This dataset completes the report of the same author (Olivier Brunner) that can be viewed at : https://doi.org/10.5281/zenodo.15696776 It is composed of three consistant CSV files - i.e. with the same columns. The report abstract is : This report presents solutions that, as of the date of publication, improve the cost (makespan) for 50 instances of the Permutation Flow-Shop Problem (PFSP) proposed by Eric Taillard or by Eva Vallada, Rubén Ruiz, and Jose M. Framinan. This report may help those who assess the quality of heuristics based on their ability to approach the best-known solutions. Three algorithmic techniques used are also presented: the super-prohibition sieve excluding paths with weights exceeding a given bound, the adaptive compartmental neighborhood with selective 2-closing, and finally the makespan calculation performed using upper-windowed multi-numbers. The same versioning is used for the report and for the dataset. The dataset was created and starts at version 5 of the report. Sources 100 & 101 : This dataset reflects an ongoing work. Further updates will be posted until every open instance has been tested. Source 0 : This dataset is based on the best known solutions sumerized and uploaded by Jan Gmys in feb, 2021. It can be viewed at : https://doi.org/10.5281/zenodo.4542886 Source 50 : It also includes the solutions uploaded by J. A. Alejandro-Soto in july, 2025. It can be viewed at : https://doi.org/10.5281/zenodo.15832204