Hmm are you running from inside the Kaggle jupyter thing ?
ReassuredTiger98 no, but I might be missing something.
How do you mean project-specific?
Ephemeral Dataset, I like that! Is this like splitting a dataset for example, then training/testing, when done deleting. Making sure the entire pipeline is reproducible, but without storing the data long term?
RoughTiger69
Apparently,
, doesn’t populate that dict with
any keys that don’t already exist in it
.
Are you saying new entries are not added to the Dict even if they are on the Task (i.e. only entries that already exist on the dict are populated ?
But you already have all the entries defined here:
https://github.com/allegroai/clearml/blob/721569bb77d89d89e5b4f32a0ed98311c4574650/examples/services/aws-autoscaler/aws_autoscaler.py#L22
Since all this is ha...
FlutteringWorm14 an RC is out (1.7.3dc1) with the ability to configure from clearml.conf
you can now setsdk.development.worker.report_event_flush_threshold
from clearml.conf
@<1541954607595393024:profile|BattyCrocodile47> not restarting the docker, restarting the Docker service (on Mac it's an app, I think there is an option on the Docker app to do that)
Ohh if this is the case, and this is a stream of constant inference Results, then yes, you should push it to some stream supported DB.
Simple SQL tables would work, but for actual scale I would push into a Kafka stream then pull it (serially) somewhere else and push into a DB
JitteryCoyote63 look for the latest RC it should have the fix (output_uri=False) 1.7.3rc1
... training script was set to upload every epoch. Seems like this resulted in a torrent of metrics being uploaded.
oh that makes sense, so basically you were bombarding the server with requests, and ending with kind of denial of service
This one should work:
` path = task.connect_configuration(path, name=name)
if task.running_locally():
my_params = read_from_path(path)
my_params = change_parmas(my_params) # change some staff
store back the change, my_params assumed to be the content of the param file (text)
task.set_configuration_object(name=name, config_taxt=my_params) `
The experiment finished completely this time again
With the RC version or the latest ?
Hi @<1523711619815706624:profile|StrangePelican34>
if I am trying to deploy 100 models on a GPU that can handle 5 concurrently,
Main limitation is Triton's ability to dynamically load / unload models. We know Nvidia is adding this capability, but I think this is still not out, once they support it, it should be transparent
try Hydra/trainer.params.batch_size
hydra separates nesting with "."
that might be it.
Is the web UI working properly ?
What ports are you using?
Hi BeefyHippopotamus73
. I checked the template task and the list of “Installed Packages” indeed does not have one of my required packages in the list.
Basically the "installed packages" is auto populated based on the directly imported packages n your code base.
Could it be you do not have import snowflake-connector-python
and this is a derivative package (i.e. required from a different package)
BTW: when you clone your Task in the UI you can edit and add the missing packages,...
Should work out of the box, maybe the only thing to notice is that you will get a Task for every local_rank 0 process
does that make sense ?
In theory it should not, in practice you could run out of space while running the experiment itself...
You can always cleanup everything from time to time (maybe worth a flag?)
JitteryCoyote63 no I think this is all controlled from the python side.
Let me check something
Hi @<1628565287957696512:profile|AloofBat92>
Yeah the name is confusing, we should probably change that. The idea is it is a low code / high code , train your own LLM and deploy it. Not really chatgpt 1:1 comparison, more like, GenAI for enterprises. make sense ?
VexedCat68 actually a few users already suggested we auto log the dataset ID used as an additional configuration section, wdyt?
Hi AgitatedTurtle16
You can find documentation here:
https://github.com/allegroai/clearml-session
Basically it uses the cleaml-agents to launch a session on one of the machines in the cluster.
In the remote session itself it install jupyterlab + vscode-server, then it connects to the remote session (running on the agent's machine) automatically over ssh and creates tunnel to these services.
Hi RipeGoose2
Yes the slide feature is definelty on the do do list (a lot of users asked for it).
Unfortunately other than actually PR-ing to the UI repo, there is no easy way to add customization (If you have an idea on how we could have an easy interface, that would be great.)
I'll check what's the status with the slider, maybe we will be lucky enough to see it in he next update 🙂
Hmm, in the credentials popup there should be a "secure connect" checkbox, it tells it to use https instead of http. Can you verify?
Hi @<1671689437261598720:profile|FranticWhale40>
Are you positive the Triton container finished syncing ?
Could you provide the docker log (both the serving and the triton)?
What is the clearml-serving version you are using ?
Could you add a print in the "preprocess" function, just to validate you are getting to the correct model version ?
although ideally i'd like to tell it exactly where to unzip it.
Ohh you can use .get_local_mutable_copy()
It will unzip it to specific folder
Hi ZippyAlligator65
You can configure it in the clearml.conf: see here:
https://github.com/allegroai/clearml-agent/blob/ebb955187dea384f574a52d059c02e16a49aeead/clearml_agent/backend_api/config/default/agent.conf#L202
Hi @<1658281093108862976:profile|EncouragingPenguin15>
Should work, I'm assuming multiple nodes are running agents ? or are you saying Ray spins the jobs and clearml logs them ?