I am new to PyTorch, and I built a custom BiLSTM model for sentiment analysis that uses pretrained Word2Vec embeddings. As the text I am passing in is variable in length, my forward method needs to pass in both the text and its length to use pad_packed_sequence:
def forward(self, x, text_len):
The problem is trying to convert this model to ONNX.
My understanding is that the code should be something like this:
dummy_input_1 = torch.randn(1, seq_length, requires_grad=True).long()
dummy_input_2 = torch.randn(seq_length, requires_grad=True).long()
dynamic_axes = {'input_1' : {1 : 'len'}, 'input_2' : {0 : 'len'},
'output1': {0: 'label'}}
torch.onnx.export(model,
args=(dummy_input_1, dummy_input_2),
f='model.onnx',
input_names=['input_1', 'input_2'],
output_names=['output1'],
dynamic_axes=dynamic_axes)
However, wouldn't I need to define seq_length first? How do I do this? And is this even the correct approach?
Related
I am working on a classification problem in a project. The specificity of my problem is that I have to use two different type of data to manage it. My classes are Car, Pedestrian, Truck and Cyclist. My dataset is composed of :
-Images coming from the Camera : they are RGB image. Here is an example :
Images obtain by projecting Lidar Point Cloud (just 3D points) into 2D camera plane and encoding pixels using Depth & Reflectance. Here are examples :
I already manage to use both modalities in order to perform the classification task by using the Concatenate function of the keras API.
But what I would like to do is to use a more powerful CNN like VGG. I used pre-trained model and freeze all layers except the last 4. I read the grayscale image as RGB because the VGG16 pre-trained model need 3 channels input. Here is my code :
from keras.applications import VGG16
#Load the VGG model
#Camera Model
vgg_conv_C = VGG16(weights='imagenet', include_top=False, input_shape=(227, 227, 3))
#Depth Model
vgg_conv_D = VGG16(weights='imagenet', include_top=False, input_shape= (227, 227, 3))
for layer in vgg_conv_D.layers[:-4]:
layer.trainable = False
for layer in vgg_conv_C.layers[:-4]:
layer.trainable = False
mergedModel = Concatenate()([vgg_conv_C.output,vgg_conv_D.output])
mergedModel = Dense(units = 1024)(mergedModel)
mergedModel = BatchNormalization()(mergedModel)
mergedModel = Activation('relu')(mergedModel)
mergedModel = Dropout(0.5)(mergedModel)
mergedModel = Dense(units = 4,activation = 'softmax')(mergedModel)
fused_model = Model([vgg_conv_C.input, vgg_conv_D.input], mergedModel) )
The last line give the following error :
ValueError: The name "block1_conv1" is used 2 times in the model. All
layer names should be unique.
Did someone know how to handle this? To be simple, I just want to use VGG16 on both type of images, then just get the feature vectors for each modality, then Concatenate them and add fully connected layers at top to predict the image's class. It works with no-pre trained models. Can provide the code if needed
Try this
#Camera Model
vgg_conv_C = VGG16(weights='imagenet', include_top=False, input_shape=(227, 227, 3))
for layer in vgg_conv_C.layers:
layer.name = layer.name + str('_C')
#Depth Model
vgg_conv_D = VGG16(weights='imagenet', include_top=False, input_shape= (227, 227, 3))
for layer in vgg_conv_D.layers:
layer.name = layer.name + str('_D')
In this way, you'd still be able to use two identical pre-trained networks.
As mentioned in the error,
ValueError: The name "block1_conv1" is used 2 times in the model. All
layer names should be unique.
Therefore use Saimse network or If use dual CNN them remember in network layer ame are unique. its better and copy the network for second configuration and change the layers name.
IStackoverflowAndIKnowThings solution gives me the error:
AttributeError: Can't set the attribute "name", likely because it conflicts with an existing read-only #property of the object. Please choose a different name.
The following worked for me (see this post):
..
for layer in vgg_conv_C.layers:
layer._name = layer._name + str('_C')
..
I have 16 letters of DNA sequence. From this 16-letter DNA sequence, there is an output value so called 'Inhibition value' which ranges from 0 to 100. When I tried using LSTM, the prediction only output a constant. Is the problem lies in the code or is it just not a suitable task for LSTM or RNN in general to solve?
I have tried to increase batch size and epochs, make the LSTM deeper, change the number of LSTM units, but none of them works.
I was also wondering whether the labeling method matters or not. I tried to use One-hot encoder at first, but it didn't work. Then, I changed it to LabelEncoder, but it's also not working. Same constant output is produced.
Below here is the code for my model structure
def create_model():
input1 = Input(shape=(16,1))
classifier = LSTM(64, input_shape=(16,1), return_sequences=True)(input1)
for i in range(2):
classifier = LSTM(32, return_sequences=True)(classifier)
classifier = LSTM(32)(classifier)
classifier = Dense(1, activation='relu')(classifier)
model = Model(inputs = [input1], outputs = classifier)
adam = keras.optimizers.adam(lr=0.01)
model.compile(loss='mean_squared_error', optimizer=adam)
return model
If anyone wondering why I use functional API instead of sequential, it is because there is a possible modification where I need to use 2 input variables that needs to be processed independently before concatenating it at the end.
Thank you in advance.
I used gensim to build a word2vec embedding of my corpus.
Currently I'm converting my (padded) input sentences to the word vectors using the gensim model.
This vectors are used as input for the model.
model = Sequential()
model.add(Masking(mask_value=0.0, input_shape=(MAX_SEQUENCE_LENGTH, dim)))
model.add(Bidirectional(
LSTM(num_lstm, dropout=0.5, recurrent_dropout=0.4, return_sequences=True))
)
...
model.fit(training_sentences_vectors, training_labels, validation_data=validation_data)
Are there any drawbacks using the word vectors directly without a keras embedding layer?
I'm also currently adding additional (one-hot encoded) tags to the input tokens by concatenating them to each word vector, does this approach make sense?
In your current setup, the drawback will be that you will not be able to set your word vectors to be trainable. You will not be able to fine tune your model for your task.
What I mean by this is that Gensim has only learned the "Language Model". It understands your corpus and its contents. However, it does not know how to optimize for whatever downstream task you are using keras for. Your model's weights will help to fine tune your model, however you will likely experience an increase in performance if you extract the embeddings from gensim, use them to initialize a keras embedding layer, and then pass in indexes instead of word vectors for your input layer.
There's an elegant way to do what you need.
Problem with your solution is that:
the size of the input is large: (batch_size, MAX_SEQUENCE_LENGTH, dim) and may not fit in memory.
You won't be able to train and update the word vectors as per your task
You can instead get away with just: (batch_size, MAX_SEQUENCE_LENGTH). The keras embedding layer allows you to pass in a word index and get a vector. So, 42 -> Embedding Layer -> [3, 5.2, ..., 33].
Conveniently, gensim's w2v model has a function get_keras_embedding which creates the needed embedding layer for you with the trained weights.
gensim_model = # train it or load it
embedding_layer = gensim_model.wv.get_keras_embedding(train_embeddings=True)
embedding_layer.mask_zero = True # No need for a masking layer
model = Sequential()
model.add(embedding_layer) # your embedding layer
model.add(Bidirectional(
LSTM(num_lstm, dropout=0.5, recurrent_dropout=0.4, return_sequences=True))
)
But, you have to make sure the index for a word in the data is the same as the index for the word2vec model.
word2index = {}
for index, word in enumerate(model.wv.index2word):
word2index[word] = index
Use the above word2index dictionary to convert your input data to have the same index as the gensim model.
For example, your data might be:
X_train = [["hello", "there"], ["General", "Kenobi"]]
new_X_train = []
for sent in X_train:
temp_sent = []
for word in sent:
temp_sent.append(word2index[word])
# Add the padding for each sentence. Here I am padding with 0
temp_sent += [0] * (MAX_SEQUENCE_LENGTH - len(temp_sent))
new_X_train.append(temp_sent)
X_train = numpy.as_array(new_X_train)
Now you can use X_train and it will be like: [[23, 34, 0, 0], [21, 63, 0, 0]]
The Embedding Layer will map the index to that vector automatically and train it if needed.
I think this is the best way of doing it but I'll dig into how gensim wants it to be done and update this post if needed.
In keras, I want to calculate the mean of nonzero embedding output.
I wonder what is the difference between mask_zero=True or False in Embedding Layer.
I tried the code below :
input_data = Input(shape=(5,), dtype='int32', name='input')
embedding_layer = Embedding(1000, 24, input_length=5,mask_zero=True,name='embedding')
out = word_embedding_layer(input_data)
def antirectifier(x):
x = K.mean(x, axis=1, keepdims=True)
return x
def antirectifier_output_shape(input_shape):
shape = list(input_shape)
return tuple(shape)
out = Lambda(antirectifier, output_shape=antirectifier_output_shape,name='lambda')(out)
But it seems that the result is the mean of all the elements, how can i just calculate the mean of all nonzero inputs?
From the function's doc :
If this is True then all subsequent layers in the model need to
support masking
Your lambda function doesn't support masking. For example Recurrent layers in Keras support masking. If you set mask_zero=True in your embeddings, then all the 0 indices that you feed to the embedding layer will be propagated as "masked" and the following layers that are able to understand the "masked" information will use them.
Basically, if you build a "mean" layer that grabs the mask and computes the average only for non-masked values, then you will get the desired results.
You can find here a way to build your lambda layers that support masking
I hope it helps.
As we know:
Keras.layers.Embedding turns positive integers (indexes) into dense vectors of fixed size. e.g. [[4], [20]] -> [[0.25, 0.1], [0.6, -0.2]]
I want to know how can I see or print the dense vector output.
Or
how to see a tensor object's output?
You can take a look here : https://keras.io/getting-started/faq/#how-can-i-obtain-the-output-of-an-intermediate-layer
In a few words :
Create a new model from your trained model with the output layer in which you are interested, then use the methode predict.
layer_name = 'my_layer'
intermediate_layer_model = Model(inputs=model.input,
outputs=model.get_layer(layer_name).output)
intermediate_output = intermediate_layer_model.predict(data)