TDT chunking data

I am using The Decoding Toolbox 3.999E2 to perform an SVM analysis of an fMRI task consisting of 60 participants in which I’d like to decode condition. It is a blocked design with one session/run and 32 blocks in total, 16 blocks per condition. With 18s per block and a fixation cross of 14.4s. I am not sure how to divide the data into chunks given that there’s only one run.

Secondly, I’m considering doing a k= 5 fold cross valuation, however, I’m not sure if this is advisable