I am trying to use the tensorboard callback in keras. When I run the pretrained inceptionv3 model with the tensorboard callback I am getting the following warning:
INFO:tensorflow:Summary name conv2d_95/kernel:0 is illegal; using conv2d_95/kernel_0 instead.
I saw a comment on Github addressing this issue. SeaFX on his comment pointed out that he solved it by replacing variable.name with variable.name.replace(':','_'). I am unsure how to do that. Can anyone please help me. Thanks in advance :)
Not sure on getting name replacement to work however a workaround that may be sufficient for your needs is:
import tensorflow as tf
tf.logging.set_verbosity(tf.logging.WARN)
import keras
This will turn off all INFO level logging but keep warnings, errors etc.
See this question for a discussion on the various log levels and changing them. Personally I found setting the TF_CPP_MIN_LOG_LEVEL environment variable didn't work under Jupyter notebook but I haven't tested on base Python.
Related
I'm using MiniGrid library to work with different 2D navigation problems as experiments for my reinforcement learning problem. I'm also using stable-baselines3 library to train PPO models. but unfortunately, while training PPO using stable-baselines3 with MiniGrid environment, I got the following error.
Error
I imported the environment as follows,
import gymnasium as gym
from minigrid.wrappers import RGBImgObsWrapper
env = gym.make("MiniGrid-SimpleCrossingS9N1-v0", render_mode="human")
env = RGBImgObsWrapper(env)
env = ImgObsWrapper(env)
The training script using stable-baseline3 is as follows,
model = PPO('CnnPolicy', env, verbose=0)
model.learn(args.timesteps, callback)
I have done a quick debug and found a potential lead. I don't know if this is the real cause.
When I tried to load the environment directly from the gym the action space <class 'gym.spaces.discrete.Discrete'>. But when loaded from MiniGrid the action space is <class 'gymnasium.spaces.discrete.Discrete'>. Any help to sort out the problem is highly appreciated. Thanks in advance
That's the correct analysis, Stable Baselines3 doesn't support Gymnasium yet, so checks on gym.spaces.discrete.Discrete fail against gymnasium.
The following post answer explains how to workaround that, based on a currently open PR: OpenAI Gymnasium, are there any libraries with algorithms supporting it?.
I am working on a fully cython compiled python package, using setuptools and some reStructured Text docstrings to generate the pdf documentation using sphinx-autodoc with good success.
The docstrings are not displayed when loading the package in Jupyter Notebooks (using Pylance language server in VSCode). I did not test but suspect that the issue exists in other IDEs or frontends either. Switching to e.g. Jedi as a language server in VSCode did not help.
Using the help(...) command on the routines/classes yields the expected docstring, so the issue does not seem to be with the cython compilation.
There are quite a few issues about formatting issues popping up when searching for Docstrings and Pylance, the issue never seems to be that they simply can not be retrieved.
Since sphinx (autodoc) does not throw any errors and generates a full-package encompassing documentation set fine, this seems a lot like a problem with the language servers to me.
However, I suspect that there may be any compiler directive or setuptools option that I am unaware of that could help - but I do not know about these.
I do have
compiler_directives={'embedsignature': True}
in the setup.py and since the docstrings do show up when using help(..), I do not think this is an issue.
I have also added
import Cython.Compiler.Options
Cython.Compiler.Options.docstrings = True
even though that is indicated as being the default, and it did not change the behavior at all.
Scavenging through the options of Pylance in VSCode was unfruitful, too - I did not expect much there anyways, since obviously the display of the docstrings does work for plenty (all?) other packages that I looked at. Everything is on defaults.
I am following example by José F. Romaniello on session management with NHibernate. It's a very good article, however I'm struggling with it having very little experience with NHibernate, Windsor and MVC.
I am trying to re-create NHibernateInstaller, however encountering the following error: Component Castle.TypedFactory.DefaultInterfaceFactoryComponentSelector could not be resolved. Make sure you didn't misspell the name, and that component is registered.
In the sample project provided this error does not crop up, even though the installer is identical and Google does not come up with any results (which is very unusual). What causes this and how can it be avoided?
it seems a problem with the TypedFactoryFacility... are you doing this?
kernel.AddFacility<TypedFactoryFacility>();
before running all the installers?
uncomment the following code in Bootstrapper.cs file.
container.AddFacility();
This happened to me when I created my own implementation of ITypedFactoryComponentSelector, but forgot to register the selector itself.
There was no indication this was the actual problem (and the kernel debug information assured me the components can be resolved) - but registering it fixed the issue.
Hope this helps someone :-)
I have kind of a newbie question. I want to use some functions from prelude, but I'm getting same error all the time.
For example:
Undefined variable "diff"
Some other functions are working, some not. I guess I have to import something, Google didn't help me... I'm using WinHugs.
Here are two sites you should bookmark, because they'll help you find functions you need -- and find out what you need to import.
http://haskell.org/hoogle/
http://holumbus.fh-wedel.de/hayoo/hayoo.html
Searching both those sites, I don't see anything that looks like the function you want. Some possibilities that occur to me are:
You're learning from a book that has examples of things you'd type at the Linux command line. "diff" is a common Linux command for comparing two files. I believe the windows equivalent is comp.
You're using some sample code that you got somewhere, but you didn't import everything you needed. If you can find the file that contains the "diff" function, import that.
You've written a function called "diff", and put it in another file. In the file where you want to use "diff", you need to import the module that contains it.
Prelude is typically imported implicitly. Are you sure that is really your problem? Undefined variable "diff" is a bit vague, since I don't know of a diff function imported with Prelude. Can you elaborate?
Here is my problem:
I have two C++ modules, A and B, which are built as dynamically-linked libraries. A offers basic math functions, and custom exception types. B is a higher level module that uses A.
B::someFunction() calls a function from A, and tries to catch custom exception A:MyExceptionFromA in order to convert it into a custom type B:MyExceptionFromB (since users of module B do not need to know about the implementation details of A).
Everything works fine as long as I remain in the C++ domain. However, if I expose B::someFunction() in python via boost python, the exception is not caught anymore in the C++ module.
I can catch std::runtime_error, from which A:MyExceptionFromA derives, and call typeid(e).name() to get the retrieve the correct mangled name, so I know the correct exception is thrown. Therefore I suspect that the problem comes from resolving this mangled symbol into the correct exception type.
I have found this link, which explains that "python uses [the insular] model to open extension modules, so that extension module writers don't need to know what symbols other extension modules might be using.". I'm suspecting this is part of the problem/solution, but I do not know enough about symbol resolution to figure out how to solve my problem.
Any ideas?
I found a work-around to my problem. Based on this and link text, I figured out that adding
import sys, dl
sys.setdlopenflags(dl.RTLD_NOW|dl.RTLD_GLOBAL)
before my includes solves the problem, by forcing python to open libraries in immediate, global mode. But I'm still hoping for an alternative solution, if there's one. As mentioned in the first link, I'm suspicious that this could have unforeseen effects (I already know that name clashing could be a problem, and I suspect performance can be affected as well, but are there other side effects?)