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
[Help Needed] - Is There An Api That I Can Extract The Metrics For Multiple Experiments? It Maybe A Feature That Can Be Considered. I Want To Compare Two Scalar Metrics With Different Graph Type (Line (Default), Scatter Plot, Tables Etc). It Should Be Do

[Help needed] - Is there an API that I can extract the metrics for multiple experiments?

It maybe a feature that can be considered. I want to compare two scalar metrics with different graph type (Line (default), scatter plot, tables etc). It should be doable since the metric is stored as a list. However, currently UI only able to plot the line charts.

  
  
Posted 3 years ago
Votes Newest

Answers 6


thanks!

  
  
Posted 2 years ago

I only find the task.get_last_metrics() API, but I would need the entire metric array instead

  
  
Posted 3 years ago

I see, I will look into the documentation of it, thanks Jake.

  
  
Posted 3 years ago

To do that you'll need to use the APIClient which allows you to make direct API calls for which there is no programmatic interface in the Task class - will that be good for you?

  
  
Posted 3 years ago

EnviousStarfish54 , a simple example code:
` from clearml.backend_api.session.client import APIClient
client = APIClient()
task_id = "<task-id>"

event_type can be "training_stats_scalar" or "plot"

event_type = "training_stats_scalar"
result = client.events.get_task_events(task=task_id , batch_size=1000, event_type=event_type)
while result.events:
for event in events:
print(event.timestamp, event.iter, event.metric, event.variant, event.value)
result = client.events.get_task_events(task=task_id , batch_size=1000, event_type=event_type , scroll_id=result.scroll_id) `

  
  
Posted 2 years ago

Hi EnviousStarfish54 , you can use the APIClient to programmatically extract metrics for a specific task. Assuming you have two tasks, you can extract data for both, than compare - will that be a good solution?

  
  
Posted 3 years ago
563 Views
6 Answers
3 years ago
one year ago
Tags