vectara

/

hallucination_evaluation_model

like 210

Hi @izimmerman can you show me how you used HHEM with text-classificaiton pipeline?

classifier = pipeline(
            "text-classification",
            model="vectara/hallucination_evaluation_model",
            trust_remote_code=True,
        )
...

classifier("sample document[SEP]sample summary") <<< errors

Hi @izimmerman please see the new usage here
https://huggingface.co/vectara/hallucination_evaluation_model#using-with-pipeline
Let us know if you have additional questions.

Note that you can no longer use [SEP] to connect the premise and the hypothesis. We now use an instructional prompt template. If you want to save the hassle, you can use our end-to-end solution here: https://huggingface.co/vectara/hallucination_evaluation_model#using-with-automodel