@<1523701205467926528:profile|AgitatedDove14> any ideas?

do I need to set the extra_index_url in clearml.conf as well?

ok i think i know why it happens!

my server isn't the latest, though..

AgitatedDove14 I noticed a lot of my tasks don't contain these graphs though...

hi SuccessfulKoala55 with the clearml server update, does it use a newer ES docker?

Hey SuccessfulKoala55 thanks for the answer.
any ideas how I can try to fix this?

I think it's still caching environments... I keep deleting the caches (pip, vcs, venvs-*) and running an experiment. it re-creates all these folders and even prints

Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.8/dist-packages (from requests>=2.20.0->clearml==1.6.4->prediction-service-utilities==0.1.0) (3.4)
Requirement already satisfied: charset-normalizer<4,>=2 in /root/.clearml/venvs-builds/3.8/lib/python3.8/site-packages (from requests>=2.20.0->clearml==1.6....

clearml-agent 1.2.3

is that what you mean?

Right, I used --services-mode 2 and it still runs more than 2 tasks simultaneously

Is this a possible future feature? I have used cometML before and they have this. I'm not sure how they do it though...

right, and why can't a particular version be found? how it does it try to find python versions?

this is the code that's giving me trouble 🤔

Hi CostlyOstrich36
I added this instruction at the very end of my postprocess function
shutil.rmtree("~/.clearml")

Because sometimes it clones a cached version of a private repository, instead of cloning the requested version

SuccessfulKoala55 Yup!

a model

i'm also not sure what this is
-H "Content-Type: application/octet-stream" -H' NV-InferRequest:batch_size: 1 input { name: "dense_input" dims: [-1, 784] } output { name: "activation_2" cls { count: 1 } }'

well.. it initially worked but now i get the same thing 😕 SuccessfulKoala55

I see, ok!
I will try that out.
Another thing I noticed: none of my pipeline tasks are reporting these graphs, regardless of runtime. I guess this line would also fix that?

Ah, so you're saying I can write a callback for stuff like train_loss , val_loss , etc.

i don't think the conf is an issue. it's been deployed for a long time and working. models from yesterday correctly display the url

well, i have run the keras mnist example that is in the clearml-serving READme. Now I'm just trying to send a request to make a prediction via curl

And then you'll hook it

I have this inside my pipeline defined with decorator

Sent it to you via DM!

i'm probably sending the request all wrong + i'm not sure how the model expects the input

it says method not allowed when i try to access that url