Function _make_response

dflash/model_mlx.py:413–426

(
    text,
    tokens,
    accepted,
    prompt_size,
    prompt_tps,
    n,
    tic,
    finish_reason=None,
)

def _make_response(
    text,
    tokens,
    accepted,
    prompt_size,
    prompt_tps,
    n,
    tic,
    finish_reason=None,
):
    return GenerationResponse(
        text, tokens, accepted, prompt_size, prompt_tps,
        n, n / (time.perf_counter() - tic), mx.get_peak_memory() / 1e9, finish_reason,
    )


def stream_generate(