# Machine Learning

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

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

**Page:** 25

---

## [How do I check my custom Env in Reinforce.jl?](https://discourse.julialang.org/t/how-do-i-check-my-custom-env-in-reinforce-jl/80808)

<div class="topic-metadata">

**Author:** [@Hupdup](https://discourse.julialang.org/u/Hupdup)\
**Replies:** 4\
**Last updated:** [May 15, 2022, 7:23pm UTC](https://discourse.julialang.org/t/how-do-i-check-my-custom-env-in-reinforce-jl/80808 "2022-05-15T19:23:26Z")

</div>

Hello all, I have written a custom enviroment in reinforce.jl and now I want to check if it is correct (i.e. if the actions and the step function work as thought). For this I wanted to call the functions one by one in …

---

## [Training a million models at the same time](https://discourse.julialang.org/t/training-a-million-models-at-the-same-time/81100)

<div class="topic-metadata">

**Author:** [@StefanPerko](https://discourse.julialang.org/u/StefanPerko)\
**Replies:** 4\
**Last updated:** [May 15, 2022, 6:12pm UTC](https://discourse.julialang.org/t/training-a-million-models-at-the-same-time/81100 "2022-05-15T18:12:04Z")

</div>

I have written my own code for doing Monte-Carlo simulations with SGD and a one-parameter model, for research. Specifically, I would draw 10^6 data sets (x,y) from a known distribution with x,y 1-dimensional and train 10…

---

## [Zygote / ReverseDiff not matching input types](https://discourse.julialang.org/t/zygote-reversediff-not-matching-input-types/80838)

<div class="topic-metadata">

**Author:** [@evolbio](https://discourse.julialang.org/u/evolbio)\
**Replies:** 0\
**Last updated:** [May 10, 2022, 7:27pm UTC](https://discourse.julialang.org/t/zygote-reversediff-not-matching-input-types/80838 "2022-05-10T19:27:06Z")

</div>

In my code, the line (y ./ k).^h throws a Zygote / ReverseDiff error but works without problem with ForwardDiff. Here, k and h are views of a parameter vector and y is a subset of the variable vector that depends on the …

---

## [Zygote returning wrong gradient on GPU](https://discourse.julialang.org/t/zygote-returning-wrong-gradient-on-gpu/80750)

<div class="topic-metadata">

**Author:** [@maxfreu](https://discourse.julialang.org/u/maxfreu)\
**Replies:** 4\
**Last updated:** [May 10, 2022, 3:11pm UTC](https://discourse.julialang.org/t/zygote-returning-wrong-gradient-on-gpu/80750 "2022-05-10T15:11:35Z")

</div>

Hi! As an exercise, I want to optimize the potential energy of a system of springs, which are coupled if they are closer than some threshold. On cpu, the gradient is correct but on gpu it is nothing. Am I doing something…

---

## [SDE estimation based on a set of features](https://discourse.julialang.org/t/sde-estimation-based-on-a-set-of-features/80798)

<div class="topic-metadata">

**Author:** [@mobar](https://discourse.julialang.org/u/mobar)\
**Replies:** 1\
**Last updated:** [May 10, 2022, 12:20pm UTC](https://discourse.julialang.org/t/sde-estimation-based-on-a-set-of-features/80798 "2022-05-10T12:20:47Z")

</div>

I would like to get started with non-linear time timeseries learning and extrapolation in Julia by learning the drift parameters in an SDE based on a pre-defined set of latent features (for this particular application it…

---

## [Linear combinations of Zygote.Grads](https://discourse.julialang.org/t/linear-combinations-of-zygote-grads/80621)

<div class="topic-metadata">

**Author:** [@5Pack](https://discourse.julialang.org/u/5Pack)\
**Replies:** 0\
**Last updated:** [May 6, 2022, 3:03pm UTC](https://discourse.julialang.org/t/linear-combinations-of-zygote-grads/80621 "2022-05-06T15:03:57Z")

</div>

Here is a question I posted on Stack Overflow, but I thought it would be good to post it here also. Thank you in advance for your help! https://stackoverflow.com/questions/72143438/linear-combinations-of-zygote-grads

---

## [Learning to Rank?](https://discourse.julialang.org/t/learning-to-rank/80527)

<div class="topic-metadata">

**Author:** [@Billpete002](https://discourse.julialang.org/u/Billpete002)\
**Replies:** 8\
**Last updated:** [May 6, 2022, 1:30pm UTC](https://discourse.julialang.org/t/learning-to-rank/80527 "2022-05-06T13:30:44Z")

</div>

Hi, I am wondering if there is any package which has a 100% Julia “learning to rank” feature besides XGBoost.jl? (my dataset would be too wide if I did that). I have checked MLRanking.jl but this is deprecated by 6 year…

---

## [SimpleChains.jl extension to Graph Neural Networks](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422)

<div class="topic-metadata">

**Author:** [@DoktorMike](https://discourse.julialang.org/u/DoktorMike)\
**Replies:** 2\
**Last updated:** [May 4, 2022, 1:25am UTC](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422 "2022-05-04T01:25:47Z")

</div>

So yesterday I attended the Copenhagen julia meetup and saw a new Neural network toolbox called SimpleChains.jl which aims to be fast for small neural networks basically. Often Graph Neural Networks are also quite small …

---

## [Flux normalization and regularisation](https://discourse.julialang.org/t/flux-normalization-and-regularisation/79925)

<div class="topic-metadata">

**Author:** [@Aminath\_Shausan](https://discourse.julialang.org/u/Aminath_Shausan)\
**Replies:** 7\
**Last updated:** [May 3, 2022, 2:34pm UTC](https://discourse.julialang.org/t/flux-normalization-and-regularisation/79925 "2022-05-03T14:34:00Z")

</div>

Hi I am using a Flux model for regression with L1 regularization like below: model = Chain(Dense(size(X\_train,1),254, relu), Dense(254, 254, relu), Dense(254, 1,identity)) losses =\[\] l1n…

---

## [Flux + GPU memory problems](https://discourse.julialang.org/t/flux-gpu-memory-problems/79783)

<div class="topic-metadata">

**Author:** [@lgmendes](https://discourse.julialang.org/u/lgmendes)\
**Replies:** 2\
**Last updated:** [April 26, 2022, 10:03pm UTC](https://discourse.julialang.org/t/flux-gpu-memory-problems/79783 "2022-04-26T22:03:54Z")

</div>

I was trying to train some models available in the Metalhead (0.7) + Flux (0.13) using an NVIDIA GTX 1080 Ti (11 GB ) using some toy data (512 images size:224x224x3). I was running the code that I show at the end of the …

---

## [RNN model converges at a high training loss](https://discourse.julialang.org/t/rnn-model-converges-at-a-high-training-loss/79941)

<div class="topic-metadata">

**Author:** [@cc672012](https://discourse.julialang.org/u/cc672012)\
**Replies:** 0\
**Last updated:** [April 24, 2022, 2:00pm UTC](https://discourse.julialang.org/t/rnn-model-converges-at-a-high-training-loss/79941 "2022-04-24T14:00:36Z")

</div>

Hi, I’m following d2l.ai’s book to learn deep learning. I implemented a character-level RNN from scratch and also one that uses the stuff exported by Flux. Both seem to give the same loss and perplexity where the former…

---

## [How can I use the ScikitLearn.utils module?](https://discourse.julialang.org/t/how-can-i-use-the-scikitlearn-utils-module/79892)

<div class="topic-metadata">

**Author:** [@pat](https://discourse.julialang.org/u/pat)\
**Replies:** 0\
**Last updated:** [April 23, 2022, 1:41pm UTC](https://discourse.julialang.org/t/how-can-i-use-the-scikitlearn-utils-module/79892 "2022-04-23T13:41:42Z")

</div>

I’m converting a bit of sklearn python into Julia. In the python, they use a a method called resample from sklearn.utils but I can’t find these functions in the Julia interface. It seems that ScikitLearn.jl does have a U…

---

## [Manual RNN on GPU](https://discourse.julialang.org/t/manual-rnn-on-gpu/79580)

<div class="topic-metadata">

**Author:** [@kailukowiak](https://discourse.julialang.org/u/kailukowiak)\
**Replies:** 3\
**Last updated:** [April 20, 2022, 12:29am UTC](https://discourse.julialang.org/t/manual-rnn-on-gpu/79580 "2022-04-20T00:29:59Z")

</div>

I’m trying to redo this tutorial in Flux. The neural net runs fine on a CPU but when I try transferring it to the GPU I get an error which is caused by indexing a CUDA array. Here’s an MRE: ## using Flux using Flux: on…

---

## [Application of DiffEqFlux to real data, with useful practical techniques](https://discourse.julialang.org/t/application-of-diffeqflux-to-real-data-with-useful-practical-techniques/79691)

<div class="topic-metadata">

**Author:** [@evolbio](https://discourse.julialang.org/u/evolbio)\
**Replies:** 0\
**Last updated:** [April 19, 2022, 1:33pm UTC](https://discourse.julialang.org/t/application-of-diffeqflux-to-real-data-with-useful-practical-techniques/79691 "2022-04-19T13:33:12Z")

</div>

Self-promotion … The following article may be a useful guide for fitting real time series data to ODE and NODE models with DiffEqFlux.jl. \[2204.07833\] Optimizing differential equations to fit data and predict outcomes T…

---

## [Multitask Learning in Flux](https://discourse.julialang.org/t/multitask-learning-in-flux/79535)

<div class="topic-metadata">

**Author:** [@Marco\_Repetto](https://discourse.julialang.org/u/Marco_Repetto)\
**Replies:** 2\
**Last updated:** [April 15, 2022, 4:42pm UTC](https://discourse.julialang.org/t/multitask-learning-in-flux/79535 "2022-04-15T16:42:29Z")

</div>

I would like to implement an MTL NN in Flux but is not clear to me how to create a different set of layers for each task. Right now I’m outputting all the tasks in the final layer and using logitcrossentropy loss as all…

---

## [Broadcasting error diffeqflux](https://discourse.julialang.org/t/broadcasting-error-diffeqflux/79417)

<div class="topic-metadata">

**Author:** [@YitaoCai](https://discourse.julialang.org/u/YitaoCai)\
**Replies:** 4\
**Last updated:** [April 13, 2022, 9:34am UTC](https://discourse.julialang.org/t/broadcasting-error-diffeqflux/79417 "2022-04-13T09:34:21Z")

</div>

Built a model according to the augmented Neural ODE tutorial as follows: @with\_kw struct Args η = 1e-3 # learning rate λ = 0.01f0 # regularization paramater batch\_size = 64 …

---

## [Lots of messages printed out when I evaluate the model using MLJ](https://discourse.julialang.org/t/lots-of-messages-printed-out-when-i-evaluate-the-model-using-mlj/79415)

<div class="topic-metadata">

**Author:** [@Sijun](https://discourse.julialang.org/u/Sijun)\
**Replies:** 5\
**Last updated:** [April 13, 2022, 5:23am UTC](https://discourse.julialang.org/t/lots-of-messages-printed-out-when-i-evaluate-the-model-using-mlj/79415 "2022-04-13T05:23:36Z")

</div>

I see lots of messages printed out when I evaluate the model: There are "error"s in the printed messeges. Are they really errors? how critical is it? or can I just ignore them? (The evaluation result looks fine thoug…

---

## [\[Flux / Zygote\] Circumvent long compilation times](https://discourse.julialang.org/t/flux-zygote-circumvent-long-compilation-times/79385)

<div class="topic-metadata">

**Author:** [@Tengrath](https://discourse.julialang.org/u/Tengrath)\
**Replies:** 1\
**Last updated:** [April 12, 2022, 8:30pm UTC](https://discourse.julialang.org/t/flux-zygote-circumvent-long-compilation-times/79385 "2022-04-12T20:30:33Z")

</div>

I have implemented a custom physics in model Flux / Zygote to benefit from gradient-based optimization of these great packages. It contains custom layers, each shipping its own ChainRulesCore.rrule. A single model contai…

---

## [Simple Flux LSTM for Time Series](https://discourse.julialang.org/t/simple-flux-lstm-for-time-series/35494)

<div class="topic-metadata">

**Author:** [@compleat](https://discourse.julialang.org/u/compleat)\
**Replies:** 62\
**Last updated:** [April 11, 2022, 7:02pm UTC](https://discourse.julialang.org/t/simple-flux-lstm-for-time-series/35494 "2022-04-11T19:02:45Z")

</div>

Hi, I have been trying for a long time to get Flux working for a basic time series. I think if I can get a really simple example working, I would be able to go from there, but when I try to do the ‘obvious’ thing, it fa…

---

## [Request for review basic introductory pages on ML, Perceptron, Neural Networks](https://discourse.julialang.org/t/request-for-review-basic-introductory-pages-on-ml-perceptron-neural-networks/79341)

<div class="topic-metadata">

**Author:** [@sylvaticus](https://discourse.julialang.org/u/sylvaticus)\
**Replies:** 0\
**Last updated:** [April 11, 2022, 11:53am UTC](https://discourse.julialang.org/t/request-for-review-basic-introductory-pages-on-ml-perceptron-neural-networks/79341 "2022-04-11T11:53:24Z")

</div>

Hello, I have a couple of pages dealing with: General ML concepts and perceptron algorithm Neural network architectures These two small “pages” are targeted to non-cs / non-math students that know nothing of ML, with …

---

## [Errors with Flux RNN set](https://discourse.julialang.org/t/errors-with-flux-rnn-set/79272)

<div class="topic-metadata">

**Author:** [@stmorg](https://discourse.julialang.org/u/stmorg)\
**Replies:** 1\
**Last updated:** [April 9, 2022, 9:42pm UTC](https://discourse.julialang.org/t/errors-with-flux-rnn-set/79272 "2022-04-09T21:42:08Z")

</div>

Hi everyone, I’m currently working on a final project that involves developing an RNN to represent a parameterized dynamic model. I have been able to do this fairly easily with a feedforward ANN in Flux but I have had ma…

---

## [Primer feasibility question - modelling soft tissue behavior using ML](https://discourse.julialang.org/t/primer-feasibility-question-modelling-soft-tissue-behavior-using-ml/76175)

<div class="topic-metadata">

**Author:** [@davide445](https://discourse.julialang.org/u/davide445)\
**Replies:** 13\
**Last updated:** [April 8, 2022, 8:41pm UTC](https://discourse.julialang.org/t/primer-feasibility-question-modelling-soft-tissue-behavior-using-ml/76175 "2022-04-08T20:41:33Z")

</div>

Get interested in Julia since seems crossing many of the themes i’m touching lately: HPC, biophysics, GPGPU, ML, with the goal to create interactive systems. Not a coder, I’m a solution designer that is deepening on dif…

---

## [Flux loss: Gradient wrt input leads to empty gradient wrt parameters or to "can't differentiate foreigncall"](https://discourse.julialang.org/t/flux-loss-gradient-wrt-input-leads-to-empty-gradient-wrt-parameters-or-to-cant-differentiate-foreigncall/79180)

<div class="topic-metadata">

**Author:** [@JanJereczek](https://discourse.julialang.org/u/JanJereczek)\
**Replies:** 3\
**Last updated:** [April 8, 2022, 1:14pm UTC](https://discourse.julialang.org/t/flux-loss-gradient-wrt-input-leads-to-empty-gradient-wrt-parameters-or-to-cant-differentiate-foreigncall/79180 "2022-04-08T13:14:51Z")

</div>

Hi Everyone, I am stuck with the following problem: I need the gradient (w.r.t. the input variables) of my NN in the loss function and cannot get it working. I have seen a couple of posts on this, however none with a so…

---

## [Training FLUX models with larger datasets](https://discourse.julialang.org/t/training-flux-models-with-larger-datasets/79057)

<div class="topic-metadata">

**Author:** [@lgmendes](https://discourse.julialang.org/u/lgmendes)\
**Replies:** 4\
**Last updated:** [April 7, 2022, 1:45am UTC](https://discourse.julialang.org/t/training-flux-models-with-larger-datasets/79057 "2022-04-07T01:45:44Z")

</div>

Hi, I’m trying to train a FLUX model (ResNet50) with a medium-size dataset (~ 30K images (512X512X3)) on a computer with 128GB of RAM and Three NVIDIA GeForce GTX 1080 Ti. I’m loading the data with Dataloader using a B…

---

## [Incorrect function evaluation in Zygote when swapping variables](https://discourse.julialang.org/t/incorrect-function-evaluation-in-zygote-when-swapping-variables/79082)

<div class="topic-metadata">

**Author:** [@goerz](https://discourse.julialang.org/u/goerz)\
**Replies:** 1\
**Last updated:** [April 6, 2022, 5:42pm UTC](https://discourse.julialang.org/t/incorrect-function-evaluation-in-zygote-when-swapping-variables/79082 "2022-04-06T17:42:36Z")

</div>

I have a relatively concise routine cheby that evaluates Φ = exp(-i H dt) Ψ for a Hermitian matrix H and a complex vector Ψ by expanding the exponential into Cheybchev polynomials. A minimal working example adapted from t…

---

## [Automatic Speech Recognition in Julia](https://discourse.julialang.org/t/automatic-speech-recognition-in-julia/79003)

<div class="topic-metadata">

**Author:** [@jbrea](https://discourse.julialang.org/u/jbrea)\
**Replies:** 2\
**Last updated:** [April 5, 2022, 8:13pm UTC](https://discourse.julialang.org/t/automatic-speech-recognition-in-julia/79003 "2022-04-05T20:13:44Z")

</div>

Does anyone have (or work on) an implementation of modern automatic speech recognition tools like conformers or wav2vec2.0 in one of Julia’s deep learning frameworks Flux/Knet/Avalon? Ideally pretrained on LibriSpeech …

---

## [Hyperparameters optimization packages for ML application in Julia](https://discourse.julialang.org/t/hyperparameters-optimization-packages-for-ml-application-in-julia/33519)

<div class="topic-metadata">

**Author:** [@Simon\_Besnard](https://discourse.julialang.org/u/Simon_Besnard)\
**Replies:** 7\
**Last updated:** [April 5, 2022, 6:37am UTC](https://discourse.julialang.org/t/hyperparameters-optimization-packages-for-ml-application-in-julia/33519 "2022-04-05T06:37:20Z")

</div>

Hi all, I am currently working with Flux.jl package for deep learning model implementation. I was wondering if some of you have had experience with some hyperparameters optimization packages (e.g. using hyperbands optim…

---

## [Dense layer field alias](https://discourse.julialang.org/t/dense-layer-field-alias/78993)

<div class="topic-metadata">

**Author:** [@Rasmus\_Hoier](https://discourse.julialang.org/u/Rasmus_Hoier)\
**Replies:** 2\
**Last updated:** [April 4, 2022, 10:28am UTC](https://discourse.julialang.org/t/dense-layer-field-alias/78993 "2022-04-04T10:28:45Z")

</div>

Hi, I am writing a custom layer and am trying to emulate the UI of Flux layers. In the definition of Flux’s Dense layer the Dense struct has a field weight, but this field also has a kind of alias W. How/where is this a…

---

## [Doc2Vec in Julia](https://discourse.julialang.org/t/doc2vec-in-julia/78907)

<div class="topic-metadata">

**Author:** [@Salvatore\_Cosentino](https://discourse.julialang.org/u/Salvatore_Cosentino)\
**Replies:** 2\
**Last updated:** [April 4, 2022, 3:42am UTC](https://discourse.julialang.org/t/doc2vec-in-julia/78907 "2022-04-04T03:42:57Z")

</div>

Hello everybody. I would like one of students to start learning Julia and move from Python. The project is working on uses Gensim module for Python, which allows for modelling semantic NLP problems. Specifically he is…

---

## [TorchDrug equivalence in Julia?](https://discourse.julialang.org/t/torchdrug-equivalence-in-julia/78910)

<div class="topic-metadata">

**Author:** [@DoktorMike](https://discourse.julialang.org/u/DoktorMike)\
**Replies:** 3\
**Last updated:** [April 2, 2022, 7:24pm UTC](https://discourse.julialang.org/t/torchdrug-equivalence-in-julia/78910 "2022-04-02T19:24:52Z")

</div>

I ran into TorchDrug a few months ago and it looks really useful. Is anyone in this community doing drug discovery using Julia? My question is mainly due to thinking about how useful/hard it would be to create a communit…

[Previous page](https://discourse.julialang.org/c/domain/ml/24.md?page=23)

[Next page](https://discourse.julialang.org/c/domain/ml/24.md?page=25)