Function _apply_chat_template

dflash/benchmark.py:103–109

(tokenizer, messages: list[dict], enable_thinking: bool)

def _apply_chat_template(tokenizer, messages: list[dict], enable_thinking: bool) -> str:
    return tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True,
        enable_thinking=enable_thinking,
    )


def _make_decode_metrics(num_output_tokens: int, generation_tps: float, acceptance_lengths: list[int]) -> SimpleNamespace: