25 questions
-1
votes
1
answer
64
views
TPU for Pix-to-Pix possible?
I have model Pix-to-Pix. Not segmentation. Transfer from texture to texture. Loss = 'mse'.
And it works properly with GPU. But if i try to use TPU i get loss = 'nan'
I simplified the model by cutting ...
- 1
-1
votes
1
answer
171
views
Does SoX internally operate in Int32 precision, not in Float32? [closed]
I noticed that even when converting 24-bit Signed Integer PCM audio to Float32, clipping still occurs even after returning to the original gain level.
sox in.wav -b 32 -e floating-point out.wav
It ...
- 61.2k
1
vote
2
answers
294
views
Float32 to float16 conversion in python and how it impacts np.isclose's atol
I am trying to compare each layer outputs of two ML models trained with different libs (tensorflow and a custom lighter one) that have the same valid performance in some sort of a "unitary test&...
- 11
1
vote
1
answer
362
views
Do bfloat types have any benefits over fp16 provided by VK_KHR_16bit_storage extension?
In vulkan api there are two extensions to use 16-bit types, namely VK_KHR_16bit_storage and VK_KHR_shader_float16_int8. So, if the hardware supports these extensions, fp16 type variables can be ...
- 376
2
votes
0
answers
263
views
Troubleshooting FP8 Conversion Discrepancy from Float32
I'm trying to convert a float32 to an fp8 float. There's some detail in the code that is causing a discrepancy in the results. I can't figure out what it is, could someone help me?
def ...
- 21
3
votes
0
answers
871
views
Clang tidy problem with _Float32, _Float64, _Float128
Why does clang-tidy give me warnings about _Float32, _Float64, _Float128? The program runs fine, but the squiggles appear even in the most basic program under the first character of any file (source ...
- 87
1
vote
2
answers
3k
views
Convert tflite model of float32 model to int8
Is it possible to convert a tflite float32 to tflite model of uint8? I read this page https://www.tensorflow.org/lite/performance/post_training_quantization, but they are only loading Keras of ...
- 11
2
votes
1
answer
102
views
The normal distribution for the computational error of the Gauss and sweep methods for solving systems of linear equations does not work
The methods for solving systems of linear equations themselves work correctly. Tested with solve. But when trying to solve a large number of matrices, some solution vectors do not converge with the ...
- 23
2
votes
1
answer
1k
views
Converting Pandas dataframe to float32 changes value of low precision number (by adding incorrect high precision decimals)
I have a very large dataset with values that don't require a lot of decimal point precision. In one test scenario, my dataframe is 102 MB, and all columns have a float64 datatype.
I was hoping to ...
0
votes
1
answer
145
views
javascript float32Array push changes value
I recently find out some weird behaviour in JS, and i might need some insight on it.
i have a js script that i need to translate into C#.
now, in the JS i have:
values = new Float32Array(n).fill(0) (...
- 806
0
votes
1
answer
178
views
Fast read/unpacking of float32 from int16 in Python
Say I have a Python script which reads some binary data, packed as int16. I want to convert this data to float32 as fast as possible.
Currently I am doing this, per file
data = np.fromfile(fid, 'int16'...
1
vote
1
answer
300
views
How to create a PyGLM array of float32 type from a list of python floats
I have this function to modify it:
def buffers_init_pnt() -> None:
global mPoints
global POINTS_VAOID
global POINTS_VBOID
global glmverts_pnt_init
floatlist = []
for item ...
- 39
1
vote
0
answers
999
views
OpenCV allowed data types
I am working with OpenCV in Python. I found that working with uint8 and float32 works in OpenCV 4. This question OpenCV Mat element types and their sizes shows some more data types but I got an error ...
- 33
0
votes
2
answers
8k
views
"Input contains NaN, infinity or a value too large for dtype('float32')There are no Nan or blanks in my data set but still wont let me proceed
Keep getting the error 'Input contains NaN, infinity or a value too large for dtype('float32')' when trying to run a random forest regressor model. I've searched my data set and cant see any infinite ...
0
votes
1
answer
126
views
Possible bug with inf or too large values?
I'm trying to train a neural network with keras and tesorflow. As usual, I replace -np.inf and np.inf values with np.nan to later run a dropna sequence and clear all that wrong data such as:
Data....
1
vote
1
answer
590
views
Three.js - How to fill Float32Array with equal number of points for different geometries
I have a project that loads various models (.obj, can be anything) and generates particles from the geometry position using Float23Array's.
Given the geometries of each model are completely different, ...
- 3,890
0
votes
0
answers
96
views
How to convert (%) sign in float data type or delete it
When I try to read the date from Excel file, I found that
the column called "TOT_SALES" the data type is float64 and all values with % sign.
I want to remove this sign and dividing all ...
- 1
1
vote
1
answer
998
views
JavaScript Buffer to Float32
I am receiving a Buffer 40 48 f5 c3 3f b4 7a e1 3f 35 c2 8f with the following values, the values are coded as Float32 :
3.14: 0x4048f5c3
1.41: 0x3fb47ae1
0.71: 0x3f35c28f
Can I map the buffer into a ...
- 875
11
votes
1
answer
2k
views
np.float32 floating point differences between intel MacBook and M1
I have recently upgraded my Intel MacBook Pro 13" to a MacBook Pro 14" with M1 Pro. Been working hard on getting my software to compile and work again. No big issues fortunately, except for ...
- 219
0
votes
1
answer
213
views
How do I get the results from argmax of [0., 1e-8]?
When I ran below code, I got x=1, y=0
a = np.array([0., 1e-8]).astype('float32')
x = a.argmax()
y = (a+1).argmax()
I already know floating point expression. But, I don't know why I can get x=1, y=0.
...
- 21
5
votes
0
answers
2k
views
TypeError: You are passing KerasTensor..... when fitting a model
I am trying to finish a project on variational autoencoders (VAE)
I have tried lots of methods and lots of help from the internet but every time I run into a different problem.
Most often it is SHAPE ...
- 51
3
votes
0
answers
3k
views
Converting from Numpy Array to PyTorch Tensor
Since Numpy array is Float64 by default. How do I convert to PyTorch tensor to give a FLoat32 type and not 64?
I tried torch.from_numpy and it gave a Float64 type.
- 41
0
votes
1
answer
193
views
Change in precision value of numpy float32 while summing up the elements [duplicate]
I have an array and I would like to calculate the sum of the elements column wise (column_sum) and divide the column elements with the column_sum, so that after division, the sum of column elements ...
- 1
2
votes
1
answer
6k
views
How to convert 0-1 image float array to 0-255 int array
import matplotlib
import numpy as np
photo=plt.imread('Feynman.png')
plt.figure
plt.subplot(121)
plt.imshow(photo)
photo*=255
plt.subplot(122)
plt.imshow(photo)
Plot results in:
Clipping input data ...
- 43
0
votes
1
answer
1k
views
how to create a prefilled array of float32 in python?
if I do:
import numpy as np
arr1=np.tile(0.75, 32)
I get a 32 element array of float64 initialized with the value 0.75.
How to do the same but get a float32 array?
- 9,991