GenerativeQAPipeline return total nonsence

[NorbertHenschl]

I have a Dataset of roughly 2000 Knowledge items which I imported in a FAISS Document store.
When I run a question "How can I cancel a Leave Request?", the retriever gets the right documents but the generator

generator = Seq2SeqGenerator(model_name_or_path="vblagoje/bart_lfqa")

creates a nonsence anwser from that.

I'm pretty lost on what I should do now.
Thanks in advance for the help.

My Program:
main.py.txt

The Result:
Result.txt

[mayankjobanputra]

Hey @NorbertHenschl,

Can you try using PromptNode instead of Seq2Seq generator with Flax-XL models or OpenAI APIs?

[NorbertHenschl]

Hey @mayankjobanputra,

Thanks for the quick reply. Is there any documentation on how to do that?

I tried this:
https://haystack.deepset.ai/tutorials/22_pipeline_with_promptnode

But the results where as bad.

[mayankjobanputra]

@NorbertHenschl What was the prompt that you used for the PromptNode? was it the same as shown in the tutorial?

[NorbertHenschl]

@mayankjobanputra, Yes, but I also tried a lot of different versions, but the change was not really significant.