A question for your Multimodal LSTM

[tzczsq]

Hi, I find every testing demo programe needs load the corresponding weights data.For instance, in this
the following test_audio_all.m, we load data '49.mat' as the weights.
My question is that how can I get the weight data like 49.mat from training data? Could you give a demo for this?
Otherwise, I can not investigate the performance of your code on my own datasets.

% addpath utils/
% addpath data/speaker_naming/

clearvars -global config;
global config mem;
gpuDevice(1);
sn_audio_init();

load('speaker_naming/results/audio_1/49.mat');
config.weights = model.weights;

load('data/speaker_naming/val_audio/1');
test_labels = reshape(test_labels, size(test_labels,1), 1, size(test_labels,2));
test_labels = repmat(test_labels, [1 size(test_samples,2) 1]);
test_samples = config.NEW_MEM(test_samples);
test_labels = config.NEW_MEM(test_labels);
% test_samples = test_samples(:,:,1:2000);
% test_labels = test_labels(:,:,1:2000);