All Questions
Tagged with performance tensorflow
211
questions
0
votes
0
answers
71
views
Keras training speed with PyTorch backend is a lot slower than with TensorFlow
I am on native Windows and I used old Keras with TensorFlow 2.10 (GPU accelerated) before. I wanted to try Keras 3 with PyTorch backend.
Can someone please help me why this model trains 10x slower ...
- 13
-1
votes
0
answers
42
views
Speed up multiple tensorflow training tasks on a single A100 GPU
I need to run multiple tensorflow training tasks for molecular dynamics. I find that a training task usually takes only 8G of VRAM during the training which is only 10% of A100 's 80G VRAM.
So I think ...
- 1,519
0
votes
0
answers
87
views
ONNX I/O Binding
I need to bind tensor input and output using I/O Binding for ONNX runtime model. but i didn't get output. the output tensor return NULL pointer. I will attached the code below.
std::vector<Ort::...
0
votes
1
answer
153
views
How to make Kivy, opencv and tensorflow work together
Hi I am trying to make kivy, opencv and tensorflow work together, however it is reall slow, so that my model can't predict anything, because the image isn't updated frequently enough. I believe it is ...
- 1
0
votes
0
answers
7
views
Tensorflow: RBF Net for measuring speed of a fluid in a tub over microphone
I need a plan to bulid a RBF net over tensorflor to measure th speed of a flluid in a tub over a microphone.
Plan is to make training samples of sound with th corresponding fluid speed.
In use the ...
0
votes
0
answers
35
views
tf.data performance for timeseries forecasting
I am training a time series forecasting model in Tensorflow. I create a tf.data.Dataset containing batches of windows of data using the approach presented in the example notebook https://www....
- 148
0
votes
0
answers
59
views
Tensorflow Real Time Significantly Greater Than CPU Time
So I have a model of tensorflow that is saved in a protobuf format. And I am trying to use this model for some real-time inference. It's performance critical so I test the inference latency by writing ...
- 101
0
votes
0
answers
13
views
Running subpart of tensorflow model on CPU
I got a model which consists of several submodels. A few of these submodels should be run on CPU (as it has to use a for loop and is not suited for GPU), all others on GPU.
I tried to call like this ...
- 33
0
votes
1
answer
16
views
Predictive Coder in Tensorflow? Can it be done efficiently?
For a compression application I am trying to implement the most simplest differential pulse-code modulation (DPCM). If you do not know DPCM, it is just a differential encoding scheme, where you ...
- 33
0
votes
1
answer
35
views
Tensorflow Model is extremly slow (due to for loop in call method?)
I got the following model, which I am currently prototyping to generalize it further once it works sufficiently quick.
class FRAE(tf.keras.Model):
def __init__(self, latent_dim, shape, ht, n1, n2, n3, ...
- 33
1
vote
0
answers
194
views
Automatic Differentiation: PyTorch vs. Tensorflow
I am creating a simple function that simulates N paths of Geometric Brownian Motions (GBM) with M discretization (M+1 if you include the starting point).
My function is the returning the values of the ...
- 257
0
votes
1
answer
829
views
Why is my PyTorch code 10x slower than the TensorFlow version?
I'm trying to learn how to optimize models and have written some simple fashion MNIST training in TF and PyTorch to see how they compare when run on CPU vs GPU. My environment is a Windows PC, i7-10, ...
- 3
1
vote
1
answer
40
views
invalid results of process_time() when measuring model.fit() performance
I use the snippet below to measure and output the time spent during model fitting.
perf_counter_train_begin = time.perf_counter()
process_time_train_begin = time.process_time()
model.fit(data, ...)
...
- 449
0
votes
1
answer
52
views
is there a way to make this more efficient, iterating through rows in a data frame
I want to iterate through the rows of my data frame and extract the values from the rows into variables and then append that into arrays. the df is very long, about 30,000.
I read online that you can ...
0
votes
1
answer
324
views
Performance difference between tf.boolean_mask and tf.gather + tf.where
tf.boolean_mask reads much nicer than then combination of tf.gather and tf.where. However, it seems to be much slower in the 1-D case:
import tensorflow as tf
# use this shape
shape = [5000]
# ...
- 8,242