---
title: "Getting started with LBBNN"
output:
  rmarkdown::html_vignette:
    df_print: paged
params:
  eval: false
vignette: >
  %\VignetteIndexEntry{Getting started with LBBNN}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include=FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval = params$eval
)
```

## Introduction

LBBNN implements Latent Bayesian Binary Neural Networks in R using the torch framework. This vignette walks through basic usage: data preparation, model definition, training, validation, and visualization.

## Setup

```{r}
library(LBBNN)
library(ggplot2)
library(torch)
```

## Data loaders

```{r}
loaders <- get_dataloaders(Raisin_Dataset, train_proportion = 0.8,
                           train_batch_size = 720, test_batch_size = 180)
train_loader <- loaders$train_loader
test_loader  <- loaders$test_loader
```

## Define the model

```{r}
problem <- 'binary classification'
sizes <- c(7,5,5,1)
inclusion_priors <- c(0.5,0.5,0.5)
stds <- c(1,1,1)
inclusion_inits <- matrix(rep(c(-10,15),3), nrow = 2, ncol = 3)
device <- 'cpu'

torch_manual_seed(0)
model_input_skip <- LBBNN_Net(problem_type = problem, sizes = sizes, prior = inclusion_priors,
                              inclusion_inits = inclusion_inits, input_skip = TRUE, std = stds,
                              flow = FALSE, device = device)
```

## Train

```{r}
results_input_skip <- train_LBBNN(epochs = 50, LBBNN = model_input_skip,
                                  lr = 0.005, train_dl = train_loader, device = device)
```

## Validate

```{r}
validate_LBBNN(LBBNN = model_input_skip, num_samples = 100,
               test_dl = test_loader, device = device)
```

## Plot structure and explanations

```{r}
LBBNN_plot(model_input_skip, layer_spacing = 1, neuron_spacing = 1,
           vertex_size = 15, edge_width = 0.5)

x <- torch::dataloader_next(torch::dataloader_make_iter(train_loader))[[1]]
inds <- sample.int(dim(x)[1], 1)
data <- x[inds,]
plot_local_explanations_gradient(model_input_skip, data, num_samples = 100, device = device)
```

Note: All chunks in this vignette are non-evaluated by default to ensure fast builds and avoid backend constraints on CRAN/CI. You can set `eval=TRUE` locally.