I'm assuming some package imports absl (the TF define package) and that's the reason you see the TF defines). Does that make sense?

I think there is a bug on the UI that causes series with "." to only use the first part of the series name for the color selection. This means "epsilon 0" and "epsilon 0.1" will always get the same color, and this will explain why it works on other graphs

Basically internally we use psutil to get those stats ...
https://github.com/giampaolo/psutil/issues/1011
See psutil version that fixed that, what do you see on the Task "installed packages" ?
https://github.com/giampaolo/psutil/blob/master/HISTORY.rst#591

Hi AttractiveShrimp45
Well, I would use the Task.connect to add a section with any configuration your are using. for example
Task.current_task().connect(my_dict_with_conf_for_data, name="dataset51")wdyt?

Hi @<1523701868901961728:profile|ReassuredTiger98>
This should have worked, and seems like conda is not fetching the correct pytorch version (even though the conda env contains the cuda version they specify)
Let's try something, reset the Task, then edit the "Installed packages" and add:

cudatoolkit==11.1.1

Then try again.
Let's see what we get.
(The idea, is that I think conda forgets it just install cudatoolkit and assumes the env is for CPU)

Basically it is the same as "report_scatter2d"

Hi ObnoxiousStork61

Is it possible to report ie. validation scalars but shifted by 1/2 iteration?

No 😞 these are integers

What's the reason for the shift?

I'm also curious 🙂

could it be the polling on the Task (can't remember whats the interval), but it will update it's state once every X minutes/seconds

ShallowGoldfish8 this call does that:
https://github.com/allegroai/clearml/blob/0397f2b41e41325db2a191070e01b218251bc8b2/examples/advanced/execute_remotely_example.py#L127

Hi @<1697056701116583936:profile|JealousArcticwolf24>
You have clearml Datasets None
It will version catalog and store meta-data of your datasets.
Each version only stores the delta from the parent version, but delta is on a file granularity not a "block" granularity
Notice that under the hood of course it uses storage solutions to store and cache the underlying immutable copy of the data. What's your use case?

Merged 🙂

Scheduled training is what I’m looking forward to

I'll try to remember to update here once we pushed into the GitHub repo, feedback is always appropriated 🙂
If in the next two weeks you hear nothing, please ping here to make sure I did not forget 😉

and the clearml server version ?

This will fix it, the issue is the "no default value" that breaks the casting
@PipelineDecorator.component(cache=False) def step_one(my_arg=""):

seems it was fixed 🙂
MagnificentWorm7 thank you for noticing ! 🙏

So yes, it creates the Task on your machine (with the name and project etc.)
Then it stops the local process, and pushes it into the exxecution queue, when the agent pulls it, and re-executes the code, it will ignore the Task.init

I looked at your task log on the github issue. It seems the main issue is that your notebook is Not stored as python code. Are you running it on jupyter notebook or is it ipython that you are runnig it on? Is this reproducible? If so what's the jupyter version, python and OS versions?

Hi ElegantCoyote26
If there is, it will have to be using the docker-mode, but I do not think this is actually possible because this is not a feature of docker. It is possible to do on k8s, but that's a diff level of integration 🙂
EDIT:
FYI we do support k8s integration

I solved the issue by implementing my own ClearML logger

This is awesome! any chance you want to PR it to transformers ?

I did change the

instead of 8080?

So this is the issue

ReassuredTiger98
(for some reason it kind of jumps over PyTorch, but then installs torchvision?!)
Could you run with the latest with --debug
We just added but you will have to install from git:
pip3 install git+Then run with --debug:
clearml-agent --debug daemon ...

You can make reports on experiments with interactive graphs

Yes, I can totally see how this is a selling point. The closest is the Project Overview (full markdown capabilities, with the ability to embed links to specific experiments). You can also add a "leader metric", so you can track the project performance/progress over time.
I have to admit that creating a better reporting tool is always pushed down in priority as I think this is a good selling point to management but the actual ...

Hi ProudMosquito87
My apologies there is still no concrete ETA ...
That said I think a good toy example would really help accelerate this process.
How about opening a PR with a nice hydra example, then we can start discussing implementation details based on the toy example ?

My pleasure

I see, by default it will look for requirements.txt in the root of the repo (the actual repo).
That said in code you can specify the requirements .txt:
Task.force_requirements_env_freeze(requirements_file='repo/project-a/requirements.txt') task = Task.init(...)Notice, you need to call it prior to the Task.init call

Click on the "k8s_schedule" queue, then on the right hand side, you should see your Task, click on it, it will open the Task page. There click on the "Info" Tab, there look for "STATUS MESSAGE" and "STATUS REASON". What do you have there?

WackyRabbit7 I guess we are discussing this one on a diff thread 🙂 but yes, should totally work, that's the idea

Hi

The Squash operation copies all the data and is no longer linked to previous commits?

Yes, basically the idea is if you have data version that relies on many parents that needs to be merged, the squash will create a merged copy and push it all as a single version, and then yes the parent versions are no longer needed

I thought this operation is like git squash but it seems to me

yeah... we did not want to actually delete the parents because unlike git, the operation is done ...

Just tried: also works with 0.17.2

Great!

You can always specify diff clearml.conf files with --config-file 🙂