# Some questions for LSTM
**URL:** https://discourse.julialang.org/t/some-questions-for-lstm/84907
**Category:** General Usage
**Tags:** question
**Created:** [July 28, 2022, 7:29am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907 "2022-07-28T07:29:29Z")
**Posts on this page:** 7
**Page:** 1
### Author:  [@zlq178](https://discourse.julialang.org/u/zlq178)
#### Post date: [July 28, 2022, 7:29am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/1 "2022-07-28T07:29:29Z")
I am trying to do batch training using LSTM for a time series data with multiple features.
Assuming I have 5000 samples and 5 features for each sample. The input uses 14 days into the past and the output is a single value on the 15th day. (My time step is 14). The size of my data is the following:
```julia
using Flux
using Flux: Flux.Data.DataLoader
X = rand(14,5,5000);
Y = rand(1,1,5000);
model = Chain(
LSTM(5, 20),
Dropout(0.5),
Dense(20, 1,σ)
)
#evaluating prediction for the input sequence
function eval_model(x)
Flux.reset!(model)
inputs = [x[1,t,:] for t in 1:14]
output = model.(inputs)
end
L(x, y) = Flux.mse(eval_model(X), Y)
opt = ADAM(0.001)
#creating batches and training
train_loader = DataLoader((X, Y), batchsize=30, shuffle=true)
@time Flux.train!(L, params(model), train_loader, opt)
```
But some errors show as follow:
BoundsError: attempt to access 14×5×5000 Array{Float64,3} at index [1, 6, 1:5000]
Errors happen in `train_loader = DataLoader((X, Y), batchsize=30, shuffle=true)`
How can I solve it?
Another question: How to show the loss during the training process?
---
### Author:  [@Karthik-d-k](https://discourse.julialang.org/u/Karthik-d-k)
#### Post date: [July 28, 2022, 10:03am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/2 "2022-07-28T10:03:24Z")
> [@zlq178](#):
>
> `inputs = [x[1,t,:] for t in 1:14]`
Answering your first question,
Problem seems to be here.
t → varies from 1:14, but at 2nd index of x, you only have 5 elements (x = rand(14,`5`,5000)).
---
### Author:  [@zlq178](https://discourse.julialang.org/u/zlq178)
#### Post date: [July 28, 2022, 10:17am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/3 "2022-07-28T10:17:04Z")
So it should be `inputs = [x[t,:,:] for t in 1:14]` or `inputs = [x[:,:,t] for t in 1:14]`,according to the problem aboved?
---
### Author:  [@zlq178](https://discourse.julialang.org/u/zlq178)
#### Post date: [July 28, 2022, 10:21am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/4 "2022-07-28T10:21:15Z")
I try to turn `inputs = [x[1,t,:] for t in 1:14]` as `inputs = [x[t,:,:] for t in 1:14]`, it also show me
MethodError: no method matching (::Flux.LSTMCell{Array{Float32,2},Array{Float32,1},Tuple{Array{Float32,2},Array{Float32,2}}})(::Tuple{Array{Float32,2},Array{Float32,2}}, ::Array{Float64,2})
Closest candidates are:
Any(::Any, ::Union{AbstractArray{T,2}, AbstractArray{T,1}, Flux.OneHotArray}) where {A, V, T} at C:\Users\zlq.julia\packages\Flux\qp1gc\src\layers\recurrent.jl:137
The same error happened when change as`inputs = [x[:,:,t] for t in 1:14]`
---
### Author:  [@Karthik-d-k](https://discourse.julialang.org/u/Karthik-d-k)
#### Post date: [July 28, 2022, 11:33am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/5 "2022-07-28T11:33:22Z")
Sorry, I haven’t used LSTM / RNN from Flux yet, so i don’t have an answer,
But there is kind of similar post where you might get some lead →
> [@Broadcasting over a Flux.DataLoader](https://discourse.julialang.org/t/broadcasting-over-a-flux-dataloader/70414):
>
> Apologies for this question since I’ve had a lot of trouble producing a MWE (which I don’t have). I’m working on an RNN using text data (each observation is a sentence - each word in each sentence is mapped to a word embedding). The problem is classification into six classes. The issue I am having I think relates to using the Flux.DataLoader type. The basic setup is… train\_data = Flux.DataLoader((data=train, label=tlabels), batchsize=42, shuffle=true) # train\_data.data.data is 1x357, wher…
---
### Author:  [@zlq178](https://discourse.julialang.org/u/zlq178)
#### Post date: [July 29, 2022, 11:43am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/6 "2022-07-29T11:43:10Z")
Sorry, I still do not understand how the method solves my error.
---
### Author:  [@zlq178](https://discourse.julialang.org/u/zlq178)
#### Post date: [July 29, 2022, 11:44am UTC](https://discourse.julialang.org/t/some-questions-for-lstm/84907/7 "2022-07-29T11:44:49Z")
Can anyone solve the error I proposed at first? Thank you very much