After creating a Keras custom layer with training weight, how can one test the correctness of the code? It does not seem to be described in Keras' manual.
For example, to test the expected behavior of a function, one can write a unit test. How can we do this for a Keras custom layer?
You can still do something like unit test by getting the output of the custom layer for the given input and verifying it against the manually calculated output,
Let's say your custom layer Custom takes (None, 3, 200) as input shape and returns (None, 3)
from keras.layers import Input
from keras.models import Model
inp = Input(shape=(3, 200))
out = Custom()(inp)
model = Model(inp, out)
output = model.predict(your_input)
You can verify the layer output output with your expected output for a known input your_input.
layer_test in keras utils.
https://github.com/keras-team/keras/blob/master/keras/utils/test_utils.py
They provide following code, which tests the shape, the actual result, serializing and training:
def layer_test(layer_cls, kwargs={}, input_shape=None, input_dtype=None,
input_data=None, expected_output=None,
expected_output_dtype=None, fixed_batch_size=False):
"""Test routine for a layer with a single input tensor
and single output tensor.
"""
# generate input data
if input_data is None:
assert input_shape
if not input_dtype:
input_dtype = K.floatx()
input_data_shape = list(input_shape)
for i, e in enumerate(input_data_shape):
if e is None:
input_data_shape[i] = np.random.randint(1, 4)
input_data = (10 * np.random.random(input_data_shape))
input_data = input_data.astype(input_dtype)
else:
if input_shape is None:
input_shape = input_data.shape
if input_dtype is None:
input_dtype = input_data.dtype
if expected_output_dtype is None:
expected_output_dtype = input_dtype
# instantiation
layer = layer_cls(**kwargs)
# test get_weights , set_weights at layer level
weights = layer.get_weights()
layer.set_weights(weights)
expected_output_shape = layer.compute_output_shape(input_shape)
# test in functional API
if fixed_batch_size:
x = Input(batch_shape=input_shape, dtype=input_dtype)
else:
x = Input(shape=input_shape[1:], dtype=input_dtype)
y = layer(x)
assert K.dtype(y) == expected_output_dtype
# check with the functional API
model = Model(x, y)
actual_output = model.predict(input_data)
actual_output_shape = actual_output.shape
for expected_dim, actual_dim in zip(expected_output_shape,
actual_output_shape):
if expected_dim is not None:
assert expected_dim == actual_dim
if expected_output is not None:
assert_allclose(actual_output, expected_output, rtol=1e-3)
# test serialization, weight setting at model level
model_config = model.get_config()
recovered_model = model.__class__.from_config(model_config)
if model.weights:
weights = model.get_weights()
recovered_model.set_weights(weights)
_output = recovered_model.predict(input_data)
assert_allclose(_output, actual_output, rtol=1e-3)
# test training mode (e.g. useful when the layer has a
# different behavior at training and testing time).
if has_arg(layer.call, 'training'):
model.compile('rmsprop', 'mse')
model.train_on_batch(input_data, actual_output)
# test instantiation from layer config
layer_config = layer.get_config()
layer_config['batch_input_shape'] = input_shape
layer = layer.__class__.from_config(layer_config)
# for further checks in the caller function
return actual_output
Related
I am following this tutorial from Pytorch for Finetuning a pre-trained model on my own dataset. I have my annotation in the COCO format in a json file, so, I first implemented the dataloader as follows:
import torch
import json
from torch.utils.data import Dataset
from pycocotools.coco import COCO
from PIL import Image
import os
import numpy as np
from torchvision import transforms
import Config
import transforms as T
from torchvision.transforms import functional as F
class CustomDataset(Dataset):
def __init__(self, root, json_file, transform=None):
self.root = root
with open(json_file) as f:
self.data = json.load(f)
self.transform = transform
self.image_ids = [img["id"] for img in self.data["images"]]
self.imgs = list(sorted(os.listdir(os.path.join(root, "Images"))))
self.masks = list(sorted(os.listdir(os.path.join(root, "Masks"))))
def __getitem__(self, idx):
# Get image ID
img_id = self.image_ids[idx]
img = next(image for image in self.data["images"] if image["id"] == img_id)
img_path = os.path.join(self.root, "Images")
mask_path = os.path.join(self.root, "Masks")
# Load image
image = Image.open(os.path.join(img_path, img['file_name'])).convert("RGB")
# extract annotations from the json file
annotations = [ann for ann in self.data["annotations"] if ann["image_id"] == img_id]
# extract labels from annotations
labels = [ann["label"] for ann in annotations]
# convert labels to integers
labels = [label for label in labels]
labels = torch.as_tensor(labels, dtype=torch.int64)
# extract boxes and convert them to format [x1, y1, x2, y2]
boxes = [ann["bbox"] for ann in annotations]
boxes = [[bbox[0], bbox[1], bbox[2], bbox[3]] for bbox in boxes]
num_objects = len(boxes)
# read the mask and include the number of objects in the first dimension
mask = np.array(Image.open(os.path.join(mask_path, img['file_name'])).convert("L"))
# Check if mask is empty
if mask.size == 0:
mask = np.zeros((num_objects, 1, 1), dtype=np.uint8)
else:
mask = np.expand_dims(mask, axis=0)
mask = np.repeat(mask, num_objects, axis=0)
# convert the binary mask array to a torch tensor
mask = torch.as_tensor(mask, dtype=torch.uint8)
# suppose all instances are not crowd
iscrowd = torch.zeros((num_objects,), dtype=torch.int64)
# convert bboxes to tensors
boxes = torch.as_tensor(boxes, dtype=torch.float32)
# calculate the area of the bounding box
area = (boxes[:, 3] - boxes[:, 1]) * (boxes[:, 2] - boxes[:, 0])
# convert id to tensor
image_id = torch.tensor([idx])
# create target dictionary
target = {}
target["boxes"] = boxes
target["labels"] = labels
target["masks"] = mask
target["image_id"] = image_id
target["area"] = area
target["iscrowd"] = iscrowd
# apply the transform if any
if self.transform is not None:
image, target = self.transform(image, target)
return image, target
def __len__(self):
return len(self.imgs)
and I am using this code for training:
import torchvision
from torchvision.models.detection.faster_rcnn import FastRCNNPredictor
from torchvision.models.detection.mask_rcnn import MaskRCNNPredictor
from engine import train_one_epoch
import utils
import transforms as T
from dataloader import CustomDataset
import Config
import torch
import utils
from tqdm import tqdm
from torch.optim.lr_scheduler import StepLR
from torchvision.transforms import functional as F
def get_instance_segmentation_model(num_classes):
# load an instance segmentation model pre-trained on COCO
model = torchvision.models.detection.maskrcnn_resnet50_fpn(pretrained=True)
# get the number of input features for the classifier
in_features = model.roi_heads.box_predictor.cls_score.in_features
# replace the pre-trained head with a new one
model.roi_heads.box_predictor = FastRCNNPredictor(in_features, num_classes)
# now get the number of input features for the mask classifier
in_features_mask = model.roi_heads.mask_predictor.conv5_mask.in_channels
hidden_layer = 256
# and replace the mask predictor with a new one
model.roi_heads.mask_predictor = MaskRCNNPredictor(in_features_mask,
hidden_layer,
num_classes)
return model
def get_transform(train):
transforms = []
# converts the image, a PIL image, into a PyTorch Tensor
transforms.append(T.PILToTensor())
if train:
# during training, randomly flip the training images
# and ground-truth for data augmentation
transforms.append(T.RandomHorizontalFlip(0.5))
return T.Compose(transforms)
json_path = 'annotations.json'
# use our dataset and defined transformations
dataset = CustomDataset(root = Config.Dataset_dir, json_file=json_path, transform = get_transform(train=True))
# for image, target in dataset:
# print(image.shape)
# split the dataset in train and test set
torch.manual_seed(1)
indices = torch.randperm(len(dataset)).tolist()
dataset = torch.utils.data.Subset(dataset, indices[:-500])
dataset_test = torch.utils.data.Subset(dataset, indices[-500:])
# define training and validation data loaders
data_loader = torch.utils.data.DataLoader(
dataset, batch_size=1, shuffle=True, num_workers=4,
collate_fn=utils.collate_fn)
data_loader_test = torch.utils.data.DataLoader(
dataset_test, batch_size=1, shuffle=False, num_workers=4,
collate_fn=utils.collate_fn)
device = Config.DEVICE
# # our dataset has two classes only - background and person
num_classes = 2
# get the model using our helper function
model = get_instance_segmentation_model(num_classes)
# move model to the right device
model.to(device)
# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer = torch.optim.SGD(params, lr=0.1,
momentum=0.9, weight_decay=0.0005)
# and a learning rate scheduler which decreases the learning rate by
# 10x every 3 epochs
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
step_size=3,
gamma=0.1)
# let's train it for 10 epochs
num_epochs = 10
for epoch in range(num_epochs):
# train for one epoch, printing every 10 iterations
train_one_epoch(model, optimizer, data_loader, device, epoch, print_freq=10)
# update the learning rate
lr_scheduler.step()
# evaluate on the test dataset
evaluate(model, data_loader_test, device=device)
This training code is as stated in the tutorial is using some helper functions which can be accessed from here. I have run the training code and the training is working for the first 10 samples in the data, but then it gives the following error:
Epoch: [0] [ 0/2759] eta: 13:29:50 lr: 0.000200 loss: -136.8811 (-136.8811) loss_classifier: 0.9397 (0.9397) loss_box_reg: 0.0017 (0.0017) loss_mask: -137.9142 (-137.9142) loss_objectness: 0.0859 (0.0859) loss_rpn_box_reg: 0.0057 (0.0057) time: 17.6117 data: 10.0775
Loss is nan, stopping training
{'loss_classifier': tensor(nan, grad_fn=<NllLossBackward0>), 'loss_box_reg': tensor(nan, grad_fn=<DivBackward0>), 'loss_mask': tensor(nan, grad_fn=<BinaryCrossEntropyWithLogitsBackward0>), 'loss_objectness': tensor(nan, grad_fn=<BinaryCrossEntropyWithLogitsBackward0>), 'loss_rpn_box_reg': tensor(nan, grad_fn=<DivBackward0>)}
An exception has occurred, use %tb to see the full traceback.
SystemExit: 1
This error is raised from the engine.py train_one_epoch function, especially from this part of the function:
with torch.cuda.amp.autocast(enabled=scaler is not None):
loss_dict = model(images, targets)
losses = sum(loss for loss in loss_dict.values())
# reduce losses over all GPUs for logging purposes
loss_dict_reduced = utils.reduce_dict(loss_dict)
losses_reduced = sum(loss for loss in loss_dict_reduced.values())
loss_value = losses_reduced.item()
if not math.isfinite(loss_value):
print(f"Loss is {loss_value}, stopping training")
print(loss_dict_reduced)
sys.exit(1)
Which indicates that the losses returned after the first loop are NaN ... What could be wrong here please? I am running out of ideas and don't know what's going wrong anymore.
I have a problem with data normalization in PyTorch when I try to execute the training. First thing you need to know is that the dataset is composed of 3024 signal windows (so 1 channel), each one with a length of 5000 samples, so the dimension of the CSV file is 5000x3024. Each signal has 1 label that needs to be predicted.
Here is the code for how I load and normalize the data:
class CSVDataset(Dataset):
# load the dataset
def __init__(self, path, normalize = False):
# load the csv file as a dataframe
df = read_csv(path)
df = df.transpose()
# store the inputs and outputs
self.X = df.values[:, :-1]
self.y = df.values[:, -1]
print("Dataset length: ", self.X.shape[0])
# ensure input data is floats
self.X = self.X.astype(np.float)
self.y = self.y.astype(np.float)
if normalize:
self.X = self.X.reshape(self.X.shape[1], self.X.shape[0])
min_X = np.min(self.X,0) # returns an array of means for each signal window
max_X = np.max(self.X,0)
self.X = (self.X - min_X)/(max_X-min_X)
min_y = np.min(self.y)
max_y = np.max(self.y)
self.y = (self.y - min_y)/(max_y-min_y)
# reshape input data
self.X = self.X.reshape(self.X.shape[0], 1, self.X.shape[1])
self.y = self.y.reshape(self.y.shape[0], 1)
# label encode target and ensure the values are floats
self.y = LabelEncoder().fit_transform(self.y)
self.y = self.y.astype(np.float)
# prepare the dataset
def prepare_data(path):
# load the dataset
dataset = CSVDataset(path, normalize = True)
# calculate split
train, test = dataset.get_splits()
# prepare data loaders
train_dl = DataLoader(train, batch_size=32, shuffle=True)
test_dl = DataLoader(test, batch_size=1024, shuffle=False)
return train_dl, test_dl
While the train method is:
def train_model(train_dl, model):
# define the optimization
criterion = BCELoss()
optimizer = SGD(model.parameters(), lr=0.01, momentum=0.9)
model = model.float()
# enumerate epochs
for epoch in range(100):
# enumerate mini batches
for i, (inputs, targets) in enumerate(iter(train_dl)):
targets = torch.reshape(targets, (32, 1))
# clear the gradients
optimizer.zero_grad()
# compute the model output
yhat = model(inputs.float())
# calculate loss
loss = criterion(yhat, targets.float())
# credit assignment
loss.backward()
# update model weights
optimizer.step()
The error that I get is in the line loss = criterion(yhat, targets.float()) and it says:
RuntimeError: all elements of input should be between 0 and 1
I have tried inspecting the X in the variable explorer and it doesn't seem that there are any values that are not between 0 and 1. I don't know what I could have done wrong in normalization. Can you help me?
Builtin loss functions refer to input and target to designate the prediction and label instances respectively. The error message should be understood as "input of the criterion" i.e. yhat, and not as "input of the model".
It seems yhat does not belong in [0, 1], while BCELoss expects a probability, not a logit. You can either
add a sigmoid layer as the last layer of your model, or
use nn.BCEWithLogitsLoss instead, which combines a sigmoid and the bce loss.
I am trying to convert pytorch model with multiple networks to ONNX, and encounter some problem.
The git repo: https://github.com/InterDigitalInc/HRFAE
The Trainer Class:
class Trainer(nn.Module):
def __init__(self, config):
super(Trainer, self).__init__()
# Load Hyperparameters
self.config = config
# Networks
self.enc = Encoder()
self.dec = Decoder()
self.mlp_style = Mod_Net()
self.dis = Dis_PatchGAN()
...
Here is how the trained model process image:
def gen_encode(self, x_a, age_a, age_b=0, training=False, target_age=0):
if target_age:
self.target_age = target_age
age_modif = self.target_age*torch.ones(age_a.size()).type_as(age_a)
else:
age_modif = self.random_age(age_a, diff_val=25)
# Generate modified image
self.content_code_a, skip_1, skip_2 = self.enc(x_a)
style_params_a = self.mlp_style(age_a)
style_params_b = self.mlp_style(age_modif)
x_a_recon = self.dec(self.content_code_a, style_params_a, skip_1, skip_2)
x_a_modif = self.dec(self.content_code_a, style_params_b, skip_1, skip_2)
return x_a_recon, x_a_modif, age_modif
And as following is how I did to convert to onnx:
enc = Encoder()
dec = Decoder()
mlp = Mod_Net()
layers = [enc, mlp, dec]
model = torch.nn.Sequential(*layers)
# here is my confusion: how do I specify the inputs of each layer??
# E.g. one of the outputs of 'enc' layer should be input of 'mlp' layer,
# or the outputs of 'enc' layer should be part of inputs of 'dec' layer...
params = torch.load('./logs/001/checkpoint')
model[0].load_state_dict(params['enc_state_dict'])
model[1].load_state_dict(params['mlp_style_state_dict'])
model[2].load_state_dict(params['dec_state_dict'])
torch.onnx.export(model, torch.randn([1, 3, 1024, 1024]), 'trained_hrfae.onnx', do_constant_folding=True)
Maybe the convert-part code is in wrong way??
Could anyone help, many thanks!
#20210629-11:52GMT Edit:
I found there's constraint of using torch.nn.Sequential. The output of former layer in Sequential should be consistent with latter input.
So my code shouldn't work at all because the output of 'enc' layer is not consistent with input of 'mlp' layer.
Could anyone help how to convert this type of pytorch model to onnx? Many thanks, again :)
After research and try, I found a method which maybe in correct way:
Convert each net(Encoder, Mod_Net, Decoder) to onnx model, and handle their input/output in latter logic-process or any further procedure (e.g convert to tflite model).
I'm trying to port onto Android using this method.
#Edit 20210705-03:52GMT#
Another approach may be better: write a new net combines the three nets. I've prove the output is same as origin pytorch model.
class HRFAE(nn.Module):
def __init__(self):
super(HRFAE, self).__init__()
self.enc = Encoder()
self.mlp_style = Mod_Net()
self.dec = Decoder()
def forward(self, x, age_modif):
content_code_a, skip_1, skip_2 = self.enc(x)
style_params_b = self.mlp_style(age_modif)
x_a_modif = self.dec(content_code_a, style_params_b, skip_1, skip_2)
return x_a_modif
and then convert use following:
net = HRFAE()
params = torch.load('./logs/002/checkpoint')
net.enc.load_state_dict(params['enc_state_dict'])
net.mlp_style.load_state_dict(params['mlp_style_state_dict'])
net.dec.load_state_dict(params['dec_state_dict'])
net.eval()
torch.onnx.export(net, (torch.randn([1, 3, 512, 512]), torch.randn([1]).type(torch.long)), 'test_hrfae.onnx')
This should be the answer.
I am new to Deep Learning and wondering how to modify my model to fix it.
It says Target 1 is out of bounds, so what parameter should I change to make it works. When the output is changed to 2, it works. However, the goal for the model is to predict 2 classes classification. Also, when output is 2, the training loss becomes nan.
The data is a dataframe with shape (15958, 4) transformed into tensor format.
Sorry Split_NN is a class:
# SplitNN
# to protect privacy and split
class SplitNN:
def __init__(self, models, optimizers):
self.models = models
self.optimizers = optimizers
self.data = []
self.remote_tensors = []
def forward(self, x):
data = []
remote_tensors = []
data.append(self.models[0](x))
if data[-1].location == self.models[1].location:
remote_tensors.append(data[-1].detach().requires_grad_())
else:
remote_tensors.append(
data[-1].detach().move(self.models[1].location).requires_grad_()
)
i = 1
while i < (len(models) - 1):
data.append(self.models[i](remote_tensors[-1]))
if data[-1].location == self.models[i + 1].location:
remote_tensors.append(data[-1].detach().requires_grad_())
else:
remote_tensors.append(
data[-1].detach().move(self.models[i + 1].location).requires_grad_()
)
i += 1
data.append(self.models[i](remote_tensors[-1]))
self.data = data
self.remote_tensors = remote_tensors
return data[-1]
def backward(self):
for i in range(len(models) - 2, -1, -1):
if self.remote_tensors[i].location == self.data[i].location:
grads = self.remote_tensors[i].grad.copy()
else:
grads = self.remote_tensors[i].grad.copy().move(self.data[i].location)
self.data[i].backward(grads)
def zero_grads(self):
for opt in self.optimizers:
opt.zero_grad()
def step(self):
for opt in self.optimizers:
opt.step()
Below are the codes:
Model set up: The Model is a sequential deep learning model, which I tried to use nn.linear to generated binary prediction.
torch.manual_seed(0)
# Define our model segments
input_size = 3
hidden_sizes = [128, 640]
output_size = 1
# original models
models = [
nn.Sequential(
nn.Linear(input_size, hidden_sizes[0]),
nn.ReLU(),
nn.Linear(hidden_sizes[0], hidden_sizes[1]),
nn.ReLU(),
),
nn.Sequential(nn.Linear(hidden_sizes[1], output_size), nn.LogSoftmax(dim=1)),
]
# Create optimisers for each segment and link to them
optimizers = [
optim.SGD(model.parameters(), lr=0.03,)
for model in models
]
Train model is here:
def train(x, target, splitNN):
#1) Zero our grads
splitNN.zero_grads()
#2) Make a prediction
pred = splitNN.forward(x)
#3) Figure out how much we missed by
criterion = nn.NLLLoss()
loss = criterion(pred, target)
#4) Backprop the loss on the end layer
loss.backward()
#5) Feed Gradients backward through the nework
splitNN.backward()
#6) Change the weights
splitNN.step()
return loss, pred
Finally the training part, also the part where problem happen:
the send function is for assigning model to the nodes, cuz this is set up to simulating federated learning.
for i in range(epochs):
running_loss = 0
correct_preds = 0
total_preds = 0
for (data, ids1), (labels, ids2) in dataloader:
# Train a model
data = data.send(models[0].location)
data = data.view(data.shape[0], -1)
labels = labels.send(models[-1].location)
# Call model
loss, preds = train(data.float(), labels, splitNN)
# Collect statistics
running_loss += loss.get()
correct_preds += preds.max(1)[1].eq(labels).sum().get().item()
total_preds += preds.get().size(0)
print(f"Epoch {i} - Training loss: {running_loss/len(dataloader):.3f} - Accuracy: {100*correct_preds/total_preds:.3f}")
The error show the problem occurs at loss, preds = train(data.float(), labels, splitNN)
The actual error message:
During handling of the above exception, another exception occurred:
IndexError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py in nll_loss(input, target, weight, size_average, ignore_index, reduce, reduction)
1836 .format(input.size(0), target.size(0)))
1837 if dim == 2:
-> 1838 ret = torch._C._nn.nll_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
1839 elif dim == 4:
1840 ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
IndexError: Target 1 is out of bounds.
Please help me. Thank you
I am trying to use a custom Keras loss function that apart from the usual signature (y_true, y_pred) takes another parameter sigma (which is also produced by the last layer of the network).
The training works fine, but then I am not sure how to perform forward propagation and return sigma (while muis the output of the model.predict method).
This is the code I am using, which features a custom layer GaussianLayer that returns the list [mu, sigma].
import tensorflow as tf
from keras import backend as K
from keras.layers import Input, Dense, Layer, Dropout
from keras.models import Model
from keras.initializers import glorot_normal
import numpy as np
def custom_loss(sigma):
def gaussian_loss(y_true, y_pred):
return tf.reduce_mean(0.5*tf.log(sigma) + 0.5*tf.div(tf.square(y_true - y_pred), sigma)) + 10
return gaussian_loss
class GaussianLayer(Layer):
def __init__(self, output_dim, **kwargs):
self.output_dim = output_dim
super(GaussianLayer, self).__init__(**kwargs)
def build(self, input_shape):
self.kernel_1 = self.add_weight(name='kernel_1',
shape=(30, self.output_dim),
initializer=glorot_normal(),
trainable=True)
self.kernel_2 = self.add_weight(name='kernel_2',
shape=(30, self.output_dim),
initializer=glorot_normal(),
trainable=True)
self.bias_1 = self.add_weight(name='bias_1',
shape=(self.output_dim, ),
initializer=glorot_normal(),
trainable=True)
self.bias_2 = self.add_weight(name='bias_2',
shape=(self.output_dim, ),
initializer=glorot_normal(),
trainable=True)
super(GaussianLayer, self).build(input_shape)
def call(self, x):
output_mu = K.dot(x, self.kernel_1) + self.bias_1
output_sig = K.dot(x, self.kernel_2) + self.bias_2
output_sig_pos = K.log(1 + K.exp(output_sig)) + 1e-06
return [output_mu, output_sig_pos]
def compute_output_shape(self, input_shape):
return [(input_shape[0], self.output_dim), (input_shape[0], self.output_dim)]
# This returns a tensor
inputs = Input(shape=(1,))
x = Dense(30, activation='relu')(inputs)
x = Dropout(0.3)(x)
x = Dense(30, activation='relu')(x)
x = Dense(40, activation='relu')(x)
x = Dropout(0.3)(x)
x = Dense(30, activation='relu')(x)
mu, sigma = GaussianLayer(1)(x)
model = Model(inputs, mu)
model.compile(loss=custom_loss(sigma), optimizer='adam')
model.fit(train_x, train_y, epochs=150)
Since your model returns two tensors as output, you also need to pass a list of two arrays as the output when calling fit() method. That's essentially what the error is trying to convey:
Error when checking model target:
So the error is in targets (i.e. labels). What is wrong?
the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 2 array(s), but instead got the following list of 1 arrays:
I may have found the answer among Keras FAQs.
I found out that it is possible to retrieve intermediate steps' output using the code snippet below:
layer_name = 'main_output'
intermediate_layer_model = Model(inputs=model.input,
outputs=model.get_layer(layer_name).output)
intermediate_output = intermediate_layer_model.predict(train_x[0])
intermediate_output
In this case intermediate_output is a list of two values [mu, sigma] (just needed to name the output layer main_output and retrieve it later)