by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Charles Riley
    @Charles70378427_twitter
    @n30sh4d0wderp_twitter IDS is an intrusion detection system and IPS is an Intrusion protection system. They both work basically to alert systems of potential threats occurring in real time.
    tsu12
    @tsu12
    Hi @ibmjedi, has the link of online ONNX workshop been sent out to the all registered attendees? I haven't got the link so far.
    Thomas Truong
    @ibmjedi
    @tsu12 link will be sent out at 5pm (PT) from LFAI event coordinator
    SreeV
    @sreev
    Hi
    It was a good virtual event without the so called now issues :)
    Almo Daved
    @CsharpIslife
    I had to skip out on the meetup :( was really dying to be there..
    Will there be some sort of recap of the discussed topics?
    Vinitra Swamy
    @vinitra
    @CsharpIslife I believe the videos of the workshop sessions will be posted in the next few days!
    Thomas Truong
    @ibmjedi
    The recordings and presentations are now available on the LF AI Day ONNX Community Virtual Meetup wiki page: https://wiki.lfai.foundation/display/DL/LF+AI+Day+-ONNX+Community+Virtual+Meetup+-+Silicon+Valley+-+April+9
    Jari
    @safijari
    I exported a model from pytorch using opset_version=11. I can check the model in python and it doesn't throw an error but if I try to load it in an onnxjs session I get TypeError: "Invalid shape: length 0 is not allowed"
    How can I best go about debugging this?
    Prasanth Pulavarthi
    @prasanthpul
    @n30sh4d0wderp_twitter for R questions, check with @terrytangyuan who wrote https://github.com/onnx/onnx-r
    @safijari i dont think onnx.js supports opset 11 (it's open source, so contributions are welcome). you can always use onnxruntime though (https://aka.ms/onnxruntime)
    @np84 checkout https://github.com/onnx/onnx/blob/master/docs/IR.md and other documents in that directory
    Almo Daved
    @CsharpIslife
    @vinitra @sreev Thanks!
    Jari
    @safijari
    @prasanthpul thank you for your response. I'd love to contribute but unsure where to start. I looked into the runtime but there doesn't seem to be an obvious way to run that in the browser. Do I need to use emscripten or something like that?
    Emma Ning
    @EmmaNingMS
    @safijari , ONNXJS can help you run the model entirely in the browser. ONNX.js uses WebGL and WebAssembly, and supports all the major platforms. welcome to contribute if any ops in your model hasn't be supported yet.
    Jari
    @safijari
    Hi @EmmaNingMS. I'd love to contribute but don't quite understand how. I tried using ONNX js and got the error message I posted above. I have to export using opset 10 or 11 because my model uses an upsampling layer with bilinear interpolation.
    Alex Leiva
    @aviel08
    Hi guys, I trained a model using Pix2pixHD and I exported it to ONNX (opset 7) but I can't seem to make it work since it requires 2 inputs. Any ideas? I created a notebook to show what I mean: https://github.com/aviel08/pix2pixHD/blob/master/Inference_with_ONNX_Runtime_%28Multiple_Inputs%29.ipynb
    Prasanth Pulavarthi
    @prasanthpul
    @aviel08 you seem to be passing in the names of the two inputs rather than a dictionary of name/value pairs. The second param to session.run is a python dictionary (not a list). So {inputname1: inputvalue1, inputname2: inputvalue2}
    Boris Kourtoukov
    @BorisKourt

    Hi folks! Was wondering if anyone has any thoughts about this issue: https://stackoverflow.com/questions/61231340/input-input-image-of-layer-63-not-found-in-any-of-the-outputs-of-the-preceed

    Basically I am getting:

    "Error reading protobuf spec. validator error: Description of image feature 'input_image' has missing or non-positive width 0.".

    When I try to work with set dynamic_axes during conversion to ONNX, and then to coreml. The code itself is in the question above.

    5 replies
    Emma Ning
    @EmmaNingMS
    @safijari , currently the opset onnxjs supports is only up to 7.
    Boris Kourtoukov
    @BorisKourt
    I've updated the SO question above based on some additional tests based on a suggestion there.
    Kiril Dokh
    @dsounded
    Hello, guys is here a good place to ask about https://github.com/onnx/keras-onnx ?

    onnx/keras-onnx#279

    I left my comment there, I can share my .h5 file if needed

    Jari
    @safijari
    Is there any way to get pytorch's interpolate working on onnx js?
    Alex Leiva
    @aviel08
    Hi @prasanthpul, you were right, I needed to specify a python dictionary instead of a list. I goes like : res = sess.run([], {"label": x, "inst": y}) Thank you very much!
    Max Allan
    @__maxallan_twitter
    My guess is that the answer will be no, but is it possible to convert an onnx model to libsvm, rather than just the other way around (libsvm -> onnx)
    Artyom
    @dodler

    Hi everyone,
    how can I store model weights as sparse tensor and convert it to dense after loading from disk?

    Also, can I remove graph initializer and store weights as separate file? Thanks

    I tried SparseTensorProto, but seems like it doesn't have a name attibute
    Ke Zhang
    @linkerzhang
    @dodler thank you very much for raising the issue - onnx/onnx#2762 was created for discussion. I think a "name" field should be added.
    Artyom
    @dodler
    Thanks a lot.
    Arjun
    @kingpin22_gitlab
    Hi folks! I was wondering if it would be possible to add the source distro for OnnxRuntime to PyPI?
    Ke Zhang
    @linkerzhang
    @kingpin22_gitlab you may ask this in ONNX runtime community. http://github.com/Microsoft/onnxruntime.
    Matthew Moloney
    @moloneymb
    Hi folks, I'm looking for feedback on https://github.com/moloneymb/FSharp.ML.Onnx It explores the use of code quotations to generate Onnx graphs. It is a work in progress but the idea is there and working.
    In theory once the training API is ready I could target it and users could build trainable Onnx models directly in code and get full graph execution speed.
    Matthew Moloney
    @moloneymb
    It's possible to interactively write eager execution code similar to how you'd use Numpy. When you want to convert it into a graph you can quote the very same code and convert it. There are limitations to the conversion, similar to PyTorch and Tensorflow, but there is enough functionality there to cover a wide variety of use cases.
    G. Ramalingam
    @gramalingam
    @dodler : A SparseTensorProto has a “TensorProto values” which has a name. Won’t it be enough to use the name in the “values” field? Do we need a separate name?
    Ke Zhang
    @linkerzhang
    @gramalingam Fair enough. a separate name is not needed here, though one line of comments may be added inplace to clarify this a little bit.
    kohei0418
    @kohei0418
    Hi guys! I'm trying to convert some tensorflow models into ONNX. Is there any operation in ONNX equivalent to tensorflow's SegmentSum (or SegmentMean)?
    Mike Smith
    @gomlfx
    hi
    Nikolay Renziglov
    @Mabanza_gitlab
    Hi there. Could you guys direct me on the following: I want to create and train a microbiological model for slipper animalcule. I have a set of images kept locally. What do I do first and then? Thanks.
    Svetlana Levitan
    @sveta-levitan
    Hi Nikolay,
    To train a model you first need to use a deep learning framework such as PyTorch or TensorFlow. Then you can export the model in ONNX and deploy into ONNX Runtime or other framework if you want. ONNX does not yet provide a full mechanism for training models.
    KaranSwatch
    @KaranSwatch
    Im trying to convert .pth file into onnx, can anyone help me? Im new to machine learning
    Rogier
    @sluijs
    Artyom
    @dodler
    @gramalingam Well I need sparse tensors to compress model with lots of zeros(actually I got about 10x reduction using pytorch sparse tensors). I think that name for sparse tensor is needed if one wants to store neural network weights in sparse format to save space (in mobile application for example). However I don't know the exact mechanics of storing neural networks in onnx
    Also, are there any tools to compress onnx model? At least when it stored in fs.
    Sindre Eik de Lange
    @SelComputas
    Hi, I have a .onnx model and would like to create some explainability functions for it, such as heatmaps, anchoring, etc., and was wondering if anybody here had any experience with this? One option is to "port" the model to PyTorch or Keras, but I don't know the name of the architecture, so it's hard to replicate it in a different framework - is there any way to port it without having this information?
    Svetlana Levitan
    @sveta-levitan
    @SelComputas Maybe you can use an ONNX to TensorFlow converter? Or write some code that would generate various perturbed inputs and use results of ONNX inference to build the output you need.