Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    vballu
    @vballu

    Hi everyone!
    After a long year with a waiting-to-be-used prescience token. I finally start to explore this tool.
    Yet, I have some issue to connect my warp10 backend to it. I always have the 'unable to fetch data from time series backen' error message. I assume that come from a misconfiguration when I add a source. But I can't figure out how to fix it because my warpscript work on a quantum instance. So I assume that the wanted backend url must be the address only, without the endpoints (/api/v0/exec). I also assumed that the data can be numerous gts even if i have tried with an only one GTS as output.

    So, I'm now clueless and I some helps would be really appreciate :)

    Maël LE GAL
    @mael-le-gal
    Hello @vballu
    • what is the ID of your prescience project please ? (you can see it once you're logged in prescience in your browser at the bottom left corner)
    • have you create a Warp10 Source or a TimeSerie Source ?
    vballu
    @vballu
    Hello @mael-le-gal
    We have tried both warp10 and TimeSerie sources.
    The project id is c5ea632f-5184-4659-8bde-ae11c303aca8
    Maël LE GAL
    @mael-le-gal
    In the log I see an error 500 received from Warp10 server saying : LIMIT cannot extend limit past 1000000
    The warp script query is probably returning too much points
    To answer your other questions :
    • yes the backend url should be filled without the http path (/api/v0/exec)
    • If you use the TimeSerie source you can only have a single GTS
    • If you use the Warp10 source you can have several
    vballu
    @vballu
    okay, that is strange because for the last tries I volontary a lot reduce the fetch span to avoid this problem.
    I will try to increase the hard limit. I let you know. Thank you for your answer
    conwei
    @conwei
    Hi everyone, I just discovered the wonderfull https://market-place.ai.ovh.net. However after registering an account, I got 504 Gateway Timeout Error on https://api-market-place.ai.ovh.net/image-car-classification endpoint using the Swagger exemple. I also tried Sentiment Analysis which works fine. Does any one encounter this issue ?
    baaastijn
    @baaastijn
    hello @conwei ! thanks for your feedbacks, i’ll escalate for this bug
    we released a « prod-ready » version of it, insite OVH Public Cloud Control panel
    It’s called Machine Learning Serving
    you can use pre-trained model (such as sentiment analysis) but also uplaod your own
    don’t hesitate to try it ! it scales much better, it’s prod-ready
    wingtkw-school
    @wingtkw-school
    Hi everyone! I just discovered this useful website. Yet, when I tried the swagger example of Spleeter, I got 504 Gateway Timeout. Does anyone encounter this problem?
    baaastijn
    @baaastijn
    hello @wingtkw-school ! yes i do
    i fw the issue to the team
    also, are you looking for specific AI models ?
    wingtkw-school
    @wingtkw-school
    Yes I am trying to make a music rhythm game in unity that allows user to import their own music and selects what kind of instruments to be used in generating the beat map for my final year project
    what i plan to do is to adopt an audio separation AI model to separation the audio file into soundtracks of different instruments first, then applying fourier transform to find the beats
    I have done some research online and seems like the Spleeter API will help in my project. If there's other AI model that can help, I would love to try it out as well.
    I have already subscribed to the API and authorize it with my apiKeys before launching the default swagger example online, but I got 504 Gateway Timeout
    I'm still a newbie to API so may be I haven't set up the thing correctly
    baaastijn
    @baaastijn
    thx for your explanations :) no that’s a issue on our side sorry for that
    we now have a solution called ML Serving where you can deploy your own models and get and API endpoint, and also find pre-trained ones, but Spleeter is not implemented yet
    wingtkw-school
    @wingtkw-school
    Oh ok
    Thank you for your prompt reply :) Looking forward to this ~
    Nicocg70
    @NicoCG70_twitter
    Hello. Some troubles with DeOldify ?
    Nicocg70
    @NicoCG70_twitter
    I confirm, direct access with curl to DeOldify doesn't work. Someone to confirm or fix it ?
    Nicocg70
    @NicoCG70_twitter
    Error 504 / Error: Gateway Timeout
    Nicocg70
    @NicoCG70_twitter
    @baaastijn maybe ?
    Nicocg70
    @NicoCG70_twitter
    I confirm the problem today
    baaastijn
    @baaastijn
    Hello @NicoCG70_twitter sorry for that
    to be fully transparent our team is focused to release a new service (AI training / allowing your t run training jobs on multiple GPU easily) for the end of the month
    i will ask them to take a look but i don’t have an ETA to fix it
    Nicocg70
    @NicoCG70_twitter
    so, bad news :\ I stay tuned... Last time, I remember it was easy to fix. thx
    Nicocg70
    @NicoCG70_twitter
    Some returns by your team ? @baaastijn
    baaastijn
    @baaastijn
    sorry no, it will not be for today sadly
    Nicocg70
    @NicoCG70_twitter
    ouch. Does it mean it could be fixed after more than the next week ?
    baaastijn
    @baaastijn
    sorry i don’t have an exact ETA, we will do our besT. Our goal is to move these models to our new platform, ML Serving https://docs.ovh.com/gb/en/ml-serving/
    it was not a high priority but we will see what can be done quickly
    if you have very few pictures to colorize, you can use Deoldify here : https://www.myheritage.fr/incolor @NicoCG70_twitter
    Nicocg70
    @NicoCG70_twitter
    ok thx, I need the API for an app
    Nicocg70
    @NicoCG70_twitter
    @baaastijn any hope for this week ? or definitely not ?
    baaastijn
    @baaastijn
    Hello @NicoCG70_twitter, sadly not for this week. We started the migration of the AI models to ML Serving and we expect +- 3 weeks of work
    after that, we will have a production-ready platform fall many AI models
    this current marketplace was free but not production-ready, we want improve the experience
    Nicocg70
    @NicoCG70_twitter
    If I understand the service is down (ok it's not for production, no worry) and the new platform will sell a production environnement with this kind of services ? but now, between the two, there is no solution ? m'i wright ?
    baaastijn
    @baaastijn
    almost right : the new platform is already here, with abilities to deply pre-trained models OR your own models
    but so far Deoldify is missing
    https://deepai.org/machine-learning-model/colorizer
    alternative via API if you are stuck
    GitHub project for DeOldify : https://github.com/jantic/DeOldify
    Nicocg70
    @NicoCG70_twitter
    You could say that before even if it was an experimental platform :\
    baaastijn
    @baaastijn

    Hello there !

    We are really glad to announce the launch of OVHcloud AI Training as a paid beta, fully integrated in our public cloud openstack ecosystem.

    You may know that data-related projects are booming, and we want to provide you all the required tools to build you own data and AI dreamed platform.

    Since 2 years we added new services (Apache Hadoop big data clusters, Data processing powered by Apache Spark, ML Serving to deploy models in production, …).

    Now, with AI Training, we simplify the life of data scientists and data engineers for neural network trainings over GPUs.

    You just have to push your code in a docker and your data on Object Storage and the platform takes care of the rest (data sync, workload deployment, user management, ...).

    If you don't know docker, it doesn't matter, you can use our prebuilt images with your favorite framework (HuggingFace, Pytorch, Tensorflow, MXNet, Fast.ai, ...) with JupyterLab notebooks and VScode.

    All of that with as always, a simple and nice pricing : 1.75€ /hour /GPU NVIDIA V100s (more flavors to come…)

    What does it solve ? lot of things ! easiness to scale, orchestration, no more infrastructure to manage, cost control, … focus on code 😊

    Don’t hesitate to try it out, to share feedback or questions ! it’s on your Public Cloud control panel (AI training in left menu) or via CLI

    Product page : https://www.ovhcloud.com/en/public-cloud/ai-training/
    Documentation : https://docs.ovh.com/gb/en/ai-training/
    Roadmap : https://github.com/ovh/public-cloud-roadmap/projects/2

    Have a good day !