按照#65中的解答 源码安装后Triton 2.1.0 后报错

起初报错内容和#65报错内容一样，源码安装后再次运行报错如下：

(moss) root@autodl-container-99da11befa-9a05a134:~/autodl-tmp/MOSS/MOSS# python moss_gui_demo.py
Waiting for all devices to be ready, it may take a few minutes...
triton not installed. Run `pip install triton` to load quantized version of MOSS.
Traceback (most recent call last):
  File "moss_gui_demo.py", line 31, in <module>
    raw_model = MossForCausalLM._from_config(config, torch_dtype=torch.float16)
  File "/root/miniconda3/envs/moss/lib/python3.8/site-packages/transformers/modeling_utils.py", line 1024, in _from_config
    model = cls(config, **kwargs)
  File "/root/autodl-tmp/MOSS/MOSS/models/modeling_moss.py", line 612, in __init__
    self.quantize(config.wbits, config.groupsize)
  File "/root/autodl-tmp/MOSS/MOSS/models/modeling_moss.py", line 736, in quantize
    from .quantization import quantize_with_gptq
  File "/root/autodl-tmp/MOSS/MOSS/models/quantization.py", line 27, in <module>
    @autotune(
NameError: name 'autotune' is not defined

已删除autotune后重新安装，依旧报错

Reproduce the failure with python moss_gui_demo.py after the source installation described in the issue. Inspect models/quantization.py around the @autotune decorator and models/modeling_moss.py at the quantize call; done means the demo starts without the reported NameError.