CostlyOstrich36

0 Questions, 4172 Answers

Active since 10 January 2023

Last activity 2 years ago

Reputation

Answers 4172

0 I'Ve Been Using Add_Function_Step In The Pipeline Controller As Of Yet. Now I Wish To Use Add_Step For One Of The Steps And Pass In A Task Id. Only Problem Is, For Some Of The Argparse Parameters, I Want To Modify The Values In The New Task. Can You Guys

You mean you want the new task created by add_step to take in certain parameters? Provided where/by who?

3 years ago

0 Hi There, Our. Self-Hosted Server Is Periodically Very Slow To React In The Web Ui. We'Ve Been Debugging For Quite Some Time, And It Would Seem That Elastisearch Might Be The Culprit. Looking At The Elastisearch Index, We Have An Index Of Around 80G Of Tr

Hi @<1523701601770934272:profile|GiganticMole91> , As long as experiments are deleted then their associated scalars are deleted as well.

I'd check the ES container for logs. Additionally, you can always beef up the machine with more RAM to give elastic more to work with.

10 months ago

0 In Pipelines. Is It Possible To Inject A Requierment.Txt Such That The Executing Node Will Install Before Running Tasks?

and just making sure - by pipeline we're talking about the ClearML pipelines, correct?
https://clear.ml/docs/latest/docs/references/sdk/automation_controller_pipelinecontroller

3 years ago

0 Hi, I Logged My Params To Clearml Like This:

Do you need to pull it later somewhere? Is there a specific use case?
Because I think you can get the params dict back via code as a the same dict

3 years ago

0 Is There A Way To Upload An Artifact I Forgot To Upload During The Task Duration To That Task After It Is Already Complete?

None

2 years ago

0 Hi. I'D Like To Try The Gcp Autoscaler.

Fair point

3 years ago

0 Hi All! I Am Trying To Run The Notebook From The Remote Server Using Aws. It Seems Like I Successfully Installed Clearml Using !Pip Install Clearml Command, But When I Am Trying To Run !Clearml-Init I Got The Following Error: /Bin/Bash: Clearml-Init: Comm

You must perform Task.init() to have something reported 🙂

3 years ago

0 Hi, I Am Trying To Upload An Optimization Task And I Have The Following Message At The Bottom Of The App. Also When I Try To Upload The Task, It Fails With The Following Error ((Private - Removed) - Is Just Instead Of The Real Id):

AbruptWorm50 , the guys tell me that it's under progress and we will be updated in the following minutes 🙂

3 years ago

0 Hello ! I Am Having Some Issues While Running A Pipeline Remotely (Same Device But Queue + Agent) . Hitting The Following Issue On My First Step:

The idea is that the server is not something that should be moved on a whim. Once you do setups you can test but once everything is setup it is expected that the server will remain in the same address.

Also, this is when you're using the files server. If you were using minio from the start this would be a non issue.

one year ago

0 Hi Everyone, I Have A Question On How To Disable Running Pipelines In Hybrid Mode, So That Agent Only Considers The Code From Git And Not From Local Repo While Running A Task Remotely. I Am Trying To Run A Demo Pipeline Remotely Using Clearml-Agent. Some

This will disable storing the uncommitted changes

11 months ago

0 Hi, My Server Is On Version 1.11, And Yesterday A Few Of My Runs (With Agent 1.5.1) Failed On

I meant that maybe you ran it with a newer version of the SDK

one year ago

0 Hi. For The Experiment Scalar Tab, There'S A Gpu Resource Graph. The Gpu Mem Used Is In Percentage, Is It Possible To Display As Absolute Gb Instead? Reason Is Because The User Doesn'T Really Know How Much Vram Is Allocated.

Hi @<1523701504827985920:profile|SubstantialElk6> , I don't think there is such an option currently. Maybe open a GitHub feature request?

2 years ago

0 Hi, I Have A Plt Plot That Is Being Automatically Uploaded To A Task And Is Presented In The Plots Section Of The Task As Expected. However, The Plot Itself Isn'T Presented Fully, And I Can Only See The Titles And Some Of The Legend, But Not The Lineplot

DistressedKoala73 , can you send me a code snippet to try and reproduce the issue please?

3 years ago

0 Hi Guys! Is There Any Way To Get Full State Of Dataset From Somewhere, Except My S3 Bucket? I Need A Mapping Of Files And Batches That Were Uploaded As A Dataset. Maybe This Information Is Also Available In One Of The Clearml Databases?

Hi @<1584716355783888896:profile|CornyHedgehog13> , you can only see a list of files inside a dataset/version. I'm afraid you can't really pull individual files since everything is compressed and chunked. You can download individual chunks.

Regarding the second point - there is nothing out of the box but you can get a list of files in all datasets and then compare if some file exists in others.

Does that make sense?

2 years ago

0 It Seems That Through Sdk Users Can Influence What Image The Agent Should Use For The Pod. Is It Possible For The User To Say How Many Resources He Requires (Vcpu, Ram) And Whether Gpus Are Needed? Or We Need To Have Multiple Agents With Different Resourc

Hi @<1857232027015712768:profile|PompousCrow47> , the resource requirements are basically abstracted via the queue. You connect agents according to the resources/queues you want to expose to users.

one month ago

0 Hi, We Ran Into A Bit Of A Problem Storing Artifacts In S3. In Python 3.10 When The Files Are Large Enough For Multipart Uploading To Kick In, Something Goes Wrong In Boto3 With Threading Causing The Upload To Fail. (Exact Error Being

Is this error originating from boto3?

3 years ago

0 Sorry For The Barrage Of Questions. I Can'T Seem To Figure Out How Best To Get A Python Script I Need To Run On An Agent. I Have An Agent Listening To The Default Queue And I Also Have The Script That I Need To Run On The Agent. Can You Guide Me On How To

3 years ago

0 Hello Everyone! I Hope You Can Help Me. I Have A Local Workstation Where I Want To Deploy The Clearml Server And Clearml Agent. I Have Prepared Two Independent Docker-Compose Files, While I Need To Explicitly Specify The External Docker Network? Or With C

@<1707203455203938304:profile|FoolishRobin23> , the agent in the docker compose is a services agent and it's not for running GPU jobs. I'd suggest running the clearml-agent with the GPU manually.

one year ago

0 Hi, I'M Trying To Upload Data To Clearml Parallelly. Is It Impossible To Use

MagnificentWorm7 , I'm taking a look if it's possible 🙂
As a workaround - I think you could split the dataset into different versions and then use Dataset.squash to merge into a single dataset
https://clear.ml/docs/latest/docs/references/sdk/dataset#datasetsquash

3 years ago

0 Hello, I Would Like To Understand Better Metrics Storage Since I Reach 1Go Very Quickly With Just 4-5 Trainings. I Am Using

Hi @<1632551554206666752:profile|DelightfulBear99> , maybe you're logging a lot of metrics or very large log files? Is it possible you have large configurations or previews to artifacts?

Is there an example script you can provide that creates a lot of metrics storage for you?

one year ago

0 Hi, We Could Only Access The Self-Hosted Clearml-Server Via

Hmmmm I think you would need to change some configurations in the docker-compose to use https

3 years ago

0 Hello. I Have An Issue Where Clearml Doesn'T Read From Uncommitted Changes. I Changed The Batch Size Locally And Started Training. When I Check It In

Hi @<1523704674534821888:profile|SourLion48> , I'd suggest connecting your batch size as a configuration parameter of the experiment, for example using argparser, and then regardless of the committed or uncommitted code, you will be able to control this value through the configuration section.

What do you think?

9 months ago

0 Hey Can Anyone Explain This, My Max Number Of Instances Is 5 But Its Showing Something Like That: Its Showing 8 Instance Which Does Not Make Sense,

How did the tasks fail?

2 years ago

0 Hi, Question About Setting Output_Uri=True For A Pipeline Step Using Decorators. What I’M Seeing Is That Training Results In No Output Destination Specified, And Logs To Local Disk (When Run Remotely In Docker Container), Where Is The Proper Place To Set

Hi ExuberantParrot61 , that's a good question. This is a bit hacky but what if you try to catch the task with Task.current_task() from inside the step and try to change the output_uri attribute there?

3 years ago

0 Hi, Are There Any Plans Or Already Ways To Deploy A Pipeline With Clearml-Serving To Triton? I Would Also Be Interested In The Support Of Deploying Pure Python Models Using The New Python_Backend Of Triton.

ClumsyElephant70 , I'm not sure. There usually a roadmap provided on our community talks so it'd be great if you joined next time to see what's next 🙂

3 years ago

0 Hello! I'Ve Been Trying To Use Clearml For The First Time, But I Cannot Seem To Run The First Serving Model. First, I Run The Following In Powershell: >>> Clearml-Serving Create --Name "June Test" Log: Clearml-Serving - Cli For Launching Clearml Serving

Its a bit confusing, please add the full list of action you took + whatever the console printed.

Preferably using the following format for logs

This is a very convenient way to post logs

2 months ago

0 Hi, I Am Trying To Request A Demo For Cleargpt But After I Enter By Connection Details I Get A Garbled Webpage And It Looks Like My Request Have Not Gone Anywhere (At Least From The Fact That I Don'T Get A Confirmation E-Mail). Does Anyone Know How Can I

Hi @<1577468638728818688:profile|DelightfulArcticwolf22> , what email did you use? Can you try again now?

2 years ago

0 Hi All! I Want To Setup A Tool That I Can Use To Analyse Clearml Logs (Just User And Experiment Logs, I Am Not Doing Any Model Serving Yet). Is There An Api Endpoint That I Can Fetch Such Metrics? Metrics Like Number Of Active Users, Number Of Experiments

Hi @<1573119955400921088:profile|CloudyPelican46> , you can certainly do this. You can find all the related api calls here - None
I suggest opening developer tools (F12) and seeing what is sent in the UI to fetch the various metrics you're looking for

2 years ago

0 Hello, I Had A Setup For Training Ml Model Of Multiple Sub-Modules, Used Pipelines From Task. Didn'T Run It For 4 Months, Was Working Fine Back Then, Now When I Run Pipeline, Steps Are Being Run And Completed Emty, No Console Log, No Stats, Nothing, What

You mentioned you are self deployed. When you deploy the server, one of containers deployed is the ES container. Did you not deploy the server via docker compose?

one year ago

0 I Think This Is A Bug In Clearml Web Gui The Issue Is The

SuperficialDolphin93 , looks like a strange issue. Can you maybe open a github issue for better tracking?

3 years ago

Show more results