allennlp.common.checks.ConfigurationError：使用SemBERT时srl不是模型的注册名称

Question

我正在尝试使用SemBERT预训练模型来获取一些句子对的包围度分数（https://github.com/cooelf/SemBERT-导航至“使用原始数据进行评估”）。

这是我的SemBERT目录的样子-

README.md
├── SemBERT.png
├── data_process
│   ├── __init__.py
│   ├── __pycache__
│   │   ├── __init__.cpython-38.pyc
│   │   └── datasets.cpython-38.pyc
│   ├── datasets.py
│   └── util.py
├── glue_data
│   └── MNLI
│       ├── dev_matched.tsv_tag_label
│       ├── test_matched.tsv_tag_label
│       └── train.tsv_tag_label
├── output
├── pytorch_pretrained_bert
│   ├── __init__.py
│   ├── __main__.py
│   ├── __pycache__
│   │   ├── __init__.cpython-38.pyc
│   │   ├── file_utils.cpython-38.pyc
│   │   ├── modeling.cpython-38.pyc
│   │   ├── optimization.cpython-38.pyc
│   │   └── tokenization.cpython-38.pyc
│   ├── file_utils.py
│   ├── modeling.py
│   ├── optimization.py
│   └── tokenization.py
├── run_classifier.py
├── run_scorer.py
├── run_snli_predict.py
├── snli_model_dir
│   ├── bert_config.json
│   ├── pytorch_model.bin
│   └── vocab.txt
├── srl_model_dir
│   ├── config.json
│   ├── files_to_archive.json
│   ├── fta
│   │   ├── model.text_field_embedder.elmo.options_file
│   │   └── model.text_field_embedder.elmo.weight_file
│   ├── vocabulary
│   │   ├── labels.txt
│   │   ├── non_padded_namespaces.txt
│   │   └── tokens.txt
│   └── weights.th
└── tag_model
    ├── __pycache__
    │   ├── modeling.cpython-38.pyc
    │   ├── tag_tokenization.cpython-38.pyc
    │   └── tagging.cpython-38.pyc
    ├── modeling.py
    ├── tag_tokenization.py
    ├── tagger_offline.py
    └── tagging.py

如github存储库中所述，我运行了以下命令-

CUDA_VISIBLE_DEVICES=0 \
python run_snli_predict.py \
--data_dir /share03/zhangzs/glue_data/SNLI \
--task_name snli \
--eval_batch_size 128 \
--max_seq_length 128 \
--max_num_aspect 3 \
--do_eval \
--do_lower_case \
--bert_model snli_model_dir \
--output_dir snli_model_dir \
--tagger_path srl_model_dir

但是我收到 allennlp.common.checks.ConfigurationError 这是错误说明-

08/10/2020 13:06:44 - INFO - __main__ -   device: cpu n_gpu: 0, distributed training: False, 16-bits training: False
08/10/2020 13:06:44 - INFO - pytorch_pretrained_bert.tokenization -   loading vocabulary file snli_model_dir/vocab.txt
08/10/2020 13:06:44 - INFO - allennlp.models.archival -   loading archive file srl_model_dir
Traceback (most recent call last):
  File "run_snli_predict.py", line 598, in <module>
    main()
  File "run_snli_predict.py", line 464, in main
    srl_predictor = SRLPredictor(args.tagger_path)
  File "/home/akib/SemBERT/tag_model/tagging.py", line 7, in __init__
    self.predictor = Predictor.from_path(SRL_MODEL_PATH)
  File "/home/akib/.local/lib/python3.8/site-packages/allennlp/predictors/predictor.py", line 275, in from_path
    load_archive(archive_path, cuda_device=cuda_device),
  File "/home/akib/.local/lib/python3.8/site-packages/allennlp/models/archival.py", line 192, in load_archive
    model = Model.load(
  File "/home/akib/.local/lib/python3.8/site-packages/allennlp/models/model.py", line 391, in load
    model_class: Type[Model] = cls.by_name(model_type)  # type: ignore
  File "/home/akib/.local/lib/python3.8/site-packages/allennlp/common/registrable.py", line 137, in by_name
    subclass, constructor = cls.resolve_class_name(name)
  File "/home/akib/.local/lib/python3.8/site-packages/allennlp/common/registrable.py", line 184, in resolve_class_name
    raise ConfigurationError(
allennlp.common.checks.ConfigurationError: srl is not a registered name for Model. You probably need to use the --include-package flag to load your custom code. Alternatively, you can specify your choices using fully-qualified paths, e.g. {"model": "my_module.models.MyModel"} in which case they will be automatically imported correctly.