openSESAME frame parser produce conll files for train, test and dev, along with list of sentences separately.
All conll files have been converted to csv files, but only frameName and Lexical Unit information is taken from conll files. I did not parse roles for simplicity. They are more easy to read in conll format.

Explaination for subdirs is as follows:

original: original data files as produced with preprocessing script of openSESAME frame parser
verbs:  data from original dir, but filtered for only 'verb' predicates for lexical units
1ExPerSent_rand01-train.conll and respective csv file: as name suggest sample file used for expansion experiments. Test and dev are same from verbs dir.