如何看一个batch的具体NER结果呢

[zywzyw]

No description provided.

[zywzyw]

I tried python evaluate.py chips_2020
then I got the following error message：
Traceback (most recent call last):
File "evaluate.py", line 22, in
model.evaluate(session,True)
File "/home/xxx/biaffine-ner/biaffine_ner_model.py", line 287, in evaluate
self.load_eval_data()
File "/home/xxx/biaffine-ner/biaffine_ner_model.py", line 280, in load_eval_data
self.eval_data = [load_line(l) for l in f.readlines()]
File "/home/xxx/biaffine-ner/biaffine_ner_model.py", line 277, in load_line
return self.tensorize_example(example, is_training=False), example
File "/home/xxx/biaffine-ner/biaffine_ner_model.py", line 127, in tensorize_example
lm_emb = self.load_lm_embeddings(doc_key)
File "/home/xxx/biaffine-ner/biaffine_ner_model.py", line 93, in load_lm_embeddings
group = self.lm_file[file_key]
File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper
File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper
File "/home/xxx/.pyenv/versions/biaffine/lib/python2.7/site-packages/h5py/_hl/group.py", line 264, in getitem
oid = h5o.open(self.id, self._e(name), lapl=self._lapl)
File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper
File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper
File "h5py/h5o.pyx", line 190, in h5py.h5o.open
KeyError: "Unable to open object (object 'batch_test_1' doesn't exist)"

[juntaoy]

You will need to run the extract_bert_features.sh first to get the cached BERT embedding for your test set.

[zywzyw]

You will need to run the extract_bert_features.sh first to get the cached BERT embedding for your test set.

I really appreciate your quick reply!
Actually, I have already done the part you mentioned and I could see the result on the Tensorboard, like Precision and Recall. But I desire to see the NER label type on one exact example, so I run the command python evaluate.py chips_2020 . the test filename is test.jsonlines, and the doc_key is 'batch_test_1'. I would like konw that if I should run python evaluate.py chips_2020 when I want to get result of the label type.

[juntaoy]

If you got the error it might mean you only cached BERT embeddings for train and dev set but not the test set, you might need to run it for test.jsonlines as well. If you have done so correctly you will be able to get the scores for both overall and per-label.