There are a few academic papers that have used keylogger data set. An example from
"A Case-Study of Keyloggers and Dropzones" (PDF link):
We found more than 33 GB of keylogger data, containing stolen information from more than 173,000 victims.
For security reasons, the authors gave the data to AusCERT (Australia’s National Computer Emergency Response Team), and it's mostly impossible that this data set is ever released. But the paper is very interesting in its explanation of how they collected the keylogger data.
Another paper, "A Metric for the Evaluation and Comparison of Keylogger Performance" (PDF link):
(The authors have) developed a framework to assess the performance of a keylogger. This paper provides the documentation on how such a study can be conducted, while the required source code is shared online.
The analysis code (python) and keylogger data (sql database format) is available as a download from the Security in Telecommunications group at TU Berlin. The data is from tablets and smartphones.
The toolkit also holds the test-data from the performed use-case study and a demo keylogger.
The git page is here, and the direct download link is here. The data files are called KeyStrokesExperiment.sql and KeyStrokesRealExperiment.sql in the "data" folder.