AbruptHedgehog21 , can you go into the network tab of the developer tools and see which request is returning 500?

From the screenshots provided you ticked 'cpu' mode AND I think the machine that you're using n1-standard-1 is a cpu only machine, if I'm not mistaken.

Hi UpsetBlackbird87 ,
If you're in the pipelines UI, you can switch to the detailed view and you can see each step of the pipeline as a node 🙂
You can see an example here:
https://clear.ml/docs/latest/docs/pipelines/pipelines

If you're running in docker mode you can add those steps very easily to the bash startup script

Hi GrittyHawk31 ,

To avoid reusing tasks you can set the following configuration in ~/clearml.conf
sdk.development.task_reuse_time_window_in_hours: 0

In any case - models shouldn't be replaced but just new ones will be created in the system

At least from the log of the agent failure

Does the other PC have the package locally there somewhere?

RattyLouse61 , SuccessfulKoala55 , I think your solution is better 🙂

Hi @<1572032783335821312:profile|DelightfulBee62> , I think 1 TB should be enough. I would suggest maybe even having 2T just for the safe side

Is it failing when running on a worker?

Hi ShallowGoldfish8 ,

I'm not sure I understand the scenario. Can you please elaborate? In the end the model object is there so you can easily fetch the raw data and track it.

CourageousKoala93 , yes 🙂

You can use https://clear.ml/docs/latest/docs/references/sdk/automation_controller_pipelinecontroller#start_locally
to run the pipeline locally

FreshKangaroo33 , I'm sorry for the delay. It looks like it will require a feature request. Maybe open a github issue to track it? 🙂

GorgeousMole24 , another note, in the comparison screen itself, you can add other experiments to the comparison view.

I think this is the API you're looking for 🙂
https://clear.ml/docs/latest/docs/references/api/events#post-eventsdebug_images
This should return what you're looking for. You can see a usage example in the developer tools when viewing debug samples in the UI

Hi HugeArcticwolf77 , did you spin new agent versions? What did you have before and what do you have now? Can you check if you revert to the previous version it works out?

Well after you finish and you select your best - move it somewhere else and then cull the rest

Hi,
From the looks of it, it always returns a string. What is your use case for this? Do you have some conditionality on the type of variable the parameters are?

Hi @<1561885921379356672:profile|GorgeousPuppy74> , you can get the key/secret pair from the ClearML UI.
I think this is the documentation you were looking for: None
You can access the settings page via the profile icon at the top right.

StaleButterfly40 , it looks like there might be a good solution for your request. In the previous link I provided, there is a parameter ' continue_last_task' that should work for you 🙂

Hi @<1546303277010784256:profile|LivelyBadger26> , how did you set the random seed? I think you can also disable ClearML's random seed override and set one with Pytorch

Hi 🙂

A task is the most basic object in the system in regards to experiments. A pipeline is a bunch of tasks that are controller by another task 🙂

TartLeopard58 , I think you need to mount apiserver.conf to the api server. This is an API configuration 🙂

Happy to help 🙂

Hi @<1638712141588467712:profile|ExuberantTurtle48> , does this happen to all of your tasks? Are you certain you fetched the right task?

WackyRabbit7 , how did you report the table? Can you please provide an example for the data structure of the table?

Hi @<1625303806923247616:profile|ItchyCow80> , can you please describe how you're running it? Is it inside a jupyter nokebook? Do you have a code sample?

How did you call the steps in the pipeline?

Do you see something in the documentation to indicate that?