Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
, This Is A Great Tool For Visualizing All Your Experiments. I Wanted To Know That When I Am Logging Scalar Plots With Title As Train Loss And Test Loss They Are Getting Diplayed As Train Loss And Test Loss In The Scalar Tab. I Wanted That The Title Shoul

@<1523701205467926528:profile|AgitatedDove14> , this is a great tool for visualizing all your experiments. I wanted to know that when I am logging scalar plots with title as train loss and test loss they are getting diplayed as train loss and test loss in the scalar tab.
I wanted that the title should be loss and under that I should get these two differnet graphs train loss and test loss. Is this possible?
image

  
  
Posted 4 years ago
Votes Newest

Answers 68


so, like if validation loss appears then there will be three sub-tags under one main tag loss

  
  
Posted 4 years ago

You can do:
task = Task.get_task(task_id='uuid_of_experiment')
task.get_logger().report_scalar(...)

Now the only question is who will create the initial Task, so that the others can report to it. Do you have like a "master" process ?

  
  
Posted 4 years ago

You can always click on the name of the series and remove it for display.
Why would you need three graphs?

  
  
Posted 4 years ago

i mean all 100 experiments in one project

  
  
Posted 4 years ago

This code gives me the graph that I displayed above

  
  
Posted 4 years ago

I didn't got it.

  
  
Posted 4 years ago

Oh I got it.

  
  
Posted 4 years ago

See on line 212 I am calling one function "combined" with some arguments

  
  
Posted 4 years ago

then if there are 10 experiments then I have to call Task.create() for those 10 experiments

  
  
Posted 4 years ago

then if there are 100 experiments how it will create 100 tasks?

  
  
Posted 4 years ago

Like here in the sidebar I am getting three different plots named as loss, train_loss and test_loss

  
  
Posted 4 years ago

and then log using logger

  
  
Posted 4 years ago

yes

  
  
Posted 4 years ago

image

  
  
Posted 4 years ago

I have 100 experiments and I have to log them and update those experiments every 5 minutes

  
  
Posted 4 years ago

and under that there will be three graphs with title as train test and loss

  
  
Posted 4 years ago

And you want all of them to log into the same experiment ? or do you want an experiment per 60sec (i.e. like the scheduler)

  
  
Posted 4 years ago

okay, Thanks @<1523701205467926528:profile|AgitatedDove14> for the help.

  
  
Posted 4 years ago

Hi @<1523701205467926528:profile|AgitatedDove14> , I wanted to ask you something. Is it possible that we can talk over voice somewhere so that I can explain my problem better?

  
  
Posted 4 years ago

logger.report_scalar(title="loss", series="train", iteration=0, value=100)
logger.report_scalar(title="loss", series="test", iteration=0, value=200)

  
  
Posted 4 years ago

Just so I understand,
scheduler executes main every 60sec
main spins X sub-processes
Each subprocess needs to report scalars ?

  
  
Posted 4 years ago

It will not create another 100 tasks, they will all use the main Task. Think of it as they "inherit" it from the main process. If the main process never created a task (i.e. no call to Tasl.init) then they will create their own tasks (i.e. each one will create its own task and you will end up with 100 tasks)

  
  
Posted 4 years ago

like in the above picture

  
  
Posted 4 years ago

Is this u meant?

  
  
Posted 4 years ago

So you want these two on two different graphs ?

  
  
Posted 4 years ago

Create one experiment (I guess in the scheduler)
task = Task.init('test', 'one big experiment')
Then make sure the the scheduler creates the "main" process as subprocess, basically the default behavior)
Then the sub process can call Task.init and it will get the scheduler Task (i.e. it will not create a new task). Just make sure they all call Task init with the same task name and the same project name.

  
  
Posted 4 years ago

main will initialize parent task and then my multiprocessing occurs which call combined function with parameters as project_name and exp_name

  
  
Posted 4 years ago

In the side bar you get the title of the graphs, then when you click on them you can see the diff series on the graphs themselves

  
  
Posted 4 years ago

No. since you are using Pool. there is no need to call task init again. Just call it once before you create the Pool, then when you want to use it, just do task = Task.current_task()

  
  
Posted 4 years ago

so what I have done is rather than reading sequentially I am reading those experiments through multiprocessing and for each experiment I am creating new task with specified project_name and task_name

  
  
Posted 4 years ago
25K Views
68 Answers
4 years ago
7 months ago
Tags