Method spec_generate

dflash/model.py:350–366

(
        self,
        target: nn.Module,
        input_ids: torch.LongTensor,
        max_new_tokens: int,
        stop_token_ids: list[int],
        temperature: float,
    )

    @torch.inference_mode()
    def spec_generate(
        self,
        target: nn.Module,
        input_ids: torch.LongTensor,
        max_new_tokens: int,
        stop_token_ids: list[int],
        temperature: float,
    ):
        self.eval()
        return dflash_generate(
            self,
            target=target,
            input_ids=input_ids,
            max_new_tokens=max_new_tokens,
            stop_token_ids=stop_token_ids,
            temperature=temperature,
        )