mac m1芯片下打开output_timestamp 内存溢出

[zhuchao941]

Notice: In order to resolve issues more efficiently, please raise issue following the template.
（注意：为了更加高效率解决您遇到的问题，请按照模板提问，补充细节）

🐛 Bug

mps with output_timestamp=True cause oom

 File "/Users/zhuzhuchao/PycharmProjects/FunASR/funasr/models/sense_voice/model.py", line 936, in inference
    align = ctc_forced_align(
            ^^^^^^^^^^^^^^^^^
  File "/Users/zhuzhuchao/PycharmProjects/FunASR/funasr/models/sense_voice/utils/ctc_alignment.py", line 53, in ctc_forced_align
    best_score[:, padding_num:] = log_probs[:, t].gather(-1, _t_a_r_g_e_t_s_) + prev_max_value
                                  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
RuntimeError: MPS backend out of memory (MPS allocated: 1.90 GB, other allocations: 81.94 GB, max allowed: 81.60 GB). Tried to allocate 256 bytes on private pool. Use PYTORCH_MPS_HIGH_WATERMARK_RATIO=0.0 to disable upper limit for memory allocations (may cause system failure).

To Reproduce

run with macbook m1 cpu

from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

model_dir = "iic/SenseVoiceSmall"

model = AutoModel(
    model=model_dir,
    vad_model="fsmn-vad",
    vad_kwargs={"max_single_segment_time": 30000},
    device="mps",
)

# en
res = model.generate(
    input=f"/Users/zhuzhuchao/Downloads/yuyin/av2/1.wav",
    # input=f"{model.model_path}/example/zh.mp3",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=False,
    # merge_vad=True,  #
    # merge_length_s=15,
    ban_emo_unk=True,
    output_timestamp=True
)
text = rich_transcription_postprocess(res[0]["text"])
print(text)
with open("log-sence.txt", "w") as f:
    f.write(str(res))

Environment

• OS (e.g., Linux):MacOS 13.2.1
• FunASR Version (e.g., 1.0.0):1.2.6
• ModelScope Version (e.g., 1.11.0):
• PyTorch Version (e.g., 2.0.0):
• How you installed funasr (pip, source):
• Python version:
• GPU (e.g., V100M32)
• CUDA/cuDNN version (e.g., cuda11.7):
• Docker version (e.g., funasr-runtime-sdk-cpu-0.4.1)
• Any other relevant information:

Additional context