h2oai_pipeline.py

from transformers import TextGenerationPipeline
from transformers.pipelines.text_generation import ReturnType

human = "<human>:"
bot = "<bot>:"

# human-bot interaction like OIG dataset
prompt = """{human} {instruction}
{bot}""".format(
    human=human,
    instruction="{instruction}",
    bot=bot,
)


class H2OTextGenerationPipeline(TextGenerationPipeline):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

    def preprocess(self, prompt_text, prefix="", handle_long_generation=None, **generate_kwargs):
        prompt_text = prompt.format(instruction=prompt_text)
        return super().preprocess(prompt_text, prefix=prefix, handle_long_generation=handle_long_generation,
                                  **generate_kwargs)

    def postprocess(self, model_outputs, return_type=ReturnType.FULL_TEXT, clean_up_tokenization_spaces=True):
        records = super().postprocess(model_outputs, return_type=return_type,
                                      clean_up_tokenization_spaces=clean_up_tokenization_spaces)
        for rec in records:
            rec['generated_text'] = rec['generated_text'].split(bot)[1].strip().split(human)[0].strip()
        return records