invalid dataset error while scanning a Gemini model

[lanonse]

Hello,

i have this basic code (mostly from the Giskard documentation):

import os
import giskard

import google.generativeai as genai

from giskard.llm.client.gemini import GeminiClient

import pandas as pd

genai.configure(api_key=os.environ["GEMINI_API_KEY"])

giskard.llm.set_default_client(GeminiClient())

giskard_dataset = giskard.Dataset(pd.DataFrame({
    "question": [
        "According to the IPCC report, what are key risks in the Europe?",
        "Is sea level rise avoidable? When will it stop?"
    ]
}), target=None)

def model_predict(df: pd.DataFrame):
    return [llm_api(question) for question in df["question"].values]

giskard_model = giskard.Model(
    model=model_predict,  # our model function
    model_type="text_generation",
    name="Climate Change Question Answering",
    description="This model answers any question about climate change based on IPCC reports",
    feature_names=["question"],  # input variables needed by your model
)

output_file = "report.html"
scan_results = giskard.scan(giskard_model, output_file)

and with or without the definition of giskard_dataset, i get this error:

ValueError: The dataset object you provided is not valid. Please wrap your dataframe with `giskard.Dataset`.

Precisely, the output is as follows:

I am probably missing something. Can someone help?

Thank you!

[lanonse]

What i was missing was passing "giskard_dataset" to giskard_scan. When i do pass it, i get another error "Detector LLMStereotypesDetector failed with error: 400 Please use a valid role: user, model." but this is connected to genai, so i close this discussion.

PS
At https://docs.giskard.ai/en/stable/getting_started/quickstart/quickstart_llm.html it is not mentioned that one needs to create a dataset (like it is mentioned in the tabular doc https://docs.giskard.ai/en/stable/getting_started/quickstart/quickstart_tabular.html) so it's not clear to me if the dataset part from the tabular doc should be present in the llm doc as well.
Thank you!