprompter = BootstrapFewShot(metric=validate_context_and_answer) compiled_rag = teleprompter.compile(RAG(), trainset=trainset)

Any idea why? I am using the signature and module from the documentation (https://dspy-docs.vercel.app/docs/tutorials/rag). What I changed is using my own article a dataset:

from dspy.datasets import DataLoader dl = DataLoader() qa = dl.from_csv("/content/ai_integration_qa.csv") splits = dl.train_test_split(qa, train_size=15) train_dataset = splits['train'] test_dataset = splits['test'] trainset = [x.with_inputs('question') for x in train_dataset] devset = [x.with_inputs('question') for x in test_dataset]

I can't figure out what the issue is...

[okhat]

Hey @jeremierostan ! Do pass keyword arguments, like @mlederbauer said.

[jeremierostan]

qa = RAG() pred = qa("What is an AI dashboard?") print(f"Predicted Answer: {pred.answer}")

Hi, yes, I have it here and it was already working:

qa = RAG() pred = qa("What is an AI dashboard?") print(f"Predicted Answer: {pred.answer}")

What is not working is the compiling:

from dspy.teleprompt import BootstrapFewShot def validate_context_and_answer(example, pred, trace=None): answer_EM = dspy.evaluate.answer_exact_match(example, pred) answer_PM = dspy.evaluate.answer_passage_match(example, pred) return answer_EM and answer_PM teleprompter = BootstrapFewShot(metric=validate_context_and_answer) compiled_rag = teleprompter.compile(RAG(), trainset=trainset)

[okhat]

Passing keyword arguments means:

Do not write pred = qa("What is an AI dashboard?")

Instead, write pred = qa(question="What is an AI dashboard?")

Apply this in every call to a module. I hope this helps.