Python 3.13.5 : keras and torch and tensorflow on python version 3.13.5 !

I am not very satisfied with the security of the system in this area, I have noticed intrusions when I was at the Vodafone provider and now I am on mobile and they persist. I should improve my software and hardware protection and restart the system, but I have the impression that it won't help much considering the hacking capabilities that exist.

For now, I'm sticking with this solution, because it's quite good for simple tests with the Python language.

Today I will show you that the involvement of developers is somewhat lagging behind, and this can be seen from the analysis of the development progression across versions of the Python language, as it is not easy to implement in newer versions. Some packages, although important, are not available in newer versions.Today I will show you that the involvement of developers is somewhat lagging behind, and this can be seen from the analysis of the development progression across versions of the Python language, as it is not easy to implement in newer versions. Some packages, although important, are not available in newer versions. Let's see these python packages:

Keras is a deep learning API designed for human beings, not machines. Keras focuses on debugging speed, code elegance & conciseness, maintainability, and deployability. When you choose Keras, your codebase is smaller, more readable, easier to iterate on.

... read more on the official website.

pip install --upgrade keras
Collecting keras
  Downloading keras-3.10.0-py3-none-any.whl.metadata (6.0 kB)
...
Successfully installed absl-py-2.3.0 h5py-3.14.0 keras-3.10.0 ml-dtypes-0.5.1 namex-0.1.0 optree-0.16.0

TensorFlow does not support Python 3.13 as of July 2025; it supports Python 3.9–3.11. To use Keras with TensorFlow, downgrade to Python 3.11, and this install will comes with:

pip install tensorflow
ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow

You can use the torch python package:

pip install torch
Collecting torch
...
Successfully installed filelock-3.18.0 fsspec-2025.5.1 mpmath-1.3.0 sympy-1.14.0 torch-2.7.1

... during the test with these packages in running it seems that I have errors. I believe that those who develop Python didn't really have a strategy regarding the protection of installed packages because I've found all sorts of junk packages that can be installed with pip or from GitHub.

Python 3.11.11 : Colab simple test with CogVideoX-5B model and default example - part 050.

I tested this GitHub project from THUDM user on my colab google account and works well with CogVideoX-5B model.

You can find the default implementation on my colab GitHub repo.

The default example comes with thjs prompt:

prompt = ( "A panda, dressed in a small, red jacket and a tiny hat, sits on a wooden stool in a serene bamboo forest. " "The panda's fluffy paws strum a miniature acoustic guitar, producing soft, melodic tunes. Nearby, a few other " "pandas gather, watching curiously and some clapping in rhythm. Sunlight filters through the tall bamboo, " "casting a gentle glow on the scene. The panda's face is expressive, showing concentration and joy as it plays. " "The background includes a small, flowing stream and vibrant green foliage, enhancing the peaceful and magical " "atmosphere of this unique musical performance." )

The speed of rendering starts from :

38% ... 19/50 [30:54<50:57, 98.61s/it] using T4 GPU

... then run at :

100% ... 50/50 [1:26:09<00:00, 109.87s/it]

when the video render was 100% somehow google give this error, but the source code run well:

OutOfMemoryError: CUDA out of memory. Tried to allocate 1.32 GiB. GPU 0 has a total capacity of 14.74 GiB of which 654.12 MiB is free. Process 26970 has 14.10 GiB memory in use. Of the allocated memory 12.97 GiB is allocated by PyTorch, and 1.00 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

I think is need to set some extra memory on CUDA but this require to parse some documentation and is not a task for me now.

Python 3.11.11 : Colab simple test with VGG16 - part 049.

VGG16 is a deep convolutional neural network (CNN) trained on a massive image dataset called ImageNet. This architecture is known for its remarkable performance in image classification tasks and is widely used in various computer vision applications.

You can find one simple example on my GitHub project.

Python 3.10.12 : My colab test with Gated recurrent unit mechanism - part 037.

This is a simple example for Gated recurrent unit mechanism known as GRUs.

You can find this in my GitHub colab project.

import numpy as np
import tensorflow as tf
import keras
from keras import layers
units = 64
tf.keras.layers.GRU(
    units,
    activation="tanh",
    recurrent_activation="sigmoid",
    use_bias=True,
    kernel_initializer="glorot_uniform",
    recurrent_initializer="orthogonal",
    bias_initializer="zeros",
    kernel_regularizer=None,
    recurrent_regularizer=None,
    bias_regularizer=None,
    activity_regularizer=None,
    kernel_constraint=None,
    recurrent_constraint=None,
    bias_constraint=None,
    dropout=0.0,
    recurrent_dropout=0.0,
    return_sequences=False,
    return_state=False,
    go_backwards=False,
    stateful=False,
    unroll=False,
    time_major=False,
    reset_after=True,
)
inputs = tf.random.normal([32, 10, 8])
gru = tf.keras.layers.GRU(4)
output = gru(inputs)
print(output.shape)

gru = tf.keras.layers.GRU(4, return_sequences=True, return_state=True)
whole_sequence_output, final_state = gru(inputs)
print(whole_sequence_output.shape)
print(final_state.shape)