AttributeError: 模块“cupy”没有属性“scatter_add”

如何解决AttributeError: 模块“cupy”没有属性“scatter_add”

当尝试运行一个模型进行 spacy NER 训练时

spacy.require_gpu()

我收到此错误：

AttributeError                            Traceback (most recent call last)
<ipython-input-6-a86608cac558> in <module>
      3 }
      4 start_time = time.time()
----> 5 train_model(data)

<ipython-input-2-10f04ac303a6> in train_model(train_data)
     82     other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'ner']
     83     with nlp.disable_pipes(*other_pipes):
---> 84         optimizer = nlp.begin_training()
     85         for itn in range(1):
     86             print("Starting iteration " + str(itn))

c:\users\aventus\envs\py38\lib\site-packages\spacy\language.py in begin_training(self,get_gold_tuples,sgd,component_cfg,**cfg)
    623                 kwargs = component_cfg.get(name,{})
    624                 kwargs.update(cfg)
--> 625                 proc.begin_training(
    626                     get_gold_tuples,627                     pipeline=self.pipeline,nn_parser.pyx in spacy.Syntax.nn_parser.Parser.begin_training()

_parser_model.pyx in spacy.Syntax._parser_model.ParserModel.begin_training()

c:\users\aventus\envs\py38\lib\site-packages\thinc\check.py in checked_function(wrapped,instance,args,kwargs)
    154                     raise ExpectedTypeError(check,["Callable"])
    155                 check(arg_id,fix_args,kwargs)
--> 156         return wrapped(*args,**kwargs)
    157 
    158     def arg_check_adder(func):

c:\users\aventus\envs\py38\lib\site-packages\thinc\neural\_classes\model.py in begin_training(self,train_X,train_y,**trainer_cfg)
    120     def begin_training(self,train_y=None,**trainer_cfg):
    121         for hook in self.on_data_hooks:
--> 122             hook(self,train_y)
    123         return self.Trainer(self,**trainer_cfg)
    124 

c:\users\aventus\envs\py38\lib\site-packages\spacy\_ml.py in <lambda>(model,X,y)
    149 
    150 @describe.on_data(
--> 151     _set_dimensions_if_needed,lambda model,y: model.init_weights(model)
    152 )
    153 @describe.attributes(

c:\users\aventus\envs\py38\lib\site-packages\spacy\_ml.py in init_weights(model)
    268         t_i = 0
    269         for t_i in range(t_max):
--> 270             acts1 = predict(ids,tokvecs)
    271             var = model.ops.xp.var(acts1)
    272             mean = model.ops.xp.mean(acts1)

c:\users\aventus\envs\py38\lib\site-packages\spacy\_ml.py in predict(ids,tokvecs)
    254                 (hiddens.shape[0] * model.nF,model.nO * model.nP)
    255             )
--> 256             model.ops.scatter_add(vectors,ids.flatten(),hiddens)
    257             vectors = vectors.reshape((vectors.shape[0],model.nO,model.nP))
    258             vectors += model.b

ops.pyx in thinc.neural.ops.CupyOps.scatter_add()

c:\users\aventus\envs\py38\lib\site-packages\cupy\__init__.py in __getattr__(name)
    873         value = _deprecated_attrs.get(name)
    874         if value is None:
--> 875             raise AttributeError(
    876                 f"module 'cupy' has no attribute {name!r}")
    877         attr,eq_attr = value

AttributeError: module 'cupy' has no attribute 'scatter_add'

用于训练模型的数据是spacy NER的注释。

运行 spacy.require_gpu() 时得到这个：

Python 3.8.3 (tags/v3.8.3:6f8c832,May 13 2020,22:37:02) [MSC v.1924 64 bit (AMD64)] on win32
Type "help","copyright","credits" or "license" for more information.
>>>import spacy
>>>spacy.require_gpu()
True

操作系统 - Windows 10
Python 3.8
spacy 2.2.4
杯子 9.0.0
Nvidia 驱动程序版本：466.11
CUDA 版本：11.1
NVIDIA GeForce GTX 1660Ti

如何摆脱这个错误？任何帮助将不胜感激。

解决方法

有几个选项可以避免这个错误：

将cupy 降级为cupy<8.0.0 (pip install "cupy-cuda110<8.0.0")
将 spacy 升级到 v2.3.5+

根据评论，我不知道您是如何从 pip install -U spacy 获得 spacy v2.2.4 的，但是如果您特别不需要该版本，则升级到没有此错误的版本是最简单的解决方案且更新版本与cupy v8兼容。

如果您确实需要 spacy v2.2.4（需要 thinc==7.4.0）和更新版本的cupy，您可以将此补丁应用到thinc v7.4.0 并从源代码安装：

https://github.com/explosion/thinc/pull/441/files#diff-742ac455f5934d8207f38852f68013706e9561e8832a5b7c5b85757fa295b5f0L47-R1090

另见：https://github.com/explosion/spaCy/issues/5380