finlay-liu / kaggle_public Goto Github PK

情感分类中，DATA_LIST = np.array(DATA_LIST)这一步会报错ValueError: setting an array element with a sequence，因为DATA_LIST原本是‘字符串+array’的形式，因此就不能再array了，不知道这个问题怎么解决，有大佬指点一下吗/

RuntimeWarning: Early stopping conditioned on metric val_acc which is not available. Available metrics are: val_loss,val_accuracy,val_acc_top2,loss,accuracy,acc_top2
(self.monitor, ','.join(list(logs.keys()))), RuntimeWarning

出现这个错是因为什么啊？keras版本？需要将val_acc改成val_accuracy 或者是val_loss吗

这个最后的评分 我用训练集拆分了一部分拿来做测试机这时候的评分也就0.8左右，但是用测试数据来评测为啥f1评分就只有0.3了啊，acc也就0.5左右

小白想请问下这个baseline大概要跑多久的？？只在CPU环境下能跑完吗？因为跑了两天都还没跑完，epochs一直在循环。新闻情感分析题。

IOError: [Errno 2] No such file or directory: './bert-chinese-ner/output/result_dir/label_test.txt

运行相关代码有以上的IOError,请问这个label_test.txt是如何生成的？

ValueError: Error when checking target: expected dense_1 to have 2 dimensions, but got array with shape (8, 1, 2)
应该是采用 categorical_crossentropy 作损失函数时，需要将结果变为one-hot形式？

token_dict = {}

with codecs.open(dict_path, 'r', 'utf8') as reader:
    for line in reader:
        token = line.strip()
        token_dict[token] = len(token_dict)

class OurTokenizer(Tokenizer):
    def _tokenize(self, text):
        R = []
        for c in text:
            if c in self._token_dict:
                R.append(c)
            elif self._is_space(c):
                R.append('[unused1]') # space类用未经训练的[unused1]表示
            else:
                R.append('[UNK]') # 剩余的字符是[UNK]
        return R

tokenizer = OurTokenizer(token_dict)

水哥，请问一下分词代码里面self._token_dict是什么意思啊，也没看到定义，程序会走到R.append这个分支吗

AttributeError Traceback (most recent call last)
in ()
----> 1 train_model_pred, test_model_pred = run_cv(10, DATA_LIST, None, DATA_LIST_TEST)
2 test_pred = [np.argmax(x) for x in test_model_pred]
3 test_df['label'] = test_pred
4 test_df[['id', 'label']].to_csv('baseline3.csv', index=None)

4 frames
/usr/local/lib/python3.6/dist-packages/keras/engine/base_layer.py in input_shape(self)
887 return unpack_singleton(input_shapes)
888 else:
--> 889 raise AttributeError('The layer "' + str(self.name) +
890 ' has multiple inbound nodes, '
891 'with different input shapes. Hence '

AttributeError: The layer "model_14 has multiple inbound nodes, with different input shapes. Hence the notion of "input shape" is ill-defined for the layer. Use get_input_shape_at(node_index) instead

InvalidArgumentError: Incompatible shapes: [8,514,768] vs. [8,512,768]
[[{{node model_2/Embedding-Position/add}} = Add[T=DT_FLOAT, _class=["loc:@training/Adam/gradients/model_2/Embedding-Position/add_grad/Reshape"], _device="/job:localhost/replica:0/task:0/device:GPU:0"](model_2/Embedding-Token-Segment/add, model_2/Embedding-Position/Tile)]]
[[{{node metrics/acc_top2/Mean_1/_3259}} = _Recvclient_terminated=false, recv_device="/job:localhost/replica:0/task:0/device:CPU:0", send_device="/job:localhost/replica:0/task:0/device:GPU:0", send_device_incarnation=1, tensor_name="edge_21926_metrics/acc_top2/Mean_1", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:CPU:0"]]

bert模型参数太多，没有显卡跑起来特别慢，你这些都是在kaggle 上运行过嘛？花费的时间是多久 ？