def _model_generate()

in lm_eval/models/huggingface.py [0:0]

• 18 lines of code
• 10 McCabe index (conditional complexity)

    def _model_generate(self, context, max_length, stop, **generation_kwargs):
        # temperature = 0.0 if not set
        # if do_sample is false and temp==0.0:
        # remove temperature, as do_sample=False takes care of this
        # and we don't want a warning from HF
        generation_kwargs["temperature"] = generation_kwargs.get("temperature", 0.0)
        do_sample = generation_kwargs.get("do_sample", None)

        # The temperature has to be a strictly positive float -- if it is 0.0, use greedy decoding strategies
        if generation_kwargs.get("temperature") == 0.0 and do_sample is None:
            generation_kwargs["do_sample"] = do_sample = False

        if do_sample is False and generation_kwargs.get("temperature") == 0.0:
            generation_kwargs.pop("temperature")
        # build stopping criteria
        stopping_criteria = stop_sequences_criteria(
            self.tokenizer, stop, context.shape[1], context.shape[0]
        )
        return self.model.generate(
            input_ids=context,
            max_length=max_length,
            stopping_criteria=stopping_criteria,
            pad_token_id=self.tokenizer.pad_token_id,
            use_cache=True,
            **generation_kwargs,
        )