predict_sentiment(text) about chinese_sentiment

[davidqing2000]

可尝试改一下以下代码：

num_words = 50000 --替换为以下-》

num_words=len(cn_model.vocab)

[bbbubblegum]

你好，对于这个情感分析任务，我有几个疑问。
1.我们将一条短评，进行分词，然后将利用预训练的word2vec构建构建这个短评的embeddding,让这个embuding作为LSTM的输入项.而LSTM输入项应该是分词词语的向量才对吧，因为只有词语和词语之间才会有时间序列上的关联，而句子和句子间是没有的。
例如下图中的 x , x , x<t+1> ,应该是词语向量，而不是一个个embedding。

2.如果LSTM输入的是词语向量，而LSTM中每个一LSTM单元的输出层都是二分类的话，是对一个一个词语去判断他的情感极性么？我觉得应该是整个LSTM循环之后再连接一个输出层二分类这样才是合理吧，然后如果是这样的话，我们对每一条短评都建立一个LSTM神经网络模型和二分类的输出层，最后一共m个短评构建出一个LSTM神经网络集群，从而去定义整个集群的损失函数sum(loss)/m么？
很抱歉，我是个初学者，在对LSTM和Word2vec上得理解得不够准确和透彻，希望得到您的解答，谢谢。

[bluesky632]

我也是初学者，我的理解是：第一个问题，输入是词语的向量，也可以是单个字的向量，第二个问题我理解的也不是很透彻，展开图里的第二层是每个输入之间都是由联系的，并且LSTM的原理里，是对信息进行取舍判断的，我觉得是在一整句话的基础上，进行极性的判断，我也不是太清楚，如果有错误的话，请包涵。

[bbbubblegum]

谢谢你的解答
我的理解是 单个字，两个字，三个字，都可以是词语，只要是word2vec上可以将其转换成词向量的，都可以将这个词用词向量来表示。
一条短评中各个词语之间是有信息之间的关系，而一条短评和另一条短评之间是没有什么什么信息之间的联系的吧？我们embedding之后，把短评句子构建的矩阵作为输入项x，那么怎么应该下 x , x,x<t+1>都应该是该句子的词向量吧？而lstm神经网络中，我们不需要输出y，因为我们不需要知道每个词语是什么情感极性或者是其他的东西，只需要用他的激活项和词向量来控制门的开闭，最后整个lstm连接一个二分类的输出层，输出句子的情感极性。
我理解得到也可能不到位，望见谅。

[Faith-Uchiha]

num_words=len(cn_model.vocab)

这是把词典中所有的词都放进去了么？@davidqing2000

[aespresso]

请查看更新的debug之后的代码，解决了这个bug，
对于分类问题，可以看看我讲的机器学习中的数学： https://www.bilibili.com/video/av53400966

[LiAI-tech]

开始训练

model.fit(X_train, y_train,
validation_split=0.1,
epochs=20,
batch_size=128,
callbacks=callbacks)

报错如下：
ValueError Traceback (most recent call last)
E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\op_def_library.py in _apply_op_helper(self, op_type_name, name, **keywords)
526 as_ref=input_arg.is_ref,
--> 527 preferred_dtype=default_dtype)
528 except TypeError as err:

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\ops.py in internal_convert_to_tensor(value, dtype, name, as_ref, preferred_dtype, ctx, accept_composite_tensors)
1295 if ret is None:
-> 1296 ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
1297

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in _constant_tensor_conversion_function(v, dtype, name, as_ref)
285 _ = as_ref
--> 286 return constant(v, dtype=dtype, name=name)
287

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in constant(value, dtype, shape, name)
226 return _constant_impl(value, dtype, shape, name, verify_shape=False,
--> 227 allow_broadcast=True)
228

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in _constant_impl(value, dtype, shape, name, verify_shape, allow_broadcast)
264 value, dtype=dtype, shape=shape, verify_shape=verify_shape,
--> 265 allow_broadcast=allow_broadcast))
266 dtype_value = attr_value_pb2.AttrValue(type=tensor_value.tensor.dtype)

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\tensor_util.py in make_tensor_proto(values, dtype, shape, verify_shape, allow_broadcast)
436 if values is None:
--> 437 raise ValueError("None values not supported.")
438 # if dtype is provided, forces numpy array to be the type

ValueError: None values not supported.

During handling of the above exception, another exception occurred:

ValueError Traceback (most recent call last)
E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\op_def_library.py in _apply_op_helper(self, op_type_name, name, **keywords)
540 observed = ops.internal_convert_to_tensor(
--> 541 values, as_ref=input_arg.is_ref).dtype.name
542 except ValueError as err:

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\ops.py in internal_convert_to_tensor(value, dtype, name, as_ref, preferred_dtype, ctx, accept_composite_tensors)
1295 if ret is None:
-> 1296 ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
1297

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in _constant_tensor_conversion_function(v, dtype, name, as_ref)
285 _ = as_ref
--> 286 return constant(v, dtype=dtype, name=name)
287

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in constant(value, dtype, shape, name)
226 return _constant_impl(value, dtype, shape, name, verify_shape=False,
--> 227 allow_broadcast=True)
228

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\constant_op.py in _constant_impl(value, dtype, shape, name, verify_shape, allow_broadcast)
264 value, dtype=dtype, shape=shape, verify_shape=verify_shape,
--> 265 allow_broadcast=allow_broadcast))
266 dtype_value = attr_value_pb2.AttrValue(type=tensor_value.tensor.dtype)

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\tensor_util.py in make_tensor_proto(values, dtype, shape, verify_shape, allow_broadcast)
436 if values is None:
--> 437 raise ValueError("None values not supported.")
438 # if dtype is provided, forces numpy array to be the type

ValueError: None values not supported.

During handling of the above exception, another exception occurred:

ValueError Traceback (most recent call last)
in
4 epochs=20,
5 batch_size=128,
----> 6 callbacks=callbacks)

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, max_queue_size, workers, use_multiprocessing, **kwargs)
726 max_queue_size=max_queue_size,
727 workers=workers,
--> 728 use_multiprocessing=use_multiprocessing)
729
730 def evaluate(self,

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training_arrays.py in fit(self, model, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, **kwargs)
672 validation_steps=validation_steps,
673 validation_freq=validation_freq,
--> 674 steps_name='steps_per_epoch')
675
676 def evaluate(self,

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training_arrays.py in model_iteration(model, inputs, targets, sample_weights, batch_size, epochs, verbose, callbacks, val_inputs, val_targets, val_sample_weights, shuffle, initial_epoch, steps_per_epoch, validation_steps, validation_freq, mode, validation_in_fit, prepared_feed_values_from_dataset, steps_name, **kwargs)
187 # function we recompile the metrics based on the updated
188 # sample_weight_mode value.
--> 189 f = _make_execution_function(model, mode)
190
191 # Prepare validation data. Hold references to the iterator and the input list

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training_arrays.py in _make_execution_function(model, mode)
563 if model._distribution_strategy:
564 return distributed_training_utils._make_execution_function(model, mode)
--> 565 return model._make_execution_function(mode)
566
567

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training.py in _make_execution_function(self, mode)
2182 def _make_execution_function(self, mode):
2183 if mode == ModeKeys.TRAIN:
-> 2184 self._make_train_function()
2185 return self.train_function
2186 if mode == ModeKeys.TEST:

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\engine\training.py in _make_train_function(self)
2114 # Training updates
2115 updates = self.optimizer.get_updates(
-> 2116 params=self._collected_trainable_weights, loss=self.total_loss)
2117 # Unconditional updates
2118 updates += self.get_updates_for(None)

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\optimizers.py in get_updates(self, loss, params)
474
475 def get_updates(self, loss, params):
--> 476 grads = self.get_gradients(loss, params)
477 self.updates = []
478

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\keras\optimizers.py in get_gradients(self, loss, params)
90 """
91 grads = K.gradients(loss, params)
---> 92 if None in grads:
93 raise ValueError('An operation has None for gradient. '
94 'Please make sure that all of your ops have a '

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\ops\math_ops.py in tensor_equals(self, other)
1334 return gen_math_ops.equal(self, other, incompatible_shape_error=False)
1335 else:
-> 1336 return gen_math_ops.equal(self, other)
1337 else:
1338 # In legacy graph mode, tensor equality is object equality

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\ops\gen_math_ops.py in equal(x, y, incompatible_shape_error, name)
3624 _, _, _op = _op_def_lib._apply_op_helper(
3625 "Equal", x=x, y=y, incompatible_shape_error=incompatible_shape_error,
-> 3626 name=name)
3627 _result = _op.outputs[:]
3628 _inputs_flat = _op.inputs

E:\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow_core\python\framework\op_def_library.py in _apply_op_helper(self, op_type_name, name, **keywords)
543 raise ValueError(
544 "Tried to convert '%s' to a tensor and failed. Error: %s" %
--> 545 (input_name, err))
546 prefix = ("Input '%s' of '%s' Op has type %s that does not match" %
547 (input_name, op_type_name, observed))
ValueError: Tried to convert 'y' to a tensor and failed. Error: None values not supported.

是因为TensorFlow版本不同导致的吗？

[LiAI-tech]

bug已解决，是TensorFlow框架问题，代码没有问题

可参考https://github.com/tensorflow/tensorflow/pull/33097/files#diff-bfba746b141112f6e8f852c873e01b84R92