ValueError：您必须包含至少一个标签和至少一个序列

Posted 2023-03-12

【中文标题】ValueError：您必须包含至少一个标签和至少一个序列

【英文标题】：ValueError: You must include at least one label and at least one sequence

【发布时间】：2022-01-13 14:24:15

【问题描述】：

我正在使用这个Notebook，其中Apply DocumentClassifier部分更改如下。

Jupyter 实验室，内核：conda_mxnet_latest_p37。

错误似乎是 ML 标准实践响应。但是，我传递/创建与原始代码相同的参数和变量名称。所以这与他们在我的代码中的值有关。

---

我的代码：

with open('filt_gri.txt', 'r') as filehandle:
    tags = [current_place.rstrip() for current_place in filehandle.readlines()]

doc_classifier = TransformersDocumentClassifier(model_name_or_path="cross-encoder/nli-distilroberta-base",
                                                task="zero-shot-classification",
                                                labels=tags,
                                                batch_size=16)

# convert to Document using a fieldmap for custom content fields the classification should run on
docs_to_classify = [Document.from_dict(d) for d in docs_sliding_window]

# classify using gpu, batch_size makes sure we do not run out of memory
classified_docs = doc_classifier.predict(docs_to_classify)

# let's see how it looks: there should be a classification result in the meta entry containing labels and scores.
print(classified_docs[0].to_dict())

all_docs = convert_files_to_dicts(dir_path=doc_dir)

preprocessor_sliding_window = PreProcessor(split_overlap=3,
                                           split_length=10,
                                           split_respect_sentence_boundary=False,
                                           split_by='passage')

输出：

INFO - haystack.modeling.utils -  Using devices: CUDA
INFO - haystack.modeling.utils -  Number of GPUs: 1
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-11-77eb98038283> in <module>
     14 
     15 # classify using gpu, batch_size makes sure we do not run out of memory
---> 16 classified_docs = doc_classifier.predict(docs_to_classify)
     17 
     18 # let's see how it looks: there should be a classification result in the meta entry containing labels and scores.

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/haystack/nodes/document_classifier/transformers.py in predict(self, documents)
    137         batches = self.get_batches(texts, batch_size=self.batch_size)
    138         if self.task == 'zero-shot-classification':
--> 139             batched_predictions = [self.model(batch, candidate_labels=self.labels, truncation=True) for batch in batches]
    140         elif self.task == 'text-classification':
    141             batched_predictions = [self.model(batch, return_all_scores=self.return_all_scores, truncation=True) for batch in batches]

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/haystack/nodes/document_classifier/transformers.py in <listcomp>(.0)
    137         batches = self.get_batches(texts, batch_size=self.batch_size)
    138         if self.task == 'zero-shot-classification':
--> 139             batched_predictions = [self.model(batch, candidate_labels=self.labels, truncation=True) for batch in batches]
    140         elif self.task == 'text-classification':
    141             batched_predictions = [self.model(batch, return_all_scores=self.return_all_scores, truncation=True) for batch in batches]

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/transformers/pipelines/zero_shot_classification.py in __call__(self, sequences, candidate_labels, hypothesis_template, multi_label, **kwargs)
    151             sequences = [sequences]
    152 
--> 153         outputs = super().__call__(sequences, candidate_labels, hypothesis_template)
    154         num_sequences = len(sequences)
    155         candidate_labels = self._args_parser._parse_labels(candidate_labels)

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/transformers/pipelines/base.py in __call__(self, *args, **kwargs)
    758 
    759     def __call__(self, *args, **kwargs):
--> 760         inputs = self._parse_and_tokenize(*args, **kwargs)
    761         return self._forward(inputs)
    762 

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/transformers/pipelines/zero_shot_classification.py in _parse_and_tokenize(self, sequences, candidate_labels, hypothesis_template, padding, add_special_tokens, truncation, **kwargs)
     92         Parse arguments and tokenize only_first so that hypothesis (label) is not truncated
     93         """
---> 94         sequence_pairs = self._args_parser(sequences, candidate_labels, hypothesis_template)
     95         inputs = self.tokenizer(
     96             sequence_pairs,

~/anaconda3/envs/mxnet_latest_p37/lib/python3.7/site-packages/transformers/pipelines/zero_shot_classification.py in __call__(self, sequences, labels, hypothesis_template)
     25     def __call__(self, sequences, labels, hypothesis_template):
     26         if len(labels) == 0 or len(sequences) == 0:
---> 27             raise ValueError("You must include at least one label and at least one sequence.")
     28         if hypothesis_template.format(labels[0]) == hypothesis_template:
     29             raise ValueError(

ValueError: You must include at least one label and at least one sequence.

原始代码：

doc_classifier = TransformersDocumentClassifier(model_name_or_path="cross-encoder/nli-distilroberta-base",
    task="zero-shot-classification",
    labels=["music", "natural language processing", "history"],
    batch_size=16
)

# ----------

# convert to Document using a fieldmap for custom content fields the classification should run on
docs_to_classify = [Document.from_dict(d) for d in docs_sliding_window]

# ----------

# classify using gpu, batch_size makes sure we do not run out of memory
classified_docs = doc_classifier.predict(docs_to_classify)

# ----------

# let's see how it looks: there should be a classification result in the meta entry containing labels and scores.
print(classified_docs[0].to_dict())

请让我知道是否还有其他需要添加的内容/澄清。

【参考方案1】：

 阅读官方docs，分析是调用.predict(docs_to_classify)时产生的错误建议你尝试做基本测试，比如使用参数labels = ["negative", "positive"] ，如果是字符串引起的，请纠正 外部文件的 值 并且您还应该选择检查它指示使用 管道 的位置。

pipeline = Pipeline()
pipeline.add_node(component=retriever, name="Retriever", inputs=["Query"])
pipeline.add_node(component=doc_classifier, name='DocClassifier', inputs=['Retriever'])

【讨论】：

啊！我的 tags 变量是空的，因为文件也是如此。不一定是您的代码，而是您的思维过程有所帮助。干杯人