How to use channel.send?


#1

Hi, I am trying to run code on my local machine and send logs through neptune api.
Is it possible to use like this?
I am executing the following code:


ctx = neptune.Context()
#params = read_params(ctx)

numeric_channel = ctx.create_channel(
    name='numeric_channel',
    channel_type=neptune.ChannelType.NUMERIC)

text_channel = ctx.create_channel(
    name='text_channel',
    channel_type=neptune.ChannelType.TEXT)

numeric_channel.send(x=1, y=2.5)
numeric_channel.send(x=1.5, y=5)

text_channel.send(x=2.5, y='text 1')
text_channel.send(x=3, y='text 2')```

#2

Hi @Denys,

You should add following line at the beginning of your code to use neptune library.
from deepsense import neptune
After that your code will be totally fine.


#3

This line is present in my code, I didnt add it here because I thought it is clear. Is it possible to use neptune not on neptune-ml cloud, but localy and just collect logs?


#4

Yes, it is!
Use neptune run command instead of neptune send so your code will be executed on your local machine. All channel values will be still reported to Neptune.


#5

When executing the code above inside my project directory with the code above using neptune run filename.py I have the following behaviour.

It seems that it starts to send all directory to cloud


#6

Yes, it actually sends your project to Neptune. If you want to avoid it you can add following line to your neptune.yaml file:

exclude: ['*']

It will not upload any sources from your directory.
You should see uploading message like one below:

> Experiment enqueued, id: SAN-1                                                                                                                                                  
>                                                                                                                                                                                 
> To browse the experiment, follow:                                                                                                                                                                                                                                            
> https://app.neptune.ml/dashboard/experiment/163965ef-cb3e-4a7e-89ad-f74e3bb38607?getStartedState=folded                                                                                                                                                                      
>                                                                                                                                                                                                                                                                              
                                                                                                                                                                                                                                                                               
Calculated experiment snapshot size: 0 Bytes                                                                                                                                                                                                                                   
Sending sources to server: 0.00B [00:00, ?B/s]     ```

#7

Yes, now it works, thank you)
Is it possible to do the same inside jupyter notebook just using python api?
When I start ctx = neptune.Context() in the jupyter notebook it says that the Ofline mode is used …


#8

Could you feed us with id of your notebook? I will look at this issue.


#9

I run it locally, but logs should be uploaded to neptune-ml


#10

Currently, Neptune doesn’t support running notebooks locally. We will consider adding this feature in future releases.