I am new to ML and W&B, and I am trying to use W&B to do a hyperparameter sweep. I created a few sweeps and when I run them I get a bunch of new runs in my project (as I would expect):
Image: New runs being created
However, all of the new runs say "no metrics logged yet" (Image) and are instead all of their metrics are going into one run (the one with the green dot in the photo above). This makes it not useable, of course, since all the metrics and images and graph data for many different runs are all being crammed into one run.
Is there anyone that has some experience in W&B? I feel like this is an issue that should be relatively straightforward to solve - like something in the W&B config that I need to change.
Any help would be appreciated. I didn't give too many details because I am hoping this is relatively straightforward, but if there are any specific questions I'd be happy to provide more info. The basics:
Using Google Colab for training
Project is a PyTorch-YOLOv3 object detection model that is based on this: https://github.com/ultralytics/yolov3
Thanks! 😊
Update: I think I figured it out.
I was using the train.py code from the repository I linked in the question, and part of that code specifies the id of the run (used for resuming).
I removed the part where it specifies the ID, and it is now working :)
Old code:
wandb_run = wandb.init(config=opt, resume="allow",
project='YOLOv3' if opt.project == 'runs/train' else Path(opt.project).stem,
name=save_dir.stem,
id=ckpt.get('wandb_id') if 'ckpt' in locals() else None)
New code:
wandb_run = wandb.init(config=opt, resume="allow",
project='YOLOv3' if opt.project == 'runs/train' else Path(opt.project).stem,
name=save_dir.stem)
Related
This is a great package for Bayesian optimization of hyperparameters (especially mixed integer/continuous/categorical...and has shown to be better than Spearmint in benchmarks). However, clearly it is meant for Linux. What do I do...?
First you need to download swig.exe (the whole package) and unzip it. Then drop it somewhere and add the folder to path so that the installer for SMAC3 can call swig.exe.
Next, the Resource module is going to cause issues because that is only meant for Linux. That is specifically used by Pynisher. You'll need to comment out import pynisher in the execute_func.py module. Then, set use_pynisher:bool=False in the def __init__(self...) in the same module. The default is true.
Then, go down to the middle of the module where an if self.use_pynisher....else statement exists. Obviously our code now enters the else part, but it is not setup correctly. Change result = self.ta(config, **obj_kwargs) to result = self.ta(list(config.get_dictionary().values())). This part may need to be adjusted yet depending on what kind of inputs your function handles, but essentially you can see that this will enable the basic example shown in the included branin_fmin.py module. If doing the random forest example, don't change at all...etc.
My goal is to create / update objects by using a csv file. csv file is fine and allready heavy used.
I wrote a complete script for that, most setSomething() works very well with text, structured Datas etc.. but i'm facing a trouble with the Images. I read the doc twice...
https://pimcore.com/docs/4.6.x/Development_Documentation/Objects/Object_Classes/Data_Types/Image_Types.html
$image = Asset\Image::getByPath("/examples/example1.jpg");
$object->setImage($image);
$object->save();
and that exactly what's i did to create / update $article
$marque = Pimcore\Model\Asset\Image::getByPath("/Catalog/Marques/".$image);
$article->setEarg($marque);
$article->save();
please note that :
$image exists,
$marque is found, and is an instance of Asset\Image
setEarg($marque) is properly defined to set an Image data field
in UI all works fine
cache cleared
Is it possible i misunderstood something ?
I also searched in passed Issues, no luck.
https://github.com/pimcore/pimcore/issues
Maybe someone did face the same problem ? Anc could a hint on how to solve that ?
Regards,
I'm confused...this code is working as it should.
It seems to a cache problem, even if i cleared it. In the UI image did appear after some minutes / hour. Now it appear immedialtly. This question can be closed, but this push me to another one... how did this cache / deffer problem happened. If i find something i will update here.
I'm new to Intel RealSense. I want to learn how to save the color and depth streams to bitmap. I'm using C++ as my language. I have learned that there is a function ToBitmap(), but it can be used for C#.
So I wanted to know is there any method or any function that will help me in saving the streams.
Thanks in advance.
I'm also working my way through this, It seems that the only option is to do it manually. We need to get ImageData from PXCImage. The actual data is stored in ImageData.planes but I still don't understand how it's organized.
https://software.intel.com/en-us/articles/dipping-into-the-intel-realsense-raw-data-stream?language=en Here you can find example of getting depth data.
But I still have no idea what is pitches and how data inside planes is organized.
Here: https://software.intel.com/en-us/forums/intel-perceptual-computing-sdk/topic/332718 kind of backwards process is described.
I would be glad if you will be able to get some insight from this information.
And I obviously would be glad if you've discovered some insight you can share :).
UPD: Here is something that looks like what we need, I haven't worked with it yet, but it sheds some light on internal organization of planes[0] https://software.intel.com/en-us/forums/intel-perceptual-computing-sdk/topic/514663
UPD2: To add some completeness to the answer:
You then can create GDI+ image from data in ImageData:
auto colorData = PXCImage::ImageData();
if (image->AcquireAccess(PXCImage::ACCESS_READ, PXCImage::PIXEL_FORMAT_RGB24, &colorData) >= PXC_STATUS_NO_ERROR) {
auto colorInfo = image->QueryInfo();
auto colorPitch = colorData.pitches[0] / sizeof(pxcBYTE);
Gdiplus::Bitmap tBitMap(colorInfo.width, colorInfo.height, colorPitch, PixelFormat24bppRGB, baseColorAddress);
}
And Bitmap is subclass of Image (https://msdn.microsoft.com/en-us/library/windows/desktop/ms534462(v=vs.85).aspx). You can save Image to file in different formats.
I am trying to use the Sphinx4 library for speech recognition, but I cannot seem to figure out the correct combination of acoustic model-dictionary-language model. I have tried out various combinations and I get a different error every time.
I am trying to follow the tutorial on http://cmusphinx.sourceforge.net/wiki/tutorialsphinx4. I do not have a config.xml as I would if I was using ConfigurationManager instead of Configuration, because there is no perceivable way of passing the location of the config file to the Configuration itself (ConfigMgr takes it as an argument to the constructor); and that might be my problem right there. I just do not know how to point to one, and since the tutorial says "It is possible to configure low-level components of the application through XML file although you should do that ONLY IF you understand what is going on.", I assume having a config.xml file is not compulsory.
Combining the latest dictionary (7b - obtained from Sourceforge) with the latest acoustic model (cmusphinx-en-us-5.2.tar.gz - from SF again) and the language model (cmusphinx-5.0-en-us.lm.gz - from SF again) results in NullPointerException in startRecognition. The issue is similar to the problem here: sphinx-4 NullPointerException at startRecognition, but the link given in the answer no longer works. I obtained 0.7a from SF (since that is the dict the link seems to point at), but I am getting even earlier in the execution Error loading word: ;;; when I use that one. I tried downloading latest models and dict from the Github repo, that results in java.lang.IndexOutOfBoundsException: Index: 16128, Size: 16128.
Any help is much appreciated!
You need to use latest code from github
http://github.com/cmusphinx/sphinx4
as described by tutorial
http://cmusphinx.sourceforge.net/wiki/tutorialsphinx4
Correct models (en-us) are already included, you should not replace anything. You should not configure any XML files, use samples as provided in the sources.
This question is related to the github issue of Neo4django. I want to create multiple graphs using Neo4j graph DB from Django web framework. I'm using Django 1.4.5, neo4j 1.9.2 and neo4django 0.1.8.
As of now Neo4django doesn't support labeling but the above is my core purpose and I want to be able to create labels from Neo4django. So I went into the source code and tried to tweak it a little to see if I can make this addition. In my understanding, the file 'db/models/properties.py' has class BoundProperty(AttrRouter) which calls gremlin script through function save(instance, node, node_is_new). The script is as follows:
script = '''
node=g.v(nodeId);
results = Neo4Django.updateNodeProperties(node, propMap);
'''
The script calls the update function from library.groovy and all the function looks intuitive and nice. I'm trying to add on this function to support labeling but I have no experience of groovy. Does anyone have any suggestions on how to proceed? Any help would be appreciated. If it works it would be a big addition to neo4django :)
Thank you
A little background:
The Groovy code you've highlighted is executed using the Neo4j Gremlin plugin. First it supports the Gremlin graph DSL (eg node=g.v(nodeId)), which is implemented atop the Groovy language. Groovy itself is a dynamic superset of Java, so most valid Java code will work with scripts sent via connection.gremlin(...). Each script sent should define a results variable that will be returned to neo4django, even if it's just null.
Anyway, accessing Neo4j this way is handy (though will be deprecated I've heard :( ) because you can use the full Neo4j embeddeded Java API. Try something like this to add a label to a node
from neo4django.db import connection
connection.gremlin("""
node = g.v(nodeId)
label = DynamicLabel.label('Label_Name')
node.rawVertex.addLabel(label)
""", nodeId=node_id)
You might also need to add an import for DynamicLabel- I haven't run this code so I'm not sure. Debugging code written this way is a little tough, so make liberal use of the Gremlin tab in the Neo4j admin.
If you come up with a working solution, I'd love to see it (or an explanatory blog post!)- I'm sure it could be helpful to other users.
HTH!
NB - Labels will be properly supported shortly after Neo4j 2.0's release- they'll replace the current in-graph type structure.