# Machine Learning

**URL:** https://discourse.julialang.org/c/domain/ml/24.md?page=6

[Latest](https://discourse.julialang.org/latest.md) · [Categories](https://discourse.julialang.org/categories.md) · [Tags](https://discourse.julialang.org/tags.md)

**Page:** 7

---

## [Delusional drone in RxInfer example](https://discourse.julialang.org/t/delusional-drone-in-rxinfer-example/117057)

<div class="topic-metadata">

**Author:** [@Andreas\_Poulsen](https://discourse.julialang.org/u/Andreas_Poulsen)\
**Replies:** 6\
**Last updated:** [September 17, 2024, 3:38pm UTC](https://discourse.julialang.org/t/delusional-drone-in-rxinfer-example/117057 "2024-09-17T15:38:26Z")

</div>

Hi I am recreating the drone example from https://github.com/ReactiveBayes/RxInferExamples.jl/blob/main/drone/drone\_simulation.ipynb. It occured to me, that the plot/animation is based on the models posteriors. Does thi…

---

## [Recover weights of logistic regression using MLJ.jl](https://discourse.julialang.org/t/recover-weights-of-logistic-regression-using-mlj-jl/119331)

<div class="topic-metadata">

**Author:** [@PeX](https://discourse.julialang.org/u/PeX)\
**Replies:** 0\
**Last updated:** [September 12, 2024, 2:55pm UTC](https://discourse.julialang.org/t/recover-weights-of-logistic-regression-using-mlj-jl/119331 "2024-09-12T14:55:07Z")

</div>

Hi all, I’m trying to see if I can recover the weights of the logistic regression from data I’m generating synthetically, but for some reason there is large error in the optimized weights when using MLJ.jl. To create t…

---

## [Whats going on in kissABC.jl package smc() function](https://discourse.julialang.org/t/whats-going-on-in-kissabc-jl-package-smc-function/119113)

<div class="topic-metadata">

**Author:** [@Sahil\_Khan](https://discourse.julialang.org/u/Sahil_Khan)\
**Replies:** 5\
**Last updated:** [September 6, 2024, 11:57am UTC](https://discourse.julialang.org/t/whats-going-on-in-kissabc-jl-package-smc-function/119113 "2024-09-06T11:57:20Z")

</div>

Hello everyone… Im working with KissABC.jl package smc() function. Based on my reading through source code for same, it doesn’t look like classic smc() algorithm but kind of hybrid algorithm. As there is not much refer…

---

## [Error while reproducing example of continuous normalizing flows documentation](https://discourse.julialang.org/t/error-while-reproducing-example-of-continuous-normalizing-flows-documentation/118957)

<div class="topic-metadata">

**Author:** [@l8l](https://discourse.julialang.org/u/l8l)\
**Replies:** 4\
**Last updated:** [September 5, 2024, 10:32am UTC](https://discourse.julialang.org/t/error-while-reproducing-example-of-continuous-normalizing-flows-documentation/118957 "2024-09-05T10:32:39Z")

</div>

I am executing an exact copy of the code in the documentation about continuous normalizing flow (CNF): Continuous Normalizing Flows · DiffEqFlux.jl However, I get an error in the line res1 = Optimization.solve(optprob, …

---

## [BatchDuplicated functions in Enzyme](https://discourse.julialang.org/t/batchduplicated-functions-in-enzyme/119001)

<div class="topic-metadata">

**Author:** [@gdalle](https://discourse.julialang.org/u/gdalle)\
**Replies:** 3\
**Last updated:** [September 3, 2024, 7:21pm UTC](https://discourse.julialang.org/t/batchduplicated-functions-in-enzyme/119001 "2024-09-03T19:21:34Z")

</div>

I’m using Enzyme to differentiate through a closure, and BatchDuplicated on the closure itself doesn’t seem to work, unlike Duplicated. using Enzyme struct MyClosure{A} a::A end function (mc::MyClosure)(x) # c…

---

## [TensorBoard for everybody](https://discourse.julialang.org/t/tensorboard-for-everybody/10978)

<div class="topic-metadata">

**Author:** [@oxinabox](https://discourse.julialang.org/u/oxinabox)\
**Replies:** 2\
**Last updated:** [September 3, 2024, 2:53pm UTC](https://discourse.julialang.org/t/tensorboard-for-everybody/10978 "2024-09-03T14:53:10Z")

</div>

I bundled up a quick proof of concept for having tensorboard outside tensorflow. It utterly cheats because it actually is invoking TensorFlow.jl to do this. It would be possible at some point to build something similar…

---

## [Current State of Hyperparameter Tuning?](https://discourse.julialang.org/t/current-state-of-hyperparameter-tuning/118701)

<div class="topic-metadata">

**Author:** [@jacobadenbaum](https://discourse.julialang.org/u/jacobadenbaum)\
**Replies:** 1\
**Last updated:** [August 30, 2024, 4:19am UTC](https://discourse.julialang.org/t/current-state-of-hyperparameter-tuning/118701 "2024-08-30T04:19:55Z")

</div>

For people who work on machine learning, I’m trying to figure out what is the state of the art for packages/methods to do hyperparameter tuning. I see there is a package Hyperopt.jl, which has a lot of contributions but…

---

## [Add coordinate transformation before first layer in NeuralPDE.jl](https://discourse.julialang.org/t/add-coordinate-transformation-before-first-layer-in-neuralpde-jl/118769)

<div class="topic-metadata">

**Author:** [@nico](https://discourse.julialang.org/u/nico)\
**Replies:** 5\
**Last updated:** [August 29, 2024, 8:16pm UTC](https://discourse.julialang.org/t/add-coordinate-transformation-before-first-layer-in-neuralpde-jl/118769 "2024-08-29T20:16:54Z")

</div>

Hi, I woud like to add a transformation to the coordinates x and y before they are passed to the first layer of a neural net defined with Lux using NerualPDE.jl. I am implementing physics informed neural networks. The …

---

## [Which function in julia.Flux is similar to "detach()" in pytorch?](https://discourse.julialang.org/t/which-function-in-julia-flux-is-similar-to-detach-in-pytorch/118757)

<div class="topic-metadata">

**Author:** [@ZC\_H](https://discourse.julialang.org/u/ZC_H)\
**Replies:** 2\
**Last updated:** [August 29, 2024, 8:59am UTC](https://discourse.julialang.org/t/which-function-in-julia-flux-is-similar-to-detach-in-pytorch/118757 "2024-08-29T08:59:36Z")

</div>

When I train a network using pytorch, sometimes I do not what a tensor participate in gradient backward. But however, I cannot find a function that is similar to “detach()” in julia.Flux. So when I use julia to train a n…

---

## [Flux: if shuffle=false in DataLoader, is Descent "not" SGD?](https://discourse.julialang.org/t/flux-if-shuffle-false-in-dataloader-is-descent-not-sgd/118659)

<div class="topic-metadata">

**Author:** [@tom-plaa](https://discourse.julialang.org/u/tom-plaa)\
**Replies:** 0\
**Last updated:** [August 27, 2024, 12:33pm UTC](https://discourse.julialang.org/t/flux-if-shuffle-false-in-dataloader-is-descent-not-sgd/118659 "2024-08-27T12:33:46Z")

</div>

Sorry for the stupid question, but it’s like in the title: When using DataLoader, if we don’t set shuffle=true, will the gradient descent algorithm from Descent always go through the batches/samples in the same order fo…

---

## [Learning curve for validation dataset](https://discourse.julialang.org/t/learning-curve-for-validation-dataset/118529)

<div class="topic-metadata">

**Author:** [@niltsz](https://discourse.julialang.org/u/niltsz)\
**Replies:** 1\
**Last updated:** [August 26, 2024, 8:17am UTC](https://discourse.julialang.org/t/learning-curve-for-validation-dataset/118529 "2024-08-26T08:17:31Z")

</div>

I want to train a ML model with MLJ, for example an XGBoost Regressor with L2 Loss. I split my dataset into three parts, train/val/test, and during training I want to train on the train dataset but also plot the loss/ac…

---

## [How to use Flux to write a Multi-Head Output Network](https://discourse.julialang.org/t/how-to-use-flux-to-write-a-multi-head-output-network/118605)

<div class="topic-metadata">

**Author:** [@ZC\_H](https://discourse.julialang.org/u/ZC_H)\
**Replies:** 4\
**Last updated:** [August 26, 2024, 3:30am UTC](https://discourse.julialang.org/t/how-to-use-flux-to-write-a-multi-head-output-network/118605 "2024-08-26T03:30:34Z")

</div>

I want to use julia.Flux to recurrent SAC algorithm. But I am facing a problem when I write the policy network modular because it is a multi-head output network, which receive a state matrix as input and output both expe…

---

## [NeuralODE training failed on GPU with Enzyme](https://discourse.julialang.org/t/neuralode-training-failed-on-gpu-with-enzyme/118537)

<div class="topic-metadata">

**Author:** [@Yunan](https://discourse.julialang.org/u/Yunan)\
**Replies:** 6\
**Last updated:** [August 25, 2024, 8:41pm UTC](https://discourse.julialang.org/t/neuralode-training-failed-on-gpu-with-enzyme/118537 "2024-08-25T20:41:49Z")

</div>

Hi, when I try to train a NeuralODE with Discretecallback using 'sensealg=InterpolatingAdjoint(autojacvec=EnzymeVJP(), checkpointing=true) ’ , I get: Enzyme execution failed. Enzyme: unhandled augmented forward for jl\_…

---

## [PennyLane - worth a read](https://discourse.julialang.org/t/pennylane-worth-a-read/34318)

<div class="topic-metadata">

**Author:** [@johnh](https://discourse.julialang.org/u/johnh)\
**Replies:** 2\
**Last updated:** [August 22, 2024, 3:18pm UTC](https://discourse.julialang.org/t/pennylane-worth-a-read/34318 "2024-08-22T15:18:53Z")

</div>

At FOSDEM I saw a presentation on PennyLane OK, so its Python but it is functionally oriented and they use differentiable circuits. I had a quick chat with the presenter and he agreed there is commonality with Julia…

---

## [Error while loading Serialized BSON](https://discourse.julialang.org/t/error-while-loading-serialized-bson/118408)

<div class="topic-metadata">

**Author:** [@Emma\_B](https://discourse.julialang.org/u/Emma_B)\
**Replies:** 4\
**Last updated:** [August 22, 2024, 7:10am UTC](https://discourse.julialang.org/t/error-while-loading-serialized-bson/118408 "2024-08-22T07:10:52Z")

</div>

I have a BSON file, I used to save a Flux network using @save modelpath model #Loss\_params\_1 batchSize NbBatch lr weigths I’m trying to load “model”, but encountered the following error: julia\> d2 = BSON.load(myPath,…

---

## [Creating a custom container layer in Lux](https://discourse.julialang.org/t/creating-a-custom-container-layer-in-lux/117979)

<div class="topic-metadata">

**Author:** [@rkube](https://discourse.julialang.org/u/rkube)\
**Replies:** 2\
**Last updated:** [August 11, 2024, 7:31pm UTC](https://discourse.julialang.org/t/creating-a-custom-container-layer-in-lux/117979 "2024-08-11T19:31:18Z")

</div>

Hi, I’m trying to implement a custom container layer in Lux, following the manual to bind a custom masked linear layer together. The idea is that each masked layer has a user-defined mask, and all masks need to be set f…

---

## [Custom layer in Lux](https://discourse.julialang.org/t/custom-layer-in-lux/117642)

<div class="topic-metadata">

**Author:** [@rkube](https://discourse.julialang.org/u/rkube)\
**Replies:** 1\
**Last updated:** [July 31, 2024, 2:42am UTC](https://discourse.julialang.org/t/custom-layer-in-lux/117642 "2024-07-31T02:42:09Z")

</div>

Hi, I’m trying to implement a custom layer in Lux that is similar to a linear layer but needs to apply a user-defined mask to the weights. It needs to compute h(x) = g(b + (W ⊙ M)(x)) where g is an activation function…

---

## [Tips for handling large Datasets with a lot of preprocessing](https://discourse.julialang.org/t/tips-for-handling-large-datasets-with-a-lot-of-preprocessing/117551)

<div class="topic-metadata">

**Author:** [@felix12123](https://discourse.julialang.org/u/felix12123)\
**Replies:** 1\
**Last updated:** [July 27, 2024, 8:24pm UTC](https://discourse.julialang.org/t/tips-for-handling-large-datasets-with-a-lot-of-preprocessing/117551 "2024-07-27T20:24:50Z")

</div>

Hello, i am currently trying to machine learn a local relation between two 2D arrays. Consider the two 2D CircularArrays A and B of the same size. I want to predict the value of B\[x,y\] with the region A\[x-r:x+r, y-r:y+…

---

## [Jacobian of a network in the loss function with Flux](https://discourse.julialang.org/t/jacobian-of-a-network-in-the-loss-function-with-flux/117077)

<div class="topic-metadata">

**Author:** [@Baptiste\_Fabre](https://discourse.julialang.org/u/Baptiste_Fabre)\
**Replies:** 6\
**Last updated:** [July 23, 2024, 1:14am UTC](https://discourse.julialang.org/t/jacobian-of-a-network-in-the-loss-function-with-flux/117077 "2024-07-23T01:14:59Z")

</div>

Dear all, for a sciml problem in quantum mechanics (strongly inspired by what I saw during the @ChrisRackauckas talk’s in Grenoble), I need to differentiate the loss function. Here is a minimal example of the kind of pr…

---

## [Universal differential equations with MTK](https://discourse.julialang.org/t/universal-differential-equations-with-mtk/103225)

<div class="topic-metadata">

**Author:** [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Replies:** 6\
**Last updated:** [July 22, 2024, 8:26am UTC](https://discourse.julialang.org/t/universal-differential-equations-with-mtk/103225 "2024-07-22T08:26:06Z")

</div>

The SciML documentation has this great tutorial on Automatically Discover Missing Physics by Embedding Machine Learning into Differential Equations In the tutorial the universal differential equation is defined as foll…

---

## [Starting With Julia!](https://discourse.julialang.org/t/starting-with-julia/117285)

<div class="topic-metadata">

**Author:** [@ZherpiaWhiskers](https://discourse.julialang.org/u/ZherpiaWhiskers)\
**Replies:** 6\
**Last updated:** [July 21, 2024, 7:14am UTC](https://discourse.julialang.org/t/starting-with-julia/117285 "2024-07-21T07:14:28Z")

</div>

Heya! I’m a bit new to Julia, so, i’ve been studying the language for a couple of days, i already know some of the syntax, i’ve learned various types of data structures and I’m going to learn algorithms by implementing t…

---

## [Dense Layers in Attention Don't Train](https://discourse.julialang.org/t/dense-layers-in-attention-dont-train/117190)

<div class="topic-metadata">

**Author:** [@Helectus](https://discourse.julialang.org/u/Helectus)\
**Replies:** 1\
**Last updated:** [July 20, 2024, 4:19pm UTC](https://discourse.julialang.org/t/dense-layers-in-attention-dont-train/117190 "2024-07-20T16:19:15Z")

</div>

I am trying to create a transformer model using Flux.jl from scratch. I have an encoder struct that has an attention struct, which includes Q, K, V, O dense layers. My problem is that Dense.weights and Dense.bias don’t s…

---

## [BoundsError in Zygote RNN gradient](https://discourse.julialang.org/t/boundserror-in-zygote-rnn-gradient/117219)

<div class="topic-metadata">

**Author:** [@FabriSR15](https://discourse.julialang.org/u/FabriSR15)\
**Replies:** 0\
**Last updated:** [July 19, 2024, 8:06am UTC](https://discourse.julialang.org/t/boundserror-in-zygote-rnn-gradient/117219 "2024-07-19T08:06:43Z")

</div>

Hi everyone, I’ve experienced some problems with Zygote regarding the computation of gradients of a loss function obtained via sampling. An RNN generates samples autoregressively and the loss function is then computed u…

---

## [LoadError: \`llvmcall\` must be compiled to be called when calling Zygote.Jacobian](https://discourse.julialang.org/t/loaderror-llvmcall-must-be-compiled-to-be-called-when-calling-zygote-jacobian/116974)

<div class="topic-metadata">

**Author:** [@Baptiste\_Fabre](https://discourse.julialang.org/u/Baptiste_Fabre)\
**Replies:** 2\
**Last updated:** [July 15, 2024, 9:28pm UTC](https://discourse.julialang.org/t/loaderror-llvmcall-must-be-compiled-to-be-called-when-calling-zygote-jacobian/116974 "2024-07-15T21:28:37Z")

</div>

Dear all, I am facing a problem when trying to use Zygote.Jacobian to train a network. Here is the function that I am using to get the Jacobian of a network function dSCustomNonLinear(model, px::CuArray{T}, py::CuArra…

---

## [Trouble setting up PPO with reinforcementlearning](https://discourse.julialang.org/t/trouble-setting-up-ppo-with-reinforcementlearning/116653)

<div class="topic-metadata">

**Author:** [@clempe](https://discourse.julialang.org/u/clempe)\
**Replies:** 0\
**Last updated:** [July 5, 2024, 10:01am UTC](https://discourse.julialang.org/t/trouble-setting-up-ppo-with-reinforcementlearning/116653 "2024-07-05T10:01:13Z")

</div>

I’m new to julia reinforcementlearning, trying to run some experiments with PPO. Tried a lot of things, somehow can’t get ActorCritic to work. Maybe someone could give me a tip? Julia Version 1.10.4 Commit 48d4fd4843 (2…

---

## [Kolmogorov-Arnold network](https://discourse.julialang.org/t/kolmogorov-arnold-network/113712)

<div class="topic-metadata">

**Author:** [@Tarny\_GG\_Channie](https://discourse.julialang.org/u/Tarny_GG_Channie)\
**Replies:** 5\
**Last updated:** [July 2, 2024, 7:08am UTC](https://discourse.julialang.org/t/kolmogorov-arnold-network/113712 "2024-07-02T07:08:15Z")

</div>

A new architecture was dropped, vastly different from deep learning. In case you were interested in implementing something in Julia, maybe this?

---

## [TypeError: Expected UnionAll, got Type{Machine{...}}](https://discourse.julialang.org/t/typeerror-expected-unionall-got-type-machine/116354)

<div class="topic-metadata">

**Author:** [@niltsz](https://discourse.julialang.org/u/niltsz)\
**Replies:** 6\
**Last updated:** [July 1, 2024, 11:31am UTC](https://discourse.julialang.org/t/typeerror-expected-unionall-got-type-machine/116354 "2024-07-01T11:31:28Z")

</div>

I want to save a trained MLJ machine (EvoTreeRegressor) on the cluster and then open it locally on my computer. On the cluster, I save the machine with MLJ.save("path/on/cluster/machine.jls", mach) and then try to load…

---

## [Simple Reshaping in Zygote Throwing MethodError](https://discourse.julialang.org/t/simple-reshaping-in-zygote-throwing-methoderror/115825)

<div class="topic-metadata">

**Author:** [@vasudeva-ram](https://discourse.julialang.org/u/vasudeva-ram)\
**Replies:** 2\
**Last updated:** [June 30, 2024, 10:07pm UTC](https://discourse.julialang.org/t/simple-reshaping-in-zygote-throwing-methoderror/115825 "2024-06-30T22:07:17Z")

</div>

Hi, I’m using Zygote to find the Vector-Jacobian Product (VJP) of a function. The MWE version of the function is as follows: function ffunc(x::Vector{Float64}) # first reshape xMat = reshape(x, (:, 2)) # app…

---

## [Why isn’t Symbolic Regression used more?](https://discourse.julialang.org/t/why-isn-t-symbolic-regression-used-more/116215)

<div class="topic-metadata">

**Author:** [@Tarny\_GG\_Channie](https://discourse.julialang.org/u/Tarny_GG_Channie)\
**Replies:** 10\
**Last updated:** [June 27, 2024, 6:36pm UTC](https://discourse.julialang.org/t/why-isn-t-symbolic-regression-used-more/116215 "2024-06-27T18:36:07Z")

</div>

Thanks to its ability to dynamically compile expression and execute it efficiently, Julia positions itself very strongly in the symbolic regression landscape. Symbolic regression offers a way to discover formulas with f…

---

## [GraphPPL RxInfer Model](https://discourse.julialang.org/t/graphppl-rxinfer-model/114255)

<div class="topic-metadata">

**Author:** [@ohmsweetohm1](https://discourse.julialang.org/u/ohmsweetohm1)\
**Replies:** 3\
**Last updated:** [June 26, 2024, 4:44pm UTC](https://discourse.julialang.org/t/graphppl-rxinfer-model/114255 "2024-06-26T16:44:04Z")

</div>

I have tried to extend the example from here to include an unknown parameter theta: @model function rotate\_ssm(y, x0, Q, P) x\_prior ~ MvNormalMeanCovariance(mean(x0), cov(x0)) x\_prev = x\_prior theta ~ Norma…

[Previous page](https://discourse.julialang.org/c/domain/ml/24.md?page=5)

[Next page](https://discourse.julialang.org/c/domain/ml/24.md?page=7)