Knitr option hook results in LaTeX output - knitr

Question

I'm creating ioslides and beamer output from RMarkdown source files and need to have variable figure output dependent on the output format.
I've generate a plot using ggplot2 which renders fine.
I want the plot to have an out.width set to 100% for HTML output and 70% for LaTeX output.
The problem is that when I set the option hook and check for LaTeX output, the tex file generated by knitr contains the LaTeX source verbatim for including the image which renders as text in the slide.
## Modify the output based on the format we're exporting
knitr::opts_hooks$set (out.width = function (options) {
if (knitr::is_latex_output()) {
options$out.width = '70%'
}
options
})
The plot renders fine in HTML output.
However, for beamer I get as shown in the image:
And the resulting output in the .tex file:
\begin{frame}{Why bother?}
\protect\hypertarget{why-bother}{}
\textbackslash begin\{center\}\textbackslash includegraphics{[}width=70\% {]}\{slide-book\_files/figure-beamer/lec-4-modularity-cost-1\}
\textbackslash end\{center\}
\end{frame}
Here's complete code for a MWE:
## Modify the output based on the format we're exporting
knitr::opts_hooks$set (out.width = function (options) {
if (knitr::is_latex_output()) {
options$out.width = '70%'
}
return (options)
})
```{r, lec-4-modularity-cost, echo=FALSE, out.width='100%', fig.align='center'}
plot.data <- tibble (X = seq (0,100), Y = 2 * X)
ggplot(plot.data, aes(x=X, y=Y)) +
geom_path()
```

Answer 1

When setting out.width this way you have to use a format that LaTeX understands right away, i.e. 0.7\linewidth instead of 70%. And you have to double the backslash in the R code:
---
output:
html_document: default
pdf_document:
keep_tex: yes
---
```{r setup, include=FALSE}
## Modify the output based on the format we're exporting
knitr::opts_hooks$set (out.width = function (options) {
if (knitr::is_latex_output()) {
options$out.width = '0.7\\linewidth'
}
options
})
```
```{r, lec-4-modularity-cost, echo=FALSE, out.width='100%', fig.align='center'}
library(ggplot2)
library(tibble)
plot.data <- tibble (X = seq (0,100), Y = 2 * X)
ggplot(plot.data, aes(x=X, y=Y)) +
geom_path()
```