你好，我把数据集换成自己的就会这样报错，请问如何能够解决或者是什么原因呢？

    train(data, save_model_dir,dset_dir, seg)
  File "C:/Users/11/Desktop/Name-Entity-Recognition-master/Name-Entity-Recognition-master/fyz_lattice_NER/main.py", line 310, in train
    print("     Instance: %s; Time: %.2fs; loss: %.4f; acc: %s/%s=%.4f"%(end, temp_cost, sample_loss, right_token, whole_token，(right_token+0.)/whole_token))
ZeroDivisionError: float division by zero

为什么模型输出要比label_list多一个呢
https://github.com/FuYanzhe2/Name-Entity-Recognition/blob/master/BERT-BiLSTM-CRF-NER/bert_lstm_ner.py#L652

@FuYanzhe2 如题，如图

你好，请教一下，fyz_lattice_NER 中文NER lattice model这个的数据集在哪里下载？
谢谢

请问BERT-BiLSTM-CRF-NER文件夹中没有main.py文件么？

输出有很多这个日志，这是干什么的啊 @FuYanzhe2 @fuyanzhe
I0823 07:37:37.097895 139764188952448 tpu_estimator.py:2159] global_step/sec: 0.522892
I0823 07:37:37.098161 139764188952448 tpu_estimator.py:2160] examples/sec: 16.7326
I0823 07:37:39.207139 139764188952448 tpu_estimator.py:2159] global_step/sec: 0.

请问在使用您的模型用在自己的数据集上时，是不是不用输入自己标注实体的tag了呢？

您好，我看了您的命名实体识别的代码，发现在评估代码的时候会出现，utf8错误，如果更改evesteps值为1，有结果但为nan，还有就是 out of index 怎么解决呀

我这边调用了这个接口，返回的都是nan 这个要怎么办

为什么验证的时候重新定义了eval_loss，原来tatal_loss此时又代表什么呢？
代码位置如下：

针对NER ,进行了修改

        def metric_fn(label_ids, logits, trans):
            # 首先对结果进行维特比解码
            # crf 解码

            weight = tf.sequence_mask(FLAGS.max_seq_length)
            precision = tf_metrics.precision(label_ids, pred_ids, num_labels, [2, 3, 4, 5, 6, 7], weight)
            recall = tf_metrics.recall(label_ids, pred_ids, num_labels, [2, 3, 4, 5, 6, 7], weight)
            f = tf_metrics.f1(label_ids, pred_ids, num_labels, [2, 3, 4, 5, 6, 7], weight)

            return {
                "eval_precision": precision,
                "eval_recall": recall,
                "eval_f": f,
                # "eval_loss": loss,
            }

        eval_metrics = (metric_fn, [label_ids, logits, trans])
        # eval_metrics = (metric_fn, [label_ids, logits])
        output_spec = tf.contrib.tpu.TPUEstimatorSpec(
            mode=mode,
            loss=total_loss,
            eval_metrics=eval_metrics,
            scaffold_fn=scaffold_fn)  #

已经按ｍｄ文件在fyz_lattice_NER下载好lattice 词向量了，工作目录ｔｒｅｅ如下：

可是不管是跑main.py或是fyz_run_decode.sh都报缺少文件，内容如下：

Thank you for your work to modify the lattice LSTM code! I have tried your lattice LSTM code, but my f1-value just achieve 0.82, much lower than the other models with the same dataset. I want to know how about your result, tks.

BERT-BiLSTM-CRF-NER文件夹中bert_lstm_ner.py中的crf损失值在迭代很多次之后会变成负数？不知道为什么会这样，还望解答一下，还有为什么num_labels=len(label_list)+1?我隐约感觉到损失值为负数与标签数是有关系的。

Traceback (most recent call last):
File "main.py", line 48, in
flags.DEFINE_string("emb_file", "wiki_100.utf8", "Path for pre_trained embedding")
NameError: name 'os' is not defined

label_test.txt文件中会出现O后面连接I的情况

想问一下大家如何取出字向量的值呢，我没法在原有的图中运行sess.run(tensor)，因为我无法定义这个sess。可是当我新建了一个图之后，运行sess.run(tensor)，会报错：RuntimeError: The Session graph is empty. Add operations to the graph before calling run()。如图我想取出embeddings中的一个batch的数据，但是当我调用这个函数后，就提示上述错误，我百度谷歌了很多答案，但是都解决不了，连问题出在哪儿都不知道，麻烦大家看看能否提供点思路，谢谢大家。

Traceback (most recent call last):
File "C:/Users/Chenxinliang/Desktop/Name-Entity-Recognition-master/Chinese_ner/main.py", line 231, in
if name == "main":
File "C:\Users\Chenxinliang\Desktop\BERT-NER_3.0_test\venv\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "C:/Users/Chenxinliang/Desktop/Name-Entity-Recognition-master/Chinese_ner/main.py", line 225, in main
clean(FLAGS)
File "C:/Users/Chenxinliang/Desktop/Name-Entity-Recognition-master/Chinese_ner/main.py", line 191, in train

File "C:/Users/Chenxinliang/Desktop/Name-Entity-Recognition-master/Chinese_ner/main.py", line 88, in evaluate
# print(ner_results)
File "C:\Users\Chenxinliang\Desktop\Name-Entity-Recognition-master\Chinese_ner\utils.py", line 69, in test_ner
eval_lines = return_report(output_file)
File "C:\Users\Chenxinliang\Desktop\Name-Entity-Recognition-master\Chinese_ner\conlleval.py", line 282, in return_report
counts = evaluate(f)
File "C:\Users\Chenxinliang\Desktop\Name-Entity-Recognition-master\Chinese_ner\conlleval.py", line 74, in evaluate
for line in iterable:
File "C:\Users\Chenxinliang\AppData\Local\Programs\Python\Python36\lib\codecs.py", line 711, in next
return next(self.reader)
File "C:\Users\Chenxinliang\AppData\Local\Programs\Python\Python36\lib\codecs.py", line 642, in next
line = self.readline()
File "C:\Users\Chenxinliang\AppData\Local\Programs\Python\Python36\lib\codecs.py", line 555, in readline
data = self.read(readsize, firstline=True)
File "C:\Users\Chenxinliang\AppData\Local\Programs\Python\Python36\lib\codecs.py", line 501, in read
newchars, decodedbytes = self.decode(data, self.errors)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa3 in position 0: invalid start byte

您好，运行您的chinese_ner 出现这种错误，请问怎么解决呢？

你好，对于crf_loss的计算有个疑问

在Chinese_ner项目的crf_loss层用的是num_tags+1对输出的头和尾加上了一个类似和的标签，用来计算转移概率。

但是在BERT-BiLSTM-CRF-NER项目的crf_loss层用的是num_labels，
https://github.com/FuYanzhe2/Name-Entity-Recognition/blob/master/BERT-BiLSTM-CRF-NER/lstm_crf_layer.py#L138
这是因为BERT输出的embedding已经包括[CLS]和[SEP]吗？

运行 fyz_lattice_NER 时报错如下：

FileNotFoundError: [Errno 2] No such file or directory: 'test_data/fyz.train.embs'

请问是不是缺失了这部分的文件？

纯萌新，因为最近的研究需要参考您的这部分代码，望不吝赐教。:D