I read that tutorial and I can see why you are confused. First, check out the HTML::Template perldoc for the section on the TMPL_LOOP tag. It shows the actual structure that is being used for TMPL_LOOP. If you want to use Data::Dumper, you can see it for yourself:

use Data::Dumper;
use CGI qw|:standard|;

use HTML::Template;

  # the fruit data - the keys are the fruit names and the values are
  # pairs of color and shape contained in anonymous arrays
  my @loop;
  my %fruit_data = (
                    Apple => ['Red, Green or Yellow', 'Round'],
                    Orange => ['Orange', 'Round'],
                    Pear => ['Green or Red', 'Pear-Shaped'],
                    Banana => ['Yellow', 'Curved'],
                   );
  print header();
  my $tmplObj = HTML::Template->new(filename=>'..\..\tmpl\fruit.tmpl')
+;
  for(keys %fruit_data){ 
    push @loop, {name=>$_, 
                 color=>$fruit_data{$_}->[0],
                 shape=>$fruit_data{$_}->[1]
         };
  }
  my $whatIamReallyPassing = { fruit_loop => \@loop };
  warn Dumper $whatIamReallyPassing;
  $tmplObj->param($whatIamReallyPassing);
  print $tmplObj->output;
[download]