---------------------------------------------------------------------UnboundLocalError Traceback most recent call lastCell In73 line 4 2 model = GPUDoc2Vecvector_size=100w

The error message "UnboundLocalError: local variable 'model' referenced before assignment" indicates that the variable 'model' is being referenced before it is assigned a value.

In the code snippet provided, it seems that the variable 'model' is defined as a class attribute in the GPUDoc2Vec class. However, in the train() method of the GPUDoc2Vec class, a new local variable 'model' is being created and used before the class attribute is assigned a value.

To fix this error, you can remove the local variable 'model' in the train() method and use the class attribute 'model' instead. Here's the modified code:

class GPUDoc2Vec(Doc2Vec):
    def __init__(self, *args, **kwargs):
        self.model = None
        super().__init__(*args, **kwargs)

    def train(self, documents, total_examples=None, total_words=None, epochs=None, start_alpha=None, end_alpha=None,
              word_count=0, queue_factor=2, **kwargs):
        graph = tf.Graph()
        with graph.as_default():
            # 创建一个doc2vec模型
            self.model = super().train(documents, total_examples=self.model.corpus_count,
                                       total_words=total_words, epochs=epochs,
                                       start_alpha=start_alpha, end_alpha=end_alpha,
                                       word_count=word_count, queue_factor=queue_factor,
                                       **kwargs)
            # 初始化tensorflow变量
            session.run(tf.compat.v1.global_variables_initializer())

By using the class attribute 'self.model' instead of the local variable 'model', you can avoid the UnboundLocalError.