Python/Keras/Theano - 索引越界

Question

我是 Keras 新手，在形状方面遇到一些问题，特别是在涉及 RNN 和 LSTM 时。

我正在运行此代码:

model=Sequential()
model.add(Embedding(input_dim=col,output_dim=70))
model.add(SimpleRNN(init='uniform',output_dim=30))
model.add(Dropout(0.5))
model.add(Dense(1))
model.compile(loss="mse", optimizer="sgd")
model.fit(X=predictor_train, y=target_train, nb_epoch=5, batch_size=1,show_accuracy=True)

我遇到此错误:

IndexError: index 143 is out of bounds for size 80
Apply node that caused the error: AdvancedSubtensor1(<TensorType(float32, matrix)>, Flatten{1}.0)
Inputs types: [TensorType(float32, matrix), TensorType(int32, vector)]
Inputs shapes: [(80, 70), (80,)]
Inputs strides: [(280, 4), (4,)]
Inputs values: ['not shown', 'not shown']

我不明白“index 143”来自哪里以及如何修复它。

有人可以启发我的旅程吗？

下面有额外信息。

-- 编辑-- 这个“索引 143”实际上每次运行代码时都会有所不同。这些数字不遵循任何明显的逻辑，我唯一注意到的是，无论是否巧合，出现的最小数字是 80(我运行了代码 20 多次)

---

额外信息

About predictor_train (X)

类型:'numpy.ndarray'

形状:(119,80)

数据类型:float64

About target_train (Y)

类型:类'pandas.core.series.Series'

形状:(119,)

数据类型:float64

Date
2004-10-01    0.003701
2005-05-01    0.001715
2005-06-01    0.002031
2005-07-01    0.002818
...
2015-05-01   -0.007597
2015-06-01   -0.007597
2015-07-01   -0.007597
2015-08-01   -0.007597

model.summary()

--------------------------------------------------------------------------------
Initial input shape: (None, 80)
--------------------------------------------------------------------------------
Layer (name)                  Output Shape                  Param #             
--------------------------------------------------------------------------------
Embedding (Unnamed)           (None, None, 70)              5600                
SimpleRNN (Unnamed)           (None, 30)                    3030                
Dropout (Unnamed)             (None, 30)                    0                   
Dense (Unnamed)               (None, 1)                     31                  
--------------------------------------------------------------------------------
Total params: 8661
--------------------------------------------------------------------------------

FULL TRACEBACK

File "/Users/file.py", line 1523, in Pred
model.fit(X=predictor_train, y=target_train, nb_epoch=5, batch_size=1,show_accuracy=True)
File "/Library/Python/2.7/site-packages/keras/models.py", line 581, in fit
shuffle=shuffle, metrics=metrics)
File "/Library/Python/2.7/site-packages/keras/models.py", line 239, in _fit
outs = f(ins_batch)
File "/Library/Python/2.7/site-packages/keras/backend/theano_backend.py", line 365, in __call__
return self.function(*inputs)
File "/Library/Python/2.7/site-packages/theano/compile/function_module.py", line 595, in __call__
outputs = self.fn()
File "/Library/Python/2.7/site-packages/theano/gof/vm.py", line 233, in __call__
link.raise_with_op(node, thunk)
File "/Library/Python/2.7/site-packages/theano/gof/vm.py", line 229, in __call__
thunk()
File "/Library/Python/2.7/site-packages/theano/gof/op.py", line 768, in rval
r = p(n, [x[0] for x in i], o)
File "/Library/Python/2.7/site-packages/theano/tensor/subtensor.py", line 1657, in perform
out[0] = x.take(i, axis=0, out=o)
IndexError: index 143 is out of bounds for size 80
Apply node that caused the error: AdvancedSubtensor1(<TensorType(float32, matrix)>, Flatten{1}.0)
Inputs types: [TensorType(float32, matrix), TensorType(int32, vector)]
Inputs shapes: [(80, 70), (80,)]
Inputs strides: [(280, 4), (4,)]
Inputs values: ['not shown', 'not shown']

Answer 1

您的X变量可能包含值143。Embedding层的尺寸为80x70。

我假设这是 NLP 领域。这意味着您的词汇量为 80 个单词，每个单词由长度为 70 的向量表示。您的 X 变量表示 119 个长度为 80 的句子(或 80 个长度为 119 的句子)，其内容表示词汇表的索引。如果它包含的单词索引大于 80，则会弹出此错误。

col 变量的更常见值是 10.000 以上。当然，这取决于你在做什么。